📚┃suno-school

[wide stone]

i need help

[neat nexus]

Where did you get this prompt from?

[unreal sleet]

Ai responds to a hierarchy of information rather than a world salad, categorising is very helpful. taking out what you actually wrote and setting them in categories such as;

Genre & Tempo: Brazilian Phonk-Funk, Funk Carioca, 130 BPM.

Instrumentation: Heavy distorted 808 kick, syncopated percussion, metallic off-beat snares, high-pitched looping synth whistle, minor key piano.

Vocals: Male rhythmic chanting, aggressive vox, heavy compression, distorted grit.

Vibe & Production: Aggressive ambient, spirited, G Minor, mega bass, wet reverb on vocals and piano, professional song structure, high-energy phonk aesthetic.

[wide stone]

tysm

[unreal sleet]

You could even add [brackets] in the lyric box where you want [metallic off-beat snares] to not clutter the style too much

I've used insanely detailed styles since i have about 20+ years of using a DAW, prompting for wavetables, khz, LPF envelopes etc on particular elements and lets just say easy and smart does the trick. Just be sure to say what you want, never tell it what you dont want is it can trigger it as a positive

Dark alt-R&B ~120 BPM; slow halftime; spacious,
Emotion: moody/cinematic/grayscale,
Plucks: sharp/bright 1-shot hits; fixed root/oct; no melody/runs; fade 10–20ms; dec 300–500ms; LP~4kHz; res 10–15%; verb ~25%; delay ~15% (1/4 pp); hit every 1–2 bars,
Pads/Chords: dark airy synth pads define progression; minor; A 50–100ms; D 1–2s; S 70–80%; R 2–3s; LP 1–1.5kHz; res 5–10%; slow mod ~0.05Hz; verb 40–50% (4–6s),
Bass/Sub: sine/triangle; follows pad roots; glide 20–40ms; A 0ms; D 600–800ms; R 50–100ms; soft comp,
Drums: minimal trap; late kick 300–400ms (soft); back clap/snare; fast closed 1/16 hats; 1/32 micro-rolls (2–3) at bar-ends; lower vel; no open hats,
Texture: digital hiss/air (filt-noise, slow LFO) + low electronic wash; non-vocal,
All timbres synthetic,
Entry: verse vocal bar 1; no intro/pre-roll; female at chorus only,
Vocals: verse male, sung-legato, low-mid intimate, dry/crisp; subtle pitch-corr; +3–5kHz presence,
Chorus: breathy female layers (lyrical only; no non-lexical)

this is an example of overkill 😂

and that doesnt include the prompts in the lyrics box xD

[paper dragon]

why does the melody suddenly sounds quiter when i replace a section in the editor, like a sudden drop of volume right in the replaced section, its very noticeable

[unreal sleet]

the editor doesnt really give you the same sound, its a new gen that is very hard to get fixed, cover is usually better for simple things

[valid flax]

I'm about to mix this final mash-up down! It's about 20 different generations combined at this point lol

[rough depot]

Ok. So. What if i want to keep the percussion of the original durng the chorus (that pumping beat) but keep the broadway stuff i added in this remix? https://suno.com/s/kwgcZzuuL2zchDB7

imo the beat during the chorus (starting with the word "functions") is the best part of the song

[rough depot]

holy fright

[quiet solar]

Not an easy task. Basically, you need to do a simpler version of what Grash0per just showcased.

If you have Suno Studio, you can split out the stems from your remake, then import the original song and split out the drums, then either switch over the drum stems entirely (assuming the song is identical length) or do a bit of tinkering to just switch them over for the choruses. Then export as a new version of the song.

In the absence of Studio, you'd have to do a similar exercise in an external DAW or audio editor. The headache you may run into there are the production effects on the song itself - Studio IIRC does an internal pass to fix those up when you save a new version of a song. That doesn't happen if you extract the original stems to work on, so you'd have to cross your fingers it all exports out correctly and you don't end up with everything sounding a bit hollow and tinny.

If that does happen, when you reimport the song to Suno you can cover it with high audio influence to restore the quality, but inevitably you may have to take a couple of attempts to get a good compromise between the things you like and the things that change slightly from one version to another.

[rough depot]

Suppose i could get the stem then

Did you try listening to Master of Mind to see what i mean" "The mechanism FUNCT-IONS" its this epic drop beat thing

I'll see about getting it holup

[valid flax]

This is the finsl result of all those stem layers, it's a psychplogical horror experience https://youtu.be/7myi4q1yliw?si=TsbDA8IQiWK1D_DB

[stiff cliff]

Hey guys, I wonder how I can make Suno break out of its hidden structure. I created a lot of songs so far and my brain starts to catch a certain pattern despite fundamentally different music styles. Suno seems to like to add a “structure” to songs (i.e there will be almost always some sort of drop or dramatic event even though the music styles doesn’t necessarily require it) and pre chorus or chorus also has this downwards trend in the melody. It’s subtle but it’s there and my question is how to force Suno to generate a more uniform song structure.

My song tags usually look like “Minimal French Touch, Dry Boutique House, Restrained Disco Pulse, Compact Drum-Machine Groove, Muted Bass Bounce, Sparse Filtered Chords, Detached Vocal Fragments, Elegant Urban Cool, Nocturnal Seoul Mood, Subtle Retro-Modern Polish”

[viscid pumice]

You could use Moises for stems. If you you just want drums they have a free tier and I get fewer artefacts and better sound than in studio

[slender urchin]

How to change the way the person is singing?
Hi, currently I have a song that I really like but there is a part where the woman growls like in a metal song. and I want her to sing like she does in the rest of the song. is it somehow possible with the replacement tool?

Minute 2:52 - 3:05

yesterday I tried

Hard dance, deep bass, Mainstream-Hardstyle, Hardtechno, euphoric soft female vocals

excluded Styles: growling, screaming, heavy metal, shouting

but then I got 3 minute replacements or the exact same screaming as it is in the original song

I tried all three methods (smart, clasic and fixed) nothing gave me the expected results

The Song: https://suno.com/song/e5814a0f-3128-46c3-8d93-bc01400cd10b

[bronze island]

I use this strategy for making Suno songs: Style, Mood, Tempo, Vocals, Instruments and Background. SMTVIB for short

[obsidian nacelle]

Hey, i am working on an Kawaii Metal / Metalcore Band for a Persona. Trying new stuff, more prompts in the Lyrics section. Tareg was to get a soft female voice with sudden Madness shifts into breakdowns... took me some time to get the high Frequenzies out of that voice, and more Bass. I would like some constructive feedback if someone cares enough XD I am pretty new to Suno.
https://suno.com/s/g0upPjXg4NZCtwkz

[turbid hemlock]

@stuck junco @unique stone are yall available?

[polar spade]

Do you use prompt not in a genres wheelhouse to swing the pendulum the other way?

You could try anything. Whisper, bubblegum, k-pop, softrock, narration. Just think of whats the opposite of what its giving. I suppose we can call it tensegrity prompting or contrast prompting, rather than rely on exclusion alone. The principle is the same. Pulling against the present focus.

The difficulty is always to avoid affecting the whole song, so as always, using extensions is a good starting point. But if already into it, then yes, using Replace is an alternate means of infusing some contrasting prompts to affect just that section to be replaced.

Or if we must, metatagging those segments or stanzas to see if Suno will pay any attention.

[polar spade]

Mate, if theres an issue , just mention what it is. There may be an answer among the community. Consider the channel as voice-mail. 😁 Just leave a message. Someone will respond.

[slender urchin]

Thanks I will try it later. I hope it will generate new parts. This morning this feature kept loading without giving any new results.

[mint nexus]

Anyone know how I can make the transitions sound less obvious when it changes sound file?

[turbid hemlock]

i gotta report someone he sent me phishing links and i just want him banned here

[mint nexus]

Sorry if im at wrong place asking for that

[polar spade]

That can be done by right-clicking in the spam message in your DM and click Report. Choose the reason for the report. If reported by multiple users, then that will help in the response to the spammed/phishing status.

[polar spade]

Remaster can help blend things. To me remastering is nice, but always touch and go, never knowing which level will work optimally. So I just start with normal and go subtle or high, if needed.

[frozen dagger]

You can also try using https://discord.com/channels/1069381916492562582/1488621561240879194. This is a new thread added in the last week or so. May be what you're looking for if it's Discord related.

[untold lava]

Seems like this should be a simple question but I can't seem to get a really good answer out of anybody SO I came here. And I want to use the cover feature and my goal is to to swap the voice and the song using the new voices what should the audio influence slider be at because it's both a cover and we have a voice applied normally you're controlling the strength of the voice with the audio influence slider if you're just generating a song from scratch and since you can't use the voice with the inspo or sample feature then the question is if I'm trying to get the voice to sound like a different voice Keep in mind the voices aren't that that dissimilar but it is different What would the best settings be for any of the sliders or for anything else for that matter

[wintry saddle]

@strong vapor be banned already

[frozen dagger]

You have a song that has singing vocals on it, and in the end you want to have a new voice singing in the song, replacing the current singer?

[frozen dagger]

the audio slider does effect both the audio we are covering and the Voice selected. It does seem counter-intuitive when you're trying to replace a voice. I usually resort to stem split and covering the instrumental with new vocals being created with Voice, because trying to switch with Cover either does little to nothing, bleeds voices together, or mixes them like a duet.

[quiet solar]

You’re looking for a sweet spot between not too low and not too high. In v5 and earlier, I’d have suggested trying around the 60-65% mark. 5.5 is more sensitive, so perhaps start experimenting around 40-50%.

[paper dragon]

hi guys im a novice suno user using pro version, a song i generated doesnt show lyrics on the right table as usual. so when i want to edit some lyrics in the editor the lyrics table is just empty. what am i supposed to do?

[untold lava]

yeah but i like the melody and pacing of the original vocals and want to try retaining it. I suppose if im going that far could just do an RVC transpose

[untold lava]

i will try

[polar spade]

If you literally got nothing after typing lyrics in Create box and hit generate, then you could open in Remix-edit/Editor , let it load up in Editor UI then click SAVE as new file. To see if it adds back the lyrics.

[paper dragon]

hey man thats a great advice i will try this immediately

[polar spade]

I deleted already generated lyrics.... so it may not work. Try it and see. I have never had Create to generate a song but not retain the lyrics before. So just trying to find a solution.

[paper dragon]

YO it worked thats amazing thank you so much I can try and fix the lyrics now

[polar spade]

workaroud #2: is to just download it and load it back up so that Suno transcribes the audio back into lyrics. This workaround #1 is basically doing the same thing when you Save the file, Suno reworks the lyrics based on the audio.

[paper dragon]

I see.. thats some tech stuff, so what i encountered is considered a bug?

tho now that i look closely i see that suno made some huge mistakes with the lyrics it shows compared to the actual lyrics sung

[polar spade]

Hopefully a glitch. Thats when it happens once or a few times but hard to or impossible to repeat, still can be seen as a problem, so looked into as maybe a server overloading, a glitchy wi-fi, maybe something on your end. Who knows, but not affecting many others. A bug is when something keeps happening to many people that the system need to be looked at to fix once and solve for everyone at the same time.

[polar spade]

Yep. Good you checked. Just comb through and correct where it needs.

[polar spade]

Bottom line is , never a problem to feedback/report on anything. So the more users report/feedback the more evident it becomes and needs to be looked into.

[paper dragon]

many thanks man i appreciate you very helpful and cool

i hope to be as helpful as you are

[polar spade]

No worries. Welcome to Suno.

[fossil remnant]

Hi there, Does anyone know how to get the sort of knocking or clicking sound out of suno music when there is lack of melody ? for example in ambiance genre.

its like the AI is trying to keep a beat so it can stay on track.
even though its clearly instructed to not do such things in 60ish BPM

[polar spade]

lo-fi could possibly get a crackle though its often more a soul or hip-hop vibe, but can be used in a DJ Mix , or grungy pop, drunk trance, chill wave groove as well.

Try also turntable, vintage, classic, nostalgia, crackle, dusty . . . and cassette tape if you want it a little wobbly.

@quiet solar will have much better ideas, so the best is yet to come.....

[quiet solar]

It's not really possible to "instruct" Suno to do anything - you can't really give it commands, only prompts to steer it in the general direction of what you'd like it to take inspiration from.

Is the clicking the only sound you can hear at the points you want to get rid of it? If so, it's a fairly simple editing job you can do inside Suno on Pro/Premier.

If not, it can be done by a more complex editing job, but you'd need Premier for Studio or an external DAW or Audio editor.

@polar spade Haha - no pressure!

[polar spade]

Oh... you want to get RID of it... lol.. sorry I missread. I thought we wanted to nudge out some crackle and pop from Suno to sound old and vintage. My bad. OK.

OK... ummmm.. stick those prompts I suggested in the exclude box. 😄

Joking aside,
Remaster on High can try to remove noise and pops, so thats an option to try.

[trim pivot]

i only found out at this moment how to control the song by lyrics. i really liked the tool

it was a mess until i understood what was really going on

but yep. I'm getting better slowly >w<

i was just confused seeing a text typing panel without knowing what it really did

now i'll probably use the tools better, but in the beginning it's really confusing to see empty textarea waiting for me to do something without some information icon saying what it really is to write

[mossy basalt]

I'm really struggling with having both male and female vocals in a song, can someone help me understand if I am doing something wrong, I had to introduce the Verse 1A Verse 1B because it wouldn't complete the whole verse in 1 voice, but even when I am explicit like this, when I start the PreChorus it kicked in with a Male vocal, I have been trying for a while now and can't seem to get vocals to match my tagging structures in the lyrics and I've reached the point of frustration:
[Verse 1A]
[Male Vocal] [Low Energy] [Dark] [Steady]
Hear the sound, fall into line
Keep the rhythm, trust the design
Hands are still and voices low
Move as one where the bells command you go

[Verse 1B]
[Female Vocal] [Soft] [Uneasy] [Rising]
Born beneath an endless night
Told the dark itself is right
Every hour shapes the will
Every silence sharper still

[Pre-Chorus]
[Female Vocal] [Soft] [Uneasy] [Rising]
But something stirs beyond the tone
A breath that does not feel like home
A light that slips between the chime
A pulse that does not keep their time

[frozen dagger]

try putting the prompts in one bracket
[Verse 2- Female Vocal, Soft, Uneasy, Rising]
Try asking an ai chat bot for music terminology; that usually helps. (Falsetto, Crescendo, Aria, etc, etc)
Just give it what you have and ask it to describe it in musical terminology.

https://discord.com/channels/1069381916492562582/1479483412250366054

[mossy basalt]

Thanks Oblmous. So it is better to group instructions in 1 square bracket per intention? I'll give it a go, thank you!

[frozen dagger]

Usually it reads it like each [prompt] is used in that area. multiple bracketed boxes gets messy and it kinda skips over them. this shows the structure clearly for how each stanza should be at the start. with it keeping those guided prompts (hopefully) until it sees the next [prompt guide] where it will try to apply what it sees near there. it's all context tho. it looks at the whole thing and then decides what you're doing. (it can still skip around and not hit every prompt, because they are just guides not commands)

[mossy basalt]

Yeah when I was editing yesterday it was picking up the tag from the next verse, rather than the one before. Is there anything I can do to stop it reading a tag that is after the lyric?

[frozen dagger]

there isn't a "one way" can place the [prompt] before where it should be going or after. what's important is how it reads. use line breaks and punctuation. have it read (even the prompts) like you want it. if it is structured with clear spacing in between and your prompts are kept together for when they effect the entire stanze, that's a good bet.

[summer pond]

I'm trying to figure out how to tell suno that for a specific part I want the vocals to do a sort of "Walk Down" part.

Example: "Sun’s climbing up over rolling green hills"

on the last word "hills" I wanted the vocals to do a sort of walk down thing for it but i'm not sure if I'm able to get suno to do it

[mossy basalt]

Thanks Oblmous, I'll give it a try

[frozen dagger]

"Sun's climbing up over rolling, green — hills..."
or

Sun's climbing up over green
[Descending scale aria] hills.

[Descending scale aria]Sun's climbing up over green hills.

Maybe something like that depending on what you're trying for?

I gotta sleep. lol. Good luck to you both. It really is all context on how it reads with how we write lyrics and prompts. No one way to do anything.

[summer pond]

not exactly what i was looking for but I know I have a suno song that has the thing I was looking for. Thank you though

[broken phoenix]

Hi - there is a fair amount of static noise with the generations this morning. more than usual..

[vapid steppe]

hi..."What’s the current 'meta' for cloning/changing voices on Suno tracks? I’m looking for an app that can take a Suno-generated vocal and 'reskin' it into a high-quality studio voice

[exotic bobcat]

How can I stop Suno AI v5.5 from generating vocals? Even though I’ve removed the lyrics and added "instrumental" and "no singer" to the style, it still sings.

[pliant trail]

My Suno has a bug. When I open its Studio mode, select a clip on one of the tracks to regenerate, it just keeps spinning and loading but won’t generate anything. Has anyone else had this issue? It’s been two days already. Any solutions, guys? Urgent!

[wintry saddle]

try giving it a bunch of instrumental meta tag instructions so it has 'something' none vocal to do

failing that, studio and delete the damn vocals

I have noticed V5.5 instrumentals often have some vocal element

[paper dragon]

hi guys, im a novice suno user with a pro plan, and i am looking for a way to change the amount of songs generated in one click, like can i decrease it from 2 each time to 1?

[sly haven]

The more I use suno the more I like however, I think the render is pretty heavy compressed. Do you guys know how to avoid that? Best.

[gloomy kestrel]

I emailed the support yesterday and now I have no answer. I bought the pro subscription but then it ended up on an account linked to my mobile number instead of the account I was logged into when I made the purchase. I have a song on the account that was left without a subscription that I want to upgrade. Support said they would move all the songs to the account that got the subscription but nothing has happened. Does anyone know how I can recreate the song? I have tried but it doesn't get the same melody or voice.

[unreal sleet]

You can do metatags above lyrics stating how u want it mixed, stereo, high fidelity, SSL G-Series mix etc

depending on genre

[lavish pumice]

Hello world! Did they change our profile's home screen by any chance? I find myself with something very different from yesterday

[wintry saddle]

it is constantly evolving for sure

[lavish pumice]

I understand this, but 10 hours ago I had the same screen, now I find a completely different one... I don't know.

[quick wedge]

Hello there. You can’t decrease the number of generated songs at all It’s always been default at two clips whenever you click create. There’s no way of changing it either. It’s always been that way since I joined Suno.

[quick wedge]

First off Suno School isn’t a support channel at all it’s more for learning about prompts and the feature tools. You can check the ticket help channel and follow the steps inside there if you haven’t yet https://discord.com/channels/1069381916492562582/1479483412250366054 Try to give them time to answer back to you. Good luck and have a wonderful day 🍀

[wintry saddle]

yeah the updates are like that unfortunately and the profile screen has been moving a lot lately

[lavish pumice]

damn dude.. well ok thx

[paper dragon]

Hi thank you for the response, is this answer accurate? i accept it may be true. I feel like i seen mention about it somewhere before just cant remember where exactly

[quick wedge]

Yes it’s accurate the other users I met generates with two songs at one time it’s never been any different.

[paper dragon]

Thank you I appreciate it very much

[quick wedge]

You’re welcome. Have a wonderful day and a good weekend 😊

[gilded falcon]

I just uploaded a song with no lyrics just me beat boxing.... BUT SUNO IS SAYING COPYRIGHTED LYRICS????????
Was my beatboxing that bad that the site got offended?
I lowered the pitch and it got through. Here is the so called copyrighted lyrics:

[Instrumental]

[Intro]
[vocal bass kick and snare enter]
[rhythmic breathing]

[Main Pattern]
[vocal hi-hats and clicks enter]
[syncopated vocal bass line]
[vocal snare on beats 2 and 4]

[Breakdown]
[vocal record scratch effects]
[rapid-fire vocal percussion fills]
[bass drops out briefly]

[Outro]
[vocal bass returns]
[fading vocal percussion]

[polar spade]

Removed lyrics?

Not sure I follow from the details, but if it was a song with lyrics before, then the quick answer is just stemming out the instruments as an exact solo track.

BTW, there is really no understanding of negative prompts in the style prompt box. Suno created the exclude box for that which still, its hard to force vocals from showing up by entering something in the exclude box if they were there before.

But removing lyrics, and leaving weird and style sliders at default is usually enough for my tests.

Then audio slider:
25% for a wide variant instrumental, but retaining melody, likely shorter
50% increase in tempo retention and melody
75% tempo, melody and similar instruments, closer to original length
90% similar and including adlibs and backup vocals or full vocals towards the end., closer to original length.

You could do one at 50% audio slider and one instrumental stem. Then mash the two for a new output, see what you get. Thats always interesting.

[polar spade]

It just recognized something familiar in the audio, perhaps.. Who knows after that. Good you found a quick workaround. Sometimes thats what it takes.

[frozen dagger]

Download them if possible. If not, allow remix, then remix on your paid account weirdness low audio influence high, no style prompt. Will always change, that's what Suno does, but can get close until get sorted out with accounts.

[bronze island]

Hey guys, what sort of strategy do you guys use for making songs?

[paper dragon]

hi everybody im a novice suno user with pro plan, still learning all the tricks and nuances with the system. I got a question does anyone know if there is a way to separate the vocal stems? I have a small issue where the end of the chorus is partially "riding" on the verse.

[mellow raven]

Hi there! can anybody tell me why V1 and V2 are kidna on mute? there's no output

Hah found it

[summer patrol]

So the last 6 generations 5 are 7:59 long with a clicking sound at the beginning, i tried addint Radio-Edit to end of my prompt, this is 5.5 model -- a bit tedious to edit all i want to keep on refining -- anything else i can do ? or is the model just flawed since its new

[polar spade]

My opinion is there are elements being "improved, mitigated or updated" that opens gates for awhile, doing things like 7:59 and double-song results. Basically, Suno's tokenizing gate is left open with no limit at the moment. In my head, this explains alot. But my inside voices can be illusive sometimes. 😊

Does 5.0 do less of 7:59s? I havent tested that direction yet.

Personally, im getting more variance and creativity at the moment with 7:59s So a bit of a win.

[summer patrol]

Ok Ok, yeah stuff like this happens, yes tested 5.0 and it didnt do it on the same frequency, i have edited a few of the 7:59 since the first part is good, edit out the click in the beginning and find a suitable spot to fade out the song and keep the rest, i like my songs to be around the 5 min mark and up, 3-3.5 i call radio edits 😉 for electronic music longer songs are a plus, but I would like to control that with an extend function as i see fit, but i guess for now i keep editing the ones i like to build further upon, thx for answering

[frozen dagger]

right click on the song (or click the ...) to open the sub-menu. From there you should see Get Stems / MIDI.

[frozen dagger]

https://youtu.be/oUOWz34VlLA?si=ahB1bKatYFdKBFSn&t=99 This video should start @1:39. Official Suno YT channel. (If you click link, otherwise faster forward to the index point there)

@mellow raven in short they are preview lanes, "extra takes". they are not part of the timeline until you choose to place them in the actual track, not the alternate takes.

[ashen vigil]

who is owner or programming team ?

I want tell some ideas for suno and problems they have?

@flint haven @flint haven

[half dragon]

#📝┃user-research-feedback #1486436501771517972 #🪲┃studio-feedback

[pearl warren]

Did inspo change? I used to be able to pick a playlist and choose up to 4 songs, now it just chooses the playlist ?

[willow dew]

Hello everyone, I’m having a really frustrating issue with Suno. I do not want any guitar, saz, bağlama (Turkish long-necked lute), or any other plucked string instrument in my songs. I clearly specify in the Style section which instruments I want, and in the Exclude section which ones I do not want. Still, in 8 out of 10 songs, Suno ends up using some kind of plucked instrument. Has anyone found a good way to prevent this?

[half dragon]

#🐣┃suno-showcase #🎤┃artist-showcase #💽┃playlist-showcase #🚀｜v5․5-showcase

[wide stone]

https://suno.com/song/4fbe782e-a36f-4496-8a30-ce33763613a6

[polar spade]

https://discord.com/channels/1069381916492562582/1154587937611190354 and https://discord.com/channels/1069381916492562582/1082775596767117312

They are not discussion channels but just a spot to make suggestions or report an issue.

[sweet ivy]

Is Suno having issues tonight. I uploaded an audio track, simple piano-vocal. Hit cover & Suno put in lyrics and instruments in the Lyrics. It also kept asking me every 10 seconds did I want to overwrite or keep current styles. Never done this before

I also have the audio uploaded in my library now maybe 30 times. I only uploaded it once

[olive night]

It's probably because of the music suno was trained on. Sounds like you're making a song in a style that almost always have some kind of plucked string instrument, so suno will want to replicate that sound someway. Stuff like silence, void, suspension, anticipation, etc can often be very hard to generate because suno will want to fill it with something. If you share me the prompt, I can try and help you figure out a solution.

[dapper void]

does anyone have any tips regarding updating songs from older versions to the new version? i want to preserve the details that are the soul of the song, like specific tonality or pronounciations of certain words, but struggle to do so, not mentioning that suno can often even start singing in different parts of the music
i just mostly want to clean up the sound, not change it

[dense arch]

whenever i generate a music theres always words that repeat in a 'glitchy' or 'stuttering' way, but I want it to be smooth. Anyone know what to do?

[hardy moth]

anyone know trick to repair volume or glitch track at some version song that we generate? actually whole generate already have best part.. but still have glitch sound or volume noise at ending song.

[near siren]

Hi, I’d appreciate any advice you can give me about a song I’ve created. I love it, and no other generation of this song matches it. The only issue is that the original generation has a female vocal in some parts, but I want it to be the same male vocal throughout.

I’m looking for a feature that takes the stems, selects the sections of female vocal, and generates a male acapella singing the same melody and lyrics. However, it doesn’t generate acapella, and the vocal doesn’t even match the original, so I can’t even strip the instrumental stems from it and use that.

Any work arounds?? Thank you

[near siren]

Relatable 😭 as someone producing hyperpop music, the glitchy vocal style is nice but suno just over does it now, i am yet to find a prompt to fix this.

[polar spade]

• It could be a feature Suno feels belongs in your song based on your styles.
• It could be your sliders are up or down quite far.
• it could be your line metering /length is a bit incompatible with an easy way to sing it, that Suno has to "fill in" the gaps with stuttering.
• If you save your prompts and lyrics. Clear your Create panel. Paste them all back in, and still get the glitchy style, it could be the above. But if we are just continuing the same Create screen without "starting over" then Suno may be just echoing previous iterations of the same issue.

[fervent pasture]

My Suno is stick generating songs that all use the same melody of a previous song. I am not making a cover, I never told it to make a cover of this song. I've tried refreshing the create page, starting with a completely blank slate and putting in all the info again. It still gives me the same melody. No songs are selected for cover, there is nothing in inspo or audio. How do I get it to stop giving me the same melody???

I tried doing a song with just "testing" as the lyrics and it gave me something different. But as soon as I make another song with the same set of lyrics and put in a style prompt it goes back to the same melody as my last published song

[dapper elbow]

DasvonMinotaur's points are good. prompt-level things that help specifically:

exclude with synonyms: no glitch vocals, no stutter, no chopped vocals, no vocal cuts, no digital artifacts. ask for the opposite on the positive side too, something like "smooth continuous legato vocal, connected phrasing, natural vocal flow".

lower weirdness. high weirdness is where a lot of the glitch character comes from.

also watch your style tags. hyperpop, deconstructed club, IDM, footwork all lean glitchy no matter what's in exclude. if any of those are in your style field, swap them out.

one more. lyrics formatting matters. fragmented phrases with dashes, ellipses, all-caps can read as glitch cues. stick to full sentences with normal punctuation.

[robust wigeon]

Anyone here knows if there is something I can do on mac to be able to preview songs while they are generating? It used to be possible, but now I need to wait for the full generation to finish before I can hit play on them. Appreciate any pointers!

[strong tangle]

does anyone know whats going on with the sharpness happening in the high end? ive done just about everything with eliminating trigger words for that, phase smearing, centering low end, vox and everything and its just coming out really insane every time. even toggling between 5.5 and 5.0 with the same minimal prompt and style of just genres and a descriptors. mixes are coming out muddy, flat, and hissy as hell.

[fervent pasture]

I have done multiple tests with this song and I cannot use the styles I usually use anymore. If I leave it blank it gives me different melodies. If I put anything in the style section like I normally would, it goes back to the same melody. I've wasted so many credits trying to figure this out. Not doing a cover or remix. No audio input. No voice file, no inspo, nothing to explain why I keep getting the same melody. I even tried just putting "testing" in the lyrics and keeping the style and it gives me the same song again but with her singing "testing" to the same melody. I feel like I am losing my mind at this point. I can't use my usual styles to create anything but this one melody over and over.

[strong tangle]

slider values are relatively neutral and im not covering or remixing anything, no voice, nada.

[strong tangle]

ive been getting the same song more often than not. i cant really use my normal workflow anymore with suno its just not working. chord progressions, keys, everything seems to just not work the way it did. unless i do like edm or something my niche is over.

[fervent pasture]

Is it like the song sounds like you're making a cover of the same song over and over?

[strong tangle]

its the same everything nearly every time. since i have to condense everything and eliminate things its just gravitating to that gen through probability i think. it did it before whenn i didnt know what i was doing on 3.5 but now i have to rachet everything back to avoid harshness and/or phase smearing.

[robust wigeon]

I'd log out clear the cache and try again

[fervent pasture]

For me it's like any of my usual prompts give me the same melody again no matter what lyrics I put in

Logged out, cleared browser cache. Manually entered lyrics, style and song name again. Still same melody

[paper dragon]

hi everybody, i am new to suno and use the pro plan, i noticed that theres this cover art thing that can be generated.. it's pretty pricey, i wonder from people who used this, how is this function, is it reliable, is it worth 100 credits per gen, whats the best approach with it to get the wanted result, is it a loop or a video thats abruptly cut?

[polar spade]

I have to ask now, since its there, but have you previously created a Custom Model. Just make sure its not defaulting to Custom. Illiminate that as a possibility at least.

Then use 5.5, 5.0 and another say 4.0, see if the same.

[fervent pasture]

Where would I make a custom model?

Poking around, see where to make those now. I haven't done that. I have just been doing what I normally do to make a new song

It does seem to be something tied to v 5.5 as it isn't doing it when I use the same inputs but on v 5.0

[half dragon]

Under Create screen - v5.5 dropdown , you'll see the option to create your custom model.
Select it and follow the on screen guide.

[fervent pasture]

Is creating a model required to use v 5.5 without getting the same song over and over?

[half dragon]

Creating custom model so you could "train" your own model from the songs you uploaded to it.
Then on Create, you don't use v5.5, and select your custom model for the vibe that your model produces.
Not sure but I think you could have different custom models for different vibes? maybe.

[fervent pasture]

So should we just not make songs on v5.5? That doesn't seem very intuitive?

[half dragon]

Your choice. Not that you should, but if you want to.

[paper dragon]

guys i am trying to get specific stems from a song thats been generated. to this point I have generated stems 4 different times and I am NOT happy with the stems it generates. how do I get the stems I am looking for? is rolling the dice the only way?

[fervent pasture]

I just want to know if using 5.5 is supposed to produce the same song over and over?

[half dragon]

The answer is No.

[fervent pasture]

So then it's still some kind of bug that is making me unable to use v5.5 to generate songs

[half dragon]

It could be several reasons... not just using v5.5
I don't have that experience, many don't either.
It could be prompts... and if you're using Cover or not.. and then sliders... many other factors...
But in general v5.5 or any model don't usually produce same song over and over.. . or we will all make the same songs!

But if you feel that you tried everything and still experienced the same -- I suggest reporting it to [e-mail] with details of each tries.

[covert abyss]

I created song with support of Suno, then published them on Spotify, but now Suno won't let me re-work on my own creation, stating that it is Copyrighted material !!! Yes, it is, and I AM the copyright owner!! Does anyone already encountered this issue? Any way to solve it?

[half dragon]

Do you still have the original song you created in Suno? or you tried to upload the final version that you worked on that offline on your DAW? did you add in anything to it?
If original as the version you downloaded from Suno, you can reupload just fine.. I did that several times.. as long as it's the same; which has nothing in it that wasn't made from Suno.
Now, saying that, it could be false positive -- which you can email [e-mail] , I think they can handle exception case by case.

[covert abyss]

thanks. I want to reuse my lyrics to put them on another atmosphere

[half dragon]

You can certainly do that. Just copy the lyric and create a new song.

[covert abyss]

and yes, I have all my original (from my initial creation on Renoise, lyrics, rework versions on Suno etc... up to the version I publihsed on Spotify) - I am just mad/suprised that the publication on spotify actually "validated / granted" my rights (copyright), and that it does not allow ME to anymore use my own lyrics. Is there a way to "tell" Suno that I have the rights to these lyrics?

[half dragon]

Where did you get this eror? on upload? on create?

[covert abyss]

No, it doesn't work, Suno tells me that my lyrics contains copyrighted material

[covert abyss]

on Create, with a new music prompt but with copy pasting my lyrics

[half dragon]

There might be a term or a line that triggers the block... if the song was created in Suno with the same lyric back then,... it's possible that the recent updates that tighten up would detect it this time.

Try to narrow down which part of the lyric to find that and rewrite it a bit.

[covert abyss]

yep, I tried, but I need to change or cut my lyrics quite profoundly for it to work. (the only trick that sometime work is to use phonetic approximation, but this is not always functionnng). THe message I get is "Couldn't generate that.
Your lyrics contain copyrighted material. Please change it and try again."
This is Very frustrating

[half dragon]

Not saying your lyric has it, but in the past, some terms are detected as copyrighted producers, names or phrases... at a glance it's hard to know but narrowring down by removing part of the lyric would help to find... yes, it'd waste some credits but no other way around it.

[half dragon]

Yeah, I know. We just have to work with the system, like phonetic spelling to guide it to pronounce correctly for example.

[covert abyss]

would you know WHO or HOW I should contact people at Suno to report this "issue", or let them tell me how Authors can use their AI to work on their own material? (otherwise many creators really won't pursue with using the tool, which wouldn't be good news for Suno)

[half dragon]

You could email [e-mail] describing the details, with links and stuffs...

[covert abyss]

Yep, I'll do that - not sure to get an answer tough 😉

[half dragon]

I'm sure you will... might not be immediately but they will.
Visit #ticket-help to follow up after.

[paper dragon]

guys i am trying to get specific stems from a song thats been generated. to this point I have generated stems 4 different times and I am NOT happy with the stems it generates. how do I get the stems I am looking for? is rolling the dice the only way?

[half dragon]

Unfortunately you have no control how Suno generate the Stems -- so pretty much so, keep rolling the dice if you can afford it.
Curious though, what are you looking for? specific instrument stem?

[paper dragon]

you could say yes its a specific instrument, but suno catches it in the leading vocals, so IDK anymore if I should keep trashing credits or give up

I can say for sure that Suno crews needs to improve the stems section, its very sporadic, unless its the way to make money

[half dragon]

So it's mixed in with other things? is it noticable?
I wonder if you can take that Stem and run through Stems again to see if it could seperate them, not from the whole song but from that specific stem.
or have you tried other 3rd party Stem seperator, see if any of them can?

[paper dragon]

actually thats a good idea, i didnt think about it. so should I just download the specific stem and then upload it and try and get stems from it in Suno?

[half dragon]

Yeah basically. But I think you don't have to download it... Stems are extracted and will be in your Library
Check your filter to enable them visible... no need to download and upload again

[paper dragon]

yes i found it the filtering of stems and disliked was on

it still doesnt catch the stems im looking for.. should I try some kind of 3rd party seperator?

[polar spade]

Are you listening to "Version 2" of these stems as well, the original and the stems-of-a-stem version 3 &4 i guess they would be as well?

[frail turtle]

https://suno.com/song/3e1e7c62-9af9-4b07-80ad-997173c8344b here is a rap
anyone wanna rate this?

[final trail]

Ready for class

[stable steppe]

Can we have an actual rule where if you want someone to listen to your music; you must at a bare minimum state BRIEFLY what genre you're doing (so someone here can be on the same page with you and "enter your head"), and seek constructive criticism?

I'm just politely asking this so this doesn't become another "Soundcloud 12.0" where it's just "ook ook, here's music, listen"

I mean.. how else did I run into @knotty merlin and we communicated on a personal level upon roughly my first day messaging here?...

[knotty merlin]

I HAVE BEEN SUMMONED

[stable steppe]

It's like come on men, just TRY thinking. Just try it

I'm not asking you to solve an antiderivative.. I just want actual people.. not just Soundcloud larpers

Or is that the status quo here? It's just "Soundcloud 9000.0" and you just randomly spat "here's music (link), listen now" and that's all?

[knotty merlin]

nah those people usually just get ignored and never come back

[stable steppe]

Oh.. so I kept him around by using the Reply feature to his message? :(((

oh.. the suno school..

the wrong chat, oops

I though this was the #💬┃general-chat :(((

[limpid spoke]

I need some big help, When using the studio and let's say I have acoustic guitar strumming and vocals, what I want to do is make a replica of the acoustic strumming only. And what happens almost every time is that the acoustic strumming has the vocal melody in it too played by the guitar. And if I take the vocals track out completely, then suno makes up a random melody along with the strumming. I need to do this because I'm trying to generate a better version of the strumming then I started with.

I'm guessing I'm either using bad prompts or this is just a problem in general.
Which I didn’t have this problem as much when the Studio first dropped.

I use simple prompts like "Acoustic guitar strumming" "Acoustic guitar" "Crispt acoustic guitar strumming"

I'll be very thankful if someone can help me with this 🙏

Also, I don't know where to put this message exactly. I put it here because I need someone to teach me.

[polar spade]

Its nice when someone mentioned the model theyre using, perhaps whether they are on the app or a browser, when asking about trchnique and help.

And as you say, when we ask for some critique of an output to mention the style we're going for, where it is on the vast spectrum of genres, as well as our intent for the song. My outputs could land somewhere about "new age experimental boho western pop". So thats reasonable. Thats likely more to happen with a dedicated song critiquing channel though, where the rules make a little more sense to be somewhat more strictly enforced to keep things running smoothly.

[echo halo]

Has anyone here started on the free plan and later upgraded to a paid plan, and then successfully gotten an exception for commercial rights or ownership of songs created earlier? I saw that “exceptions may be made on occasion, under various conditions,” and I’m curious if anyone has actually experienced that and what the process or requirements were?

[stable steppe]

It's just I figured the entire point of this discord is to have a more "in-depth" discussion, not just tell people "listen to muh music". It's kinda like soliciting, "are you trying to sell me something?" or looking for actual constructive criticism.

because just saying "listen to it and rate it" is quite literally no different than a business giving me a free sample and saying "gibe me 5 star wating pweeese c: "

[stable steppe]

Look, I get it, I have the same exact feelings too. But I still do it the "right way" vs. walking up to someone, throwing a product in their faces and saying "listen and rate". It's just.. not sure how to explain it besides it comes off as disingenious

[twilit frost]

When I generate a whole song, the interlude ends up shorter than I’d like. So I try entering prompts like “16-bar interlude,” but it doesn’t seem to work. How do you guys specify that you want a longer interlude?

[dapper elbow]

Specific bar counts don't really work — Suno doesn't think in measures. What works better is describing the feel and length in words.

In your lyrics/structure, try [Extended Instrumental Interlude] or [Long Instrumental Break] instead of "16-bar." The word "extended" nudges Suno to give it more space.

In your style prompt, adding something like "on Extend shift into an elaborate instrumental breakdown with evolving textures" helps too — it tells Suno what to do with that section instead of just making it longer.

[final hazel]

@livid hill please communicate in english. If you want to communicate in your own language we have #1154779408989687848 & #1230630513299947581
Thank you for your understanding!

[livid hill]

ok, mas as minhas explicações foram deletadas ?

[olive night]

A picture says a thousand words..

Use the auto generated thumbnail suno gives you to your advantage, if it depicts something that isn't quite right, try and think about how to make it give you the picture you want.

If suno understands your prompt correctly, it WILL give you a good image for it..

Use the image to dial in your intent.

[polar spade]

Yes. We agree again on point.

Posting for constructive criticism seem to follow suit to us honing a craft. True.

On a broader spectrum, its cathartic just to present something you helped think up. Just that dopamine alone is nice -- and enough -- for many. And often worse when someone says anything negative or even so, constructive, that our crafted output needs changing somehow.

So it comes down to just having a fridge and a magnet to stick our stuff up for awhile to have folks look at it sometimes. When I was 5 and solicited mom to stick my drawing up, it was a big boost. We're all still big kids in those things fun.

Yes, and there are other who just want to get some food from the fridge (ask questions, learn some pointers) without having others posting their art on the door. Thats when we are big adults and express tolerance.

So maybe the point is not to have an indepth discussion until a point where we make friends and gain some respect to actually get indepth with, then that "song critiquing" can be done more one-on-one than a require protocal for everyone. Let others have their sense of accomplishment and fame for a minute just to post something to listen to or not.

Best of both worlds without judging.

As long as its properly moderated, it seems to have worked now with showcases and challenges and contests without any critique or literal judging, for three years.

I can see either side as a valid argument.

---

Welcome to Suno's Discord by the way. Have fun.

[south vector]

I'm trying to do a cover of a comp that I've edited exactly the way I want it to go, but no matter how many meta tags I use to tel it to end just like the reference, the covers never end and give me 6 minute songs with no end, and with every cycke if the endless outros, it slows down. Is this a bug in 5.5? It's got to be. With all the instructions I'm giving it using every linguistic description for it to END the song, it just won't END, literally.

I'm schooled on prompting, and using SUNO, so, it's not a matter of not knowing how to direct my prompts. I've searched and utilized all the prompt-hacks for more directional control via Grok. So, the prompts I'm using are perfectly clear and very descriptive.

I'm racking my skull and still don't have a SIMPLE good hard-stop ending with a ring out, which is all I need. But no, it insists on repeating the outro endlessly and slows down like it's dying and then stops with a cut-off.

[final hazel]

I think there’s a bit of a mismatch here. It’s somewhat the other way around to expect everyone here to use a translator to understand what you’ve written.

If you’d like to explain something in this space, it’s generally expected that you do so in English so everyone can follow along, including moderation. So it would help if you could adapt to that.

Thanks for understanding.

[polar spade]

• My translation of someone's language would not get the nuance of what they are saying.

• Having this Discord as English only, given that everyone has a means of translating, keeps a homogenous discord for everyone to participate.

• Being a Moderator in many instances, I find its better to let others express themselves by translating into English as best they can, to let them be liable for such expression, than to have me misinterpret what their native language is trying to express. More effort for them than me, yes, but for a group of social channels of pre-adults, something necessary, I think.

• Most websites are empirialistic. Whoever created it , makes the rules. Some rooms i have to translate my comments into Mandarin, some into Hindi, Vietnamese or German. Some I have to even struggle to translate into Australian. Its expected and respected. I dont push my English on their rules.

If one were to comment " Il avait la moutarde qui lui montait au nez quand ils lui ont dit non." That would not translate well into English. Let the user express themselves within the parameters of English ( or whatever language the format has set) for consistency , continuity and accountability.

[polar spade]

Am I sounding too AI today? I've got to stop dating Grok. 😆 well to be honest, shes starting to some more like me.

[quick wedge]

You just have a lot on your mind today. You’re not a bot must be the Grok influence 😄

[polar spade]

Beep boop

[paper dragon]

yes i am listening to both, i notice in the 2nd one theres more stems extracted. but still is not catching what i am looking for

[polar spade]

OK. Cool. Just checking . Making sure they weren't missed.

[paper dragon]

hi everybody, i am new to suno and use the pro plan, i noticed that theres this cover art thing that can be generated.. it's pretty pricey, i wonder from people who used this, how is this function, is it reliable?

[polar spade]

Reliable as a video AI generator? Well if we consider those , like nano-banana can take a few iterations to iron out the kinks, I guess Suno is in the same level.

So, if the credits seem costly in comparison [nano-for instance is about 350 credits under the site I use -- converting the monthly fee into credits of course] it seems Suno does a good job within ONLY a 9:16 format.

Personally, im happy to say, I like Suno's vid outputs, and used a few over what I could get for free even, elsewhere, but not having a choice of aspect ratio where I can then crop to a 9:16 or re-use elsewhere is sometimes a cause to use something else.

[celest drift]

https://suno.com/s/Zl1dZ8YDQeP6KsJY

5:24. I have been trying to replicate this sound for an extremely long time now.

i know it's retro-synthwave as i never encountered it before i started using that tag, but i have no clue how to prompt for such an effect.

persona won't work as it's only capturing voices since the recent update. anyone have an idea?

[fair adder]

help, v5 don't seems able to trigger a propre end. i write [End] both in the lyrics and style box but still it generates me longer music. i only want to trigger less than 10 seconds. should i try to push still slider to 100%? or am i just doomed?

i'm also trying to crop/remove on a clip i liked and just appears to have a short silence time where i can cut, but remove just doesn't want to stick to the timestamps i set it in and for crop i always have to listen the full song before making any change. using legacy editor cause the normal is still not accessible to screen readers

[stable sandal]

Question. This was once 1 song, I split it into 2, How do I take the second track there and make it into its own song?

[half dragon]

Maybe right click on it and download the wav file then upload it back to create new song from it?

[stable sandal]

I didn't think of reuploading it

[echo halo]

I have a big question. Has anyone here started on the free plan and later upgraded to a paid plan, and then successfully gotten an exception for commercial rights or ownership of songs created earlier? I saw that “exceptions may be made on occasion, under various conditions,” and I’m curious if anyone has actually experienced that and what the process or requirements were?

[polar spade]

• Also can mute the main track. Then saving should just save all unmuted tracks.

• Also can select the track and save "all selected tracks".

[stable sandal]

Good to know, but right now the song isn't saving hmm

[polar spade]

Well, thats odd. Save, to mean clicking the Export button, right?

[stable sandal]

Suno is having an anurism right now, doing all sorts of weird stuff

[polar spade]

I see. Just to compare: I just now tested and it exported "Entire " song as track w which track one was muted.

And it exported "selected time range" of track 2 after I highlighted it.

[stable sandal]

Yeah, the export is working, its everything else. I can't load workspaces or playlists thats why I didn't think it worked the first time

Now its throwing weird html errors at me

It just made a lier out of me lol

[polar spade]

Well, we just compared is all. It could be a backend problem and just affecting users at different times.

[stable sandal]

Oh yeah

I did get it to work so thanks to @polar spade and @half dragon for the advice!

[night eagle]

Hi, I sent an audio file. I'm not sure if it's what you need. Maybe you can decompose it into stems to see if this effect works for you, or I can send the instructions I used to create some of these effects

[celest drift]

Ahhh yeah i could splice it and download then re upload the actual audio huh

[night eagle]

I guess that could be done, although I've never done it myself, but I wanted to help you in any way I could.

[dapper elbow]

That 5:24 texture sounds like it might be the detuned chorus-heavy lead. Try adding "analog chorus synth, pitch drift, warm oscillator detuning" to your Styles alongside retro-synthwave. The key is usually in the exclude field too -- something like "exclude: modern production, clean mix, digital synthesis" really pushes Suno toward that degraded analog character.

[rough depot]

sorry has anyone had any luck preserving specifically the background beats of...damn how do i phrase it.

Listen for when he says "Functions". I want to remake this in broadway style but keep that sick beat on the chorus https://suno.com/s/zRFDXx4Jyw8ifNMB?time=105

1:45 in case it didnt properly copy my time

Dont get me wrong i am loving the broadway gens but I am sorely missing that beat

(er also how to remove breath-ins?)

[worn flicker]

Are you using Studio? Or stems?

[rough depot]

I was doing covers

[worn flicker]

Ah. I dont know what is or isn't behind a pay wall anymore. But using the background music stem as an upload may work for you.

[rough depot]

so the hard way. ok i need to work on a different song for now

[worn flicker]

Yea it does end up there. The specific beat tends to be the first thing to go in covers. In my experience.

My question:
Anyone have a remix inspired pop up on their profile from a song that has remixes turned off?

[dapper elbow]

Broadway pop hybrids work way better than straight broadway for this - try "theatrical pop, broadway belting, driving drums, dramatic chorus" in the Styles field. Add exclude: "ballad, slow, stripped, acoustic" or the chorus beat will get softened into standard musical territory. The percussion stays punchy when you keep a pop/rock anchor alongside the theatrical elements.

Put "breaths, inhales, vocal breaths" in the exclude field, that handles most of it. If you're still getting them, adding "clean vocals" to your Styles can help reinforce it.

[bronze island]

Lol at first I read autistic 😂

[quiet solar]

That's a conversation you'd need to have with Suno directly. It's not a standard process, it's something that in principle you can ask for.

But the standard position is that ownership and commercial rights are laid out pretty clearly in the TOS.

[slender fable]

How can I get a song to be in a certain key? For example, there is this song that I uploaded that’s in the key of E minor, but when I make covers of it, it always ends up in a different key (usually G minor), even if I set the audio influence slider to 100% and add “key of E minor” to the list of prompts.

[neat nexus]

It has happened. You'll need to actually send an email to [e-mail] and ask about it to see what you might need to do and what the steps are.

[unborn star]

is there a way to prompt 2 male voices in one song which have something like an duet?

[quiet solar]

Cover does like to switch up the original key if left to it's own devices, but usually prompting for the key signature and nudging audio influence up a little keeps it in check. As with all things Suno, it won't always do that, but I'm surprised to hear it's changing key more often than keeping it the same, I haven't run into that issue personally.

I would say be wary of putting the audio influence slider at 100% - especially in v5.5. You may find you get better results at less extreme levels; c.70-80% in v5 and before, or c.50-60% in v5.5 are usually more than enough to keep it locked in to the original audio, and overweighting sliders can cause odd side effects. It honestly might be worth seeing what the behaviour is with a lower audio influence setting.

[quiet solar]

Not easily.

If you're doing it by prompting alone, the trick is to prompt for the two vocalists to be as highly distinctive as possible; different vocal ranges (e.g. alto vs bass) and very different singing styles (e.g. powerful operatic singer vs fast-paced rhythmical rapping) can help a lot. But it's more challenging to get two distinct male voices for a full duet than it is, say, male and female vocalists - expect multiple failures along the way.

Depending on the details of your duet, if you're not looking for harmonies and don't mind exactly who sings what, a possible alternative is to do a version of the song with just the one male singer, then cover it using a Voice for a different male singer, and mashup the two versions.

As a final option, if you have Studio or an external DAW, you can use the 'cover and replace the singer' approach and replace the vocal stems of one singer with the other as needed throughout the song.

[fallen pond]

If you've listed before thank you. If you are interested in studio, editing tracks:

I could use some extra ears. Does the enunciation of Bad pitty at 0:41 sound right? I added the woof in through studio, the song didn't complete it on the remix, I think that's ok.

I don't know is it some kind of layover effect? If you list a little before, it seems to trail and the "d" doesn't seem to enunciate, but it you start right on the time, it sounds perfect.

Anyone experienced correcting this? Does it need to be, is it just my ears?

https://suno.com/s/8oHAxvc7zu31qfjD

[slender fable]

I only use v4.5-all.

[quiet solar]

Personally? I'd say you can leave it. The 'd' in bad is implied rather than enunciated, much like " comin' " and " steamin' " drop the harsher 'g' sound. It sounded perfectly natural to me - I wouldn't have noticed it if you hadn't pointed it out.

[quiet solar]

Ah, okay. v4 models weren't quite as good at taking on board key signature prompts as the v5 models.

If you have access to it (ie, you're not on free tier), try using 4.5+ and see if that makes a difference.

[slender fable]

I’m on free tier.

[quiet solar]

Then good luck. Does it matter if the song is produced in a slightly different key?

[slender fable]

I’d like it to be in the original key, not a different key, especially since I specified for it to be in the original key/be highly influenced by the original audio. AI should listen to the prompts it’s given.

[obsidian ocean]

i mean it does for the most part

[slender fable]

There are still shortcomings.

[obsidian ocean]

you guys every use the lyrical section for a instrumental prompt? then cover the song together and lower the style up the same as original thing?

[obsidian ocean]

yea it does like to vear at times tho. 🙁

[quiet solar]

The models develop in capability as time goes on.

While I appreciate your frustration at v4.5 not being as good at that one particular point (it can definitely do it, just not as reliably), I don't really have a better answer beyond 'keep trying' except to suggest putting it on ice until the v5 models are released to the free tier in due course. Or temporarily upgrade to Pro to fix up this specific song.

[slender fable]

It’s terrible at making drumless music, making acoustic guitar music without it becoming electric guitar, and apparently keeping certain songs in their original key in spite of specifications for it to be in the original key.

[obsidian ocean]

hmm so you are trying to get it to do just guitars only like a nice strum melody?

[quiet solar]

Not quite sure what you mean

[slender fable]

Some of the V5 models are going to be released to the free tier in the future?

[obsidian ocean]

ok idc bout views or likes, but if you check out this playlist i got 500 ish beats remixable or copy and paste w.e, but you can see in the lyrical section what i did on the later ones.
https://suno.com/playlist/f6cdf96e-6da2-4f7f-bfa0-72b8657bb1ec

if it makes any sense but skip near the bottom to 5.5 stuff

[quiet solar]

I can't say for sure, but that's been the approach over time with every other model. v4.5 was exclusive to Pro until the release of v5.

[obsidian ocean]

style for me at lest give it the idea, lyrical section directs the beat a little better then cover with min changes gets me closer.

i can try for an all acustic see if it works.

[quiet solar]

Mate, there's 420 songs...could you send a link to a particular one that has what you're referring to? xD

[obsidian ocean]

they all mostly have it i lest it open so people could see but ill go find one

[quiet solar]

Okay, that might just have been bad luck on the two I clicked into. xD

[slender fable]

Well, I was listing the different things it often fails to produce.
But yes, there was an instance where I just wanted to make a piece with acoustic guitar/strumming and it always made it electric guitar.

[obsidian ocean]

5.0?

not the best example but to show what i mean

https://suno.com/s/c4pdpyURq5lvT1NU

[quiet solar]

It also struggles with duets, tends to make electric guitars over-bright, occasionally swallows the ends of words, mispronounces homographs...

All of which pales into comparison to what it can do. I've found it an incredibly useful and powerful, if sometimes predictably quirky, tool. I found it helped to recognise where you're pushing a model's limitations and see that as a challenge rather than a frustration. And learning when to compromise, or put a song to one side and come back to it with a future model.

[quiet solar]

Gotcha.

Wow, that's more stage direction in the lyrics prompts for an instrumental than I usually see from people. Do you find that it works well?

[obsidian ocean]

more or less ive been getting closer and closer at times and others it messes around on me

but i also try to keep an open mind when the ai plays a little

cause i figure its pros they hired to make this right so helps keep it more stable how i see it

i could be wrong tho so take it for what it is

but i do notice in my last few set of collections is really does play the beat with the words, i have gotten it to cut, chop with the vocal and the beats

more darkness, demonic voices, choirs

so..

this one is a good example when its covered after and the vocal play in the begining
https://suno.com/s/GLgIH30WGWwywfR8

bout 12 seconds in.

sry if im spammin yall a bit

[obsidian ocean]

here is worked but sometimes it gives you 7 minutes, sometimes you gotta redo
https://suno.com/s/gMn6OX1tEFNDSooc1
https://suno.com/s/2CL6QL6qaysfRkcD
that what you were saying?

if you like it lyrical prompt everyones new favorite word im guessing.
also another friendly tip if you dont want the bot to vear as much and stick to as close to the prompt you asked

its the wierdness and style miss parts i think adds the drums

hope that helps you fellas, copy and past those two tho and play with it, dont worry about credits or anything. have fun!

[quiet solar]

The solo guitar piece is really nice.

I'd be a little wary of putting the sliders into 'here be dragons' territory at the extreme ends - it can introduce problems. Some of the Suno devs themselves have suggested to avoid doing it unless results nearer to default just aren't working for what you're trying to do.

[obsidian ocean]

well gotta break the mold to find something new

so it does get wierd at times but it shows me some cool stuff, but i would listen to the devs if they dont suggest it.

[quiet solar]

Oh, totally. If you're deliberately pushing it, push away and have fun. 🙂

[obsidian ocean]

idk i think the tool is pretty amazing, anything i do is in the interest in discovering or pushing what i love to do more. figure the option was there so.. i know i risk my points on a loss for it.

but the 85% range for cover does blend it better most times for me

and ill remove the style on the cover to save it from infulencing the song again different so it has to work off original.

lot more but again i dont wanna piss anyone off spammin

[fallen pond]

Thanks for double checking. After a while the ears start to mess with ya!!

Appreciate it.

[keen lotus]

I just made a circus 🎪 beat about my work lifestyle and how a Manager has been asking other people about me wearing blue shoes 👟 instead of black for 1 of my weekend days. I got ahead of that one and emailed management and H.R. to write me up if The want but my Manager has created a "Snitch Culture." A good song but I try to release 3-4 songs per week. This week was light humor

https://suno.com/s/3UEBcvscrCSdFsPv

I signed up for pro studio in November. I've not tried Sample yet, but I like mashups.

I need to change my ID picture for Discord to my psychadelic chill image. I'll get that done after work

[polar spade]

Was that meant for me?

Cool that you're exploring a bit more. I find every tool has its purpose regardless of whether I use it often or not. Always good to have options.

[slender fable]

Oh my God, what an experience. Brilliant idea to make a song about it!

[keen lotus]

Oops I'm in school not general chat. My apologies

[slender fable]

It’s okay.

[slender fable]

Thank you!

[polar spade]

Its all good. No worries.

[keen lotus]

I just wish I got alerts for who I follow when I have my favorite followers release music. Because they seem to be regulars . I'm working on WAV file format for SoundCloud, but that's another chapter down the road. I really appreciate the software.

FYI if you want to sell to friends and family and their circle. I built a questionnaire and fill in the blanks. Funeral music seems to land as Ode to Music. I throw in a surprise USB drive. It's small money, but people like personal albums for them along with instrumentals are very popular

[obsidian ocean]

Your Welcome Friend! Thanks for listening

[keen lotus]

This qualifies as education as a former teacher and intermediate Producer and writer

[keen lotus]

Facts. Sometimes one song is 3 minutes and 2nd version will be 7:30 minutes

[obsidian ocean]

yea, at this point soon as i see that circle taking longer then usual i get the aww.. feeling lol

[slender fable]

I tried the prompt and title and it worked! (I edited the title afterwards to the current one) https://suno.com/s/eCzc3ZKZV0MSAgo9

Don’t mind the strange cover art, haha. A coffee cup with two handles on one side? Okay then.

[quiet solar]

For the first coffee of the morning, when you've got the shakes so bad that you need two hands holding the mug to bring it to your lips. 10/10 design feature, would recommend.

[obsidian ocean]

aye congrats man! sry was eating a late night dinner was afk

[slender fable]

Thank you! No need to apologize. Hope your meal tasted delicious ^^

[slender fable]

Ah yes, very helpful indeed.

[obsidian ocean]

wife made it so always does

glad it worked out, and if you want to do it with another instrumental just be specific in the prompting that you do not want anything at all on it and for the lyrical part try to give a small direction in the structer.

[wintry jacinth]

Can someone please let me know any tricks to stop the stupid ad-libs that Suno makes for Lyrics.

[left hornet]

Hello, is anyone else experiencing voice inconsistency with the cover feature?

Lately when I do covers, the consistency is roughly like this:
Intro – Verse 1 (100%)
Chorus – Post Chorus – Verse 2 – Pre-Chorus 2 (50–75%)
Chorus 2 (0%) — the voice changes a lot.

What I’ve been doing is cutting out the chorus parts where the voice sounds off, then re-covering those sections, then the voice absolutly good again. After that I stitch everything back together and cover again.
I’ve tried both V5.5 and V5, but there’s basically no difference.

[quiet solar]

When you say 'the voice changes a lot', do you mean that it sounds like a completely different singer, or that it's the same singer but different vocal tone (e.g. softer, or just sung differently from the original but to the same melody)?

[unborn star]

I´m wondering if I get the same 2 voices when prompting them in Studio. Not sure if I replace one voice if I get the same voice back. So I was afraid to use it

[torn crane]

hi guys, i wanted to ask if this is the E F note??

[polar spade]

Looks like it. On a bass clef. More context would help. But I dont consider myself a musician. Id assume you'd know better than I.

Two dotted whole notes, E2 /F2 left and right on a stem so played simultaneously for .... 6 beats.

[polar spade]

Thats kind of been a thing since Suno began. So, with each model they work for consistency. 5.0 and subsequently 5.5 (with better "nuance" according to Suno) should have made the vocal more consistent. Ive stuck with correcting with extensions and adding "persona" when things go astray.

Your way, if it works, sounds fine too.

Depending on how varied we want each chorus in regards to singing style (runs, vamps, key, notes, melody) we can even just duplicate the first chorus and replace it also into any of the other choruses. Cover again or just Remaster.

[left hornet]

Completly different singer, and yeah sing differently from the original with same melody. And its always on slow romance ballad song that have build up tense, dunno why

[quiet solar]

No need to be afraid to use it - worst case scenario is it doesn't work and you burn a small number of credits in the experiment.

Personally, I'd do the cover outside of Studio to begin with, just because it's easier to work with from the Library for something like this, and start by switching the vocals using a male voice from the same broad genre. Then import both songs into Studio for the stem extraction and switching, which is where Studio excels.

[quiet solar]

Okay, genuinely never run into that issue before.

Assuming you haven't done anything unexpected or unusual with Voice or taken your sliders to the extremes, that sounds like a bug and might be worth reporting.

[left hornet]

Yeah I see,

The inconsistency mostly shows up when I do slow romance ballads with a build-up, which is kinda funny since that’s like one of the most common styles, especially in J-pop.

That’s where it tends to fall apart for me, especially going into later choruses.

Hope suno can improve every update especially in consistency voice

[left hornet]

Just want to share, the different is like this. The 2nd one is the most near my voice

The option is same, 100% on sound, 0% on others.

[polar spade]

They may be relying on +Voice to add the consistency moreso than the generative process can do alone.

Since Suno refeeds on itself basically, it can get muddled or diluted as it goes along, trying to predict what comes next by review everything that came before it, , at the same time staying consistent.... so it stays in its lane and mellows out ( gets bland and not staying crisp to the vocals and instruments) because of it.

[polar spade]

You mean:

Audio slider at 100% ?
Style slider 0% ?
Weird slider 0% ?

[left hornet]

Ahh yeah that actually explains a lot.

Especially the part where it starts to mellow out over time, that’s exactly what I’m hearing. It kind of loses that crisp vocal identity and just blends into something more “safe.”

Makes sense why it struggles more on longer build-up ballads then. But when it comes to city-pop, without build up tense, the consistency is really good.

Yes, thats audio slider i use, and, you're who suggest me that 😄

[polar spade]

Hmmm... not I.

I never suggest putting slider to their full extremes. I often discourage that. Thats why I asked.

If it's working, no harm, of course. Use them that way if you prefer.

[rugged hedge]

Hi, im making a short and i would like to know if someone can help me with the prompts of my music, im not able to achieve what i want. Is there any good practices to get movie score themes? My themes always look similar, for example, if I try to create a chase, it always sounds electronic, if i want to create horror, it always sounds kind of comical, like mars attacks ost :S, im a bit clueless regarding this

Also it would be a way to control the duration? like if i want a specific song to last 20 seconds, or maybe some part of the song that later i can isolate in postproduction?

[polar spade]

As for duration. Short lyrics can indicate a short output. More difficult with nothing in the lyrics box. But you can tag for a "jingle" or even " radio spot" 📻 as long as the prompts are strong.

But using [interlude] instead of [instrumental] may get a short bit as well.

Adding tags and styles to your liking after those primaries, to get your vibe.

As for genre, try being more descriptive within the genre, rather than horror (yes otherwise you get a therapin , ghostly sounds and a bouncy cello)

So more like creaky, distant echoing , low sub-bass drone, distorted demonic whisper, high-pitched metallic screech, reverse cymbal swell, eerie dissonant strings, heavy breathing texture

https://suno.com/s/4ZngfOxYNWEDJ2ZQ

[torn crane]

yesyes a bass clef, so it's indeed an E F note then

[rugged hedge]

many thanks for the tips! :)))

[torn crane]

thankyou @polar spade

[polar spade]

Hmmmm.... testing us? 😁

[torn crane]

nope, i was being real since I'm self studying learning and reading music and music sheets

i just wanted to confirm

[polar spade]

Dont trust me to confirm. Lol. Like I said. I know just enough to get in trouble but happy I was right.

[torn crane]

yesyes thankyou

i thought it would be a 1-5 note and not a beside note like E and F

i tried simply piano before but i forgot

[slender fable]

I have five credits left. How the heck do I use them? Can’t I just make one song? I don’t know how I ended up with five to begin with. All I did was make some extensions and save two of them into full songs. But I can’t make any more.

[shell flicker]

I need export my song to sheet. How it can i ?

[quiet solar]

There's not a way to export a song in sheet music form from Suno.

There are various tools online (of varying quality) that can attempt automatic transcription of music to notation. Generally they work best with simple transcriptions, or the song will need to be split out into its component stems first - which is something you can do in Suno on a Pro/Premier account.

[slender fable]

Also, I specified no drums and lullaby instrumental sleep music etc in my prompt and it still added stupid drums halfway through the song anyway.

What a load of malarkey.

[rugged hedge]

is there a way to mix two songs? like not making a remix, but cut one song and cut another and merge them?

[quiet solar]

Ahh...so combining Voice and Cover, and using extreme slider settings. Yeah, those are both separately common culprits for things going a little odd.

I'd suggest not using 100% audio influence and 0% weirdness as a go-to setting - especially not in v5.5, which is more sensitive to sliders. As Dasvon hinted at, heavily overweighting or underweighting sliders often introduces other problems. That gets compounded with Cover + Voice, which are both affected by the same 'audio influence' slider, even though you're actively trying to replace one with the other.

For v5.5, try something like weirdness 30% and audio 50% as a starting point, run a couple of generations and see if that makes a difference. Then if needed, gently nudge them closer towards extreme levels. For the purposes of what you're doing, you can either leave Style on the default 50% but blank out the style prompt, or set it to 0% if you prefer - with a blank prompt, it shouldn't make too much difference either way.

For the other half of the equation, if it's still struggling specifically with vocal replication despite moving the sliders out of the extreme zones, try creating a specific Voice for that song based on you singing that particular song and see if that helps.

[quiet solar]

Mashup is Suno's AI-assisted version of blending two songs together into something new.

If you mean actually taking, say, the verse of song 1 and having it go into the chorus of song 2, then...yes that's possible with Suno Studio or an external DAW. It will help a lot if the two songs are in the same key signature and tempo, but if not you can still stitch together a Frankenstein's Monster of a song and then cover it in Suno to more properly fuse the two songs together into something that sounds deliberate rather than literally cut and pasted.

[quiet solar]

As a general rule, I'd suggest to avoid putting "no (xyz)" in the style prompt as Suno doesn't handle negative prompting well and will often read that as "drums", which is the exact opposite of what you want.

I'm not sure if you have access to the 'More Options' dropdown in Create as a free user, but in there is an 'exclude styles' field in which you can put 'drums' or 'percussion' and that may help a small amount - although it's typically more effective if you specify the instruments you do want in the style box and simply leave drums out of it. The catch-all tag 'acoustic' can also be a useful style prompt to help with this.

Getting Suno to produce a drum-free track is a little tricky, since so much of it's reference material will include drums of some description. In addition to the above, you can try introducing other genres or musical styles where no drums are common to try to steer the references Suno looks up.

It's worth remembering that prompts don't work as commands that Suno is obeying or disobeying, but indications of what it should be looking to reference when creating the song, with some limited ability for Suno to contextualise what you're putting in to your prompt.

[frozen dagger]

Double check the ... More; bottom left of screen. Some have an option to "earn credits". If ya want to create another without waiting for the next 50 free daily credits.

[slender fable]

Is this on the website version?

[frozen dagger]

yeah. the ui diff for every1 it seems. and my "earn credits" is tapped out, hasn't refreshed for a good minute. but could be worth a quick check if ya just needed some quick creds for another gen.

[slender fable]

The “no drums” thing worked for DeltaWolf when producing a song with acoustic guitar only, as well as when I used that same list of prompts. I have been told both to specify no drums and to not specify no drums by multiple people and it works with some songs but not others. I honestly don’t know what I should put.

[slender fable]

I’ll look on the website version. Thank you ^^

[oblique cairn]

Hi,
Anyone tried to cover some psytrance music?
I dropped some old files i made 20+ yrs ago (very poor mix on the source files)
Im getting strong Reverbs and some noises in the background

Im on the pro plan so no studio option

[polar spade]

It likes adding drums or percussion to soft music. I think it lowers the priority of excluded instruments as it move through its auto-regressions. I have a tough time with that too.

Ambient, as a prompt helps a little, but once that 2 minute mark hits, it does regress into random/standards sometimes.

Anyway, instead of new age or sleep, try Spa music as well. But likely, id say, even Spa music has episodes of percussive rhythmic repetition that could convert to drums as Suno gets tired.

You could try -less which doesnt seem to be negated by Suno as well. Beatless, timeless.

Otherwise use strong two to three word prompts that encourage the opposite.
“deep ambient drone, soft pads, no rhythm, pure atmosphere, zero percussion, meditative stillness”

[slender fable]

Thank you, I’ll try that when my credits replenish.

I also posted an issue I’m having under ticket help with my username/handle, in case I didn’t put it in the right place.

[quiet solar]

It's not a straightforward answer, I'm afraid!

Suno can pick up context to a limited extent, so it's certainly possible 'no drums' could work correctly. However, because of the way Suno boils prompts down to keywords, it's a recognised risk that including something you don't want in the prompt can lead to Suno reading it as a keyword prompt for something you do want.

On top of that, since Suno doesn't always attach the same weight to any particular thing that you prompt, it's not easy to be sure what's going on based on a handful of generations. It's possible to get a drum-free track without mentioning drums at all, so you have to factor in the counterintuitive scenarios where you put 'no drums' in the prompt, Suno attached little or no weight to the prompt, and still produced a drum-free track anyway because of other prompting elements.

So the question ultimately becomes whether adding 'no drums' makes it more or less likely that you'll get a drum-free track. My own experience in previous models suggests it makes it a little less likely, which isn't the same as saying it will always fail. Others have experienced the opposite, which suggests to me it probably doesn't make a massive difference either way.

As far as I'm aware, the main problem is that percussion is so common to a lot of music that it carries a high innate weighting in anything Suno makes. Which is why as a general rule I'd always suggest focusing more in the prompt on what you do want, as that's consistently more effective than exclusions in any case - and it avoids the risk of accidentally having Suno read "no drums" as a prompt for drums, which to me is a plus.

But if using "no drums" works for you consistently, great - keep on using it.

[slender fable]

Thank you, I’ll keep that in mind as well.

[polar spade]

Ticket help is fine for user name and account issues. But ultimately, you need a response from Support. So go ahead and review this link to send an email as well. Get in the email queue. #support-guide message

[slender fable]

Alright, I emailed support as well. Thank you.

[pure panther]

I finally got a song I like using v5.5 but it has a near constant high pitched trill in the background. I figured out that you can remaster it on subtle using v5 to get rid of that.

[polar spade]

Great!

Remaster can help those occassionally. Sometimes removing it , sometimes just clipping everything below a certain volume or frequency, sometimes turning it into a violin, or clicks into a percussive instrument. Not perfect, and we may need to play around with combos of either 4.5, 5.0, or 5.5 amd which strength: subtle/normal/high. We might hit something nice on first try or after a few tries.

[pure panther]

I tried remastering it both with v5.5 and v5 and V5.5 consistently keeps the trill, v5 consistently removes it (although sometimes it turns it into white noise, which is more tolerable if you ask me) I think there's an issue with the model creating that super high frequency "shimmer" in the background.

[polar spade]

Downside has been to "smooth" out vocals when we want to have that crackley smokey gritty vocal or remove audiences (or clip them as they fade and pitter off) and sound FXs altogether .

[pure panther]

Unfortunately, it seems to be in the vocals. I expect that if I separate the stems to use the non-remastered version, it will transfer the shimmer over. I'll probably try it anyways though. For science.

[polar spade]

Yes! Take one for the team! For science and beyond!! .🚀 🧑‍🚀

[pure panther]

I'll let you know what I discover 😄

[pure panther]

Oof... 10 minutes later

[limpid epoch]

WHAT is easiest way to get captions file for a song?

like lyrics

[frozen dagger]

Download > Video? Suno's auto-lyric video (not perfect)

[pure panther]

What are you trying to accomplish? A lyrics file for something like Spotify?

[limpid epoch]

i just need the lyrics on a video

I tried auto-captioning with capcut but coz the song is in french its not working

[pure panther]

Do you have Vegas version 22 or later?

[limpid epoch]

no

[pure panther]

Ahh. That's what I use, it has an ai subtitle genereator that works reasonably well if you feed it the vocal stem. I'm sure there are other things like that out there, probably not free though.

[pure panther]

Did you give it only the vocal stem or the whole song? I find auto captioning only works with just the vocal stem.

[pure panther]

I don't know how everyone else's experience with the Suno generated video's lately, but I've given up using them for the lyrics. They used to be wrong every once and a while, but now they're messed up 100% of the time for me.

[oblique cairn]

Try this
https://editor.videobolt.net

I used the free account added lyrics it made the captions, then I asked claude opus 4.6 to try and download the srt file for me and it succeeded, i should have the script on my pc dm me if u want ill send it to u

[frozen dagger]

@pure panther Ya, I agree. I dunno how they generate those. It looks like a combination of what is in the song details and transcribed, because it has errors all the time. Dunno if it is a sync issue or what. I noticed it does not include ॐ in my lyrics videos either. Kinda dissapointed there. @limpid epoch just mentioned French so Suno could be iffy. Other options I can think of use a form of audio transcription without much room to edit them for accuracy. (producer.ai for example) maybe i missed it. But I've been looking for a way to streamline a workflow to create Typography styled lyric videos myself, and not found a decent regular auto-lyrics yet. Most full of errors.

[pure panther]

I've found Vegas to be really useful for it. It's not perfect, but it transcribes most of the lyrics right and puts the text events at the right time most of the time. It's definitely easier than 100% manually adding them. It usually goes on sale on Humble Bundle or Fanatical a couple times a year. I got v22 for $60 about 6 months ago.

It has to be at least v22 or above for the auto caption feature

I've been using Vegas since v14 though, so I'm really comfortable with it and a bit biased haha

There's probably a better program/method out there

[frozen dagger]

https://www.vegascreativesoftware.com/vegas-pro/
is this what you are referring to?

[pure panther]

Yep. It's a solid video editor. I wouldn't recommend paying full price though, since older versions go on deep discount so often.

I'm always 2 or more versions behind

[frozen dagger]

cool. thanks. i will check that out, and keep that in mind. Last full editor I was using was DaVinci. But with the low use I been doing switched to just using Canva. =\ I would like a full editor again tho.

[pure panther]

I tried to use DaVinci a while back, I couldn't get used to the workflow since it's different than Vegas which I'm already used to. If you got used to DaVinci Vegas may feel clunky to you. I would definitely try the free trial first to be sure.

[frozen dagger]

Seeing you in chat reminds me... I was at work yesterday thinking... ya know, I bet I could sing in the Marilyn Manson voice style... XD I need to try that...haha

[pure panther]

For sure! That's one way to get the nuances that would be otherwise impossible to describe with prompting.

[pure panther]

@polar spade It seems to be both in the vocals and the drums, it's pretty well embedded. However, I did discover that the "shimmer" lives in the 4000 - 6222 frequency band (not sure if that's the correct term) and cutting those entirely gets rid of it without effecting the rest of the song very much. So perhaps cutting that range from only the vocal and drum stem would do the trick with even less effect on the rest of the song? Also I cut it out completely for the sake of finding where it lives, but I bet reducing it could be sufficient too.

[polar spade]

Thanks. Extra info that comes in handy next time. Appreciate your detailing it out. 🙏

[severe narwhal]

Hi all, I am playing with suno, just for fun (no artistical experience etc.), and using v5.5 pro. Now I was doing a remix and using a sample, something weird is happening : the song is 8 minutes, and it starts with clear sounds and vocals, but at the end, it looks like the A.I. fell asleep, and the sounds comes from a tube or something like that. Another song with a length of 3-4 minutes is normal. I've used default values for weirdness and style influence. Is this something for the bugs department? Best regards!

[frozen dagger]

would sumbit the bug report from inside Suno. Right click the song and submit bug report and select check boxes that apply. This happens from time to time. Reporting them like this is quick and recommended. If big bug or persistent submit in app and on Discord with details in the bug report threads.

[sleek stump]

This is somewhat typical. If you are putting your weirdness slider really high or really low that has been know to cause these kind if issue.

[severe narwhal]

thx! 👌

[sleek stump]

@unique stone has done quite a bit of testing on this particular thing and came to a conclusion that some slider settings are a major factor for it.

[severe narwhal]

thx, i have encountered that too, mostly i don't touch the sliders anymore but start always with defaults, thx for the info, I'll check with @unique stone

[sleek stump]

That said, it's not uncommon for Suno to hallucinate any number of things and glitch. Some 8 minute tracks are bugged with repeated song sections after the actual end, and then just cut out, or goe well beyond the 8 minute limit. We have seen quite a few of those.

[next kettle]

Is it possible to create a 20s-30s song for a short ad with suno? It always gives me 2-3 minutes..

[limpid epoch]

is there a good way to make the suno - do 1:30 minute song or i just type , length 90 seconds, ?

[quiet solar]

Same

[unborn star]

It's possible yes

But I can't explain it right now, but you can do commercial stuff, I did a lot

[next kettle]

Ok, that's good. But how? Does anyone know?

[limpid epoch]

when i use tag "short" it usually drops 15 to 30 secs

or the term "jingle"

let me know how those work for u

[earnest agate]

https://suno.com/song/21e071c1-a9bf-46b9-b3ba-093bc39b1820

[next kettle]

Works about 50% of the time, good enough for me. Tnx!

[slender fable]

Yes! For science!

[polar spade]

5.0 and above tends to recognize "jingle" as a prop.

My prompt is...
.
" A 30-second radio jingle with cinematic music style, Features 1–2 seconds of intro music followed by a vocal hook, sung lyrics start immediately and carry through the full length, short intro, high energy,"

But you can do most anything probably.

Referencing a few simile prompts that seem jingle-like. I often put in "memo" to keep it thinking "short".

Mostt are around 19-20 seconds. Tweak lyrics until you get a longer result closer to 30 seconds.

https://suno.com/s/IhxrcJdA8npjLoJR

[dapper elbow]

This message has no context — "But how? Does anyone know?" without a prior thread or topic means there's nothing to reply to meaningfully. Per rule 10, this is a skip. Share the surrounding conversation thread and I can write something genuinely useful.

[sly jetty]

helloo

is there issues with extending songs on v5.5?

[pure panther]

@polar spade High pitched shimmer defeated! 😄 I jumped through many hoops, but in the end the most useful thing was remastering on subtle in v5.
https://suno.com/s/W7MJ805MZoFhRyPU

[delicate karma]

This tool is literally a Suno credit destroyer

[tropic frost]

hi all, I am trying to generate instrumental music in Suno that is more rhythm-driven (pulsing / groove-based) rather than melody-driven.
I want to avoid strong melodies, lead instruments, or solos, and instead focus on repetitive rhythmic patterns and texture.
Does anyone have tips on how to prompt for this?

[limpid epoch]

any way to get 90 secs?

[polar spade]

First rule of Suno. We cant get exact time.

Lyrics tend to control the length. Tweak it and test.

Generally a verse amd chorus may be about 2 minutes. Not knowing what youre going for, Its easier to go two minutes and cut back / edit what you can to get 90 second or 1:30.

Cover or Remaster to smooth it out.

The other version is to start then with the short 20 seconds and extend to get close to 90.

[polar spade]

You've been on Suno awhile so ...... "SUNO Enhancer" ?

Honestly, I've never seen this layout/format for Suno. Interesting. Does VN get a different layout?

Are you paying a Pro subscription at $10 USD a month (about 263000 VND). Just making sure this is legit.

[polar spade]

What's the mp3s for mate? Need help with anything?

There is a showcase channel to post songs if you wish. https://discord.com/channels/1069381916492562582/1083129341870878780

[old lion]

Hi Fam has anyone successfully moved from Yearly Pro plan to the Monthly Premiere plan, Premiere is a bit steep to go all in without a bit of play? 🙁

[delicate karma]

I'm using a one year Suno Pro account, and the Suno Enchancer tool I'm using was also created by me, with added features and 4 language options for the UI interface.

[vivid cedar]

Where to find the remaining resources on the latest GUI (pro user)🙂

[delicate karma]

Just one simple prompt line to create a quality piece of music.Just one simple prompt line to create a quality track. I used to delete 80% of the tracks I generated because they didn't sound good, but now I feel like it's really good

[regal charm]

i dont know how to fix all those new studio issues since the newest update. new tracks created are off-tempo, new tracks created are not in the style given by prompt (mostly meaning: vocals are like a full track, solo instruments come with drums, stuff like that) - any1 has some fix for those things?

[polar spade]

Thats how its done, my friend. Simple seems better.

[polar spade]

Here is fine, but you may want to ask in studio-chat as well, for a more focused chat.

[regal charm]

oooh you can activate channels, i thought they removed some of them haha sorry

[vivid cedar]

Any1 How to spot the remaining credit of the month🤨 For some reason quit well hidden

[delicate karma]

Do you think it's okay?

[vivid cedar]

very strange this is no longer visible - I used to check for credits here

Im on Safari. I'll try other browsers

[delicate karma]

Are you using any ad blocker? Try clearing your entire cache or opening an incognito tab

[vivid cedar]

Tried firefox and Duck Duck Go - same result. I'll try your advice later. Thx for now

[quiet solar]

There's now a toggle in your Account Settings to restore it

[vivid cedar]

Yes! Would never have found it Thanx. Strange for Suno to toggle off. I certainly never did 🙂

[jolly frost]

Don’t do it. I’m now trapped. I had a yearly pro plan and signed up for the monthly premier plan. However that won’t be applied until the yearly plan runs out, next October

[jolly frost]

I think you need to turn it on now. In the account settings IIRC

[old lion]

thanksbuddy I got it sorted. Just send an email to accounts, they can pro rate your account and allow you to get the premiere instantly 🔥

Anyone have a work around for stuck stem extractions? I have 4 extractions going on spinning for minutes now 🥲

[hoary stream]

hello, is there an official list of meta words ? and symbol for usin suno ?

[polar spade]

No. Not really. But there are a few in the workflow channel to get you started. Otherwise, you can browser search and find quite a few sites with lists.

Personally, it quite easy to just ask an AI for some suggestions based on popular posts and what styles youre looking for.

[coral glen]

I second this, personally I have simply asked my favorite LLM/AI for help with my SUNO prompts so far, but the workflow channel also has an excellent (very thorough) document called Suno-AI-God-Mode-Manual in it.
For the sake of not violating the "no promotion" rule, I'll keep the exact LLM to me but it's a popular one, so not hard to find.

[scarlet crater]

I have seen several approaches about prompting, the main ones are :

• Simple style + Lyrics with local instructions (the one I use thanks to LLM)
• Complex style with precises instructions + Lyrics only
• Simple style + Lyrics with a huge bloc of instruction before lyrics

Curious to know which one do you use and why ?

[quiet solar]

It depends what I'm doing, but if I have a very clear and specific musical vision in mind then I'll tend to do slightly more detailed style prompts. If it's a vague sense or trying something out, I'll keep the style prompt simpler with some broad brush suggestions for Suno to focus on.

I'm a little wary of overloading the style prompt, since Suno won't assign equal weight to everything you put in (early prompts are more important than later ones) - so if you put everything and the kitchen sink in there, it's likely Suno will skim over a lot of it.

I'll use metatags in the lyrics if there's something specific I want Suno to do at that point, or for directing the vocalist(s) in things like duets. Suno is even more likely to skip over directions here, so again, I find focus is more effective than overload.

The block of instructions before lyrics is an interesting one - based on experience, my expectation is that would act as a less effective style prompt than the actual style prompt, but I'm aware some people swear by it. Where initial instruction can be helpful is in giving a steer for the intro to help set/reinforce what's going to happen in the rest of the song.

[solar wolf]

Hi!
I need some help. I made a really great song using v5.5 and I remixed it and it sounds amazing however…. It’s taken 1/2 of the song and stuck it on the end and it abruptly ends mid song. How do I fix it? I can upload it for y’all to listen and give me some advice. If this is t the right sub let me know.

[prisma hound]

I developed a GPT that injects lots of DaW-like commands and which is based on several datasets of styles, instruments, etc. This really works well, together with a very detailed positive style description (and negative style always). Nevertheless, simple and clear must always remains at the core.
As an example:
https://suno.com/s/Ba4XPGhZW2ujoKyk

[scarlet crater]

Okay thanks, do you use standard negative style ? (such as for image generation using things like "bad fingers, poorly drawed face, etc")

[prisma hound]

No, to my experience what works best is to list style and instruments you don't want, separated by a coma. But you can indeed says you don't want noise or whatsoever

[quiet solar]

Could you crop out the randomly extra added portion of the song to fix it?

Also, sure - send through a link and tell us the timestamp where it goes haywire, that'd be helpful.

[wise summit]

Did they remove the persona creations? i can only record my own voice now?

[quiet solar]

You can still create Personas the old way (they're now called voices too) from the Library or from the new create screen, although that's a bit more involved.

[solar wolf]

https://suno.com/s/0It5Kc2oqFWpFlEx at 5:11 it’s supposed to end. Let me k ow what you think.

[wise summit]

I need to verify with my voice when i do it, and damn., im not that good singer

[quiet solar]

Yeah, don't do it through Create - pick out the song from the Library, use ... -> Create -> Voice

[worn flicker]

I agree about the lyrics prompting. It was a maybe thing in earlier models that seemed to work for some things. I ended up using it when making instrumentals rather than trying to steer the song that way. Tags and style are way more effective in my opinion. Also about the big list in style, at one point using capitalized words helped weight in style, but once it started accepting full sentence structure I believe that went away. Regardless, I tend to put the most emphasis at the beginning. Main style, vocal type, secondary style, then at the very end ill use some instruments or other core style words.

[quiet solar]

Yeah, super easy fix.

Go to the song in the Library, select Remix/Edit -> Crop and set the timers to 00:00.0 and 05:12.0

[quiet solar]

Oh yeah, instrumentals are a whole other ballgame. Using metatags in the lyrics to steer the song can definitely be really helpful.

[solar wolf]

I’ve been pouring through my lyrics and wanting the song to come alive. Thanks for the tips.

[solar wolf]

I see a lot of remix options. Remix: extend. Remix: cover. Remix: reuse style. Which one should I use?

[dense hinge]

hi, if i do a cover of a song, but use a voice persona, how do i make the end result of the singer sound like the persona voice and not the original track voice?

[normal ridge]

Hit and miss on this one, it’s like the two vocalists are fighting to emerge haha

[frozen dagger]

either stem split to remove the vocal and replace entirely, or try covering covers to push it out.

[dense hinge]

yea i cover covers. no such luck. but someone in the main chat advised to put in the instrumental, and do a recording and persona on it

[quiet solar]

None of the above. You're looking for Crop - I'm not sure if it's available via the mobile app, so it may be something you have to do on desktop:

[quiet solar]

Main tip is don't turn the audio influence slider too high, as it influences both the existing audio and the persona. In v5 and before, somewhere around 50%-70% was usually a good range. I've not tried it in v5.5 much, but given it's general higher sensitivity on sliders, maybe try something like 40%-50% as a realistic maximum. If the original voice still comes through, turn the audio influence slider down a bit more.

[dense hinge]

this audio influence thing is so vague. What does it affect? The persona or the cover? turn it too high, the voice sounds like the original, turn it too low, they change the melody of the original, yet keep elements of the original voice.

[quiet solar]

The audio influence silder weighting affects both the original audio being covered and the voice persona, as they're both audio inputs. Which is why what you're looking to do is a tricky exercise, since Suno doesn't psychically know what you're trying to keep and what you're trying to change.

Too high audio influence will tend to make it stick closely to the original sound, including the original sound of the voice, which is why you need to lower it if that happens. Melody changes are a likely side effect.

If you want to keep the exact same instrumentation while switching the voice over, you really need to be looking at it as an editing exercise, rather than just running it through cover. I can give you some suggestions for that if you have Suno Studio or an external DAW and are open to audio editing.

[dense hinge]

Thanks.

[old lion]

Has anyone had the spinning circles of death when doing the extract all stems in studio? Seems like no new attempts generate anything and no way to cancel? Any workarounds?

[polar spade]

I think worth mention, as a note if not known already, its good to check your Library immediately (also clicking Unhide Stems -- in the Filters) after clicking for stems as the first source of stem generating. When they are done. Go back into the stem module (or Studio) and discover if the stems are all loaded there as well.

So also be familiar with how to go get back into the stem module by finding "See Extracted Stems".

[fluid sentinel]

When extracting stems (wav-tempo locked) how do you find out what the tempo is? Also, if you don't use "tempo locked" will the stems match up with the original gen, or will there be some minor changes?

[old lion]

Unfortunately no bueno, nothing in library view with filter off or in see extracted stems view, I guess this is known issue? what happens with these stuck projects?

[polar spade]

They do have tempos, but kind hard to find. You can at the least, go to the editt UI and check the bottom readout to find the tempo.

Also, you should click the tempo cog and choose "Manual Tempo" to get the solid average BPMs.

[fluid sentinel]

I just saw that the downloaded stem folder has the tempo listed in the file name 🫠

[fluid sentinel]

I figured out my queries. Cheers

[polar spade]

Im lost a bit. What's a "downloaded stem folder" . Could you screenshot that to show the .... tempo... in the ... what.... filename?

[polar spade]

Oh cool. Great. Glad you worked it out.

[fluid sentinel]

When you "Extract all Stems" if puts it into a zip folder with the tempo at the end of the file name

[polar spade]

Thats what I thought you meant. OK. Great. Yea. It shows up here and there. Assuming we're not downloading it, there are the other ways. Good to know the options.

Downloading a single stem won't show it. So there's that.

[fluid sentinel]

Loving these new features. Absolute game changer.

[spiral wave]

Mornin. I'm trying to extend a song with an instrumental outro but it keeps hashing out the previous lyrics. I've put in [outro | guitar solo ] something like that. Selected a bit at the end to replace. Feels like I'm missing something here. Tips?

[fossil tiger]

Hi everyone, I need some help.

I usually record songs directly into the Suno app on my phone (just guitar + vocal), and then use prompts to generate versions as a starting point while I’m writing.

Normally, Suno saves:

• the original recording (exactly as I recorded it), and
• two generated covers (often instrumentals)

But recently, something has changed. My “original” recording is now being saved at a slower tempo and with my voice pitched down (it sounds like a deeper male voice), even though I recorded it normally.

I’m not applying any effects manually, and this is happening automatically after recording.

How can I:

1. Get my original recording back in its true form (normal tempo + my real voice)?
2. Prevent this from happening going forward?

Has anyone else experienced this?

Thanks so much 🙏

[polar spade]

Thats more sounding like a bug. The recording should only be that. A recording. Same samples rate, same tempo. Something in the pipeline is stretching it during the processing, it seems. Report as a bug. 🐛

[young geode]

Official, no, but I've compiled a lot of tags in my work flow thread

https://discord.com/channels/1069381916492562582/1382019029266600067

Are you looking for a specific kind of metatag or control, or just how to typically format it?

[hoary stream]

both, i'm completly new at it

[young geode]

Go ahead and read the guides

[hoary stream]

i'm on it it's very instructive merci !!

[covert granite]

Is there any way to get suno to generate more than 2 1/2 minutes. I have a 5 min song and it only generates 2 1/2 minutes of the song smashes it all together and refuses to make it longer. I'm feeding it 5 min.... why can't it finish the song. if I feed it in pieces hey don't match.

[quiet solar]

@covert granite Can you give a bit more information on what you're trying to do, and the steps you're taking in (I presume?) covering the song

[covert granite]

Sure.... I put in an old recording from a midi song I made 10 yearss ago...,. I want to cover it change the insrumentation...... let's say..... the song has 3 major sections....... it's 5 min. I say cover the song..... it does 2 1/2 minutes...... sometimes I get more but when I do the extra minute it's terrible like extend doesn't match anything. I've rarely gotten it to generate over 4. so it either skips the last section completely or it mashes them together with out the musical continuity that exists between them. even if I have a 5 min piano piece solo piano .... I say cover this with a fender rhodes piano let's say it will only do 2 min or so.

[quiet solar]

@covert granite Cover does take up some of the processing time that would be allocated to the max length of a song, but not usually that much. It sounds like there's something which is throwing Suno off when it tries to replicate it, given this seems to be a consistent problem.

Which Suno version are you using for the cover? And would you be willing to share a link to the midi upload - ie, basically the upload in Suno you're trying to cover (DM if you prefer not to put the link in here) so I can test on it?

[covert granite]

I'm speaking by way of example. I've been fighting this problem for like a month now. Yes, I'd be willing to walk through one with you with a real song. I just feel like I'm missing something big and I've prompted it a hundred ways and never gotten any consistency. It's mostly instrumental music I have this problem with. working with vocals seems to be better and I've tried to play with prompts in the vocal section even for instrumental pieces and it has some influence. But in general I'll do 10 different covers trying to get it to go over 2 1/2 minutes or so and it just refuses......

[quiet solar]

Yeah, it's definitely an easier exercise when you've got lyrics to work with in the lyrics box, but I've never known Suno be this problematic with instrumentals before.

[covert granite]

ok let me get something simple and see if yuo can get it to cover the entire song.

[covert granite]

ok here's one that's 4 min it covers 2:27 or 2:57 strips off an entire minute. how do I get you the original mp3?

[quiet solar]

If you DM me here on Discord and attach it, that's the easiest route.

Alternatively, if that's a problem for any reason, you can go to the Suno upload of it in your library, chose the three dots and then Share -> Copy link and send me that.

[covert granite]

ok standby...... on he way.

[floral solstice]

Hey, relatively new to Suno, got a question. Im trying to edit an existing song in the Editor. Im trying to add a new section. Clicked the plus, pasted in the lyrics (lyrics are the same as the hook, basically I just want another variation of the hook between verses) but somehow it always ignores the lyrics and just remakes the verse that comes after? I also marked it with "[Hook]" but it always ignores it.. Am I doing something wrong?

[quiet solar]

If you're trying to insert a new hook between two existing verses, then personally, I'd suggest doing that through cover. Cover the original song with style prompts unchanged and add the new hook in where you want it in the lyrics, then (if you're on Pro/Premier) set the sliders under 'More Options' to something like:

w15% s50% a70% for v5 or before, or
w35% s50% a50% for v5.5

It'll likely make a few underlying changes to the arrangement of the original song, but I suspect it'll be much less painful than using the Editor.

[old lion]

how do you get soundeffects,I tried to put this tag at the very end of the song and its ignored? [warm notification chime tuned to song key, a breath, silence]

[floral solstice]

Thanks, made the song almost the same, worked!

[quiet solar]

Suno doesn't reliably do sound effects as part of songs (it wasn't trained to be a SFX AI).

There is a separate part of Suno under Create called 'sounds' which is a bit more free form and can work better with sound effects, but you would need to blend those into the song yourself through editing.

[polar spade]

Notwithstanding the attempts to get a sound effect simply by prompting it in the song....

I think I mostly started with the sound effect...
Then started the song part of it by extending off the sound effect with the most silent of pauses between them, and then the extension starts with [intro] or [inyerlude] as if we are creating the new song from the extension. THen edit the song and move the sound effect where needed, but knowing that it wont blend.

Works OK for maybe a car racing, a gunshot, things that are meant to be separated anyway.

[echo citrus]

I’ve always got my sound effects by putting it detailed in the prompt first and adding it in brackets additionally in the lyric section. That usually gets the sounds I want,

[polar spade]

" . . . Tuned to song key..." ? I admit, thats asking alot. 🙂 Luckily getting something random is the better odds, but we get those sometimes.

In the metaphysical world of Suno , you could try using Sounds tab and creat the exact sound you want a few times, then try creating the sound afterwards , either separately dor as an extension and see if Suno adds the same or similar sound effect by osmosis (hoping something is still in a buffer somewhere).

Obviously the right answer these days is to use Studio and just patch it all in.

Final note is to make sure you know there is a Sounds FX tab in the Create panel? That does sound effects.

[polar spade]

Yes. We have crossed our fingers and hoped. From most posts over the years on attempting sound effects, there were only a few that seem consistent, such as gunshot, babbling brook, maybe wind, a creak, a beep, thunder, rain, and a few more.

If you usually get what you prompt for and have tested a variety of SFX then you are most definitely gifted. I'd love to see/hear some results of some obscure sound effects beyond the basics from prompting in just Basic or Pro. That would be cool to hear. Maybe a playlist? Ill study what you do. I love to learn.

But yes, your way of prompting has been the more reliable for those few primary sounds. Prompt in the style prompts and tag in the lyric box at the same time seems best.

[echo citrus]

For sure I will work on something’s today. I tend to do that when I’m working on tracks for my machinima projects ( story line sims ). Anything not put I edit in later but to be able to get it in prior is definitely a time saver. What I’m really working on now is being able to layer the track with sounds in Suno.

I will definitely post any successful soon effects I have gotten or may get.

[civic fossil]

anyone make songs where the instruments / beat are beat boxed? 🙂

this has proven to be tricky, it always sneaks drums in

[past sandal]

Question: when I replace a section of a track in suno and ask it for something slightly different, it does what I ask but it doesn’t seamlessly connect it back to the original track?

[frozen dagger]

that's pretty normal. usually more remixing is needed. Either doing a cover of the whole song, or editing with a DAW.
It normally doesn't just fit perfectly. At best it fits really well, but has clipping (or volume differences) between the transitions.

[past sandal]

Hmmm, thanks! Weird though, because when I first used the feature it worked really well. Not it doesn't even attempt to sound good

[frozen dagger]

If you're doing the replacement inside Suno Editor, double-check the checkbox for "smart / fixed / classic" options in the gear icon bottom left.
The length of the section we replace is one factor, the surrounding audio matters as well, and how we are replacing.

[past sandal]

Okay thanks

Ill give it a go

[fossil tiger]

Hi everyone, is there a way that once a song has been made and a lyric is wrong to fix it without having to regenerate the whole song?

[polar spade]

The new Edit UI is the best source for that then. You would use Replace tab. It would still need to regenerate a section of a few lines, but you can get it down to about 6 seconds (sometimes 3) to just replace the one word. Start by choosing the Fixed option below the styles box as your blend filter. You may need to widen by using Smart. Then Classic if the other two dont do well.

[obtuse lark]

is it possible to make a music with Dual voice?

[civic fossil]

yes

[obtuse lark]

how

it keep glitching

[civic fossil]

I usually type "male voice for verses" and "female voice for chorus" you also have to figure out many artists you want, it starts adding too many voices

[civic fossil]

someone mentioned the newer model might be glitchy for the time being

[frozen dagger]

multiple methods. seach through https://discord.com/channels/1069381916492562582/1252267444773326848 to see some user submitted methods. (If on PC use CTRL +F and search keywords; duet; vocals)

[obtuse lark]

thanks]

[frozen dagger]

There is prompting methods and creating Voices and combining those methods and more.

[civic fossil]

there's one more thing, once you have a song that has the voices you want, make a Suno voice from it, select the two voices

this is what I do, that way I guarantee the same two voices

I stick to duets its easier than managing too many voices

[drifting palm]

I made a song a while back, before this new voices thing, and Ireally like the sound ofit but the voice sounded kinda silly. Is there a way to replace the voice in the song using the new voices feature, without regenerating the song?

[civic fossil]

studio probably

if you get the stems, you can make the voice from the song you liked match the tempo of the newer songs instrumentals

havent done all of that fully yet, but I was playing around with the voice settings yesterday in the studio

[frozen dagger]

stem split the song, vocal and instrumental. cover the instrumental stem with the Voice feature.

[drifting palm]

yeah the new voices are kind of awesome, works better than the old personas

[drifting palm]

not a bad idea, ill try that, thanks

[civic fossil]

<@&1082792303976001547> bot scam

[dim glacier]

💀

[fluid robin]

removed one in chill chat as well

[dim glacier]

Thank you 🔥

[rugged hedge]

i have a dumb question but its killing me not be able to do it, in studio, how i can restart from the moment in which the song has stoped and not from the moment i selected ?

also how i can see the duration of my selection?

[frozen dagger]

the playhead always goes to where we clicked in the timeline. i ran into similar issue trying to play and make edits in real-time while listening, have to create a split in the track if you want the playhead to jump to that point when you select/click on that section it. (If we click on the selection the playhead goes to start of selection; where you made a split. Otherwise, if you click on the timeline (not the header of selection) the playhead goes where we click.

[rugged hedge]

😢

[frozen dagger]

for duration of selection. i dont know that it shows anywhere in the UI the duration of time for the selection. you have to math it out by checking the timestamp at the start and end of the selection. (as far as i know)

[rugged hedge]

gosh

[frozen dagger]

people smart(er) lol than I are probably able to get a good idea of the duration of time just by knowing the BPM and the time signature. Changing the grid would help i imagine in this case. I look at the timestamp myself. =\

[limber stump]

Hi all. QQ: With this new voices feature release, does it mean persona's will no longer be a thing? Right now the option to create a voice from a song generated by Suno is greyed out. I only have the option to record my own voice (and i sound like a dying cat 😂). Are we no longer able to create a persona from songs created by suno?

[frozen dagger]

Make sure the song you trying to make a Voice from doesn't have a Persona/Voice on it already. (There are some other issues that can cause Suno generated voices to be greyed out. Derivitaves of uploads and remixes for example).
*Try opening the song in the editor and "Save as new song" this will create an exact copy of the song, wait for that song to completely load (refresh web page) see if you can now create Voice.

Voice is Persona, but the UI has some weird hiccups still. Different routes needed to create a Voice with Suno generated voices, and our own voice. But the funtionality remains the same.

[limber stump]

Thanks for the advice. I'll do this and see how i get on. If i was a singer, the new voice option would be the one to use. I'll give this a go 👍

[frozen dagger]

The "Save as a New Song" removes the Persona/Voice icon from that you see attached to the original and allows you to create a new Voice.
This trick fixes a few things. If ever run into an issue this is quick, easy, and doesn't cost credits. So usually worth a shot.

[frozen dagger]

and using your own voice is AMAZING! Best feature imo. There is so much more you can do now. And it's you in there. It's great.

[limber stump]

I dont have a persona attached to the song so the 'save as a new song' option may be the trick 🤞

[frozen dagger]

It won't help if the audio is a derivitave (cover/remix) of anything with uploaded audio (with vocals), or uses audio from another users Suno song. (I think)

[limber stump]

I did use my own music as the audio input..... its not a remix of an existing song on Suno but it's using uploaded audio (my own track). So it may not work if i used my own track?

[rugged hedge]

is there any way to update the selection in the moment the song is playing?

in studio

[frozen dagger]

correct. uploaded audio may not work. I'm not 100% on this because things changed over time. I think it matters if it "detects" vocals. Which could be background noise even that isn't audible to ear.

[limber stump]

Hmmmm.....only one way to find out lol I'll give your suggestion a go and see if it works. Bit of a shame if it doesn't because i don't sing and need a AI generated voice.

[frozen dagger]

1 way to find out is right. 🙂 your voice can become an ai generated voice as well. play with it. Covering a song with vocals already in it with a Voice does more blending and switching rather than replacing. While counter-intuitive, it can create new blended voices (your own and suno blended).

[fair adder]

hello, is there a way to prevent the my tast text box to work? i mean, i've disabled the style prompt generation or i don't remember the exact name of the thing and i'm too lazy to check right now, but the text field with all details is still there and i don't want it. i have the feeling it's falsing my generations, especially when i go out of my comfort zone/just the stuff i'm generating, and since it takes a lot of gens, suno thinks it's my normal and usual muci genre and stuff

[drifting palm]

this was for sure the move, thanks for the tip

[rugged hedge]

am i running maybe into a bug? in the doc it says the pause button keeps the actual position, but when i pause the song the marker goes directly to my selection

[frozen dagger]

To disbale the "My Taste" click on your profile picture, and then 4th option down is My Taste, opens new window where a toggle for "My Stles" will turn this off.
It should only be effecting what appears when you click the blue magic wand, the wand in the Style box.

[rugged hedge]

it makes basically unusable studio for me... :/

ijust need the pause button to behave like a pause button.

[frozen dagger]

that does sound like a bug. if i press spacebar to pause, the playhead should stay where i paused and play from that location if press play/spacebar again. (without making any other selections in between)

[fair adder]

... yes, that's what i meant. i've disabled that, but the other field is still filled somehow, even when i try to delete it. it just won't stay empty. wanted to know if this was normal...

[rugged hedge]

ah gosh, so its only happening to me? how i could fix it? should i contact suno?

[polar spade]

Are you specifically talking inside of the My Taste Box? You can select it all and delete it by putting in just a period / stop dot or a slash or anything you wish and click SAVE. Then nothing will ever be there until you click "Revert to Generated" button found beside the SAVE button.

The trick is you have to put something in the box and SAVE it. Then that "nothing" empty box is now your custom taste, which is nothing.

[frozen dagger]

just checked... the playhead is at my selection point, press play and it plays, i pause and press play the playhead is going back to the selected origin point. o.O

[rugged hedge]

my problem is that when i tap on PAUSE the playhead goes back to my selection, so i cant track at what point i was when i paused

[frozen dagger]

this was not operating like this before .... i dont think. can't trust my own brain. but i should have footage to check lmao

[rugged hedge]

its not when i tap on play, that i would understand, what i dont understand is why it goes to my selection when i pause the song

[fair adder]

oh! imma try this right away, thank you so much for the tip

[frozen dagger]

no yur right its same

[rugged hedge]

so its a bug?

it decides to appear today that i upgrade my account.. 😄

[frozen dagger]

@rugged hedge I don't know what I said that triggered Woodhouse, but..
I just checked old video. Pressing pause used to stop the playhead on the timeline where we press pause, as one would expect. It did not jump back to the selected point.
Is this a bug, or just how it works now? Dunno, can't say. Not expected behavior for the pause button tho.

[rugged hedge]

its probably a bug

[limber stump]

@frozen dagger hey i tried that and it didn't work 😞 not sure what the issue is because i have another track done the same way (uploaded my instrumental to Suno and created a new song from it) and i can use the voice off that track. But i cant use the voice i want which is on a different track. Both made the same way using 2 different audio tracks but one lets me use the voice to set a persona the other doesn't. Is there a way to contact the Suno team? This may be a glitch that needs looking at.

[frozen dagger]

Not sure which route would be best for contacting the team, probably https://discord.com/channels/1069381916492562582/1479483412250366054, but it is kinda swamped with tickets and could be a while before you get a response.
You would likely find a good workaround before something could happen with that specific track being able to create a voice from. Try to create a new song with Suno in same styles and try to recreate similar voice?

[limber stump]

It's a bit of a weird one because some tracks let me create a persona and some don't. Yet they've all been made the same way: Upload instrumental > add lyrics and style > generate track. 🤔 i can imagine they've got loads on their plate right now

[frozen dagger]

Curious. That save new song trick solves most issues (can't create Voices from songs that were exported with Studio either, but saving as new song in editor removes the Studio icon and saves as the model selected).
I don't have hopes of it working, but I would try extradct stems and see if i can create a Voice from the Vocal stem.
Downloading the stem would let us edit it, but if we uploaded the stem back into Suno we can't make it a voice becuase it's an audio upload and needs to pass voice verification.

[limber stump]

That's a good idea. The things we have to do eh 😅

[fossil tiger]

Ok thanks but I’d still need to learn how to do that. What about remastering the track? I did that before after adjusting the lyrics and the new version vocal used the adjusted lyrics however that doesn’t seem to work anymore?

[tame venture]

https://jray.me/Suno101.html

[frozen dagger]

https://discord.com/channels/1069381916492562582/1252267444773326848 put it ther so it doesn't just float away with the chat.
but if it selling somethign just don't post it. 🙂

[tame venture]

Nope all free advice

[old lion]

Is there a trick to generating new tracks in studio to get the new generations to line up with the main song? I select a segment that runs whe whole length of the song and the new instrument track generated is unaligned with the main?

[frozen dagger]

I think the solution to your issue lies (maybe partially) in checking the clip settings for the track, "Original" or "On Beat".
Set one track to Original and another to on beat, change the BPM at the bottom and watch the track grow and shrink. It's adjusting the waveform visually as it is matching the set BPM at bottom. There is an option to Follow track down there. I don't fully understand it enough, but that is most likely why it's not aligned. 🫡

[old lion]

Thanks it might be a bit like that, but also I assume what ever is not muted is context for the generation to follow? It generated a bit more intro (outside of bounds of the original track) and it was significantly unaligned. But thanks you gave me something to try.

[frozen dagger]

Suno does that, it's iterative. The context window can be adjusted. Most miss that. Little arrows at the very top when a selection is made. Expand the window to use the advanced sliders to try and increase audio influence (if not already).
Click here to see an image someone else showed to see how to expand the window in Studio if not familiar: #🪲┃studio-feedback message

[hoary stream]

hello !

[quick wedge]

Hi Flo. How are you? Hope you’re doing well

[hoary stream]

doing fine

i'm trying to nail down the time of voice liek in this song : https://suno.com/song/22da6a7b-ad45-4dc9-a028-c58397691675

but i just can't reproduce it as crispy and clean, and even my best attempt doesnt stay stable all along the song, i have no idea how to maintain the quality of the voice all along...

if someone could help me in DM

[normal root]

Hey! Just wanted to share this—I made it based on your lyrics and vibe. Let me know what you think!

https://suno.com/s/7DfX2PhXlrlEtHB2

[hoary stream]

let me see

[sonic folio]

is it possible to mix genres? for arguments sake, if i want more melodic guitars but grindcore/deatcore drums. can suno blend that? i know for a fact by testing that once you add "melodic" anything in style-prompt you go from deathcore down to metal :/

[obsidian ocean]

one way to find out, its blends the ones i play with pretty good.

[sonic folio]

How do you prompt that? Or can you link?

[obsidian ocean]

like how do you get it to blend genre and different things together lke that?

lemme go figure out what deathcore is

[analog flume]

Hey all! I've been using Suno Studio Pro for a bit now and I'm struggling to figure out/learn something. I am using Suno Studio as more of a DAW (GAW as they call it), and generating individual parts to fit into a skeleton recording I did with my guitar and vocals. When I generate Drums for the verse for example, and I end up liking a version of it - now when I want to generate drums for the Chorus, it is a different sound because it is a different generation. How do we get the same "seed" almost and get the same drum sound/style/etc. such that it feels like the same drummer played the entire song on the same kit, and not disjoint drum sounds?

[obsidian ocean]

Use the same seed (or reuse/extend the original generation) so all sections keep a consistent drum sound and style? my guess

[analog flume]

Yes, that's what I'm trying to figure out. But not sure how to do that. How do we find out the seed? Where do we mention it? And how do we "extend" a part?

[obsidian ocean]

reuse the style i can ask my little bot see what it says. this is the answer i got for ye "You can’t directly view or manually set the seed in Suno, so to keep consistency you need to use the “Extend” or “Continue” feature on the original clip (or reuse that same generation as your base) instead of creating new separate generations."

[obsidian ocean]

idk much bout deathcore but this was my attempt to try
https://suno.com/s/9riCoSu5fFIq5Yer

if that was close can do w.e with it, but i left what i did there.

[sonic folio]

this is very close to what i want. now the problem is that if i get the instrumental part good, i cant get harsh vocals through out the entire song....suno always finds some tiny part to make clean vocals

[sonic folio]

Do you think it's better to make an instrumental track then add the lyrics?

[old lion]

try adding prompts in the lyrics if its a performance style you want to encourage

[obsidian ocean]

I do in most cases

[obsidian ocean]

are you using army terms like angry, scream, yell lyrics. alot of times playing on the words may do it.

[sonic folio]

haven't tried that and not eve sure how to prompt that 😅

[obsidian ocean]

hmm

brutal deathcore with grindcore drum intensity, relentless blast beats, downtuned crushing guitar chugs, chaotic tempo shifts, raw aggressive energy, heavy screamed vocals, loud and dominant vocal mix, high intensity fry screams and deep guttural lows, layered vocal stacks, shouted gang-style accents in breakdowns, vocals pushed forward in the mix not buried, explosive vocal delivery during peaks, controlled quieter tension moments before sudden loud scream bursts, contrast between silence and violent vocal impact, breakdown sections feature slow crushing riffs with aggressive shouted phrases, melodic guitar layers remain underneath but do not overpower vocals, modern heavy production, punchy drums, thick low-end, dark twisted atmosphere, high energy, intense, replayable

with the prompt adjust if you cover or remix it think cover works better then put those it should kick in at part, no promise but can try it

maybe play with style infulence bar a little see what it does.

[sonic folio]

thank you very much 😊 i will start playing around and see what it gives me 🙂 i will @ you later 🙂

[quiet solar]

Extend is a little bit hidden in Studio with the UI changes. Right-click the track you're looking to extend and choose Remix Selection. The Create window will pop up, and under the Cover dropdown you need to select Edit -> Extend.

Worth pointing out that Extend is a bit troublesome at the best of times, and is particularly so in 5.5 at the moment, so might not deliver the consistency you're hoping for.

From the way you've been doing it in pieces, if you want consistency across the entire song then at this stage you're probably best off continuing to create the individual pieces so you have a full drum track, then selecting the entire drum stem and using Cover from within Studio - assuming you're using v5.5, in the Cover popup under the advanced option sliders, try a setting like weirdness 35%, style 50%, audio 50%. That should re-perform the drums close to the original, but with a new consistent sound across all sections.

If that new cover introduces things other than drums, just extract the drum stem from it.

[hoary stream]

i'm lost, what is the correct phrasing when you want to specify stuff in a part, everything in the same bracket ? but separated with what ? coma or - ? like : [verse 1 , mellow guitar accent , harmonica notes] or is it with - like [verse 1 - mellow guitar accent - dirty harmonica]

OR

is it like [verse 1] [guitar accent] [ adirty swamp harmonica notes]

i don't know what works

what is the correct way ?

[polar spade]

AI likes to say using dashes are the way. Users are usually spread out even on their use. You can even try " | " which is common too. Most any way works fine.

[hoary stream]

wow ok it's... i don't anderstadn why there i no correct way... i just can't tame that AI, it's doing whatever it wants

[blazing zodiac]

Y usually do
[Verse 1]
[Instruction: Arrangement you want to add]

[hoary stream]

ok and if there is several instructions you separate them with - ?

[blazing zodiac]

No, in different brackets

[hoary stream]

oh ok each instruction in a bracket

and you step a lign everytime or you write them continuously ?

[blazing zodiac]

Yeah, example:
[Verse 1]
[Instrument: cello solo]
[Vocal: intimate, close mic]

Not sure if is the best way, but atm worked quite fine for me. Probably Suno will not differenciate if you do them continously instead of different lines, as it reads the bracket itself as independent instructions (in theory)

[hoary stream]

merci i'll try

[old lion]

What the hell is song radio? I just used, did my song accidently go public? Its on a list how do I undo it 🙁

I thought it was like remaster for radio or something ugh

oh its to make a song list, what strange place to put that function :/

[polar spade]

Again, hoping @blazing zodiac version works well, it still shouldnt matter much which version you use.

Note that Suno uses all the lyrics , all the metatags, considered your prompts, the length of lyrics, and any edit tool youre applying to create a generative song, that means making its best optimal guess of what everything is telling it to create.

If "nothing is working", it may not be thats tag your trying.

Ill point out, harmonica is not a style, but an instrument, so you might begin with [harmonica: swampy, New Orleans blues, dirty]

But what you have may be fine as well , as I should mention that harmonicas are a harder instrument to prompt for.

Here's a try, and incorporating a bit of @blazing zodiac suggestion as well.

Dont forget to reference styles in the prompts section as well.

https://suno.com/s/xVXiZxyEG6s8NxUL

[polar spade]

Not public.
Not saved in History.
Not saved in Playlist.
It just makes a list of like-style songs of other users outputs to listen to.
Just click back and its gone.

[blazing zodiac]

If you have troubles with harmonica, try to add it "attached" to musical style where commonly uses it or also defining a type of harmonica, not sure if it will work but could help Suno to understand it

Never tried an harmonica but maybe something like:
[Instrument: Celtic harmonica], 'Folk harmonica' could help better. Or just "Diatonic harmonica"

[polar spade]

Thanks. Im good, mate. My output did just fine. You might want to address or ping @hoary stream within the comment.

[blazing zodiac]

Oh, true, mb lol

[quiet solar]

There is no 'correct' way, as Suno works on interpretation rather than reading metatags like a programming languge.

Anything you put in [ ] in the lyrics will be recognised as a style prompt and may be taken into account when making the song. The exact format of how you arrange that within the brackets, or across multiple brackets, makes next to no difference in my experience.

[polar spade]

It is though suggested to NOT use more than three separated prompts within one bracket as a metatag. Though many do just that and dont have odd things happen, it also may simply be that Suno ignores or homogenizes most prompts anywhere that go over about 7 to 12 prompts.

But more than three separated tags, such as [vocal] [soft] [giggle] [yodeling] might start singing the additional tags. In this example it may start singling "yodeling" as a lyric. Just keep a watch if it does.

[blazing zodiac]

Yeah, probably the format using inside brackets could be more like personal organization, more than something really useful, I also have not found differences on the format used inside brackets, just personal preferences. As the "core" itself are the brackets. Maybe have little influence on the way Suno reads the prompt inside (what 'recognize' with characters like : - or | , but the difference should be residual I guess)

[calm tusk]

I wanted to make a song for a 5 member girl group. But everytime I do that I hear only one voice in the full song? Is it even possible to make something with 5 different vocals and if it is how can I do that?

[obsidian ocean]

im not 100% sure to me it would be something you would add a section at a time, idk if this prompt helps for the style welcome to try it.

Cinematic emotional pop-rock track with evolving female vocal ensemble, start with a single soft whispery female voice, gradually introduce multiple distinct female vocalists with clearly different tones: airy breathy soprano, warm emotional mid-range, raspy gritty alto, powerful belting lead, and a haunting ethereal high voice, use call-and-response between voices in verses, alternate lead vocals between sections, layer harmonies to simulate separate singers, build each section by adding a new voice, chorus features all voices stacked in wide stereo, choir-like but still individual, bridge isolates 2–3 contrasting voices, final chorus explodes with all 5 female voices overlapping, dynamic contrast, emotional and powerful, clear vocal separation, rich layered production, cinematic build, strong identity shifts between voices

like i said idk if it will work but my shot at it..

[calm tusk]

Thank you!

[calm tusk]

How would it work if I wanted to cover a song?

[obsidian ocean]

no idea lol, all i can guess it replace the style with it? take the bare for style and bump it to 85% ish then turn up your style infuence a bit.

big maybe there

i forget the name of it the bar to keep it original at the bottom.

mine been popping off some 7 minute ones here and there for some reason. your not alone on it

seems to do it for me when im doing instrumental and alot of bracket stuff so idk.. but i do get it on songs might be a bug brother

[hoary stream]

hom to you extratc all the stem in studio mode ? is it the option "get midi" ?

[obsidian ocean]

wish i knew im on dat budget membership without studio lol. hear its like a daw

well idk i cant complain aside from a few hundred points gone poof on me (from ghetto renders)

the vocals have been much smoother and more expressive cleaner. some are ruff.. and it still goes botty on talking parts.

[hoary stream]

i got it sorry

im stupid and overexited lol

[obsidian ocean]

all good so you can get midi files? can you replace stems to on the studio version?

[lost badge]

@flint haven garbage scam / spam ...

[sly jetty]

yo, does suno v5 feel different now for yall?

now that 5.5 released

[quiet solar]

Not easily, depending on exactly how you saw that working. More than 1 recognisable same gender vocalist, harmonies between vocalists, and having specific vocalists sing specific lines are all difficult exercises in their own right. Combining them gets increasingly challenging unless it's a genre that consistently operates with multiple harmonic singers (e.g. barbershop) and starts to become something you can only do with a blend of prompting and editing.

If you have Suno Studio, you can build up the voices one by one and use that to edit who sings what - although getting them to harmonise can be quite a tricky exercise. If you don't have Suno Studio, you can do much the same exercise in an external DAW using multiple covers to create different voices - just a bit more involved.

If this is all sounding like a complex nightmare, then prompting for a girl band (especially in pop/k-pop/j-pop genres) and for things like 'ensemble, backing vocals and gang vocals' may help as a compromise, provided you're not fussed whether it's exactly 5 singers and that one of them will typically be a lead singer throughout the song.

[calm tusk]

Thank you! Yeah I don't have suno studio. And it does sound like a nightmare lol. So I will probably try that third option! Thank you for helping me though, that's really nice🤗

[quiet solar]

You can - that's one of it's most useful features, that you can (re)generate things at stem level in Studio, without needing to regenerate everything else at the same time as in normal Cover. The second most-useful feature from an editing perspective is that you can quantize to remove small differences in tempo between different covers/songs so that they match perfectly for editing purposes.

[daring minnow]

Does the personalize style button go off of your most common styles you've made in the past? 🤔

[clever lily]

Can I also remix an existing song with Suno?

[hoary stream]

after 50 gen i got the track i wanted, like perfect, and it bugged at the end, missing 5 sec of lyrics... and now the damn thing doesnt want to produce the same thing... is there a solution ? for "ixing" the track ? i try "extend" but it make i dont know what a smaller song that has nothign to do

[limpid epoch]

so... anyone got that 90 second track prompt? really kinda need this length track

[solemn topaz]

Yeah, you can try Remaster first—that sometimes fixes missing or broken endings without changing the core vibe.

If that doesn’t work, go for a Cover instead and reuse the exact same lyrics prompt (including all instructions) + the same style prompt. That usually gets you closest to the original “lucky” generation you lost after 50 tries.

[polar spade]

Just to mention, I left two reasonable responses to this request . . . #📚┃suno-school message ,
#📚┃suno-school message
But if anyone wants to follow through with an alternative, the table is open.

Hope you find which answer you seek. Quick jingles arent very hard to do.

[limpid epoch]

do you have good experiences with extending ?

i appreciate ur help with these posts

[polar spade]

Things change with the models a bit , every now and then, so "good experience" is relative to everything. Its like the weather, just wait a bit and it should change.

But sure , I used extend often, and still do when im prepping a song, such as a duet. Mostly though, ive just used Studio recently for those things that it can do in lieu of Extend.

[limpid epoch]

is extend the same regardless what ver was used to prompt the song?

[polar spade]

Yes. And more, ytou can extend in any version as well. So if you have a 5.0 song, you can enmxtend the ending in 3.0. Or vice versa.

If you like certain version models for what they can do ytou can mix and match. The fidelity wont be good between different models , so we would need s final cover or remaster.

[frozen dagger]

Personalize, as in the blue magic wand/my tastes? Ya that says it is created by what we create, like and dislike, (Mostly from create we imagine) or we can edit that prompt box in My Taste to whatever we want. It only effects the blue magic wand to make the prompt be what we have in the style box mixed with "my taste".

[hybrid river]

question for any premier users, does the studio give you better control over a custom voice? I'm on pro and the custom voice I made sounds great...until about 2 min or so in then is starts to morph into a generic suno vocalist lol

[frozen dagger]

we can't use custom Voice feature directly in Studio atm, so surface layer of the question is no. because Voice specifically still uses the Create tab module and not the Studio module.
Now that doesn't mean that Studio isn't good for tweaking/editing.
Generations seem to do that with not just Voices, but with following most audio it will be close to source then drift. If extend was working properly that could be way to fix, but it gives me a headache rn. Studio becomes the next best place to do that, but we are still using the create tab on the left side of our Studio screen when we need to use a Voice from there. (We can place a voice stem on timeline and it will create something similar usually, but not consistent like Voice feature.) I think it almost always does that drift away from the source audio tho. (for better or worse)

If you are using Cover with Voice to cover a song that already had (different) vocals. That will make it feel worse as far as sticking to the Voice. I usually extract the stems and cover the Instrumental Stem in this case. If I want to cover a song and stick to my Voice, not the vocals in the song. (It blends and layers the 2 vocals, not full replace if covering with vocals in it already)

[meager moon]

Why is my own music getting copyrighted? and Why when i Remix or Reupload a song generated by SUNO gets tagged as copyrighted? but when i check the community, someone literally uploaded the original Michael Jackson's song Billie Jean with the same lyrics with no issues.

[slender fable]

What the hell? That’s ridiculous. I hope they address that issue.

Also I can’t get Suno to make effing calm classical remixes of a certain (originally symphonic metal) song without it adding stupid percussion and chugging sounds back in.

Oh so percussion and chugging sounds are calm, huh?! That’s your definition of calm, gentle, quiet, and “acoustic and classical instrumentation only”?!

Boy, I’d love to see your version of hyperactive. My ears would probably bleed.

Stupid AI.

[meager moon]

Fr, i asked Suno to make a Jazz version of a song i figured out with my Electric guitar... it was indeed, not a Jazz song... it almost became alternative metal at the end. "Lo-fi Jazz Ballad" -> "Alternative Black Death Metal of Darkness and Desperation"

[slender fable]

Oh my God.

That is a massive facepalm moment

[frozen dagger]

it doesnt really take one thing and turn it into another (with one gen) it layers that upon what exists and transforms it toward it.

[slender fable]

Even with the style influence slider set to 100%?

[frozen dagger]

yeah i mean the style influence can be 100% but the source was still something else

[slender fable]

So I have to make a remix of the remix to get it to eventually become what I want?

[frozen dagger]

or strip down the parts out of the source that you dont want present and then remix into something different.

either with (multiple) remixes or edit /extracting stems or sampling audio samples to create a different sample

it's more like layering, less like replacing

[slender fable]

Huh.

Holy crap, I went to create a song using only the simple prompt “sleep music” and it adds freaking beats???

[hoary stream]

i guess it doesnt like to sleep with the same thing as you^^

[slender fable]

I’m about to find an alternative to Suno at this point if I can’t get drumless music.

[solemn topaz]

Hmm 🤔, just to clarify—are you looking to remix an existing symphonic metal track into a calm classical version?

[slender fable]

At least an alternative for drumless music.

[slender fable]

Yes.

[cyan jewel]

https://suno.com/s/7hSzW7tGG0v5lKnx at 0:48 the sound gets dunked under water for a second like there's a low-pass filter applied to the track. this happens to a massive amount of my outputs, regardless of what I enter in the style prompt. What causes this, and can I prevent it? It happened a lot early into v5's launch as well

[slender fable]

This is the song https://suno.com/s/hMgL4WocI0yPGmmh

[frozen dagger]

the ambient percussionless stuff is a bit weird one. i have to describe the prompt in a very weird way. like nautical sumbersion or whatever more poetic descriptive and try to get it to go where I want.

[slender fable]

Have you had any success getting ambient percussionless music?

[solemn topaz]

Alright, try creating a cover of the track using new or existing lyrics, along with converted calm classical-style instructions. Keep the same style prompt in the style field, then generate it. You can use ChatGPT or Google Gemini for this—it should improve the chances of getting a better result.

[frozen dagger]

i had a song do this same thing earlier. lemme see if i can find it

[frozen dagger]

I think I've had more than one gen that did this today. Cus this track does it, but I don't think it's the same crunch I was thinking of. One I remember was worse. Coincidentally right around the 0:48 marker as well.
https://suno.com/s/SCoYqaWFsSyEbGmT

[slender fable]

It’s an instrumental piece (save for some wordless vocalizations). Are you saying for me to put the instrumental cues in the lyrics section?
For example, I tried putting this in the lyrics section:

[Instrumental]

[Intro]
[A cappella classical female voice vocalizing gently on “ooh”]
[vocals only]

[Main Theme]
[Piano and strings enter quietly]
[acoustic and classical instrumentation only]

[Bridge]
[Piano drops out]
[Emotional string quartet section]
[acoustic and classical instrumentation only]

[Outro]
[Piano returns]
[acoustic and classical instrumentation only]
[Other instruments drop out; piano only]
[Final gentle piano chord]

And it still gave it those chugging sounds and percussion.

[slender fable]

Also, I’m all out of credits.

I have to wait until tomorrow evening.

[hoary stream]

seems that sometimes suno loose the isntructions more the song is going

[slender fable]

Yes, indeed.

[frozen dagger]

Ya, song and vocals drift as it goes on.

[slender fable]

It’s like it starts listening to your prompt at first and then gives up halfway through the song. And that’s usually the best case scenario.

[hoary stream]

can i add a paragraph of instruction mid song to make it remenber ?

[solemn topaz]

Sorry for the delay.

Here’s what I tried:

1. Cover → Custom Lyrics:

[Soft, breathy male or female vocals, slow delivery]
[Gentle piano intro, sparse notes, lots of space]
[Verse 1 – intimate, low volume, emotional phrasing]
[Light strings enter, sustained violin & cello]
[Chorus – slightly fuller but still soft, layered vocals optional]
[No heavy drums, only subtle ambience or soft hits if needed]
[Verse 2 – slightly more expressive, maintain calm tone]
[Bridge – minimal instrumentation, piano-focused, emotional peak but controlled]
[Final Chorus – warm, slightly layered, still restrained]
[Outro – piano + fading strings, long reverb tail, peaceful ending]

2. Style:
   Calm Classical, soft piano & strings, slow tempo, emotional, minimal percussion, gentle dynamics, reverb-rich, intimate, cinematic, peaceful atmosphere

3. Settings:
   Weirdness set to 0 and Style Influence set to 100

Here are the results I got:

https://suno.com/s/fBGRTHcOCfeLvYfE

https://suno.com/s/tk45bWs5Ox949Bkg

[slender fable]

No need to apologize. Thank you so much!

[obsidian ocean]

one of these days im gonna have to play with that, there are so many times a stem or instrumental has sent a song to the death list lol. Sry for delay layed down earilier before i seen it. The midi thing is really interesting to, can we move the notes around without regenerating the whole song will it do that to?

[solemn topaz]

You’re very welcome—I’m glad it helped!

Also, the links have been refreshed, so you should be able to listen to them now. Let me know what you think 🙂

[slender fable]

Oooh, I like them!

[quiet solar]

No, it's not possible to edit at note level - on a technical level, that isn't how Suno is generating music. The MIDI export works in a similar way to existing stem splitting - it does a best guess breaking out of the sections retrospectively, then does a best guess transcription of that to MIDI.

[winter wind]

hey sunos, does anyone have/know of a nice gpt or agent/template for generating style prompts?

oh I see threads in #1252267444773326848 nvm

[nocturne tulip]

https://suno.com/s/EbHaEJGg3djukZxn can anybody tell me if this is good or if I need to work on anything

[solemn topaz]

If this matches what you had in mind for the song, then you’re already on the right track. At the end of the day, the only real benchmark is whether it satisfies your vision—if it does, that’s all it needs.

That said, I noticed you mentioned “cartoon ad-libs” in your style prompt, but they didn’t really come through in the final result. Honestly though, the output still sounds quite solid even without them, so it works either way. If those ad-libs were important to you, you might want to push that element a bit harder in the prompt—but if not, the current version already stands well on its own.

[obsidian ocean]

ahh ok i see, but the ability to swap stems completly would be nice, is sound effects a thing as well?

[gilded falcon]

Suno is cheating me out of me stems. Only loading one version of them.

[tight scroll]

Are there any prompts you can add to stop 5.5 randomly deciding to not use the voice?

[covert nimbus]

Is there a way to reliably prompt out the "Mmm" and "yeah" at the beginning of your songs

[manic meteor]

Right? I get this. For instance, how all the RnB songs start out with whining and moaning.

[sleek stump]

You can try [Intro - Instrumental], that usualy works for me.

[unreal sleet]

Im trying to make death metal and melodic death metal, whatever i prompt, i always get a stock AI male singing (clean) in the hook, anyone knows anything about this? 🤔

[dense inlet]

Not my genre of choice, but you can try adding this in your Style of Music: " Intense death metal, thunderous growling vocals, chaotic riffs, relentless double kick drums, violent and apocalyptic mood"

[polar spade]

Suno is an auto-regenerative style of production. So it re-churns what has come before it in order to prioritize keeping the same vibes consistent, first, and following subsequent prompts within the song, second. So, theres a result of things getting homogenized over time. Models are tweaked to encourage against this. And we try to prompt or produce around it.

[unreal sleet]

no difference

I just wanted this friday for making some new music and ofc I have to battle next problem after the other

[sleek stump]

in the hook tag specifiy the vocal style, something like [chorus - Screamo] You can use scream, growled, ect.

[unreal sleet]

im well aware of how to prompt

[sleek stump]

Thats fine, only trying to help, and the example might help others..

[dense inlet]

Check out some Death Metal sounds on this track. https://suno.com/s/kzt6ZO7xu4yyOaew Here is the Style of Music used " Blackened death metal, black metal, power metal, intricate, bombastic, cello, strings, Jews harp, melodic, harmony, clean vocals mixed with harsh growls"

[quick wedge]

If you haven’t yet you can try Spock’s workflow prompts for vocals if that’s something you wanna try out later https://discord.com/channels/1069381916492562582/1478152870791483608

[unreal sleet]

I have used that one and another one for gemini already -_-

[unreal sleet]

copied styles, still happening

[quick wedge]

Maybe Cheddar might know with his suggestions if he’s around later. Thought that might help since I don’t pick through specific prompts 🤷‍♀️

[unreal sleet]

exact same ai voice that sounds like a stock voice

can anyone try to do some "melodic death metal with deep guttural growl" or something and see if the problem lies on my end?

[sleek stump]

Can you post a link to one of the gens you did where this is a problem?

[unreal sleet]

https://suno.com/s/M87sGcQok83kUGdS

go to 1:40

i think i might have found it

nvm

didnt

[sleek stump]

Ya, thats a bit out of place, That switched over to a more 90s heavy metal style vocal...

[unreal sleet]

yeah

its been in every different prompt the past 60 songs, every. single. one

[sleek stump]

Roll the next try in v5....

[azure crane]

Trying to make some ambient music without drums, but v5.5 adds drums to every track. I know you can separate stems, but there must be something that makes v5.5 not add drums.

[unreal sleet]

I tried in v5 it didnt even scream anywhere in the tracks

[sleek stump]

I have a feeling its a 5.5 issue

really? hmmmm

[unreal sleet]

gonna try to put it in caps lock

nope

im getting really sick of that voice haha

[sleek stump]

Ok, I "think" I see what could be causing it...

[unreal sleet]

give it a go

[sleek stump]

In your styles prompt you have "clean vocals mixed with harsh growls" Take out clean vocals.

[unreal sleet]

I have done that as well

the negative prompting idea

[sleek stump]

I have found the the exclude box gets mostly ignored

[unreal sleet]

yes i tried with empty exclude box as well

i kinda saw what was deleted

I have put genre tempo, vocal style etc above

[Vocal Style: Single Vocalist: Deep Guttural Lead: High-pitched Fry Bursts: No Clean Singing]

[Intro: 200 BPM: C Standard: Twin Guitar Harmony]
(Harmonized melodic lead)
(Tremolo picked minor scale)
[snare roll]
[full blast beat entry]

etc

tried removing (Harmonized melodic lead)
cause it could be read as a lead singer rather than guitar

[dense inlet]

DM me...my message gets deleted each time I post it !!

[unreal sleet]

need new ideas 😔

[sleek stump]

Ya thats baffling... I have tried quite a few different ways to prompt this and it deos teh same thing in the same place every time.. I am going to call this a thing with 5.5...

[torpid hollow]

I must be using style and instrument exclusion or rule definitions incorrectly... I can't say why, but the 5.5 model is stubborn... a bit arrogant, like it's saying, “I'm the one who decides.” It arbitrarily adds instruments that have nothing to do with the style... like djembé and conga percussion, etc... The kind of instruments that scream “retro”—it ruins the song… and it’s impossible to remove them… What’s the solution? The “Update Instrument” button or “Reset” with SUNO 5.5 doesn’t work at all…

Thanks in advance for your advi

[unreal sleet]

guess im done then

[sleek stump]

One thing I didnt try is to drop teh "pre chorus tag"

[unreal sleet]

yea thought about just replacing it with bridge or something

[sleek stump]

Dropped teh pre chorus tag, doing one more gen before I head off to work..

[unreal sleet]

nope

dont bother

doesnt change

[sleek stump]

Well, I think that actually helped...

[unreal sleet]

did you leave it blank?

[sleek stump]

I just took the entire prechorus set out

Ya, no more 90 metal vocal

[unreal sleet]

bridge didnt work but lets try without anything

lol i still get it

gonna remove some exclude stuff

[sleek stump]

https://suno.com/s/pYmUNF53GFPutk4J

Blackened death metal fused with black metal and power metal: intricate bombastic arrangements, harmonized melodic leads, tremolo-picked minor-key guitars, and a distant-rising intro that erupts into snare rolls and full blast beats. Crushing low-end, SSL 4000G punch, wide stereo imaging, and angry guttural male vocals throughout.

and removed pre-chorus tags

Let me know when you review this and i'll delete my test @unreal sleet

[unreal sleet]

I copied yours

I still get it

im gonna try on tor browser

[solemn topaz]

In v5.5, for my latest track, I experimented with quite detailed directions.

For the lyrics, I structured it like this:

[Intro – 10 seconds minimal, near silence, faint wind, distant metallic ambience]
[Sound – very thin, high-pitched, detuned whistle, unstable pitch, slow, broken, long decay, no rhythm]
[Background – extremely soft, dry, close]
tch… [pause]… tch…
tch-tch… [long pause]… tch…
[Whisper – extremely soft, breathy, close-mic]
[Echo] అరిస్తే… ఇంకా బాగుంటుంది… AND SO ON...

And for the style, I used:
Dark cinematic funk, psychological horror, Telugu male baritone, breathy whisper intro, spoken delivery, detuned whistle lead, shutter vocal percussion, glitch vocal stutters, deep sub-bass hits, delayed percussion entry, 90 BPM halftime, heavy dappu, distorted brass drops, analog reverb, intimate mono intro, wide stereo drops, strong silence contrast, breathy laughter, selective echo, low-end impact.

Although I didn’t specify exactly where elements like laughter should appear, after generating multiple versions, one of them turned out quite satisfying.

From my experience, Suno doesn’t always follow precise timing or placement instructions for instruments and SFX. Sometimes it gives unexpected results—both good and bad. Running a few more generations or making manual edits can help until the tool improves further.

Here’s the track if you’d like to hear how it turned out— https://suno.com/s/7YtIoSR0ffpjFoPO

(The idea was to create a South Indian BGM-style track suited for a psychopathic character.)

[unreal sleet]

used yours, on a different browser and still happens

man i wish there were people from suno actually engaging with its users

downloaded a fresh browser. is my account cursed?

cant even get growls on v5

[torpid hollow]

Reading your comments, I'm starting to understand.

Suno 5.5 is actually incredibly prescriptive: if there are gaps, it will fill them in (which is a key semantic point in the file—the fact that it had to make a decision: calculation). So when editing a piece of music, commands like “remove,” “subtract,” or “replace” are completely useless... SUNO 5.5 seems to have evolved from version 5 (“quasi-fully generative → diffusion”); 5.5 has a procedural aspect... To correct a song with 5.5, you have to speak its language: target the track and inject instructions into the timeline that will take precedence over the current instruments. Specifically, if I want to replace something, I must target the drum/percussion track or another (for this, see the precise stem names), then at the same spot in the lyrics inject tags like “heavy percussion, tambora, etc.” with high precision, which will significantly exceed the “Conga” value (and SUNO 5.5 will replace it based on weight and value comparison).

Interesting—this replaces the concept of subtraction... I guess I was completely wrong ^^ I’m going to test it .

[neat nexus]

I wanted to follow up on this post I made in the #🔊┃sound-showcase channel. Here's a quick and dirty example of what can be done from using Sounds as mashup sources.

https://suno.com/s/rNvIc8G1s4LRcIcM

If needed I can post the intermediate stages. But basically I went from Mashup with two acoustic guitar samples to covering with a Voice to get what you have there. Really simple. Like the song itself which is just for demonstration purposes.

[solemn topaz]

I noticed remix permission was turned off for your track, so I wasn’t able to create a cover. I did try working with your lyrics and style by making a few adjustments and testing the output.

From what I observed, fully continuous growling throughout the song doesn’t translate very well—it tends to lose impact without some variation or melodic contrast, and can feel a bit flat over time. Introducing some dynamics usually helps the overall flow and keeps it engaging.

That said, if your goal is to keep the growling consistent, you might get better results by editing specific sections directly in the editor and applying or replacing parts with growls more selectively.

Here’s the result I got using the modified version— https://suno.com/s/50tEiEIddt77hf8A

[unreal sleet]

"fully continuous growling throughout the song" was something we only used when we tried to fix the problem. even using [deep guttural growl] in front of every section does nothing

yeah I can hear the same ai voice still layering with the deep growl in the chorus

your result was pretty much every song ive made for the past 3 hours that was bad

did you not listen to the chorus? 🤔

to clarify since its kind of a wall of text up there, its the chorus that is the problem 🙂

[solemn topaz]

Yeah, I did listen to the chorus—that’s actually why I mentioned it earlier. What you’re hearing there is kind of a limitation of Suno right now. Even when you force tags like [deep guttural growl], it still tends to layer its default vocal tone underneath, especially in choruses.

It’s not really following instructions the way a human vocalist would—it tries to “balance” things, which ends up reducing the intensity and emotion you’re aiming for. That’s why it keeps sounding similar across attempts.

At this point, you’ve got a few options: either adjust the style a bit to work with what it can handle, manually tweak sections in the editor, or just wait for future updates to improve vocal control.

Also, from what I’ve seen, this issue shows up quite often with metal/rock styles in general—so it’s not just your setup.

[solemn topaz]

That’s actually a really interesting way to look at it—thanks for sharing your perspective.

I think I get what you mean: instead of trying to “remove” something directly, it’s more about overriding it by introducing stronger or more specific elements in the same space. That does seem closer to how Suno behaves in practice.

At the same time, I’m not entirely sure how consistently it works in real use. From what I’ve seen, even with very detailed prompts, Suno can still be a bit unpredictable with timing, layering, and placement. So while targeting specific elements and reinforcing them might help, it doesn’t always guarantee precise control.

Still, your idea about “speaking its language” and guiding it through stronger weighted instructions is definitely worth experimenting with.

I’ll try that approach in a few generations and see how it responds.

Curious to hear how your tests turn out too!

[unreal sleet]

It works if u prompt with a female in the chorus lol

[neat nexus]

Another example where Sounds can help?

Sounds:
https://suno.com/s/L4j9ZvFvOtXFXt5j
https://suno.com/s/7MXL4KcTptiXGOpZ

Mashup:
https://suno.com/s/NpUMMfE4kIB4lDSd
https://suno.com/s/bQskRbqcaBrXVXrk

Again I didn't really work on it, just a demo.

[sleek stump]

Did my test sound more like what you are after?

[unreal sleet]

I don't remember I'm on my phone atm
Was it a pure growl in the chorus or could you hear it layered on top?

[sleek stump]

All of the lead was growled, but the backup was more 90s metal, it was still the proper lead though.

[unreal sleet]

Ok so not as desired

[quiet solar]

No different to the current models - however, you can import sounds created in the Sounds part of Suno into Studio to work with and manually put in place.

[unreal sleet]

Sad when I finally decided to make my favorite genre and it doesn't work

[onyx roost]

Hey Kanthric, do you have a work email or any way to contact? I have an inquiry related to Suno

[quiet solar]

You're welcome to DM, or post any queries related to Suno here.

[torpid hollow]

Could someone give me some feedback privately? Since I'm French, it seems like my use of English figures of speech might be off or not understood—I'd like to get some feedback ?

[azure crane]

Any tips on how to avoid drum beats with an ambient track? Tried everything but nothing works with v5.5.

[quick wedge]

You can ask in the global French channel https://discord.com/channels/1069381916492562582/1154779533627621518

[torpid hollow]

The idea is to check my English phrasing (idiomatic awkwardness—in English, a French mental image might not convey anything)

[unreal sleet]

can you stitch up a bunch of vocal samples in studio and use it as a persona or model and have any decent result with it?

or am i wasting time

my problem is the female vocalist is only doing what she is supposed to at the first part of the songs so figured i took like 5-6 of that good part

and put them into 1 in studio

[quick wedge]

I’m not sure then I thought it was something else

[torpid hollow]

to chat with a native English speaker to check my text

[quick wedge]

There’s plenty in general chat you can ask there as well

[lapis loom]

Hey help me out guys.

Can I upload an AI-generated voice to try to make that voice into a singer?

[unreal sleet]

can you stitch up a bunch of vocal samples in studio and use it as a persona or model and have any decent result with it? female vocalist is only doing what she is supposed to at the first part of the songs so figured i took like 5-6 of that good part
and put them into 1 in studio and made it a model or persona?

[quiet solar]

In principle there's every reason to believe that will work, but I'm not sure if it would be a solution to the 'vocal drift' issue that some people have mentioned where the song deviates from the voice of the original singer the longer and longer it gets, especially when a Voice is being applied.

I suspect that's an underlying challenge with Suno trying to blend influence from the audio prompt (e.g. Inspo, Cover) with consistency with what it's generating. Just setting expectations, since it sounded like that was the problem you were trying to solve with this.

[quiet solar]

Not easily with a voice that was created outside of Suno. When you upload a voice to create a unique voice persona, you'll be asked to speak a phrase to verify that it's actually your voice that's being used (e.g. to prevent you uploading Ed Sheeran's voice and going "Yeah, of course I'm Ed Sheeran, honest. Now let me fake a load of songs and pretend he sang them...).

[quiet solar]

It's not an exact science - drums are very common across many genres, so even with good prompting Suno will still sometimes reach for a percussion track. However, some general advice as a starting point would be:

• include genres that are not drum heavy (e.g. ambient, ethereal, acoustic, underscore)
• specify a couple of non-percussion instruments to get Suno focused on those (e.g. strings, woodwinds, synth)
• include vibe prompts that steer towards softer, slower tempo music (e.g. slow tempo, peaceful, sleep)
• add "drums, percussion" to the exclude styles field in More Options (doesn't have a big effect, but every little helps)

[unreal sleet]

yeah it couldnt fully cover the vocals

[quiet solar]

@unreal sleet On the occasions it's cropped up, I've found a workaround to be splitting the original song into two (e.g. using Crop) and covering both separately using the same style prompt and Voice with slightly increased audio influence (e.g. c.50% on v5.5, c.70% on previous models).

Provided the vocalist is reliably similar in both, you can then stitch the two halves back together in Studio or an external DAW and reupload it, then use Remaster or a high audio influence Cover to smooth out differences in instrumentation. Typically the voice is pretty stable in high audio influence covers, and not needing to add the Voice as an audio prompt on that final pass should help avoid vocal drift, since it's giving Suno fewer things to have to keep track of.

[unreal sleet]

The problem is that im doing metal now which suno struggles with the most so even getting growls correctly is something ive been doing for like 14 hours straight today and i always return to square 1

im trying different knobs atm

i havent prompted for clean singing, there's no clean singing in the voice i created for it, still clean singing..

when asking chatgpt about female but for deathmetal, they say suno currently have a problem getting it right

[frozen dagger]

i like some of these 5.5 female metal voices myself. Not heard anything really like my fave Otep or anything but for ai it pleasantly surprises me. heard a french female one other day was really good.

[azure crane]

I've done that and nothing helps. So frustrating. I mean a completely ambient track has to be the easiest thing you can create for a music AI.

[acoustic egret]

is there a way to know what processing suno used to make the stems sound like the master bus?

im really digging v5.5 mixes actually

[torpid hollow]

difficile

[blazing zodiac]

Stems .. that useless thing that makes your song sound worst than the original master

[polar spade]

Didnt we have a chat on this last month. No, that was on vocals. I had a "drumless" chat recently with someone. Maybe I can find it.

I rememeber now. Someone was working on a meditative kind of dark industrial focus tracks suitable for coders.

[polar spade]

Have you tried older models? 3.5 gave a couple. Got a few others starting well in 4.5 but the dreaded 2 minute mark its starts percussion.
https://suno.com/s/urSFOgykCWwgu5b0
https://suno.com/s/qAFfccUW7LKiDVxl

Honestly, most i were getting had a nice meditative Kitaro (Silk Road era) feel to them, and just needs the drums stemmed out of them and they'd be fine.

[solemn topaz]

Yeah, difficult 😄 but interesting approach.

[quiet solar]

They can be problematic, but it depends on how you use them. If the original song has quite heavy production-style effects like reverb, those sometimes don't come across cleanly when the song is split into multiple stems (it's less of an issue with a simple vocal/instrument split), so you'll need to either reintroduce them in your DAW, or reapply them in Suno by uploading the song again after you've done whatever it is you were doing with the stems, and then covering or remastering the uploaded version.

Similarly, you can also deliberately work at stem level in Studio, which helps a lot with issues caused by retrospectively splitting the song into stems afterwards.

It's not a one-quick solution for perfectly splitting stems, but then there isn't any solution that can do that perfectly cleanly from a song after it's all been mixed together. Your mileage may vary, but IMO it's better to have the tool than not.

[quiet solar]

Someone else was asking this earlier today - this is an example I worked up with them based on those exact same principles. Completely drum free.

Style
Ambient, ethereal, cinematic underscore, subtle melody, weightless, fluid, soft legato strings, breathy woodwinds, warm analog synth pads, shimmering textures, minimalist, reverb-drenched, slow tempo, peaceful, floating, outer space, underwater,

Additional notes: "drums, percussion" added to exclude styles, style influence set to 60%.

https://suno.com/s/OnLwMo3mBQ6VV8E7

[acoustic egret]

yeah but i wanna rerecord the instruments dude

[stoic grotto]

How did I do?

[paper dragon]

Hi guys, novice pro suno user here, Im trying to gen some songs but not very happy with the results, I wonder if there is a point to go a version or two back? or is using the v5.5 is the best decision?

[azure crane]

Yes, the problem is the v5.5 version. It works fine with the 5 and older models. The problem with v5.5 is that the production sound is so much better so I really want it to work…

[quiet solar]

It depends what you're not happy with. Every model is different and they have their own unique quirks, as well as some things being fairly consistent across all models.

What's the problem you're finding with the songs you've been trying to make?

[paper dragon]

i feel like they have too many random hype woos in between lines, sometimes off beat, beat is losing itself and ramps up the bpm then goes back down to normal again after like 4-5 quick beats and repeats this pattern in later verses - struggles to keep consistency, lyrics fall off beat, lines dont meet beat as a bit too early, rarely too late, maybe someother things i cant think of right at the moment

[quiet solar]

So the random vocalisations are more common to the v5 models - v4 models don't have that so much, but do tend to like echoing parts of the song with the backing singer.

Minor tempo variations can happen from time to time. They're not usually noticeable, and those which are can be easily fixed if you have Studio or a DAW to export to, but changing to other models probably won't help with that.

Lyrics falling off-beat is likely just how Suno is handling the vocal arrangement - it's not unusual for real world singers to be slightly off-beat if the line isn't rhythmical, or do things like deliberately starting the line on beat 4 of the previous bar, and Suno has that in it's training material. There isn't a way to directly steer that by prompting that I know of, but you can sing into Suno and have it use that as a vocal guide for covering if there's a particular way you want the vocalist to perform.

If you want more precise control over the output, you're going to get nudged towards editing or giving audio guidance rather than tweaking prompts sooner rather than later.

[leaden plank]

Hi everyone. I am trying to write a song that contains dialogue. I tried everything to trigger the right voice in the correct line but keeps messing up. Does anybody knows a solution for that?

[solemn topaz]

Hey, could you share the specific lines and your style prompt? Might be easier to spot what’s going wrong there.

[leaden plank]

Sure

Give me one sec

[polar spade]

I have just gone with the focus of "making it work" , without Studio, but with 5.5, yes, it does like to rhythm things back up again.

Of course, rather than spend too much more time on running prompts over and over, or extending in short 1 to 2 minute bursts, which is a way to keep it drumless ( which would have been my first choice) i did the alternate, . . .

• I Covered my one-run 4-minute 3.5 song into a 5.5 Cover and got 5 minutes.
This 5.5 added in some rhythm, so then.....
• Did a multi-stem extraction with Get Stems..
[STUDIO OPTION]
• Put in Studio. Silenced the persussions.
• And saved the file

Then remastered in 5.5 High. Although the remaster in 4.5 also sound just as good, I picked 5.5 High as it warmed down a guitar part better.

https://suno.com/s/4C1Ali3z6weCIYgM

[PRO Version ]
So, using only in Pro..... we can not SAVE a completed file (without drums) back to the Library as we can with Studio.

[Old School]
But we can do old-school and just run the stem module with the drums silenced and all the other stems playing at once, while recording the stream live to our DAW, computer or an audio recorder program, or patch out to whatever works.

[DAW version]
Download the WAV files from the Get Stems module. Put them all in a DAW, sync them up, enhance, and export/save ( without the rhtym drum track) to a WAV .

[leaden plank]

[Intro: Slow, Eerie Church Organ, Layered reversed choir]

[Verse 1: Low Baritone, Cold, Slow]
I have torn meaning from every scripture of stone and star.
I have bled through every discipline that promised revelation,
Yet still I remain untouched by what I seek.
Is this all knowledge becomes?
A crown of ash upon a mind still starving?
The ink is dry, the candles flicker low,
And everything I learned is everything I don't know.

[Bridge: Percussion Build, Rising String Tension] (Baritone) I call to the air! I call to the deep! (Baritone) Break the seals of the secrets you keep!

[Chorus: 180 BPM, High-Speed Blast Beats, Full Symphony]

(Latin Choir) Vox in silentio! Lux sine nomine!
(Baritone) Who steps through the geometry of my silence?
(Latin Choir) The answer! The answer inside the question!
(Baritone) I see a shadow that casts its own light!
(Latin Choir) Would you walk beyond the borders of sight?
[Dialogue: Operatic Duet, Tempo Drop to 110 BPM, Heavy Bass] (Baritone) Name your price! I have no years to waste on shadows! (Latin Choir) There is no price. There is only recognition.
(Baritone) Recognition of what? Speak, or return to the dark!
(Latin Choir) Recognition that you were never the master of the spark.
(Baritone) If beyond lies the truth... then I give what I am for what I may become!
(Latin Choir) It is done. It was always done.

[Guitar Solo: Fast, Neoclassical Harmonic Minor, Dual-Harmonies]

[Verse 2: Mid-Tempo, Heavy, Rhythmic]
The world bends... it obeys... it listens to my breath.
I have conquered the limits of life and of death.
The stars are my map, the earth is my floor, I am the scholar who opened the final door!

[Outro: Sudden Drop, Fading Piano, Dissonant Echoes]
(Latin Choir) No... you are the fracture.
(Baritone) What have I... what have I become?

[solemn topaz]

Honestly, this is really well written — the issue isn’t your lines, it’s how Suno handles voice switching. It doesn’t reliably separate dialogue just from tags. You might get better results if you split the dialogue into separate sections or simplify the prompt structure a bit. Also, could you share your full style prompt?

That plays a huge role here.

[leaden plank]

Yes one moment pls

Symphonic Black Metal, Progressive, High Contrast, 110bpm to 180bpm, Clean Male Baritone vs Deep Death Growl, Orchestral.

[solemn topaz]

Got it — thanks for sharing. I think the issue is your style prompt is giving Suno conflicting instructions (baritone vs growl + wide BPM range), so it can’t consistently assign voices—especially in dialogue.

You might get better results by simplifying it to one dominant vocal style and locking the BPM for the dialogue section.

[leaden plank]

I will try that. Thanks a lot. I appreciate the help

[solemn topaz]

Hmm, you could give something like this a try—just to simplify things a bit and help with clearer voice separation in Suno AI.

Style prompt:
Symphonic black metal, dark cinematic, 140 BPM, orchestral strings & choir, deep male voice, controlled harsh transitions, spoken dialogue sections, minimal vocal layering, high contrast dynamics, eerie atmosphere, clear vocal separation, no overlapping voices

Lyrics:
[Style – Dark cinematic tone, deep male voice focus, spoken + sung contrast, minimal layering, eerie atmosphere]
[Intro – Slow, eerie church organ, reversed choir, distant ambience, mono feel]
[Low whisper, breathy, close mic]
I have torn meaning from every scripture of stone and star...
[Verse 1 – Deep male voice, clean, slow, cold, minimal instruments]
I have bled through every discipline that promised revelation,
Yet still I remain untouched by what I seek.
Is this all knowledge becomes?
A crown of ash upon a mind still starving?
The ink is dry, the candles flicker low,
And everything I learned is everything I don't know.
[Bridge – Rising tension, percussion build, low strings, voice intensifies]
[Half-spoken, half-sung, controlled grit]
I call to the air!
I call to the deep!
Break the seals of the secrets you keep!
[Chorus – Full band, blast beats, orchestral lift, harsh vocal layer added subtly]
[Vocal: deep male, controlled harsh edge, not full growl]
Who steps through the geometry of my silence?
I see a shadow that casts its own light!
Would you walk beyond the borders of sight?
[Chant Layer – Background choir, distant, not overpowering]
Vox in silentio… Lux sine nomine…
[Dialogue Section – Tempo drop feel (still 140 BPM), minimal instruments, heavy bass pulses]
[Voice 1 – Deep male, spoken, dry, close mic]
Name your price. I have no years to waste on shadows.

[Voice 2 – Whispered, airy, wide reverb, ghost-like tone]
There is no price. There is only recognition.
[Voice 1 – Spoken, slightly strained]
Recognition of what? Speak… or return to the dark.
[Voice 2 – Soft, echoing, calm but unsettling]
Recognition… that you were never the master of the spark.
[Voice 1 – Emotional, rising intensity, still spoken]
If beyond lies the truth…
Then I give what I am… for what I may become!
[Voice 2 – Fading, layered whisper]
It is done… it was always done…
[Instrumental Break – Neoclassical guitar solo, harmonic minor, orchestral backing]
[Verse 2 – Mid-tempo feel, heavy rhythm, confident tone]
The world bends… it obeys… it listens to my breath.
I have conquered the limits of life and of death.
The stars are my map, the earth is my floor,
I am the scholar who opened the final door!
[Outro – Sudden drop, piano, dissonant echoes, fading ambience]
[Voice 2 – Whisper, echo feel]
No… you are the fracture.
[Voice 1 – Broken, fading, soft]
What have I… what have I become…

[leaden plank]

It might work. I will add you to special thanks section when i go live. Thanks alot for you time. I hope it works. I try to bypass this wall about a week now and nothing that i tried worked

[solemn topaz]

No worries at all, happy to help 🙂

I totally get that—this kind of thing can be really frustrating to get right in Suno AI. Hopefully this gets you closer, or at least gives you a different angle to work with.

And hey, no need for special thanks—just glad if it works out for you. Let me know how it turns out!

[leaden plank]

@solemn topaz unfortunately did not work. Tried to tweak still nothing. I will keep trying if i find a breakthhrough i will share it in case somebody else finds it usefull

[solemn topaz]

Just out of curiosity— which model did you try this with in Suno AI?

Might help narrow down what’s going on.

[leaden plank]

5.5

I enabled a persona to sing lines still didnt work. I will try with the persona name

[quiet solar]

Yeah, you're up against a fundamental challenge that metatags in the lyrics are treated as suggestions rather than instructions, so it can be difficult to get that type of precision through prompting alone.

Inf3rno's given some good advice. There are also things you can do with the style prompt to make it more likely to flow the way you want, and you should definitely look at using 'call and response' as a tag in the style prompt and lyrics given the structure of your song. But it is fundamentally going to be a roll of the dice. If you need to specify who sings what to this level, you may need to do some editing work.

If it helps, I put together this workflow based on v5, but it's still true for v5.5: https://discord.com/channels/1069381916492562582/1466598465969590465

[solemn topaz]

I tried this on v5.5 as well—set Weirdness to 0 and Style Influence to 100—and this is what I got:

https://suno.com/s/AoArmDeDMLEGGbt1

https://suno.com/s/ohiCvqk77gLcrIr9

Still not perfect, but sharing in case it helps or gives a slightly different result.

[leaden plank]

Thanks a lot i will read it carefully and i will come back if i get somewhere.

[solemn topaz]

Yeah, that actually explains a lot—especially the ‘suggestions vs instructions’ part. Makes sense now why it keeps blending instead of separating voices properly.

Call and response is a really good idea too, I might try restructuring the dialogue that way instead of forcing roles.

And yeah… starting to feel like this might need a bit of editing outside Suno AI instead of trying to get everything perfect in one go. I’ll check out that workflow as well—appreciate you sharing it!

[leaden plank]

Ok i tried to force it a bit. I used the name of my persona and the baritone male prompt. It worked. But after a while about have the road it breaks. At least is promising

[leaden plank]

Omg i got it. A bit of work around but got it.

@solemn topaz @quiet solar [Verse 1: Male Baritone, Spoken, Cold]
I have torn meaning from every scripture of stone and star.
I have bled through every discipline that promised revelation,
Yet still I remain untouched by what I seek.
A library of ash... a starving mind... seeking the light, but left behind.

[Bridge: Accelerando, Heavy Percussion Build]
[Male Baritone, Intense]
I call to the air! I call to the deep!
Break the seals of the secrets you keep!

[Style: Aggressive Black Metal, 180 BPM, Blast Beats]
[Male Baritone, Harsh]
WHO STEPS THROUGH THE GEOMETRY OF MY SILENCE?

[Dark Gregorian Choir, Deep Chant]
THE ANSWER! THE ANSWER INSIDE THE QUESTION!

[Male Baritone]
I SEE A SHADOW THAT CASTS ITS OWN LIGHT!

[Dark Gregorian Choir]
WOULD YOU WALK BEYOND THE BORDERS OF SIGHT?

I used a choir for second voice

[solemn topaz]

Nice, that looks solid—if it’s working for you, that’s what matters 👍

[leaden plank]

Not exactly what i had in mind. I compromised for now. I will keep trying to find a way at least i have a working route

[unreal sleet]

https://suno.com/s/yUvqt79NIwKZrsfj
Does anyone have an idea what triggers the stutters in the lyrics and sometimes in the instruments? Its kinda like a glitchy effect which feels intentional to the song, not an error. just wonder what could cause it. Could it just be the genre alone? around 1:30 to 1:50 its doing it a lot

[solemn topaz]

Possible reasons for the stutter/glitch effect:

Genre influence: Deathcore / nu-metal often use stuttered or choppy vocals for aggression

Fast section cue: “180 BPM + blast beats + minor 2nds” pushes the AI toward chaotic delivery

Fry screams: These emphasize consonants (D, T, K), causing “D-D-D” type repetition

Repeated lyrics: Writing the same line twice can trigger rhythmic splitting instead of clean repetition

Tempo contrast: Switching between fast (180 BPM feel) and groove (120 BPM) can cause timing “fills” like stutters

Conclusion:
It’s not a bug—it’s the AI intentionally adding a glitch/stutter effect to match the aggressive style.

If you want to reduce it:

Add: [smooth / continuous vocal delivery]

Avoid heavy repetition of the same line

Tone down “blast beats / chaotic” descriptors

[unreal sleet]

so its a product of what i prompted that i wanted, but got something unwanted

repeating lyrics is big in deathcore, its not doing it when fry screaming, changing bpm in deathcore is key,

im guessing this is a genre that suno have had some of the least training with too

blast beats is one of the defining beat usages in deathcore too

so everything suggested works against the genre so its prob just that

[solemn topaz]

Yeah, you’re right about the genre side—repetition, blast beats, and tempo shifts are all core to deathcore, no doubt.

I think the difference is that while those elements are intentional in real music, Suno sometimes doesn’t execute them as cleanly. Instead of tight, controlled aggression, it can overdo it and turn that energy into stutters or glitchy phrasing.

So it’s probably not the genre itself causing it, but more how the model is interpreting and rendering those genre traits.

[unreal sleet]

indeed, gonna play around more with different brackets and see what i come up with 🙂

but unsure if i should use 5.0 optimized tags if that works better than 5.0, they do differ a bit

[solemn topaz]

Yeah, sounds like a good direction 🙂

Sometimes just experimenting a bit leads to the best results.

[unreal sleet]

gemini offered me some stutter transition tips lmao

for 5.5 😄

im putting metatags in notepads so i have a collection of stuff, gonna do a "DO NOT USE" section for stuff like how to make stutters haha

[solemn topaz]

Lol nice 😄

I usually cross-check between Google Gemini, ChatGPT, and Grok—compare answers, pass results between them, then mix and filter what actually works best.

[unreal sleet]

yup thats what im doing! 😄

[shy badge]

https://gemini.google.com/gem/1FTqNPg6SbYvJCXaHWsCSWmA6OZeIXoWW?usp=sharing

[paper dragon]

Hi, sorry but what did you mean by giving audio guidance? im kinda new to the system and not familiar with the professional terms

[quiet solar]

No worries. Under the Create screen, you can upload audio into your library.

So for example, you can record yourself singing the song in the way you see it being performed, upload that into Suno and then Cover it to have Suno use your voice as the starting reference point for how the song should sound and build instruments around it.

[paper dragon]

sounds cool honestly, have u tried this function before?

[quiet solar]

Yes - it's a very useful tool in the toolbox.

I've uploaded songs I've performed on an acoustic guitar and singing and had Suno cover it, and got some really good results. You can also do things like uploading an input for a drum track based on very basic beatboxing, basslines by humming, vocals and harmonies based on singing, etc.

I've added a new bridge by recording what I wanted the singer to do, pasting that into the middle of the song with a DAW, then getting Suno to cover the uploaded version. The singer just took that guide and ran with it quite happily, filling in the instruments in the background.

[pliant lance]

Hey everyone, I know this has been asked a lot, but when I use cover and change just one word, the vocals lose some quality. Any fix?

[quiet solar]

There isn't a magic button fix, but some things you can try with prompts:

• Move 'production/mix/mastering' steers to the start of the prompt (e.g. Close-miked, modern production, soft reverb)
• Be wary of the temptation to push the sliders too far from the default. In v5.5, a balance like 30% weirdness, 50% style 50%-60% audio is usually good for a close replication. In v5 and before, 20% weirdness, 50% style and 70% audio can be a good starting point.
• Try multiple times, since the perceived quality will vary from one generation to another

[twilit basin]

I am experienced in songwriting for a couple of years, but I have the feeling, that AI isnt that good in generating lycris yet..i need to rely on my own abilities here haha

[polar spade]

Lyrics and music are two different things.

Chatgpt, Gemini, Grok, Claude all are just as familiar with lyrics of new released songs as perhaps Suno is, but its seems about 1 in 200 of getting a full and reasonably good song to start with.

From my perspective, the biggest fall back for AI is to use too many clichés, too many go-to metaphors and focus on rhyming than focus on having cutural relevance.

When they can write their own jokes and the jokes are funny, is probably when the songs will be good too.

[quiet solar]

@polar spade AI lyricists also have a surprising affinity for static, neon, digital dreams, igniting sparks and breaking chains. 😉

[polar spade]

. . . And city lights.

[fluid robin]

Echo Beating heart Concrete floors + adding in instruments as words in the song though if you give the ai the correct promt it tends to stay away from that word library that it like to use

[polar spade]

Drinking coffee from blue chipped cups and her sweater that smelled like perfume and despair.... please AI, just stop already.

[pliant lance]

I tried, but it’s still slightly different and it still bothers me.
I even tried the new voice creation feature.
I give up 😄
Thank you very much for your attention.

[frozen dagger]

If the vocals were satisfactory before making the word change, could try editing the word fix vocal piece into the song with a DAW if up to that. Replacing the smaller section of vocal with the edit would maintain everything and only change the one part.
If that is not an option, and the prompt methods mentioned to try while Covering aren't working, then I would maybe trying to just extract the stems (vocals & instrumental) on the song with the best music I liked to Cover only the instrumental. (This would be a quality of sound fix to try and replace the vocals, not an adherence to structure fix for vocals.)

*It's like trying to recreate the vocals for the music now that we have the lyrics correct, but we want the vocals removed because it effects the output. (particularly if using a Voice)

[pliant lance]

Thanks, but I’m not really a studio or DAW guy.
I should learn it sometime.
Big thanks🙏

[frozen dagger]

The extract stems and cover the instrumental stem could be helpful for you, maybe not. No DAW needed there. Sometimes pulling the vocal completely out and covering just the instrumental until we get a vocal we like is the ticket. (Also can help with Covering using Voice so that it doesn't reuse the old voice; either switching vocals or layering on top of.)

[rough depot]

WHY

Its right at the start prechorus 1

Always with a stupid pause between hearts and aligned

https://suno.com/s/O0y8zp1wci3UqwtV?time=68

Also no matter how low the bpm i set it still sings too fast

Instead of a steady pace, it speed runs through half the song and then it starts melismating all over the goddamn place

[hoary stream]

what "remaster" will do to your song exactly ?

[novel osprey]

anyone know about this?

[broken phoenix]

it's a different person or persons