📚┃suno-school

[polar spade]

For the vocal to begin a line or whole stanza , we can use [vocal]. That's all.

As it's just one line, we likely dont even need the tag [vocal]. Putting the line in quotes will likely get it sung or spoken. No quotes , it would be sung, solo or it might be treated as choral refrain (like a call and response since its just one line.

-- but if seems needed, then try [vocal | solo]

[Break] - if "break" is working for you, then keep it going. I know [interlude] usually gives me almost exactly 20 seconds each time for me. Occassionally it might go for 2 minutes. Lol. But the longer it goes, the easier I can crop it back if needed.

[NO....this/that] suno builds the arrangement first, so it can't build from "NO" and ignores it. It may read the next bit and give you more of what you dont want.

In this case, specifically you just want silence.... but.... with the vocals being sung and not spoken. That's harder as Suno like melodies to have accompaniment.
• we can go with [vocal | solo] again... hut even [spoken] [melodically] (baritone) might also have a chance.

Since its to be without music, there is actually a common solution from past DAW work that requires Studio or an outside DAW. Just do the whole song. Extract the vocals and instrumental tracks (2) not (12) (so far that can be done in Suno Pro tier). Then load in a DAW and just clip out the instrumental parts so that only the vocal is singing.

[polar spade]

Suno might recognize HTML style code but really, its not necessary.

And Suno thinks most songs deserve an intro, so its hard to find any genre that does start with an intro of something. Though close miked emo or pop or musicals might start straight in. Those prompts might be stuck about 6th to 10thbin the styles to see if they an influence.

Does this have an intro? Or just starts in as a vocal with music?

So... using choice is [break] or [interlude] something like this is worth a go:

Here's a few samples from these prompt.

https://suno.com/playlist/587bbe68-1826-404f-8269-40ccc8549399

[elder heath]

no , no intro, thanks for the link i will check it out!

[quiet solar]

Suno won't psychically know what you're after, so your prompting will be important in terms of what you're expecting in terms of production quality and consistency of instrumentation.

The audio influence slider is more likely to be the critical one, but you'll have to strike a balance between not going so high that it tries to replicate the things you don't like and not so low that it improvises too much on things you want to keep. 100% audio often isn't a good place to be on the sliders, especially in v5.5 - same goes for 0% weirdness. Ultimately, it's something you'll need to experiment with to find out what works best for you.

If it's to create a sample for editing elsewhere, you don't need it to be a capella in the sample - I've done editing with vocal stem extracts from covers of the original song multiple times over the last couple of years, and it's often possible to get those replacements to be invisible to the listener.

[polar spade]

They are just reference to minimal intros or none as all, different variations of half spoken/half sung lyrics without any music or maybe just a sustained chord. Its not much to prompt for. But some things as you will hear, such as suno not ending the song, is just Suno being Suno. So at a certain point of you getting close (not knowing your style you're going for) then we might be better to start splicing it down to what you want.

[polar spade]

Some have intros, but prompted [with emo and musical] to be minimal, that its easy to splice right before the vocal comes in. I think only the last one has no music intro at all.

[sweet ivy]

Wondering if I can fix sibilance on Studio. Zoom in to where the sibilance occurs, like the word start, select just that small area and using the EQ function to drop the high frequencies. Would the EQ adjustment be just in that selected area, or across the entire track. Any other ideas on fixing sibilance in a few spots?

[smoky heart]

Hey gang! Let me know if this is the wrong place for this question. I'm ecstatic about what I'm getting out of Suno, but I am noticing that in a lot of the quieter passages there is a hiss/distortion sound. It reminds me a lot of when my cell phone would be sitting on top of the PA in my rehearsal space. Have any of you experienced this and or found a way to suppress this phenomenon? Attached is an example. It's extra prevalent right at the beginning of the track. https://suno.com/s/OIh2isMlmaMjojtd

[sweet ivy]

Awful quiet in here & I'm not getting a response from [e-mail] which fires back an email within a few minutes

[polar spade]

One thing to consider is that not only is AI trying to emulate a certain sound, vocal, instrument , its also replicating sounds of the analogue recording studio, the live production, mic bumps, breaths, AirCons, audience, whatever it get trained on from poor recordings of the same. Then, Suno devs are trying to find that sweet spot and call it a Model that makes most replicated genres sound like what we would most likely like.

In my opinion, bottom line is that we're going to get these hisses and hums either because the model needs some tweaking, but even still, given all things perfectly tweaked, sound might not be perfect solely from AI's understanding of what perfect means.

So, one may simply load your song into an EQ , give it a low pass and be happy. Others, might throw it in a spectrograph and look for making tweaks in specific ranges, others may play it on their car radio, headphones, floor speakers, bluetooth it to some boombox and realize the "hiss" version sounds best overall than the "cleaner" version.

If it's constant , in most all quieter passages, then thats hard to say what's of anything can help. Throwing in "de-hissing" prompts can make it more muffled, most likely , from my experience.

Some standard prompts that ive seen used are :

Clean studio vocal, intimate close-mic recording, warm analog warmth, noise-free, crystal clear production, crisp intimate vocals. But none of these might really address a hiss.

So you may try:
Ultra-clean capture, hiss-free recording, studio silence

Sneaky onee-and-done trick: if you dont mind losing some high end and separation on an intimate kind of production, but still has a good warmth, ....AND..... have Studio.... you can simply run it through FX Removal tool there and can, at times, come out unscathed with a nice , less hissy (basically dry) output.

Compare original to FXRemoved version.

https://suno.com/s/KXHypZIUo8Zrnpak crispy
https://suno.com/s/pyNOS5a0d9f2xDj0 less hiss-- dry

[smoky heart]

Thank you. Well said. I was just kind of wondering if there was a simple trick, but based on my knowledge of AI, this explanation absolutely makes sense. Seems like if I re-roll a few times, sometimes I get a take that doesn't have the noise, but often the performance on that take doesn't feel as good. Which makes it a lot like actual recording.

I've got 11 songs built from voice memo acoustic guitar and vocal scratch demos that I really like the takes on. Based on my previous AI experience I had the bright idea to remix a few of them with this prompt:
"Masterpiece, highest quality, played with the best instruments and recorded in the best studio by the best engineers and producers, with the best microphones, to 2 inch tape and a fully analog, noise free, signal chain."
I have to say, the sound quality was noticeably improved, but there were often these little artifacts. I'll probably stick with the takes I liked and do some mastering in my DAW to get some of that extra shimmer out of them.

[glad orchid]

@polar spade I prefer a dry mix and specify that most of the time to get my preferred output, all with minimal input, leaving room in the prompt for more direction.

I figure some shimmer will bleed through anyway, so I steer in that direction to reduce its presence.

[young geode]

@polar spade

How experienced are you with the sampling tool and studio?

[fluid sentinel]

Is there any way to create a persona/voice of a “cover”, of an uploaded song? I generated a cover of one of my own songs and want to apply the same drum and bass tone to some other tunes.

[coral elk]

Hey guys! 🤘 Curious about your workflow with Suno + Melodic Death Metal. I keep running into the infamous "AI shimmer" (that screeching, high-end artifact).
​I separated the mix into stems (using UVR5) and the issue is clear: the noise is entirely on the guitar track. The problem is that a standard EQ/Lowpass filter just makes the riffs sound muddy and dull if I try to kill the hiss.
​What works best for you to fix this specific guitar artifact? Is it worth subscribing to Neural Analog, should I try a targeted de-esser/noise reduction, or just run a new Cover in Suno with cleaner prompts? I welcome all tips and "black magic" audio tricks!

[polar spade]

STUDIO Im familiar.
Sampling tool , I've just used experimentally, so not much other than trying it occassionally.

[young geode]

How experienced are you with Studio?

I'm looking for a good start on it.

[polar spade]

Its just a sequencer mostly with use of Create and Editing within the tracks.

So, we can start with a goal of patching different output segments together.

Or we can build a new song from scatting a tune or tapping a beat, then Cover or Replace those , using the floating style and prompt bar at the bottom, to create instrumental tracks. Adding vocals would be the same.

All the basics would be found in the intro video to all the tools. https://discord.com/channels/1069381916492562582/1468677589752352799

As far as effectiveness, keep in mind, Studio 1.2 is designed to focus on integrating creations within Studio back into the original track, in the sense of adding back in accompanying instrumentation and vibe on each new track. So, as well, to keep everything in sync, we'd use the tracks created in Studio to remain in Studio. Once taking out to a DAW and brought back in, either location, tracks may get regenerated out of sync from the original project in Studio.

There is a time stretch and squish and quantizing tool as well to play with for that. I havent needed the quantize so I havent tested on it to any measure.

[polar spade]

Sampling, I just know it takes the sample and randomly integrates it into an output, sometimes pepper throughout, sometimes it might just show in the intro or ending , but seemingly the idea is it gives a motif feel of the sample, with maybe the weirdness slider and the style prompts enhancing any noticeable " preset" effects to it.

So, I personally havent needed it for any reason.

[polar spade]

With zero questions Ive seen posted on quantizing, it doesnt seem yet as popular a tool as it is in a DAW. Thats a bit odd, so teaching to quantize an uploaded file or even a recorded scat before turning into an instrument or vocals, to show its benefit, might be a good video to make.

[quiet solar]

There isn't, sorry - not from uploaded audio or any of it's derivatives.

You could extract the drum/bass stems and use those as Inspo for a future song. If you have Studio, you could silence all of the other stems and export drum and bass combined - without Studio, you'll have to pick one or the other, or put them together in a DAW (or similar) outside of Studio and reupload them.

[fluid sentinel]

It's all good. I ended up reverse engineering it with Superior Drummer. Got pretty close to what I was after.

[young geode]

Is it worth learning?

I spend a lot of time testing prompt methods and iterative changes, just brute forcing it all, but at this point I think I've really developed all I can through this method (short of testing mixing prompts/tags).

[rain lance]

Sometimes it'll let you do it, others it won't, it's strange that way. You could also try covering the cover with a new prompt and a low audio similarity. I got good results that way once, but also it wasn't exactly what I wanted, and you can't use another upload this way. You could also maybe stem out the drums/bass and use them in a new upload, unless the pacing is far off, but low audio similarity + decent weirdness level could fix that for you while maybe keeping the sound similar. Honestly, Personas are hit or miss or miss, so it wouldn't be a miracle result if you managed to get it to work.

[unborn star]

Well I´m still testing a lot since 1 year, and I´m not finished yet. so many possiblity, so many levers. I wish I had 7500 characters in Lyric Box and 2000 in Style Box, to test really specific things. But it looks like I´m going to use Studio soon, because there you are quiet limitless.

[sleek stump]

Didnt really see any response to this so I'll touch on it for you. The answer is no and yes. You can not highlight a section and make an EQ adjustment, it will affect the entire track. However, if you make splits and create a section out of your area of concern, you CAN set EQ levels individually for each section on that channel line. Additionally, you can make other setting changes as well that will affect only that section on the line.

[young geode]

My method is fairly developed by this point: it's about as information dense and placement meticulous as it can get. Additional characters would only allow me more input, but not necessarily more control.

Studio looks like it's a good next step, but I don't wish to blow 24/month for a tool I only may use.

[unborn star]

I can feel your point, but I can also say you are not right. I should read your guide completly to say it with certain. But I guess you didn't find everything. For example (I won't explain, because I need to sort the information for me before) but to limit your track you create by ~1 min even when you type more than usually or even getting cue songs, which is for my instrumental gaming a huge plus.

But I can show you an example if you want, just hit me up on pm.

I didn't deep dive yet, but maybe you can even get 2 min or 3 min songs out of it.

but I get constantly ~1 min if I want

[young geode]

The only inputs controls I've yet to test are Track Length, EQ/Mix and types of Recording.

I've tested and refined Structure, Tag Placement, Tempo Control, Vocal Control, Narrative Control, Genre Selection/Target/Fusion, Era Selection, Instrument and Model selection/transposing, Scales/Play Patterns, Beat/Groove patterns, Meter line use, and near everything under those categories.

Short of controlling the generation of a specific stem, I'm merely limited by characters at this point.

If Studio is worth using, then I may go play with a DAW for a week and then up my sub for Studio: but only if it's worth learning in the first place.

[polar spade]

Well, on that, my philosophy is that i can always work harder to make $28 more that first couple of months month than I can afford to risk $300 and not like something the first week. If I like it as something to learn then I continue to work harder to make $20 a month. If I need it as something that is making me money, then I may fork out the full year.

[young geode]

I don't make anything off of this, it's purely for the love of the game and my own enjoyment.

If I did get an income from it, I'd likely just shell out the cash. However if it does make working with SUNO that much easier and smoother, I'd shell out the cash just to waste less time.

[unborn star]

Well if you want to use Studio we can have an agreement. I would like to know how studio works exactly. Since you doing like me a lot of in-Depth stuff, I would willing to pay you a month and share data about how studio works for you 🤔

[polar spade]

@young geode The editor should not only replacing lyrics, but has a style box. So its no different than extending. It can just do extension in between things. If we can extend a song and prompt to get just a couple of instruments, then Replace in Editor should do the same -- but without sliders, so there is that.

Of course , then the question is . . . *

• can we still make solo breaks and interludes with extensions.

[polar spade]

If Extension uses the previous context to add an extended and blended track using contect from whats previous to keep the same vibe, then we've just explained what Studio does when we create.

Add an EQ, a Time stretch and squisher, and quantizer.

The rest of what it does is already possible with Pro or Basic. Edits and stems.

The only big additional that Studio does is be a sequencer and allow whole or segments of any output to be layers and spliced together.

From that, then it becomes skills already learned to prompt and 🛝 slide, and knowing what cant be done. That's a learning curve whether learned through starting with Basic and working up through what we can budget, or going head first in with Studio.

[cyan wedge]

how do i get music without drums? anytime i prompt "no drums" i still get some sort of percussion. @ me for suggestions

[spark crypt]

So let's say I have recording of myself that I'm trying to cover. I'm trying to get suno to make me talk faster. No matter what I do, I can't get it to speed up vocals. It sticks so close to my original tempo

[polar spade]

Suno requires using either their Exclude tool, or we use a clever means of prompting for styles or ideals that dont normally include drums. Even "drum-free" might help.

Excluding just "drums"in the exclude tool isn't quite the trick either. Its better, in my opinion, to exclude those style that are predominantly with drums. Well, with drums, thats almost every style. But doesnt hurt to include rock, metal, world music, orchestral and a few other.

So for non-drums, it may be even better to prompt for syles that feel drumless: ethereal drone, ambient synth, cosmic pads, dreamy atmospheric melody,, interpretive ambience, drum-free soundscape.. drop these in about 8th position in the styles list.

But , bottom line, its seems that Suno simple sees the word "No" or similar relatives as a neutral term and ignores them , when in our prompts. Possibly, this can mean it reads the rest and actually ADDS drums to the mix.

[cyan wedge]

That is exactly the style I'm promoting but I'm on 4.5 free and about halfway through the song drums kick in

[polar spade]

If you're just Pro, there is "Adjust Speed" and "keep pitch" which will..... well... adjust the speed but keep the pitch. This may be enough.

[flat needle]

I am in trouble with tempo and instruments, first off, I forced Suno to give me fast tempo, 150 BPM, aggressive, but instead it either did a fast intro and slow down verses or both verses and choruses were slowed down, the only solution to this was to write very fast tempo, however, now I feel like it is too fast, and the verses are still guitar inactive, I wanted a constant rhythm, I wanted to add a countdown to the song, but nothing seems to work, [silent band, drummer countdown, only hi hat]

[flat needle]

I see no such thing in my creation tab, do I have to be Premiere?

[frozen dagger]

right-click on the song and look for Adjust Speed in the Remix/Edit tab

[flat needle]

great, I unlocked new feature

[polar spade]

Which thing? Adjust Speed? Thats in the menu remix-edit/Adjust Speed of an output.

There is also an adjust in bpm in Edit UI which can set a steady BPM, if needed (save file when done)

[frozen dagger]

if you toggle "Keep Pitch" you'll notice that the slider changes. I think you can make smaller adjustments with it toggled on, then you can with it toggled off.

[flat needle]

I got the answer, thanks

[desert vector]

i just removed a break that was inbetween these two sections of the song, it was too long so i made the song shorter, how do i now make it sound smooth good?

heal edits?

[polar spade]

No. Heal edits is just another way to say "cTrL z" it just reverses steps.

[desert vector]

do u know how to do what im asking?

the intro was too long, i removed about an 8 bar

so to smooth the transistion

[polar spade]

Where did the gap come from to begin with? Did you add it?

Also, sometime we accidentally drag an end over the waveform and hid things.

We can move the two segments apart so we can grab a handle and open the waveform more to see if it as far as it goes....

But if a splice point, both sides now should extend and can overlap each other as well by grabbing the handles. If just Pro and edit, we might drag the inside handles to where we see a duplicate bit on both sides, then grab a segment and over lap it , mostly visually, until it sounds good aurally.

[desert vector]

so suno makes a song after i give it my audio

but the intro is too long

so i delete an 8 bar

now to surgerically connect the remaining pieces

you catch my drift?

tbf what ive done so far doesnt sound that bad, maybe in sunos workflow it automatically connects them in a way

[polar spade]

Got it. Well, to start, if the splice is close to the zero cross edit line on the waveforms on both sides, so there's no pop or click (as much) you might simply run it through a Remaster subtle or normal setting. Let Remaster blend it.

[desert vector]

thanks ill try, and also is there a way to like change this outro? like id want it to fade out here instead of going back into another drop

i can add a section inbetween

but not toutro

[remote ivy]

you need to describe what is going to keep time and rhythm if your going no drums....in terms of other instrument's or vocals for the genre ur targetting

[polar spade]

• Click there.
• Click the menu dots and choose splice.
• Click the head of the segment to the left of the splice, the one youre keeping, and you'll see the Fade handle as you can see in your screen shot of the far right segment.

• drag fade slightly left or as much needed.
• delete the right segment (?)

[polar spade]

And also, go ahead andclose/hide that little 'training/ how-to/ introduction to editor" thing above that. Its in your way.

[desert vector]

ideally not actually fade

are you a producer

i mean it just extends the song but fades out

[polar spade]

You need a button ending? No. I dont claim me as a producer /musician .

[desert vector]

fair enough

[polar spade]

You want it to then go into a final long chord and fade? Or a just a stinger chord... that would just be some lucky prompting. But extending with model 3.5 has does some good in those.

[polar spade]

And whats at the end of the last bit? If its already a good ending, then you'd be in luck to just splice and tag it to the same style seem earlier. There's a couple of ways to go about it.

Have you tried Remaster yet, just to see how it sounds?

[cyan wedge]

No vocals, light percussion is fine but the songs are teetering into the realm of techno sometimes

[desert vector]

hmm right

yeah would be good if i can just select parts

like imagine selecting verse 2 or the outro and then saying, remove the vocals, slow things down, no more drums, fade out

i tried add a section but it just sounds very OFF and not belonging in the song

[polar spade]

Yea. Much of that can be done on an upgrade using Studio.

[desert vector]

oh id have to test that out

[quiet solar]

You can't quite do it like that without Studio, but if you export the song as-is, you can do an Extend from the point where you're looking to fade down and try prompting for something like [Instrumental Verse] [Instrumental Chorus] and various other tricks to try to get Suno to avoid bringing the singer back in. Then back in the Editor, you can run a much longer fade over that so it doesn't sound like a quick fade-out.

[remote ivy]

testing something out

[polar spade]

There is no 3.5 ver in Edit UI. So you would take the last saved song out to Create and extend there in 3.5

Even just a prompt of [end] could be enough to give a 2 to 20 second ending. Sometimes it can also be 4 minutes long, so we always understand it could take a few generations to get one that works.

Have you used extension in the Create panel before? There is a bit of a touch needing to place the timemark from where we are extending . 3.5 may need to come in a couple of second at a zero crossing point or with 5.5, I think we're can get closer to setting the extend point very lose to the end.

[remote ivy]

whats the exact prompt your using?

[ripe galleon]

is there a way i can prompt suno ai to add like metallic hits?

like as if they came straight from a scrapyard?

i cant get v4.5-all to do it

[cyan wedge]

Here are some examples

Not on PC atm

[tame leaf]

anyway reason some people arent allowed to use voices anymore on Pro plan

[remote ivy]

oh yeah i went through ambient no beat struggles a few weeks ago, i f you go the ambient style list you should see maybe 4 of my ambients and their style prompts, they use exclude lists though, although they actually maybe 4.5 also (just because everything else wanted to give me beats). i show-up as I'myourhuckleberry, i believe the lyric metatags are hidden on most of the them, just because I didn't thoroughly vet them)....for brevity's sake you can see if you reuse those style prompts as bones.....If i recall I did still get slight beats on some gens but managed to get no beats atmospheric ambient there

[remote ivy]

okay i looked at this no beats ambient...the meta tags that aren't shown are mostly instrumental direction in the section format in the lyrics: https://suno.com/s/ta5c0jMBblqEFJMl and the sections are filled w/ stuff like this:
[Section A]
[genre: atmospheric cosmic ambient, deep-space drone, suspended voidscape]
[mood: empty, hushed, suspended, far-reaching]
[energy: extremely low]
[instrumental: drifting motion, nearly static clouding, vast atmospheric wash, fleece-soft haze]
[instrumental: low-register tones in a narrow field, wrapped in haze, cushioned in cottony layers]
[instrumental: descending wisps, distant echo bloom, downy trails]
[instrumental: hesitant entries, long silences, cavernous air, plush reverb]
[instrumental: stepwise motion with brief upward flickers, feather-light threads]
[instrumental: breath-like fragments, veiled in spacious reverb, soft and fluffy]
[instrumental: motifs repeat imperfectly, buried in mist, nestled in haze]
[instrumental: low drones hum, static shimmer, distant tonal haze, velvety dust]
[instrumental: sub-bass rumble, soft spectral flicker, padded glow]
[instrumental: wandering harmonic dust, trembling resonances, woolly overtones]
[instrumental: dim pads drift, crackling hiss blooms, cotton-fine noise]
[instrumental: rising glassy swells, light void pulses, airy plush shimmer]

i would say try reusing just the prompt and if you still gets beats after 2 creates, try adding this section in and see if that removes it, then you can shift the details from there using the lyric magic wand in edit lyrics mode to shift the meta tags (as my theme here was cottony fluff as should be apparent from some of descriptors)

[old ginkgo]

I cant seem to get hard or heavy bass for some reason. Im tryin to replicate 1990's bass 305. I have tried all sorts of prompts from all sorts of places including chat gpt. It seems like suno doesnt understand ultra low frequencies idk. Any ideas?

[remote ivy]

while you wait for someone to respond if you know the 90's genres, you can pretty much type https://suno.com/genre/90s bass or something similar if you know what the genere would be named, and do some browsing for prompts

pretty much if you can imagine a genre you can type it in after /genre/ and find something there

[unborn star]

@polar spade maybe you can unser my question, it looks like you worked quiet a bit or you understand how it works?

I wanted to know how uploaded songs interact with the song you want to create. Is there any behaviour you have analysed?

[remote ivy]

https://suno.com/genre/305 is the 3rd one down the sound you want?

[old ginkgo]

Yes! that type of bass and a little deeper but yes 🙂

may be some more sustain on the tones as well

[remote ivy]

you don't know if that is a single gen or string of covers......were you just simply missing the 305 in the style prompt or is their prompt kind of emulating the sound w/ everything combined?

well if you get nowhere near the sound, then you can assume they have layers below, just move on the list i guess

[old ginkgo]

I will keep looking...the bass style is almost right but the rest is nowhere near LoL 🤷‍♂️ I may be too picky. I thankyou for your help 🙂

[remote ivy]

seems miami bass is synonomous w/ 90s 305,,...try exploring w/ 90's miami rap, i swear in one of my searches...miami came up as part of the genre name on the suno search...so its got it kind of associated, i just don't remember what i put in search though

[old ginkgo]

Look for BASS 305 - The Science Of Sound (Sub Woofer Movement Version) on yt and youll see what I mean

or bass 305 Musik of the Future (Sci Fi Bass Test)

[mighty umbra]

Is there a good way to keep my own voice consistent in songs? All my songs lose my voice 1-2 mins in and revert to a regular generic voice.

[polar spade]

I havent noticed any correlation. My whole songs Ive uploaded would be easily recognized if I felt a melody or vibe showed up in any recent outputs since an upload.

The closest I've heard are on a couple of suno songs from 2024 that Ive played about 20 times more that others over the past three years, have I heard a yell or riff of the Suno output, come, through on some sample songs I did, so far.

[polar spade]

Do any of the other instruments graduate the same way from, say, a synth into a violin? Theoretically, either you'd get the same effect with an instrument too, or several instruments just seem to evaporate after the 2 minute or 4 minute mark and we get a smaller arrangeement by the songs end.

Id say its not always easy to control, since Suno is autoregressing every now and then to keep things consistent, but at the same times needs to allocate more and more tokens toward finishing the piece. Things might get sacrificed in the process. So bagpipes, illean pipes, harmonicas, brass horns, some highly produced distortions and custom or persona/+Voice voices included probably take a hefty punch towards depleting tokens. Characteristics and instruments themselves may get sacrificed. So that custom voice may meld into a more of a Suno sounding voice as you've heard.

Thats my take on it , at least. I could be all wrong.

Make every effort to keep prompts and lyrics, tags and even sliders away from maxing out, if at all possible.

In addition, practice using extend. Finding stopping points to clear the Create panel , start over, dropping back in the bit you've done so far and extend from a point, while running your custom voice again, to get Suno to start fresh from a later point in the song and maintain the things we want as priority.

"Extend is our Friend"

[quiet solar]

I'm equally in the dark on the actual mechanics, but from observation would agree that beyond a certain point in the song Suno is more likely to begin losing track of what went before, and that when you layer on audio influences like Cover or Voice (or both) that point tends to kick in sooner.

It's not something that can really be controlled to any great extent, but using shorter prompts and shorter audio influences definitely helps to avoid it.

If it's unavoidable, then Extend is definitely a good go-to option. Cropping the song and then covering it and extending to the end in the lyrics is another fallback option that can be worth trying if Extend doesn't play nicely - although probably not ideal if you're looking to layer a Voice on top.

[cedar vortex]

is there a way to get a ~30 seconds instrumental song?
i'm trying and i got some nice songs but they are 3 minutes and i want less lol

[hoary sandal]

@polar spade @sleek stump May I have either of your assistance?

[sleek stump]

?

[hoary sandal]

I wanted to take a break from my normal album progression
Do an acoustic album, based on past songs from past albums
But how can I apply accoustics/acoustic instruments to songs without changing the vocals and their rhythms?

(Aside from maybe screaming parts I won't need)

[sleek stump]

You might want to use all of them to build a custom model. I am using a custom model to finish a current album project I am working on.

[hoary sandal]

How do I do that?

I have a persona?

And also have the songs to reference from

[sleek stump]

In create, click on the model selector and at the very top you can make a custom model.. You can use as few as 6 tracks and as many as 24 to train it.

[hoary sandal]

I will try that

Thank you

[sleek stump]

It does an unbelievable job in keeping the entire sound of both the band and the vocalist... I personally use my vocalist persona with the model.

It will take the back end a few minutes to generate the custom model, but its well worth the wait.

[hoary sandal]

Reckon 56 songs will be enough?

Considering those are my 6 albums so far lol

[sleek stump]

Well, chose the best 24 of them 😄

[hoary sandal]

Oh

[sleek stump]

Best results are achieved at 24 tracks

Although I have 100% results from 7 tracks

[hoary sandal]

Well, picked 24

Was easy to take out my first album at least

OMG It so good

[sleek stump]

Hmmm, I need to back up a bit... if you are changing genre types, a custom model wont help, it will moreso solidify the sound of the training songs. You could try to use it and prompt for acoustic versions and new song with it, I really dont know if that will work or not...

[hoary sandal]

Its working, trust

[sleek stump]

Wait, did it do what yoyu wanted?

Awesome!

[hoary sandal]

Yep, used a different style prompt

Same persona, new model

[sleek stump]

Ding ding!

[desert vector]

is there much difference using studio studio vs editor
likw how good

[sleek stump]

They have different goals... if you have Studio, some things are much easier, but some are still better/easier in teh editors.

[desert vector]

hmmm

need to know if worth upgrade

unless we can trial?

[polar spade]

One way is to not invest in it until you run into roadblocks with a Pro level that you feel can be only answered by upgrading.

I mean, before Studio, those who had a personal DAW would do what they could to download tracks and do all editing off-site. So maybe those users would be good to ask as well.

But it comes down to what your final goal is. I bet there are thousands who have never touched a DAW and so love the fact they can upload a song, scat some lines and add a new track of different instruments. Then select small sections and enhance them. And then we can add sound effects and Foley sound for making that holiday spooky song or just a nice rain or windy beach effect to our song or maybe even a cool 5 minute ASMR track of mixed effects.

Up to you to consider though. Personally, I cant recommend or discourage.

[desert vector]

when i do the fade out thing it literally does not fade it out lol

[desert vector]

as in like download stems and put into our DAW?

and yeah im very experienced in daws i work in music

ableton for example

i guess i could download stems, i cant tell if the quality gets worse

[polar spade]

Before Studio, that was the opted move when Suno could extract stems first. Now many can option for using Studio than a DAW for certain things.

Again, depends on the final goal.

[waxen merlin]

🇵🇱polska mentorka/mentor poszukiwany 💃🏽🕺🏽

[lost falcon]

hi I'm Hui sara very happy to join

[young geode]

Have you tested/evaluated your own vocal range and qualities?

SUNO recognizes these as inputs, in the style section and as metatags.

If you know your exact, and near exact ranges/qualities, make sure your give those exact qualities within the style prompt, and use the negative prompt to remove near/similar ranges and qualities.

Then use [Vocals: Male/Female, Range, Qualities] within your lyric box, and stich it throughout the lyrics, so as to keep SUNO's focus.

[quiet solar]

You might get some headaches with exporting stems in 5.5 due to the much better/heavier production effects that Suno layers onto songs, which sometimes don't quite come out as you might hope, since the stems are created retrospectively rather than being actual stems that the song is built from.

It's not something I've had much of an issue with in previous models, but in 5.5 I've generally done editing work in Studio as it seems to handle that issue far more effectively.

That aside, the main benefits I find to Studio over the basic editors are;

• You can work at stem level within a song as in a typical DAW, but with built-in options to cover or replace sections
• You can pull in multiple songs to one place in order to extract samples, combine different covers, etc.
• Basic DAW features like crossfading and some EQ balancing are available

Going to say up-front that I was never a fan of the previous editors. I always took songs and samples to an external DAW rather than use them, but I appreciate some people still swear by them. 🙂

[delicate karma]

Testing Suno v5.5 limits: Epic Battle Music
I tested multiple subgenres: Viking, Mongolian, Gufeng, Vampire, Dark Fantasy, samurai ...
https://www.youtube.com/watch?v=rriSLfVDnJk
What do you guys think it's still lacking?

[desert vector]

thank you and @polar spade for your advice

yeah ive used logic fl ableton for years, now predominantly ableton so i could just export in stems (which im doing now) but i am curious how far i can get in studio if i upgrade if its easier almost

[polar spade]

Think of both as complimentary to the other.

[desert vector]

1. being able to remove sections of the song and then blend together the cut up parts
2. being able to remove crazy vocals but keep everything else, cause it seems when i remove vocal sometimes it kind of empties the entire song

[desert vector]

im downloading stems as tempo locked wav to use in ableton, if anyone has better sugestions im all ears too

if theres a good way to find tempo/key easy too

thanks

off rip i can hear the stems quality isnt as good as the audio file in suno

so may be best off just cutting up the audio file

[sleek stump]

Dont be fooled by what you hear in a single stem, when you are playing all the stems together, theres normally no discernable difference either sonically or spectrally.

Of course there will be some loss, thats kinda the nature of stem splitting in every platform currently. No splitter can get it 100% right for everything it detects.

[polar spade]

As a general reference, the workspace files in studio should show the tempos of files, if that helps.

As for key, as Suno generates by tokens and time constraints, there may be songs in a key, but doesnt assure perfect pitch and may drift a few cents. Thats kind of expected.

But , yea no, theres not one area to show a key that I know. That would be helpful for 5 gens that all sound the same but only one is in G , then thats the one you work with. I get that.

As you would know, there are listening apps to find keys that are helpful if you have two phones or a phone lying by your desktop. So its using all tools available sometimes to keep things moving.

[desert vector]

no im playing all together, definitely a difference, im in a professional music studio, monitors and studio headphones

but its fine cant be perfect lol

[desert vector]

no worries thanks

very impressive stuff suno lol

the new age

its going to take over the music industry once people become custom to it

right now its like the start of tiktok no one wanted to touch it

my guess anyway

[polar spade]

How to do it old school with a guitar tuner. This song is in C#. 😁 Handy if producing other people songs that walk in your studio. Otherwise, we work on our relative pitch detection.

[unreal sleet]

if I say the sample that goes "woop woop" that has been used since like early 2000s, do u know which sound I mean?

its like a chant

if i knew it was called mellinial i would have said it 😄 but maybe? xD

watched a video about it and its not that from that example

its basically the exact same sample of Woop

[remote ivy]

crop and fade? else what prompts you used..i.e did you try something like short instrumental or 30 second instrumental near the beginning of the prompt...also if you do something tell it to fade out between sections that could make easier to crop

[unreal sleet]

https://suno.com/s/rJ8eW3xoqZzmUc3Q listen at 0:17 xD

[half dragon]

https://tenor.com/view/whoopthereitis-whoop-there-gif-4957750

woops, wrong woop 🙂

[unreal sleet]

😄

im trying to find some way to exclude that sound as its being repeated 50 times per song with this genre, i know i have triggers such as classic techno elements and crowd (chanting), and the genre itself used to have fx/samples like that. but every song is almost built around that sound and its becoming kinda annoying

[polar spade]

Yep. 80s.

[polar spade]

Just the prompt [jingle] can often achieve it. Screenshot just now of a 4 second and 13 second gens from just [Jingle] in lyrics and styles box.

If needed, throw in two lyric lines to get it to extend. Then take out the vocals.

[vast night]

What's the difference between voice and sample?

[polar spade]

Sample involves an output (song) of your choice. You select a small snippet or the whole song. Suno then tried to pepper it about like a sample into the new output.

+Voice : A means of taking an audio of your voice and create a "code file" of information about its characteristics and sound. Its save as a reference file to attach to a new creation.

This is used to enhance a new song with intent to insert your voice into the new output.

You can +Voice an existing output as well and save a "code file" from that. VOICE files are save under +Voice , found tab in the upper menu of Library page.

https://discord.com/channels/1069381916492562582/1486745513847033966

https://discord.com/channels/1069381916492562582/1486479683934617611

[vast night]

Yeah just started using the sample, so far loving it ☺️

[polar spade]

Slide the weird slider up more on Samples and see what it does.

[vast night]

So lower is more to the original"file"?

[polar spade]

Not saying that. But often weird may get too weird. With samples it , often has a creative edge and often nice result.

Try one. 80 to 90%. And if too janky, just drop it back to default.

[polar spade]

It all depends a bit of how much you Sample. A large about it seems to blend a medley. Smaller segment it peppers in more a motif with repetitions... depending on your prompts.

[vast night]

Yeah noticed it function at 15-85%

[nocturne drift]

Hey guys is it possible to make instrumental covers of real songs?

I want to make a classical cover of a song for a wedding

[polar spade]

Sorry. No. We can't load public songs (of any era) and any derivatives into suno. No songs that already exist. Just our own exclusively owned pieces.

[polar spade]

Its about not putting public songs into Suno. So why not write something similar(but not the same). Even hum something interesting and ask Suno to come up with a wedding type song.

[summer patrol]

im struggling to make sense of the extend function, its not exactly intuitive. i dont want to replace, but maybe add a min or 2, but all i get is wierd sounds in the transition or the dreaded 7.59 min songs. does extend work and if so how lol

[polar spade]

I could only suggest that the timemark is at a zero cross point... lowest point on the waveform (or what you hear) and come into the song at least 3 seconds to 6 or so if just attaching to the end, leaving the sliders at default.

[summer patrol]

logically the slider should be movable in both ends, so you could set a point where exactly you want to start then the other end how long you want to extend, buts thats what i feel it should look like 😛

[polar spade]

Sorry, I wasnt clear. The audio sliders of weird, audio and style.

[summer patrol]

yes, no i understood 🙂 i mean the extend slider

the others i left default now

[polar spade]

Ahh
... but yes. Your idea is good. Extend is not made for creating intro "before" sections. But in Pro that can be done to some measure in the Edit UI.

The original Extend was one of the first tools and been around since March '24. So its handy and quick and does one thing.

[summer patrol]

that did improve the transition leaving the sliders as default, seems turning it up to 100% audio confuscious things 🙂 and i played with the edit ui a bit already, especially if the first half of a 7.59 song worked i been cutting it using fade to xcreate endings

but it changes the song a bit to much because the default is 25% audio 50% style 50% wierdness -- so end result is not that great

[polar spade]

Im sure it can be subjective. So , play with the sliders to find the sweet spot or prompt to get less energy and instruments. Who knows what the best combo would be.

[acoustic egret]

Yo how come when I crop a song it CREATES A NEW ONE? how do I just override the existing?

[polar spade]

Suggest that in the suggestion channel. You're not wrong. Its just how its designed.

There may be an economic reason for only being able to do some actions one at a time. i.e. Not save the edit directly to the editor, but save to a file instead and close the editor.

You then bring it back into editor to do some of the edits one at a time.

Only the devs would know.

[polar spade]

Sorry maybe I need to start over.

Crop in New Edit UI , does crop and then remain as a new Edited/cropped file in the editor to make more edits. Then click save file. .

In Legacy Edit, it would often save but zap out of the editor once done so thats the bit I was talking about, not staying in the editor interface.

And in Studio , there isn't a crop option that I know of since its designed to move things around ans we just mute or delete sections at will. In the end , selecting and saving what we want. The rest we manually keep for re-visiting later as an option.

Sorry if I missed the point.

[small rover]

anybody getting errors on prompts and cant generate

[sleek stump]

What kind of errors, moderation?

[small rover]

yes tried prompts i used before same error

[sleek stump]

Whats the error say? It should tell you why its blocking you if moderated

[small rover]

prompt contains inappropriate material

[sleek stump]

Understand that its a GPT agent that does all the moderation, theres not anything any of us can do about that.

Is your prompt or lyrics full of NSFW stuff?

[small rover]

thats nuts theres nothing there

no but ill try a rewrite

[sleek stump]

What is your prompt?

[small rover]

hiphop rap, bouncy, boom-bap, 808beats, drumline, soul-samples, synth, snares, high-hats, cymbals, tweeters, deep drums, deep kicks, strings, adlibs, hip-hop atmosphere, boom- bap pads, studio clarity, west coast-funk elements, master-FX, male rapper

nothing crazy ill redo the lyrics and see

thanks for your help

[sleek stump]

Must be your lyrics, that prompt isnt being moderated on its own

[small rover]

got you thanks

[elfin lagoon]

Hey, is anyone else having their computer crash when they are extracting stems on Suno? Not sure if its my browser, computer, or Suno, but this is the second time in a row this has happened.

[half dragon]

Stems extraction happens on the server side. Browser makes a request and they are done on the server side.
So it might be coincidental with something else... maybe..

[elfin lagoon]

Thats what I thought. Might be the browser. Any known issues with server and Opera?

[half dragon]

I use Chrome mostly -- check #🪲┃bugs-web for other Opera users?

[elfin lagoon]

Hmm. Happened on Chrome as well. Computer completely froze.

And I wasn't even trying to extract, simply viewing the extracted stems.

[half dragon]

No issues here -- maybe some Add-On is causing the browser issue? Disable them, or try in Cognito mode and see any differences.

[elfin lagoon]

I'm not sure what. As far as I know, I don't have any browser add-ons enabled.

[half dragon]

Have you tried to reboot the computer? that would help in most cases.

[elfin lagoon]

I have to reboot the computer every time it crashes, because everything completely freezes and becomes unresponsive.

[livid vine]

Most likely causes @elfin lagoon:

GPU hardware acceleration / graphics driver issue
Very common with Chromium browsers.
Media waveform rendering or decoding can trigger GPU hangs.
A bad driver can freeze the entire desktop.
RAM or VRAM exhaustion
Large extracted stems can be memory-heavy.
If the browser leaks memory or allocates huge buffers, system lockups can happen.
Faulty browser codec / extension / experimental flag
Less likely since they said no add-ons, but still possible.
Corrupted or malformed media files
Possible if the extracted stems are invalidly encoded.
Could trigger a decoder bug in Chromium.
Underlying hardware instability
Overheating GPU
unstable RAM/XMP
failing SSD
driver corruption

The fastest way to isolate it:

Disable hardware acceleration in Chrome/Opera.
Chrome: Settings → System → “Use graphics acceleration when available” OFF
Try opening the stems in:
VLC
Audacity
Windows Media Player

Test another browser engine:

Firefox (important because it’s not Chromium-based)
Check whether only specific stem files trigger the freeze.
Re-download/re-extract the stems.

Hope it helps 🙂

[bronze island]

Yeh you should be careful with the lyrics you write. There are minors here yk

This is why I only write family friendly music

[low musk]

ello

[polar spade]

Do you have a go-to tech troubleshooter (person or company) to go through steps, beyond rebooting the computer, that you can go step by step to "clean and refresh" your system so there is a fresh slate to start with before trying to figure out why your computer crashes when using a file heavy program. Probably a good 20 minute walk-through.

Otherwise we can try giving a list of steps that everyone gets and things to try on your own.

Quick question: when you say reboot, does that also include a cold start, shutting down the computer, and pushing the power button to restart?

And have you run a verified malware/virus check (another 20 minute routine) before asking for help?

[dense agate]

How to create cover with exact same melodies? That's frustrating, the result is always makes improvement that I don't want. Sometimes in chorus. I've tried V4.5 to V5.5, same case happened.

It used to be cover only change style and vocal not melodies.

[verbal knoll]

Man, The playlists on the web, are super laggy

I hope the Playlists Improvement comes

But also mobile we can have more than least songs we max have

On playlists

[past phoenix]

v5.5 cannot do instrumentals anymore out of a track with vocals, no matter what you do for a direct cover. And in some cases, 5.0 will also fail to remove vocals. They will always be there, as pig latin/hallucination. So here's a guide for those who wanna do that:

case 1 - it's your own track/a previous render: simply export to wav, reimport it, and it's most likely gonna display detected lyrics, whereas the audio editor will have totally not detected them if they're really stuck. Keep the structure that Suno detected, but remove the detected lyrics. I've still had some rare bits of vocals bleeding through, but it seems to work on 95% of a song

case 2 - it's someone else's track that you're covering. It could technically work as case 1 but you're not allowed to. So just make a 5.0 cover with [Instrumental] as the lyrics, then do a 5.5 cover on the result. Now you're out of luck if those are vocals that even 5.0 fails to detect & remove. But it generally works with 5.0, while it simply never does with 5.5 (only for the first 1/3 of the song)

(but maybe someone has an even better solution)

[sleek stump]

Or you could just split stems in simple mode and discard the vocal track and keep the backing track, then cover that if you really need to. Unless you are doing DAW work on a track, I cant see a valid case where exporting and reimporting a track is any kind of solution for anything.

[past phoenix]

yeah but it's likely that it's gonna fail the same way. For the problematic tracks that I had, that is. (it also costs more credits than for case 1, you still have to reimport the track & cover anyway)

and don't ask me why exporting & reimporting works, I thought that opening a track in the editor was detecting lyrics the same way as when reimporting, but apparently not. Whatever Suno does differently in the background, for some reason it just works

[sleek stump]

Why do you need to export and reimport at all? if you are splitting STEMS you will have teh instrumental sitting in your library ready for more work.

[past phoenix]

I'm sure you've had improperly detected stems with vocals put in another track along with another instrument
And this is even for non-problematic songs, so it's worse for the ones I'm talking of.
Also that can't apply to case 2, which is most of the cases (making a cover of someone else's track)

[quiet solar]

Case 2 isn't allowed under Suno TOS in the first place, unless you have permission from the owner of the track to upload it to Suno yourself.

[past phoenix]

(because generally when you're making a cover of your own, you have some previous version with non-stuck lyrics to start with)

[quiet solar]

If you're talking about Remixing someone's else song on Suno that has remix enabled, that's a different position.

[past phoenix]

exactly, it's possible but not allowed, which is why I explain that the only solution is to make a 5.0 cover with [Instrumental], and then a 5.5 cover. This is allowed

[past phoenix]

of course

[quiet solar]

Just to clarify, are you trying to generate a karaoke-style version where the lyrics are removed but the instruments are unchanged, or trying to create an instrumental cover that brings the melody across?

[past phoenix]

Trying to remove the original vocals and to make a cover in another style, or other lyrics. Generally with 5.5, totally removing lyrics is impossible, but replacing lyrics fully replaces the old ones. But with the problematic songs I'm talking of, even replacing lyrics can still leave bits of the original lyrics stuck and interpreted in pig latin/hallucinations.

[quiet solar]

What slider settings are you using?

[past phoenix]

and 5.5 is special in that it really wants to place vocals. If your track has even a little bit of "ah" voice in it, for some reason 5.5 seems to think it needs vocals and will even hallucinate full lyrics (as often reported)

[quiet solar]

Because it sounds like you might have done something like turning audio influence up to 100, which has a tendency to ghost the original audio in to the cover, e.g. bringing in lyrics that aren't included in the song, but were sung in the original track. Or not switching the vocalists even though you've prompted for a vocalist change or applied a Voice.

[past phoenix]

of course I don't do that
(I also don't set it to 0% because, well, there's not much left of the original song then. I do like 20% for 5.0 or 10% for 5.5)

[quiet solar]

Odd. Do you have an example you could share? This isn't a problem I've personally run into, despite doing a lot of 5.5 instrumental covers of songs I've made.

[half dragon]

I usually cover my songs to an instrumental version; the way I do it: Remove all the lyrics, but leave the meta tags and set the audio influence to 70% range -- results are very nice.

[past phoenix]

sure, try to fully remove the lyrics from this: https://suno.com/s/lbt1mp1LTpiBpTVJ
for this one, even 5.0 could not. It fully sounds like japanese, but bits of it are actually french with a japanese accent lol. And this passes through quite some remixes.

[past phoenix]

yeah this would be my case a), but you need proper tags in the first place. Leaving the lyrics empty or [Instrumental] does not work with 5.5 like it did with 5.0

[half dragon]

The tags I kept from the previous songs.. no more no less

I just added Instrumental into the same style description, that's all

[past phoenix]

well I've never seen that work with 5.5
and I've spent so many credits on that

[sleek stump]

Ya its really important to start the style window with "This instrumental"

[past phoenix]

it generally results in: 1/3 instrumental, 1/3 some lyrics bleeding, 1/3 suno stops caring and it's full lyrics

[past phoenix]

well:

and I've also had "instrumental", "no vocals", "no lyrics", etc in the styles

[sleek stump]

Thats not the STyles window. Thats the lyrics and those are meta tags.

[past phoenix]

give it a try yourself with my link above

[sleek stump]

I have had 100% success when starting my styles prompt with teh type of track it is intended to be.

[half dragon]

Is there a full song link? I can give it a try

[past phoenix]

yeah I've posted it above
https://suno.com/s/lbt1mp1LTpiBpTVJ

it has instrumental in the style AND lyrics as you can see

and it has passed through 5.0 & 5.5 several times

[half dragon]

I meant the original one

[sleek stump]

All I see is "melodic trance, d&b, instrumental" as a style prompt

[past phoenix]

it's pretty vulgar in my native language lol so I'd avoid that
but it doesn't really matter - this is still a valid song with stuck lyrics

[past phoenix]

yeah, so?

[sleek stump]

As with the introduction of 4.5, style prompts benefit greatly with more descriptive prompts. That prompt may have been perfect for v4 or older but not so much for anything 4.5 or newer.

[past phoenix]

what could be more descriptive than "instrumental", "no lyrics", "no vocals"?

[quiet solar]

First try - two vocal-free instrumental covers:

https://suno.com/s/P5wVKXUdIwLCplFg
https://suno.com/s/o6KfU90DCZmisVbT

Prompt: Cello and acoustic guitar, close-miked, dry and intimate
Sliders: W40 / S50 / A50

Maybe try toning your style slider down to more normal levels, see if that helps? Putting any sliders to max or min tends to cause odd side effects.

[past phoenix]

interesting
do you think it's the total style change that forced it?

it can't be the audio influence, because for the ones I've tried:

[quiet solar]

Impossible to know without having the original to test on.

Yeah, okay, looks like it's something triggered within the prompt. I've tested using "melodic trance and d&b" and it's bringing the vocals over, despite the lyrics box being blanked.

[past phoenix]

what do you get if you keep the "melodic trance, d&b, instrumental"?
I'd be amazed if that removed the lyrics for you

ah

well you've found a third way then: totally change the style. It's more lossy I guess, but if it works & keeps the melody, that's fine

[sleek stump]

Heres a test I did, , completely instrumental, slider setting W60 S80 A25 https://suno.com/s/t8zazS6pBbDfM1xW

[past phoenix]

here's my guess: for some reason Suno, for styles full of synths, somehow maps the vocals to a synth-like instrument

[sleek stump]

Change your styles prompting

[quiet solar]

If it's doing that, keeping the audio influence at 25% or below and specifying instruments should fix it.

Ginger's route also looks to be highly effective either way.

[past phoenix]

that's even better
minus the guitar but I assume it's "accoustic" that triggers it

[quiet solar]

Generally if you're doing instrumental covers, you really benefit from specifying the instruments

[past phoenix]

well:

[quiet solar]

Yeah, but "melodic trance" isn't an instrument.

[sleek stump]

YA we never recommend using 100% or 0% for any slider

[past phoenix]

both of yours have guitars, that's very possibly what makes suno behave

[quiet solar]

I've done similar covers with piano and no guitar, it's definitely not that.

You could actually try what we're suggesting, of course. 😉

• Keep sliders off 100 and closer to defaults
• State the instruments you want
• Use description rather than just genre prompts (click magic wand if you're lacking for inspiration)

[sleek stump]

Running a couple more tests... If theres a bug here, we def want to ID it and push it to the engineers.

[past phoenix]

(I am, it's generating)

[past phoenix]

well @quiet solar has already confirmed that the vocals were bleeding through with no vocals in the prompt - it only depends if you call that a bug or not

I tried your prompt minus the "accoustic", because I really think that's THE keyword that fixes it

https://suno.com/s/0ivK8KE5feRc16O9

and the vocals are bleeding through

so it seems like "accoustic" is a valid fix (if you can live with the guitar - it's also weird that Suno forces an accoustic guitar for what is just a generic term if you ask me)

[sleek stump]

Ya, I'm trying EDM and its adding the vocals... Honestly, 5,5 is not my currently prefered model, I am still predominantly using v5 for most of my work.

[hard beacon]

5.5 for EDM = noooooooooooope 🙂

[past phoenix]

oh, for THIS song, 5.0 has problems too
but yeah 5.0 generally removes vocals, only not for this rare case

also if you noticed, the first 1/3 to 1/2 of the song, with 5.5, is a proper instrumental, they only kick in later

[sleek stump]

There are other work arounds, you could cover it in 4.5+ and then remaster or cover again in 5.5... have heard this work for some.

[past phoenix]

(but well in this case the starting song has already wiped out the first 1/3)

[sleek stump]

@unique stone Any thoughts on this?

[past phoenix]

sure but we're back to my cases a & b
honestly if "accoustic" only forces a guitar and even keeps it as edm, I'm fine with that, assuming the guitar itself won't get stuck in future covers (because yeah that is another problem with 5.5, I've noticed some plucked instruments that simply don't wanna go away once they're present, even with the "exclude styles"). It's nothing intuitive though

like, if you generate a song with japanese influences and you get koto in it, good luck getting rid of it

[sleek stump]

I have found the exclude window is mostly ineffective, through every model.

[past phoenix]

yeah I've never had it do ANYTHING

[sleek stump]

Honestly, I am using Studio now for all of my work and advanced STEMS for everything, its been super easy to wrangle Suno into what I want.

[polar spade]

I may be misunderstanding the premise of :: no instrumental from a song with vocals:: but have screen-vid live my making an instrumental by covering an audio of a song with a vocal.

There may be other techniques which dont vent as well, so I'm happy to consider those as well.

Just noting the initial chant of "cant do instrumentals anymore out of track with vocals" wherein we are in the Suno-school, so my presumtion is we can use any tools avail to us.

If in Studio, Create panel is part of Studio so there should be no difference to make the instrumental there and drag back into Studio.

[quiet solar]

I was assuming this was a non-Studio approach.

If Studio's on the cards, I'd open the cover there, delete the vocal track and then cover the rest. Can't bring across vocals if there's no vocals in the audio input. xD

[past phoenix]

first time I see this work with 5.5, but it could also be that it's a 1min song. As I wrote, generally the first 1/3 to 1/2 of the song is a proper instrumental. Perhaps it doesn't mean the first 1/3, but the first minute

[rough depot]

Tf is happening https://suno.com/s/mjavkZsDTO641PzE

[sleek stump]

If you dont have access to Studio, I can see how frustrating this could get.

[rough depot]

Its been doing this a lot at the end of songs

[sleek stump]

Or just split the vocal stem off and cover the instrumental... thats a pretty direct and easy route.

[past phoenix]

That's assuming the vocals are properly stemmed out for the problem track I posted. I'm ready to test a 2-track stemming but not a full one as I'm running out of credits lol

(also I don't like Suno's stemming at all, the stems don't add up to the original, for mastering offline it's not great)

[past phoenix]

I've heard worse lol

[tender geode]

Perhaps you are misunderstanding this problem . Here it is: I've created a song in Suno. Expanded it into stems in Studio. I think "I need to add a marimba track" - so I press "add track" , and give it some prompts, to create a marimba track that will follw the original recording. It spins for an hour, and then fails. Every time. The "Add Track" button does not work. Simple as. 😎

[unique stone]

I see the issue described here is that v5.5 (and v5?) is wanting to hallucinate lyrics.

If the requirement is to eliminate the vocals from a specific song I would run a vocal+instrumental stem split, then pick the best of the instrumental stems and run a remaster or cover on that. If there's no v5/v5.5 specific features desired (e.g. a persona or Voice) I'd use v4.5 or v4.5+ for the post-split cover or remaster.

[polar spade]

I'm going off the initial general premise that something presently cant be done, and showing , live, that it can be done. Sure, there can be extenuating circumstances in all attempts and outcomes -- so thanks for allowing me the time to debate the options of the basic means.

Its good to know the basics are still sound.

[unique stone]

This happens if the sliders are too high/too low. This specific extension was created with these settings. In this case I suspect the audio influence is way too high (default is 25%).

[sleek stump]

This was pretty much my first thought and reaction. Ive been around long enough to find solutions to model deficiencies rather than brute force it.

[polar spade]

My response was specific and toward another subject in these recent posts.

As for your issue of a stuck progress spinner on track creations from another track or stem , that has been discussed for you in Studio-Feedback and a guide on how to resolve it by notifying Elliott.

Things are a bit confusing as we're having several convos at once with different users.

Thanks.

[past phoenix]

I think the hallucinated lyrics is another thing - those really come from nowhere, and there's real hallucinated text behind

here I believe it's more suno not even seeing the lyrics as lyrics, but as an instrument that it wanna keep. It's all pig latin, a bit like choirs in orchestral, that for suno aren't lyrics at all, and don't map to any real language, they only "sound alike"

Because yeah, if you ask for choirs in an instrumental, for suno it's just an instrument (and that's great, because choirs in latin aren't supposed to be understood by anyone, it's sung but it's not "lyrics")

So I think it's exactly the same here, it's not orchestral, but it's "a choir instrument" for suno internally

[polar spade]

BTW, I think that was my very first song ever written in Suno back in 3.0. Just wanted to give it its debut here in Discord. lol

[past phoenix]

(here's an exemple of 5.5 hallucinating full lyrics like 3.0 & below used to a lot, but never 5.0. I had reimported the track for suno to read the lyrics, it's funny)

unlike the choirs, those are very clear & in proper english/a proper language. It's a different "problem"

I've covered @sleek stump 's render with this style, in order to know if what I was suspecting would happen: the guitar wanting to stick the same way as the vocals:

and.. yeah, they do.. but fortunately only for one of the 2 renders:
https://suno.com/s/5BnO5kz4Bjfto7CA

the other one still has some bits of guitar but it's mostly plucked synths
https://suno.com/s/qA7kfpBlCLQNYnG5

so yeah it's creating another problem. But it's still valid enough. Perhaps there are other keywords that force suno to go no vocals that don't introduce accoustic guitars

[sleek stump]

I think the bottom line is, theres a few ways to tackle this issue, but none of them are really super convenient.

[past phoenix]

• the fact that plucked instruments are very sticky, hard to get rid of, in 5.5

[sleek stump]

You arent stuck using 5.5, you can use other models as part of a production process. Again, not really convenient, but I digress.

[past phoenix]

I know many don't like 5.5, but the sound is really in another league, especially the sharpness of the drums. Hard to get back to 5.0 now. My only probs with 5.5 are the many bugs and the vocals zipper noise.
(it's also true that 5.5 is less "creative" in that it tends to give samey results to everyone, but I don't even mind that)

[quiet solar]

But you can get round that by doing it in multiple steps. Either do a cover in whatever version gives you an instrumental profile to follow with no vocals, or split the instrument stems from a cover (as Furrybeans suggested), then cover that version in v5.5 until you get what you're after.

[past phoenix]

yeah I totally do that
but there's a loss in the process, which sucks when you get the best 5.5 render

also going back from 5.0 to 5.5 is still gonna result in zipper noise (like what I'm just getting right now covering the contest song. And when it's a contest cover, no offline fixing allowed)

[polar spade]

Fortunately, any slight issue in fidelity are probably not factored in, as for the contests. (i.e. not voted on by users)

[past phoenix]

oh I don't care for the results, it's fun & exposition
but the fact that there's a lot of zipper noise in what people publish is also a problem (as a listener), it's very annoying

[small bridge]

Does anyone know if there is a way to see my stems and only stems all of them maybe even deleted ones or disliked ones all at once in a list. I lost this banger instrumental I had and I want to write lyrics for(a year later lol procrastination who?) just the stems.

[half dragon]

Stems, once extracted, are always in your Library... but unforunately, there's no way to see just the Stems.

[wide cosmos]

So... how can i end the subscribtion in suno ai? I bought around 3 months ago the subscribtion on suno ai and i want to stop or... idk how to say right

[half dragon]

Click on your Avatar > Subscription > Manage (top right) drop-down > Cancel Plan

[wide cosmos]

Oh...

Thank you!

[fiery prairie]

Has anyone been able to get extends that don't force random vocals? The current one I'm working on begins with a spoken word line and has some remixed background gibberish, but that is part of what I want to keep. The problem is that whenever I try to extend, and I want it to only be instrumental, it just inserts its own made up lyrics no matter what I prompt. Doesn't matter what I put in lyrics/style/negative, doesn't matter the slider positions, doesn't matter what version. I've tried getting around it with stem+remaster but that just ends up making it sound terrible. Is there just no way to do instrumental extends anymore?

I've had the same problem with other songs that had no voice/vocals in them whatsoever as well, it just adds in its own lyrics out of nowhere. I normally don't have many problems when working on my songs, but this has been driving me crazy for several weeks.

[small bridge]

I hate search feature so much Ty I thought so!

[neat nexus]

Did you try dropping to v5 to see if that helps?

[fiery prairie]

yeah, I tried that and 4.5/4.5+, it seems I pretty much have to add my own lyrics or it will just do it itself no matter what

[neat nexus]

Wait are you not adding the lyrics that resume from when you are extending?

[fiery prairie]

I was not intending to add any lyrics, and there were no lyrics there in the part I was extending from

[neat nexus]

I don't understand. Was it an instrumental?

[fiery prairie]

the first part opens up with a spoken line, then goes instrumental for a bit, and I wanted to extend the instrumental, but it doesn't allow me to

but I've had the same issue with 100% instrumental ones as well

it has been an ongoing issue since 5.5 released

[neat nexus]

You're not wrong there.

[fiery prairie]

don't get me wrong, I actually really like 5.5 for the most part

just been having a lot of issues with this specific thing

[neat nexus]

No, no...I use v5.5 for everything but yes, it's a pain point. I feel it too. I used to be able to work around it easier but lately it's just been a pain.

[fiery prairie]

yeah, I decided to just go ahead and write some lyrics for that part so it doesn't try to make up its own

the hallucinated vocals actually don't sound too bad on some of the extends, if the lyrics made sense lol

[neat nexus]

Depending on your instrumental, you could try...

`(...)

[Instrumental]

(insert rest of the lyrics or wherever you're going here)`

[fiery prairie]

yeah, I tried that and I also tried

[Vocals end]

[Instrumental begins]

just to see if being more specific might help, but I didn't notice a change

it is usually pretty good about doing what I want, extend has just been off for a while

and I gave up on replace a long time ago lol

[neat nexus]

Unfortunately if I'm mucking about with Replace, then something has gone very wrong. lol

[fiery prairie]

I appreciate the help, I think I can find a way to work around it, I was just getting a bit frustrated and was starting to wonder if there was some specific thing I was missing, but I think it is just something that will have to get fixed in time

and if worst comes to worse I will just move on to a new project and revisit this one later

[analog drift]

A few questions if anyone can help
1st- shameful of me (I know 🙈😂) but when I started SUNO I cheaped out and instead of getting a subscription I instead opened up a million emails for the free credits…
Now the issue is I bought the subscription but can’t figure out which email I used for the actual current account no matter which buttons I press 😂
Can someone help direct me?

2nd- I’m trying to figure out how to nail that sweet spot for a cover using another voice
Even if the voice is heavy rock but the original is let’s say acoustic- it will sound super similar without changing the vibe of the song to my new voice… how do I correct that?

Thx 🙏

[half dragon]

To see your email on the account - click on your Avatar > Account > You'll see the email and other account things.

[analog drift]

Oh dang that was simple😂
I feel dumb lol
I’m pretty sure I checked that tab 100 times and never say that before lol

Thx!

[final hazel]

Also.. It's officially not allowed to have more than 1 free account. 😉 Just so you know.

[analog drift]

I know 🙈🙈🙈
That was me being dumb🫣😂

[hoary sandal]

If I move all my uploads I used for my custom model in a Workplace
Will it move them away from my creations list I WANT TO SEE without affecting the custom model?

Cuz I really need to hide those 24 uploads

Its driving me nuts to scroll through them right now

[quick wedge]

If you needed help you could’ve asked as we wouldn’t mind helping you out. It’s a learning process around Suno but you’ll be okay 🤗

[half dragon]

Not sure what it means by "affecting the custom model" and if they are currently in your workspace you can move the uploads to another workspace...
Then you won't see them while you're working on creating songs.

[hoary sandal]

They're not in any workplace atm

I'm just wondering if I still need them after the model was made

Because if I still need them for the custom model I made to work, then I'd rather have them hid away in a workplace until a exclusion filter is made for uploads lol

[honest comet]

Hi group. I want to use Replace on the web version of 5.5 to regenerate an intro. How exactly do you select the section you want to replace? With the blue triangles above the waveform? Or with the white brackets IN the waveform? Or some other way?

No matter how I try to select the intro section, it winds up “replacing” a section that comes later, and deletes part of what comes before.
Very strange!

[frozen dagger]

I moved 24 clips i used to create a custom model to it's own workspace. it seems fine to move them. just can not delete them.

[hoary sandal]

That's fine, as long as I can move them out of my hair

I had like 60+

(Made 3-4 models)

[polar spade]

"Moving" a file from Library, to a Workspace folder should only place a copy (or rather , a visual link) in that folder that will open the same file.

If its in your Library , I think its always there.

Someone can check me on that.

[polar spade]

And they disappeared from Library?

[hoary sandal]

Oh I'm not worried about the library

I never touch the library anyways

I'm talking inside where I generate songs

I needed them out of that place and hidden away somewhere

[frozen dagger]

Moved from (i believe) My Workspace to New Workspace. that's all. He just wants to clear the clutter.

[hoary sandal]

Cuz I was doing covers and it was tedious to scroll through

[polar spade]

OK. Im getting senile. What's the "Creation" list then. Or rather , where is it.

Do we mean the main Workspace list? Yes, that can be whittled down to nothing. Is just another workspace folder. Right. I guess that's it.

[hoary sandal]

I was making covers of songs in my workplace tab

But the problem was

I made a 3-4 custom models

Which ended up being like

60+ uploads

And it was like

A chore to find songs I wanted to do covers of

Because I had do a long ass scroll to get there

[polar spade]

Yes. Got it. I worked it out that the "creation" page where the files were showing was the main My Workspace list. I was just thrown a bit by "Creation page" but its all the same thing. 👍 Thanks for the visual.

[sleek stump]

As far as the uploaded model tracks goes, its OK to just send them to the trashcan, they will be out of your hair and out of your library and still not affect the custom model.

[smoky heart]

I'm trying to use an old track of mine that was released through streaming channels and I get the "Uploaded audio matches existing work of art.
Please upload a different audio file." Is there a way to assert that I am the rights holder in order to use that track in Suno?

[timid lion]

@spring gust

treble, mosh-core big-room ultra-tritone-glitch haunt-halftone-electronica mosh-grave-edm grave-wave electro-mashup breakbeat-progressive-electro ghast-phonk, trap-minimal dark-break-house, trap-house goa-break dubwub-horror, Rap lyrics in drill-electro, bounce-minimal-house, sections utilize edm form adaptations sidechaining transitions, lyrics rapped  pan-echoing reverb, chopper drill rap, as instruments, bass should be shallow and suppressed, hard horror distorted ethereal elements, progressive electro buildup tuned to section melody, rapped vocal style, some lines metal-scream hauntly, then shift to immersive breakdown, sublines have break transition cliffs, Meter:  4/4 initial meter, Intro pads include 3+3+2 eighth-note polyrhythmic displacement, midsections with intense polyrhythmic claps employ a metric modulation from 4/4 to 7/8 for feel, super section resolves to 4/4 as 2nd midsection metric modulation equating measure length, ‑rock, ‑orchestral, ‑opera, ‑acapella

[spring gust]

Copying this right away, shiii

[tacit axle]

Anyone have any prompting tips to get Suno to avoid big swings in dynamics? Like if you have a first verse that’s guitar, strings, and vocals, it seems to reallllly want to bring in a full band for either the chorus or the 2nd verse. Are there any helpful keywords to prevent this in favor of a consistent energy level throughout?

[empty umbra]

Me too. Cant upload anything.

[worthy kraken]

Perhaps, someone has any kind of advice for editting out small errors? Like mispronounced word, scream that suddenly cuts, or even just bad trasnition. Basicly talking about few seconds moments.

When i use standard or smart replacement, it always adds a lot of extra random junk and basicly ignores my prompt

When i use she short option, i am sometimes getting the result, but it is unsusable due to bad transition that always ends up making a huge difference in volume

[worthy kraken]

I also have a feeling that suno lately just ignores most of your prompts. Example - what do you think it does here? Of course, continues music for several more minutes.... There is really no way to give it any kind of directions anymore? It just will be doing its own thing?

[hollow crater]

You might have better luck separating the stems and editing the vocal track in the legacy editor.

[worthy kraken]

Sadly, not an option, because not only it costs a lot, but always generates stems that are not the same as the track, adding even more errors that need fixing.

But i might try legacy editor, forgot about it, thanks.

[quiet solar]

There isn't an automatic route, it's something you'd have to discuss with the Suno team directly.

I'd start by dropping a line to [e-mail] and asking what your options are.

[quiet solar]

I've never got on with Replace or the Basic/Legacy Editors, personally.

My preference for a highly fixable error would be to Cover the song with the error and correct it that way. If I'm that attached to the track that I don't want it reinterpreted, then...

Prior to 5.5, I used Extend or Cover on high audio influence to create vocal samples from within a similar-sounding track, then did vocal/instrumental splits (which tend to be less problematic than full instrument splits) and made any necessary edits/replacements in Studio or an external DAW.

With 5.5 bringing in a lot more production effects into the music, that approach became tougher, as exported stem splits are a lot less clean outside of Studio. Extend is also a lot more unpredictable than it used to be. So my workflow today would be fundamentally the same, but using Covers and, if at all humanly possible, making the change in Studio.

[remote ivy]

i would have said change lyric w/ phonetic and remaster...and it has worked for me, but the cost of remaster was always losing to much of the original dynamic sound...so I always ended up unpublishing eventually...remaster can get rid of glitches too...and I have stuck w/ some of those remastered versions. I don't think it matters so much if your just on suno.

[smoky heart]

Thank you!

[neat nexus]

@bronze spruce The only place for account related issues on this Discord is #ticket-help please see that for how to proceed. Thanks.

[bronze spruce]

Thank you

[desert vector]

so even when we exclude all of these, suno still brings in drums vocals etc lol, anyway to precisely do what we say or is it just always hit n miss?

[sleek stump]

Anything you prompt or tag is just a suggestion. Generative models like Suno are not word processors for Music, they lean heavily on the norms of the training they received, and very often disregard any specific direction a user gives it. There are ways to nudge Suno into doing off the rack things, but it takes a lot of experimentation and researching styles and genre, and which ones fuse together to get what you want.
It is widely known that the exclude window hardly ever works, I have just stopped attempting to use it completely in my own work, as have many other users. If you dont want certain elements in a song, find the genre or sub genre that would most closely match what you are trying to achieve, then you will have better success with style and tag direction overall.

Some users have found great success in heavily technical prompting generated from AI agents, but even those are not guaranteed to work.

[hoary sandal]

So hey, wanted to ask a question

Is there a reason I can't do a Persona of a suno original song after its been downloaded, used in audacity, and reuploaded?

Because I kind of needed this song to have a persona lol

[desert vector]

ok thanks

[sleek stump]

Yep, it because of the vocal fraud preventors.

Its also why, when you upload your own vocal recording to a Voice, it verifies your voice via microphone.

[hoary sandal]

I guess I can still use the base version then I made before I downloaded and edited right?

[sleek stump]

Yes

[past phoenix]

It would be nice to get a clarification on that. Because I have a cover of a fully public domain compo (1893) based on a 1923 recording itself in the public domain as well. Plus quite some covers of Shchedryk from Suno's own official contest. Meaning that, on Suno, yes there are a few thousands Shchedryk covers endorsed by Suno, from a public domain compo and recording.

All I found is this in Suno's TOS, which doesn't seem to prevent covers of public domain material.
I've asked GPT to read the TOS, he went to the same paragraph, and interpreted it just like I did.
Basically we have all of these concerning public domain content: consent, permission (implied for public domain stuff), power and/or authority (since it's legal for everyone) [necessary to submit and use].

I'm asking because I'm witnessing people's accounts being locked, sometimes for legit reasons, sometimes for use of sample packs for which they quite obviously got permission to use from the company they came from.

[sleek stump]

The problem most people miss here is that you are giving Suno a sublicense for something you cant legally give a license for. That is a legal stickign point.

[past phoenix]

but that's not what the TOS says
I'm fine with that, only it has to properly state it in the TOS

[sleek stump]

It says it in the pic yopu posted above.

[past phoenix]

no it does not

[sleek stump]

The allow us to use part is the problematic statement. In the past, Suno, when trying to upload any media, popped up a window that stated you must have "exclusive" rights and grant Suno a license to use what you are uploading.

[past phoenix]

could be, but it's the TOS that's binding us

[sleek stump]

I dont make the rules, or write the TOS... if you really want to debate this, it needs to be with Suno directly.

Mailing support would be the first thing to do.

Also, GPT will give you really bad legal advise, take anything it says with a spoon of caution.

[past phoenix]

you have, or have obtained, all rights, licenses, consents, permissions, power and/or authority necessary to submit and use (and allow us to use) such Submission in connection with the Service

do I have:
-all rights? no. Public domain, no one has.
-licenses? no. They can't exist because are not needed
-consent or permission? implied for public domain (the author cannot not consent anymore)
-power or authority? YES: public domain - everyone has

[past phoenix]

but if you're asking users to report all public domain songs, I think you would also want to know

[past phoenix]

I do, it's just a second check. But the basis on a legal mention like that is just interpretation of a proper sentence. The "and/or" is very important here (and it can't even be just "and" as you generally can't get ALL of those)

[sleek stump]

Sorry, who's asking what?

[past phoenix]

it's in the post I originally respond to

[sleek stump]

Ah I see... Volunteer role members have been encouraged to promote reporting of copyrighted materials or inappropriate material.

[past phoenix]

btw the new pop up window doesn't say that anymore

[past phoenix]

no I meant this

[sleek stump]

I see Dasvon typing, I'll let him comment.

[past phoenix]

I mean I'm fine with Suno mistakingly blocking public domain or even private material (I got the message for a simple, homemade sine sweep the other day) as copyrighted, it happens, and that's a forgivable technical problem. Only if it's starting to block accounts for this reason, that's a completely different problem

[sleek stump]

How have you determined this is actually whats happening?

[past phoenix]

I've just witnessed it in another room. The guy said he was using sample packs and that it was his own material using those packs.

[sleek stump]

There are many reasons I would be skeptical of statements like that...

Also, any account actions Suno takes is never discussed anywhere.

[past phoenix]

it totally makes sense technically, lots of false flags - try to upload this to suno & see

[sleek stump]

Again, anything of any legal nature needs to be taken up with Suno directly. If yopu are concerned abotu things that you are hearing on the internet, then asking the source is the best course, or consult with an attorney.

[past phoenix]

so you don't think it should be public knowledge, and benefit everyone, to know if Suno really forbids covers of public domain songs, but rather that it should be learnt from private discussions with Suno?

[sleek stump]

None of teh volunteer staff here or team members can or will discuss legal topics or account stuff..

[past phoenix]

well I'll send an email. Can I post the reply here if I ever get one?

[sleek stump]

Thats really for Suno to determine. Its all legal stuff that is far above any of our purview.

[neat nexus]

I think I may have discovered an Easter egg?

If you go to ReMi and put in Alien chant what language do you get?

[half dragon]

French

[neat nexus]

Same as me. It's consistent.

[half dragon]

Interesting.

[past phoenix]

because chant is a french word, most likely

actually, alien is old french too
interesting, because in today's french alien has been re-imported and only means E.T.
but it did mean foreign in old french

[half dragon]

yeah Chant alone will always be French in ReMi -- not Alien
anything with Chant morelikely get French as well.

[past phoenix]

I use chant quite a lot in style prompts. But as "bulgarian chant" (which is very specific & suno rarely does it properly)

[desert vector]

is there any relation to listening to a song before its finished generating, that itll be worsE?

say if i click create, i keep listening to ideas soon as theyre ready after 15-20 seconds or so whilst still generating

but could they be better if i wait?

[half dragon]

Myth as far as I know. I do that all the time, playing while it's generating.
No one knows for sure but I highly doubt it.

[sleek stump]

Short answer is no. it does not affect the track generation in any way.
It can take longer to gen a track fully than it takes to play it, so yopu may get puases or stutters in teh playback, which go away once rendering is completed.

[chrome rampart]

So how does a person get verified on suno?

[sleek stump]

verified?

[polar spade]

To avoid being censored on plausible uploadable songs? There is no such thing.

I havent heard of anything to get someone's songs whitelisted.

As for someone saying they are you. User names are pretty interchangeable at the moment. I suppose if any issue, we verify that we are the real one by our email and/or phone and billing info.

[loud kestrel]

Is the lyric replacement in Editor just completely broken?

[polar spade]

Completely? No. I mean, to compare... I replaced two lines and a couple of words at the same time in a verse yesterday. And it blended smoothly in, with nothing more to do......But .......it did mispronounce one of the words.

[loud kestrel]

The gerbils must be underfed today, I guess. I've tried replacing "saw me first" with "looked at me" about 30 times today, and it hasn't given me anything even remotely right.

[polar spade]

The most consistent error is it seems to try to sing a word or two that exists somewhere in the chain. So if the original was "spot", and you change to "keep", it might get keep, but sometimes then revert back to a "sprakt" kind of word. As it it cant figure out which its suppose to say.

So I get 98% there OK. And then i just save the file, maybe even copy it again to try to start with a fresh "original" and thrn just change the one word. Thats been my workaround.

Pop

[loud kestrel]

Thanks, I'll give that a shot.

[chrome rampart]

Some accounts have an orange check mark

[sleek stump]

Where are you seeing this?

[sleek stump]

Interesting, never seen those before... wonder if its from a recent change.

[polar spade]

Hmmmm... maybe they attached a screenshot of the actual text. Hmmmm i have no idea.

Not one of the random top 10.

[chrome rampart]

Alright

[sleek stump]

Question has been posed in our back channel

[chrome rampart]

Thank you

[sleek stump]

It appears that these artists with checkmarks are official Suno Collaborators. They are not "verfied" they designate a Suno partner, is what we believe. Still waiting on official words on that but this seems to be what they are.. Each one of these artists have been featured for remix contests.

[chrome rampart]

Ahh ok, I would love to be able to verify my account as an artist blue check would be fine. But maybe later down the line it would happen.

[sleek stump]

Many of us would, but thats not a thing.

[spark crypt]

Weird problem. When I try to upload recording of myself singing, it's saying it's existing work. I'm definitely not singing anything "real"

[sleek stump]

Sing it directly into Suno

[rough depot]

Still ugh

https://suno.com/s/6hDJ6azyC2rZSE0d

Still getting gsng vocals. Ffs

[quiet solar]

It's worth bearing in mind there's another angle to this, which is that specific recordings of public domain music have copyright protection that's separate to the underlying song.

So for example, if the Berlin Orchestra records a song by Mozart for an album, you could play your own cover and not be affected by any copyright claim against the underlying song - but not be allowed to just upload their recording on the basis the original song is public domain, since you don't have the right to reproduce it and grant Suno permission to do the same, and that's necessary for a Suno upload. And it's very likely that the Berlin Orchestra version will have been uploaded into copyright databases

It would be a notable additional technical headache for Suno to distinguish between "I recognise this sound profile as something in the copyright database, but hold on - the underlying melody matches a song that's in the public domain, and the specific version is a recording that is no longer under copyright protection itself."

Not least because there isn't, to the best of my knowledge, a trustworthy universal database of "This music and specific recordings are all 100% guaranteed to be in the public domain in every country in the world". In the absence of that, facilitating an automatic technical solution on Suno's end is essentially impossible. So annoying though it may be when you did everything right, Suno will still block any public domain works that have a recording included in the copyright databases it references for copyright checking.

[old ginkgo]

Sorry to ask, but what setting is it that one humms to get a song beat/influence? Im still learnin'

[torn badge]

Hello Suno AI Team,

I hope you are doing well.

I want to make a humble request regarding the cover song upload feature on Suno AI. Earlier, we were able to upload songs and create AI cover versions easily, but now almost every upload is showing copyright restrictions, even for short 1–2 minute clips.

Many users in India mainly use Suno AI for creating cover songs, remixes, and creative experimental versions for personal and non-commercial use. This feature was one of the biggest reasons why people loved using Suno AI.

We completely understand the importance of copyright protection, but please do not remove or fully restrict this feature. If possible, kindly allow limited-duration uploads or provide some flexible option for creators and remix artists.

Suno AI has become a favorite platform for many music creators, DJs, and remix artists here, and we truly appreciate your amazing work.

Please consider this request.

Thank you so much for your support and understanding.

Best regards,
Chaand

[cosmic fog]

Not so sure about that:

[quiet solar]

On desktop, under Create choose +Audio and then Record.
On the mobile create screen, there should be an icon of a microphone near the bottom. Tap that.

Happy humming. 🙂

[polar spade]

Once done, Suno now throws up a style analysis of what it thinks you just did. So if you hummm or scat, it may say that "this is an a cappella vocal humming...." , and the cover of that will have humming and scatting. So, review the analysis or write your own prompts to try for musical instruments instead of your voice, if you prefer.

[past phoenix]

Sure, the recording itself must also be either custom/private or in the public domain, I've addressed that.
and yes, recognizing copyrighted material is a nightmare, with or without this (I mean apparently right now it blocks pretty much anything). But I'm not saying that Suno can't block any upload it wants, I'm just saying that whatever it allows us to upload, and is fully in the public domain, should in no way jeopardize an account since it follows the TOS.

And I actually think that it's way easier to have a database of public domain recordings, as there aren't "that many" public domain recordings out there, it's nowhere near the 30+ millions new songs that pop up on Spotify each year (mostly thanks to AI itself lol). Because yeah, Suno also now has to fight against what it iself generated and comes back as copyrighted (while it faces the problem that it may randomly generate very similar results for another user).

[brave cedar]

hi guys, getting this warning when making voices Uploaded audio matches existing work of art.
Please upload a different audio file. beenm using nursery rhymes and still getting it but even when i change the words im still getting the warning

[final hazel]

Do you have copyright of the audio?

[brave cedar]

for the record im recording using a mic directly into audacity then removing background noise

[final hazel]

Do you have copyright of the melody? Is it an existing melody?

[brave cedar]

i mean its like twinkle twinkle little star.... nobody has the copyright for it

[final hazel]

Well.. yes and no.. If I make a record with twinkle twinkle little star, and if I release it, I have copyright of that version.

[brave cedar]

pretty hard to come up with a melody thats never been sung before just for it to learn my voice... im not a musician lol

thats why imn using nursery rhymes lol

maybe you can give me tips... excuse my crappy voice lol

i mean if you have one... i even changed the wording but it didnt help

[final hazel]

I would post it in #🪲┃bugs-web
There a Suno dev is actively learning and helping people how to improve their copyright system. Maybe they can give you an answer.

[brave cedar]

sure will let em know my issue, thank

[past phoenix]

don't worry it's not you, they've just messed up and pretty much everything is now recognized as copyrighted. I mean, if this sine sweep is, everything is

if what suno detected as protected is the underlying melody, it's not gonna work better

[brave cedar]

lol i struggle xD its embaressing enough singing out loud when i try to riff diff lyrics i stumble all over the place, like even this failed

[brave cedar]

dang my headphones were too loud lol

[past phoenix]

yeah
"warning" lol

(I wanted to know what kind of compression was applied on uploads, a sine sweep was handy for that. Fortunately white noise isn't flagged as copyrighted yet)

[polar spade]

Im saying this more as levity, but from a serious position as well. I hope they never allow another derivation of "Jingle Bells" to be uploaded or trained on. Lol.

I have so much trouble having to prompt AROUND those dang three-notes that are clearly Suno throwing in a Jingle Bell vibe, everytime I prompt "Christmas", along with the fact, I dont need the literal ching ching ching of Jingle bells beginning and ending every song! 😁

[past phoenix]

ah, I don't know, I've actually never had a jingle bells vibe from Suno. In fact we couldn't even have bells at all from Suno, I haven't heard one in the carol of the bells contest(?)

(and I've really tried, to get bells, in other songs - why can't Suno do that?)

[polar spade]

I think that probably was more choral (and a waltz), than Christmas.

[past phoenix]

I just love the pig latin chorals that 5.0 was doing (and that 5.5 can still do, but does much more rarely now, when you ask for choirs).

And you can't really force those in, if you write a real text in latin, it's gonna sing it properly, with a clear articulation that's very different from the "choir as instrument" that it does

[worthy kraken]

I don't have the button to acsess song editor anymore, anyone else?

[polar spade]

Look further down the menu , now. I think near the bottom. Or you mean the actual button. Yea, maybe not there.

[past phoenix]

yeah that sounds more like a tamborine which Suno probably uses to mimic the actual jingle bells (that it probably can't do - I mean who needs them?)

[worthy kraken]

Is there any other way to launch it now?

What is even going on, bugs everywhere, suno falling apart...

[polar spade]

OK. Outside of that hair splitting. I mean the ching ching ching more than the authenticity of bells. You got me there.

[past phoenix]

yeah but you gave me hope that Suno could now do bells (& thus maybe the good ones too). I'm giving it a try now, maybe 5.5 now can

(I mean church bells, for new age & stuff)

[polar spade]

Well... I mean... you still have a Remix button and the option for editor is still there in menu... Just relocated.

Took me awhile to find Move to Workspace as well.

[worthy kraken]

Ok, thank you! I thought it was removed

[past phoenix]

btw I was told this wasn't a bug, but I'm pretty sure they just forgot to rename "voice" to "create voice" when they moved stuff around. Otherwise it's "remix -> voice", that makes no sense.

when I searched for the voice creation, I thought that this one was to add a voice to a track

[polar spade]

Yea. I never know if they are just test marketing, or if this is their final decision

[past phoenix]

well: 5.5 still can't do church bells I'm afraid. It maps to the same lullaby ones you're talking of
https://suno.com/s/u3CWWSNm6MqJm0QS

(and I tried lol)

[polar spade]

Keep in mind.... if it helps.... sound effects do a better just of at least creating different types of bell sounds.

Get a nice sound effect then extend off of it for your song? One option.

[past phoenix]

yeah possibly. But have you ever heard proper bells in any song on suno?

[polar spade]

Like the deathknell from Hells Bells. Not sure that I have.

[past phoenix]

it's still odd that it's not in its list of instruments while it can do all the weird ones. I mean I live surrounded by churches and I hear bells all the time here

searching for songs with bells I hear a lot of steel drum - that I've used in the past, suno does pretty well

interesting, it definitely tries, I mean the sound sucks but it's definitely an attempt so it's something suno does recognize (assuming it's not an upload)

[polar spade]

No not upload. Just prompted deathknell church bell

[past phoenix]

ah I hadn't seen, you used "sounds". Ok

it's never gonna do church bell with melodies then

(I listened to a couple of songs that had church bells in their styles, none of them had any)

well if suno can't even do good bells with "sounds", that's most likely why it's not doing them at all as instruments. It's odd because it's really not the most complex thing to synthesize (especially in the frequency domain)

yeah it's the same kind of glockenspiel it uses for all lullaby songs

[polar spade]

OK. Well. Best get my breakfast in before they go to lunch.

[worthy kraken]

Gods, this is infuriating. Somehow fixed mode now does not generate new replacement of the song, but instead ignores changes to lyrics and each time makes carbon copy of the original.
And if you use any other (smart on standard), instead of selected space and written lyrics, it will add up around a minute more of the rest of the song, again ignoring the prompt and selected zone for changing.

It feels like anything i do is doomed to fail because of bugs. I see no way to regenerate certain section, to just have new version of that section and nothing else.

Did anyone else encountered this? Did you found the way to solve this? Any way at all, but without covering entire song, just to change a single section?

[past phoenix]

my experience: it's one of the tools I tried at the beginning, wasted lots of credits on, for results that were always sub-par, and that I don't touch anymore (I just do full covers now). It's like "song extend" & "remaster", they're features on the paper, but they're not good (or.. were still not good the last time I tried)

And I mean, even without the bugs, when Suno is asked to insert something to an existing song, it generally sounds out of place, like a band aid

(and yeah it sucks when you got THE render that you wanna keep. But fortunately most 5.5 renders are good (aside from the zipper noise) whereas it was a lot more random with 5.0)

[polar spade]

Two days ago, I change a couple of lines in a verse as well as three specific words, at the same time, using Fixed. All but one word (it garbled) were properly replaced and the blend was flawless.

Have you used Classic but reduce the blend parameter down to zero? I dont know if that is ultimately the same as Fixed.

[past phoenix]

can I listen? (just to see if I can spot the blend)

[worthy kraken]

Where can i selected the blend parameter?

Unfortunatly i cannot change lyrics. But i tried to add vocalises between it, and add commas and dots to change pacing. All were ignored

[polar spade]

Ummm... hmmm... when choosing Classic it should show a popup of measures which you can expand or narrow to zero. Maybe a cog or the word "advanced". Ill go look.

[worthy kraken]

Thank you. So i should make it zero?

It can only be brought down to one

[polar spade]

I havent tested zero in awhile. Just saying it might function similar to Fixed.

Only to 1.... ok... sorry, then, it will vary the edges (or hopefully not) then when it blends.

[supple moon]

So, I wanted to a different version of a hymn.
I double and triple checked that it was public domain. It is.
But Suno keeps saying it’s copyright.
Is there anything I can do?

[mental parrot]

In my experience it tends to block just about anything published, regardless of its copywrite status or age.

[worthy kraken]

Damn, it almost worked. Still a really bad initial transition, but at least no more extra minute ahead. But, sadly, it still generates around 10-15 seconds on extra lyrics that is not in the prompt... Seems like zero could solve it, if it were possible.

[polar spade]

All a guess on my point. But glad it did better.

[teal jay]

I'm not sure if this is the correct channel for this query.....I'm reasonably confident that "general chat"..................isn't ;0).....but has anyone else noticed using the "My Voice" option can start off pretty good in a track but it gets more and more like the source vocal as the track progresses? I wish there were a "replace vocal" option too as covering my songs that I'm already happy with to get the vocal sound I want added is problematic..... especially since in V5.5 the cover version mainly insists on adding additional instrumental sections I neither want ot need

[sleek stump]

This has been a widely noted observation. I have heard many track where the users voice was clear but transitioned into some other voice. What I have been seeing a pattern of this happening is when Suno takes the vocal out of the users vocal range, say for a falsetto, and after that it’s just not it.

[wet maple]

Did Suno change something with audio uploads? I'm getting a constant error for different vocal tracks when trying to create a voice. Even singing gibberish is still getting the "audio matches existing art..." error

[sleek stump]

Please post in the Bugs-web channel, this has been an issue over the last day or so.

[wet maple]

TY Ginger will do

[strange oriole]

Yeah same, I guess they tightened the detection or something

[wet maple]

More like over-tightened. I can literally sing "blah blah....I just farted...blah blah blah.....it stinks." and it still gets flagged as "upload matches existing art.."

[strange oriole]

Yeah hope they reverse that

[sleek stump]

Folks, its a known issue, if you have any more detailed input that would help Developers with the issue, post them in https://discord.com/channels/1069381916492562582/1082775596767117312, and please refrain from just venting here. Thanks

[polar spade]

@teal jay

What Ginger said makes sense. Being autoregressive for the most of the generative process, it basses much of whats next on whats immediately, and further back , has come before it.

Adding to that, the less you have regarding prompts, lyrics,and tags ,probably the more tokens it can allocate to maintain the characteristics of your voice throughout the song. The more you have, Suno may forfeit a bit of the vocal vibe in order to just finish the song length

[zenith coyote]

I'm making a sultry duet and I can't get suno to respect the female and male lyrics. Any tips on getting suno to not have a male sing about his black dress?

[azure crane]

What’s happening?

[sleek stump]

Wait a few and try again? Its the internet, and services can hiccup . If its a persistent issue over hours then post in one of the proper bug channels.

[sleek stump]

You might want to do a search in this channel, duets have been routinely discussed. You can also investigate the Workflows channel, there are many how-to for duets.

[past phoenix]

didn't even know there was a workflow channel. I see 3/4 of the channels here are hidden by default, that's weird

[final hazel]

It's more connected with onboarding and what people choose to be interested in when they join the server. But you can always turn them all on via de server settings:

[past phoenix]

ah, I must have missed that

[gaunt dew]

Hi guys, i'm trying to make spanish music but the vocals are like rushed or unnatural placed in some spots, how can i make sure this doesn't happen and the song actually sounds native?

[bronze island]

I think you have to write it phonetically so that it has an accent.

[bronze island]

3/4 time signature?

[cold zenith]

My covers are not starting from the beginning....

[spark turtle]

@final hazel long time no see !

[final hazel]

Still lurking around.. mostly ninja style 🥷

[remote ivy]

do a cover w/ the lyrics removed w/ the persona( w/out a persona may also work, i've only done w/ on both covers), then pop the lyrics back in and cover again...seems to work for me for covers..im sure there are things you can do additionally suppress hallucinating old lyrics..but I haven't needed to do that myself

[old ginkgo]

I managed to get the sound I wanted for the world cup cover contest but now it doesnt recognise it as a cover so I cant enter it what do I do?

[remote ivy]

did you trace it all the way back to see if you in fact broke the cover chain by using something other than explicit cover option?

[old ginkgo]

I think I cropped a 7m error song 🙄

[remote ivy]

legacy or new editor?

also i thought they fixed that issue, at least in the legacy editor and at least for contest songs

definitely breaks the cover chain for non-contest songs

[old ginkgo]

Im in pro and I just used crop 🤷‍♂️ sorry, Im still new at this

[remote ivy]

did you try to publish it or you just don't see the remix of after crop?

[old ginkgo]

It doesnt say cover by the title, I can try to publish

[remote ivy]

i think for contest songs you may see the option when you try to publish it....seems they have special handling for them..i wanna recall i croppped and published for the bloomer contest which was middle april?

[old ginkgo]

Ill try and see what it does

Nope its broken, it even has the download option 🙄 ah well, sad part was its really good LoL

[frozen dagger]

last resort, since it was good, could try using the cropped version to create custom model (same song, at leat 6 times) turn the cropped song into a Voice as well. Then try covering the contest song using the Voice and Custom Model, along with your lyrics and style prompts.
Won't be the same, but could help try to recreate the magic.

[old ginkgo]

Cool!! Ill try that! 😁

[timid lion]

The Circuit

anhemitonic wuyin config, meter: 5/7 high energy downtempo memphis-cloud-drill exotic-club-latin dance-anthemic disstrack rap-artist-production-ensemble collab dance-party motif through-composed upon the track throughout, beach-hype-party-trap fusion, sapphic-hardcore-bossgirl macabre-rap with swelling ostinati development and dissolving meter with dynamic rubato-syncopation beat escaped subgrids, reesebass-squarewave-subbass envelopes with bi-traversal-bass-panning, wide-bright hi-hats and snare configuration, percussion is complex and layered with stacks and incorporate polyphonic dense deep-ostinati, reverb conversational to chest-voice-vocals, ‑yodel, ‑cuts, ‑simple arrangement, ‑slow, ‑gospel, ‑arabic, ‑indian, ‑pop-sheen, ‑croon

@thick flax

[remote ivy]

oh you know what, i don't use crop, i use the remove option, I'm just retesting it for sanity at the moment

[remote ivy]

yeah sorry its remove that i use...you won't see the remix of link in your workspace, but it's still linked to the contest as you will see the submit option when you go to publish...i 've mostly used the legacy editor of recent so , i won't claim the new editor isn't broken. if you can crop, then you can likely make the song work w/ multiple removes instead

[old ginkgo]

Ok, thank you very much for your time 🙂

[acoustic egret]

hey how come sometimes when i generate remixes the song necver finished generating????

[sick lily]

timestamp? i think there's a limit

[acoustic egret]

Limit of remixes?

[distant escarp]

Here's my Chibi Comics Course on Suno. Any feedback?

[lament blade]

Sometimes certain words and certain prompts don't accept so it doesn't go through.

[polar spade]

Very nice

IIt would scroll by here. Why not post in the workflow section so it can be referred to and seen by others.

[distant escarp]

I'm afraid I don't know where the workflow section is...

[polar spade]

Here is a link. https://discord.com/channels/1069381916492562582/1252267444773326848

You can scroll up to Browse Channels, as well, to click on Workflow channel and others so they show in your list.

[warped tide]

Hello everyone, I'm posting here because I'm generating audio tracks with Suno for my Lofi channel, but I always get the same flow despite changing the BPM. I don't know much about music... Do you have any advice? Thanks in advance.

[polar spade]

You can set the BPM as a manual tempo on the bottom bar, but you might need to click on the track and check the clip info panel on the right to check that the track is set to "Follow Beat" . Or, you might its better to choose "original" to have the track play at its original BPM when uploaded.

[hardy thunder]

Might be a silly question, but where can i select the track length when generating a track? would like a 4/5 min track but getting 2 min tracks currently. can i specify this in the "Styles" Box?

[polar spade]

Setting an actual duration is not an available tool.

Consider that the time you get is an optimal calculation for the best instrumental and vocals possible based on what prompts were fed to Suno.

So you can add fillers such as [break] or [interlude] in Edit UI as one option or actually write more lyrics and try again.

[spark turtle]

Ayyy , that’s Great to hear !!

[polar spade]

Before inferring that its a tool we need, most reasons to hit a certain time is when time is most important. More than creative instrumentation, clever lyrics, expert arrangement.

So, if length is priority, then it does seem the more words or prompts you toss in , the closer you can get to a designated timemark.

Optional trick then is to do extends until you get to about 4 minutes.

[neat nexus]

You can also use tempo names to help with setting the range you’re aiming for.

https://en.wikipedia.org/wiki/Tempo#Approximately_from_the_slowest_to_the_fastest

[bronze island]

Tempo is my favorite musical element 😍😎🎵

[acoustic egret]

Ahh I see

[inland frigate]

Hey everyone — maybe I posted my message in the wrong channel before (I think I posted it in studio-workflows), because I can’t even find my post anymore 😅

Can I ask here a question regarding the impossibility of uploading one of our own tracks into Suno v5.5? The system keeps rejecting it as copyrighted content even though it’s actually our project.

[barren cedar]

Hey, how come when I extend an instrumental even if it's after 3 minutes, it just adds lyrics to it, instead of keeping it instrumental??

[waxen merlin]

Szukam polskojęzycznego mentora 🇵🇱

[quiet solar]

Suno recently made some changes to the copyright identification filter - they're aware it's a bit more aggressive on giving false positives than in the past and it's something the team will look at, but no ETA for a fix as of today.

[dapper quail]

is this normal to generate under 1 min instrumental? wasted so much credits..

[versed creek]

Yoo

[brittle furnace]

hi all, i tried to release a song using suno on tunecore today and immediately got rejected saying that it's detected having unlicensed datasets, but on the other hand since they have partnered a couple of months ago with google flow music, it's the only approved ai tool for music distribution on tunecore at the moment. does anyone know whether we can release our suno songs on tunecore again? can you recommend me which distributor that can accept suno songs? thanks

[lyric lotus]

Aggressive... That's putting it lightly lol You can't even take a song made in Suno, export it, use UVR to stem it and then upload the stems without it triggering the moderation flag now.

Atleast from my understanding from some others. I dont really mess with uploads that much, but from what I'm reading the filter is on wombo mode right now 🤣

[inland frigate]

Hi Kanthric, and thanks for your reply...
You were very helpful, at least giving me a reason.

[inland frigate]

Well, anyway, it seems like a good limit to me... yes, aggressive... in the sense that it leaves you no way out...😅

[polar spade]

Probably no overall reason. If its hallucinating with vocals (gibberish) then its making guesses. Without knowing anything more, that could be from long prompting, prompts that may be contradictory, or even a blank lyrics page. A blank page is supposed to indicate "instrumental" but Suno simply read that as an open canvas sometimes. Using characters as metatags or peacemakers, such as //%$///// or having one or more sliders at 100% or 0% , may nudge Suno to sing or speak as well.

Throw [instrumental] onto the lyrics box as an option. Otherwise, I cant exactly tell.

[echo prawn]

I wanted to cover a military song, but when I uploaded a sample of the original song, it wouldn't let me. This copyright management is a bit of a joke.

🤣

Bro, I don't think I care about American laws, and I don't want to know.

Sometimes the laws there are too nitpicky.

[neat nexus]

Not caring about the law doesn't make them stop applying to companies who are in those legal jurisdictions or change the presented terms of service contract you agree to when using Suno. You don't get to violate them because you dislike them. You either accept and honor the terms or you agree not to use the software.

It's that simple.

[echo prawn]

However, its policy on copyrighted songs is a blanket ban; it doesn't allow anything, not even a song from 1930 that's already in the public domain.

[neat nexus]

Which song? Because last I checked only very few actually were.

[granite basin]

any tips to stop the gasping for air it makes during singing.

[neat nexus]

Break up the lyrics a bit to provide proper airflow for your Tamagotchi singer to breathe?

[granite basin]

have an example of what you mean by breaking it up

I notice that it is much louder when I use my voice file vs using the voice provided by suno

[echo prawn]

It's an American country song called "Will the Circle Be Unbroken" (the original version is indeed in the public domain).

I'm using the original version, not the modern version.

[neat nexus]

Are you sure? Even tiny modifications can tilt you towards a later arrangement and cause issues.

[granite basin]

Any tips to stop Suno from dropping the music and pausing into singer-only lines near the end of a fast heavy-metal song?

I’m trying to keep the groove and energy going all the way to the end. The issue happens around late Verse 2, Coda, or final sections where the band suddenly drops out for a couple vocal lines.

I’m using tags like:
[fast tempo 180-200 BPM]
[Fast neo-classical guitar and electric metal violin]
[Heavy rhythm guitars and double-kick drums remain active]

What tags or wording work best to keep rhythm guitars, bass, and double-kick drums active through the ending without an emotional vocal-only drop?

[quiet solar]

It's more of a model quirk than something you can prompt for Suno to 'not do'.

You can try putting tags like [high energy] or [full band] or other suggestions that intensity shouldn't drop at around the point where Suno usually goes for a change in pace, but it's something Suno is picking up from genre norms, so it might take multiple attempts to get what you want.

[polar spade]

Suno 5.5 is designed to sound more natural, organic and authentic, so it actually tried to add more breathing. So , to remove it on every spot before singing.

There's no good term other than "removing breaths" but you could try "de-breathing". I would have my doubts that would work.

You may need to experiment (i have not) on how often it will add noticeable breaths in other models. 3.5 might not have many at all.

Yes, then we have to cover or maybe just remaster in 4.5, then remaster again to 5.5 to update to a higher model and for better fidelity and hope it does NOT "add" breaths back in.

If you have studio, possibly try FX Remove as the breaths are lighter, to see if it removes some. Again, I doubt it would work that way, but thats what I would test.

[polar spade]

Oh. Is it exaggerating breaths that YOU take in your self-recorded audio? They may be loud enough, as you say , that Suno emphasizes them instead of muting. Physically turning your head can help. We look funny that way, turning on every breath, but it could help. Also, having anything as a pop filter in front of your mic (or even phone) is helpful, as breaths are inhaled and not forced past the filter.

[frozen dagger]

"Chocolate Rain"

[polar spade]

Oh yeah. I remember that video. Truth!

[sacred hare]

Hello, need help very urgently for my work which SUNO
for the past two days your system has stopped accepting my audio files because they are supposedly protected by copyright.
I understand that perfectly well, but you should know that I am the one who protected them. These are my personal compositions, registered with SACEM. I am the sole owner of the rights to these files, as the producer, author, and composer. It is absolutely essential that I be able to work on my audio files, that is the sole reason I subscribed to SUNO—to improve my own compositions, for which I own all the copyrights. Please take the necessary steps urgently, and thank you for your response.
Sincerely,

[frozen dagger]

#🚀┃updates message

[quick wedge]

<@&1082792303976001547>

[eager kelp]

how do u guys write ur hooks?

Hook — a cappella, one piano chord, 808 drops on thread, then i break in the lyrics

what i did in my last one

[versed creek]

Yoo coach Dasvon, help me please.

So the default setting on Suno is 50% weirdness and 50% style influence right?

I don't understand this, should is it better to leave it like that as the default or change it to 0% weirdness and 100% style influence?

[quiet solar]

It would be worth e-mailing Suno Support ([e-mail]) to explain the situation and explore options. You're in a different boat to most people getting copyright flagged as your work is copyrighted and on the copyright databases being checked against, but you're the copyright holder.

As Suno doesn't psychically know that you're the owner, it's working as intended to protect your copyright and block that audio from being uploaded. You'd need to engage with Suno directly to evidence ownership and have that be recognised.

I would say it's likely Suno Support is absolutely deluged at the moment as a result of the copyright detection tweaks. They work on a queue system, so it might be several working days before your request gets to the top of the queue to be picked up.

[quiet solar]

Not Dasvon, but official advice from the Suno devs is to leave the settings at their defaults (they're set there for a reason) unless you're deliberately tweaking them for a specific reason.

Suno can introduce glitches and errors at extreme settings on sliders. Sliders are a feature that's made available for use but not widely tested to ensure it works well at 0/100 type levels.

[versed creek]

I see, so the default is the best? allright then, thanks Kanthric!

[indigo tangle]

i have my "voice" that i add to cover my song that AI helped me with (from my original song I uploaded). but, I can't get my improved voice to be used.
what are the settings to use my voice?

my original song had no drums, but Suno keeps adding. I have exclude styles as drums, percussions -- still shows up. ugh

[lament canopy]

I wasn't sure where to post this, but I've struggled with the newest version of suno to get a particular sound. I basically use bluegrass / old time string band style. The newest version doesn't do it very well. Anyone got tips?

[spiral island]

I've used [steel guitar] and [banjo] meta tags to add strings to bluegrass stuff. Sometimes it works, other times Suno seems to ignore my meta tags.

[lament canopy]

Yeah I've noticed it too. That and almost every upbeat song has an annoying cheesy yell to start it off. I can remove with stems but still. It seems like 5.5 weights your lyrics more heavily than style prompt

[polar spade]

50% , or "default" is the optimal settings for the model you are using while following your prompts , voice, persona or lyrics. 4.5 defaults are not the same as 5.5 defaults.... so, that is in part why Remastering in 5.0 of of a 4.0 song might give two quite different sounding outcomes. Maybe nice, maybe different. So often its good advice to Cover 4.0 as a 5.0 before Remastering simply for the defsult fidelity settings are different for each.

As for sliders, Increase from that default and it will strengthen the styles use of what you're prompting for in proportion to the others in a way that widens the options more as for each style, so trying to stay within the style but varying their intensity or variance, I would best presume.

Increase weirdness, and Suno specifically says it opens the variations of options available (in one of their videos) doesnt stick to a standard zone for the genre or styles, so it may alter things like voice melismo, chord progressions, new instruments and so, to either be interesting or just quirky. Lessen Weirdness towards 0% and its an obvious tightening of creativity, staying pretty tight on what prompts and genre are asking for without taking any chances on too much creativity.

Anything near max at 100% is mostly maxing out the variation of outcome, not maximizing optimization -- and often finds no top limit by glitching or warbling and hallucinations. For a few situations, some come across just fine, so if you get one, write it down.

Think of your volume or potentiometer dial on an amplifier that gets dust crackle if we keep it down to 0 or up to 10. Those are just two settings we probably should not use. So, only Spinal Tap can set dials past 10 [100%]. Most others of us should never even dial it to 9 (90%) .

KANTHRIC said the same thing. I just gave the long version. 🧱 👨‍🔬

[polar spade]

If you dont add it also into stytle prompts, about 7th or 8th position , or 10th or 12th, depending your tastes of only one riff or it throwing instruments in for a full motif , it might not latch onto anything in the metatags either.

If you did that, then who knows, exactly.

[spiral island]

I have tried both ways. i had to generate 3 times to get a Banjo solo to work, had to use the meta tag after the second verse. Right now i can't get a duet to work. even when putting [male vocal] in front of the verse, sometimes it just go female. The last line of the song has [Duet]: for the last line where i want them to do the line together, and it has never taken after 10 tries between yesterday and today. I'm using 4.5 but was thinking of upgrading if i can get it to work properly.

[polar spade]

"......If you did that, then who knows, exactly......"

Make small changes. Test different variables. Putting Blugrass as the first prompts, may only give the perfect nuance, or it may take over the whole piece. I also try that. Sometimes its perfect. Most times, I do about 30 generations of tweaking before getting something 1) works and 2) I like.

Tried that yet?

[polar spade]

Yea. Cheesy yells, yee haw... oh baby for kpop, Mmmm for hip hop.... we are all working on that good workaround that shows promise for NOT getting adlibs or producer tags for 5.5

Not suggesting this, but I get less by tossing in "emo" or "musical" while trying to think of genres that often start on the first lines with no music. Again, dont try that. Im just talking in my sleep.

[lament canopy]

30 generations, yeah I have. I did a lot, i mean a lot, of testing the other night with 5.5. The issue is that 5.5 aggresively fills in the blanks to make a complete song. So in my case, it feels in space that doesn;t really belong in bluegrass. One thing I found out in studio, is that suno inteerupts the mandolin chop and bass notes as drum beats. So suno mimicks the mandolin chops with a drum beat.

[polar spade]

You're right in one way, its been needing lyrics more perhaps in order to get more of the vibe and nuance and emotion or how the singer should sing it (which isn't conveyed well in style prompts)

[quiet solar]

Percussion and an audible beat are so prevalent across so many genres that it's quite challenging to exclude them. Exclude styles doesn't have a particularly large influence - it's almost always more effective to prompt positively as to what you do want.

One somewhat reliable way around it is to include in the style prompt genres and other influences that'll steer Suno's references to places that are typically percussionless - even if they're not directly linked to what you're after.

For example, this prompt is something I worked with another user on and it was reasonably effective at getting percussion-free tracks:

"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"

[versed creek]

I see... so the default setting is truly the most optimal right? thanks coach!

also I face a new problem, every time I create a new song, I realized it's not loud enough, I realize I need to increase the volume, it called mastering right?

But the problem is I don't know what's the most optimal way to do it.

what I do usually just use the -14 LUFS max on cacut and maximize it manually by adding more volume like +5db or +10db.

https://discord.com/channels/1069381916492562582/1464896698508775665

I also found a method here by using some github mastering tools

what do you think is actually the best solution?

[quiet solar]

Duets are a whole other headache. At the core of the challenge is that tags in the lyrics box (or prompts in the style box) are best thought of more as suggestions to Suno than instructions that will be explicitly and consistently followed. Genre influences can overrule them when the model is pulling a song together, making duets where you specify the singers quite challenging.

There are things you can do to help, and routes that will definitely get you to the outcome you're after if you're open to a bit of editing work. Some more info here if it's useful: https://discord.com/channels/1069381916492562582/1466598465969590465

[spiral island]

Awesome, thank you!

[lament canopy]

This was funny, ngl. pre-war family string band” → somehow became orchestral
“solo unaccompanied male sacred vocal” → added piano + drums
“archival field recording of a lone rural male singer” → stomp/clap folk-pop

[polar spade]

You're probably doing just fine, if the outcome is as you want it. Within Suno, I wouldn't know one magic bullet to get the volume or LUFS perfect each time to sound good in a car, headphone and dance club. Ill leave that up to you.

Studio, does have a couple of volume sliders to play around with and may be enough, but an outside aid is just fine.

You might ask again later and get some more precise opinions. Obviously, if working up an album of songs ytou want then to be consistent with each other. Thats a tough row to hoe in Suno. So in that case, outside Mastering is pretty handy

[polar spade]

Its weird what Suno decides it is.... the only helpful thing is we can get a little inside backend info on how Suno is interpreting the combination of our prompts or the audio we upload. I Do find that fascinating.

[lament canopy]

I got another one I can show. I got two completely different song types. The only thing i changed were the lyrics. It read the lyrics and drifted into a darker tune no matter what i did. I changed the lyrics and got the vibe I was looking for. One prompt gave me appalachian lament/ dark ballad. When I changed the lyrics it gave me a more bright and uplifting prompt.

[versed creek]

Yeah I think it's pretty hard to optimize in every type of device, what do you think about BandLab mastering tho? I think it's very legit since I listened to the before & after mastering example, do you recommend it since it's free?

[polar spade]

Others would know about BandLab. Generally theres not much in making recommendations of this or that here. Most seem to be fine in utilizing the tools we need. I have subscription in different places for just one tool or the other. Its just the nature of the beast for what I specifically need. . Everyone has their own goals.

There's enough needed to just help with questions about Suno, in my opinion.

[quiet solar]

Just to pitch in, if people know absolutely nothing about mastering but want to do some basic mastering work on their songs, I've suggested Bandlab before as a low-effort entry point.

It won't be as good as tailored mastering, and it wasn't designed for the quirks of AI-generated songs in terms of audio profile - so it's not a silver bullet, but if it works for you then why not use it?

[brisk nova]

check the updates - there is a form posted regarding your exact problem. It won't fix it, but at least get someone's attendtion

[polar spade]

I leaned to play better at 9 years old than most of my friends , on n my $10 guitar that my uncle gave me, while others bought a $300 guitar. Santa originally gave me a plastic one for Christmas which i taught me Malguena and Alice's Restaurant (haha) After two years, I worked and bought a $175 dollar one at the pawn shop.

I feel using the free and cheap sevices, until you're superior and expert at using every tool, can be the most beneficial and inexpensive thing we can do.

[brisk nova]

Is Adobe Podcast Enhancer worth it? I ran a segment thru it and can't see that it did anything. I'm NOT that good!

[versed creek]

Allright, Thanks for your help coach Dasvon!

[versed creek]

thanks! Yeah I find BandLab very practical, even tho it's just allow 5 song/day😭

[dense nexus]

Is there a way to delete Voices after creating them?

[neat nexus]

Go to the voice you created, then select the three dots menu (hamburger menu?) and select move to trash.

[dense nexus]

Thank you. I knew it had to be in there somewhere.

[neat nexus]

I went nearly insane once trying to find it myself, so I completely understand.

[hoary sandal]

@polar spade

Didn't know if you could see the full or not

[polar spade]

I didnt look close. Not Mature advisory... but Parental advisory. Ya.. my bad

[hoary sandal]

No worries haha

You'll probably like whole playlist when finished

[spice mortar]

Beach Kites

Help. Edited lyrics don’t change . What am I doing wrong?

[hoary sandal]

Anyone able to help me with a cabasa prompt?

(Its like a metallic bead shaker)

I'd even work with a maraca if it works

[polar spade]

Smaller instruments are harder to prompt. Sounds might get one, but can use best if you have Studio.

[polar spade]

That sound like one? ... https://suno.com/s/GSoCtKYHEPUpbh8s Rhythmic Cabasa

[hoary sandal]

Just trying to get a Rattlesnake sound effect

I'm sure my prompts are working, but the metal music just overrides them lol

[polar spade]

Isn't that a vibra-slap or slap-rattler you need?

[hoary sandal]

Well, I generated a rattlesnake sound effect, just editing the song in audacity

[stoic grotto]

Brothers, is there an administrator? Come out and take a look， Why did I only upload a very simple piano, but it looks like this? Is the official responsible person or community administrator here?

[half dragon]

#🚀┃updates message -- If you believe its flagging was error - fill this form with details
https://forms.gle/bwJyV18WM2ynYLwY6

[lost forge]

Do we have a working solution for the piercing / 3-6k hz frequency range noise in v5.5? For me nearly all generations are hurting my ears starting at about 3mins.

[wide temple]

This is the new norm I'm afraid. Like I get it, but I'm not submitting forms and waiting an undefined amount of time in order to possibly have my uploads approved. The point of the app is I use it when I'm feeling creative and have ideas I want to play with when the moment hits. This ruins that.

[coral elk]

Hey everyone, how do you get rid of that high-pitched siren/whine artifact in v5.5? 🚨 It always creeps into the second half of my tracks under the heavy guitars. When I split the stems, it's completely baked into the 'other instruments/guitars' track. I've tried Covering and Remastering, but nothing fixes it. Is there a magic prompt to prevent this from generating, or a specific post-processing tool you use to kill it?style is industrial metalcore

[lost forge]

#📚┃suno-school message
Chiming in here as well. Same question.

[upper solstice]

How to make crowded people sing in Suno AI

[stark python]

“crowd vocals", "gang vocals", "group chant"
Try all of these seperate. Usually best to describe where they sit in the mix as well.

[coral elk]

and u have a solution? 🫣

[lost forge]

No. Did extensive research but came up with nothing but using EQs on the output (earphones or speakers). Remaster Subtle sometimes does this trick, but 1 out of 10.

[coral elk]

i made like 50 remaster, cover, different prompts, on the same track but cant kill out that siren artifact sound

[lost forge]

Yes. Same here. I don't see a solution besides waiting for a new model or a fix from Suno.
Unless some wizard here knows the trick?

[stark python]

Realistically the only way to control that considering it’s baked in is using a DAW and EQ it, or using something like the Z-noise plugin by waves or the X-noise plugin. Not much u can really do to “get rid of it” when it’s cooked in. But there are ways to suppress it after the fact until an update comes out.

[remote ivy]

do remasters from v4.5 to v5.5 tend to introduce beats that weren't there or does it just improve the sound w/out structural changes like beats

[sleek stump]

Remaster isnt supposed to change things around, but some changes do happen because of the nature of the platform.

[pliant kraken]

So I record my music with me singing and an electric or acoustic guitar. I then feed that into Suno, attach my voice, making a cover of it, generate a bunch, then a cover of a cover that gets close and keep going until I get what I'm looking for. Is that the right way to go about it?

[stoic grotto]

I don't understand😅

[near flume]

Try the UnChirp plugin by Zynaptiq.

[polar spade]

Be willing to experiment in creating songs from extensions as well. It was a necessity in 3.5 and 4.0 when we had 2 and 4 minute outputs so we could extend songs past that point. But we found the benefit, once we had 8 minutes, is we could get around some shimmers and some degrading cause by Suno basically pittering out of tokens and tiring from autoregressive drift. As it keeps trying to guess whats next it can create tiny errors that snowball and show up towards the end of outputs.

Extensions, last I heard, reach back 2 minutes into an original or to the beginning of a previous extension. This lessens the amount of info to carry forward. It may benefit for getting segments that are cleaner, as we progress forward with each extension.

Otherwise, not all things are equal. When we prompt for a introspective Folk song, with about 5 prompts and just a vocal and a guitar , its true that the sound will probably be nice and crisp all the way through. This can differ when we ask for some full orchestrated , high distorted, epic hardcore fantasy 4999 character lyrics song and max out the limits of what Suno is designed to do.

That said, yes, we can get noticeable "things" often at 1 minute, 2 minutes, and 4 minutes which then may continue to the end.

If we're looking for a one-and-done single-click output, at least shoot for something in the less than 5 minute range to help improve the odds of less work for Suno trying to predict every next nuance and moment prompted for in our heavily instrumental piece and allowing for allocating enough tokens to cover those challenges before it starts drifting before the output ends.

[half dragon]

Basically to report your case using the form.
They may or may not be resolved but at least helping Suno to understand if it was indeed an issue or not.

[viscid sandal]

any mixing and mastering engineers stress testing suno for proper mixing dynamics either by prompting/method?

[stoic grotto]

Ah, I feel that the officials are unlikely to consider the users， Although there are indeed copyright disputes， But, I just used a demo I wrote myself. After all, there are some things in songs that are universal， This is indeed something that I cannot understand， But I don't know if what I said is right?

[stuck blade]

I'm running into some frustrating audio quality issues with the lead vocal for most of my songs with v5.5. Often as soon as the arrangement gets dense (sub-bass, wide harmonies, cymbal washes), the lead vocal audio degrades.
Here are the specific issues I'm hearing:
• The "Underwater" Phaser: The vocal loses its solid center and sounds like it’s washing around through a phasey, watery filter.
• Metallic Sheen: When the vocalist hits a louder or strained note, it gets a harsh, robotic/vocoder-like ringing on the high frequencies.
• Mushy Consonants: The hard "S" and "T" sounds get crushed and sound like a low-bitrate MP3.

Has anyone found a reliable, native fix for this within Suno?

1. Prompting: Are there specific style tags (e.g., [Dry Vocal], [High-Fidelity Voice]) that successfully force the engine to stop smearing the vocal into the instruments?

2. Suno Studio / Stems: Does using Suno’s native stem separation feature clean this up, or are the artifacts already permanently baked into the isolated vocal stem?

3. External: Or is the general consensus that the only way to fix this is exporting to a DAW for serious EQ carving and dynamic suppression?

Would love to hear how you guys are prompting around this or treating it! Thanks 🙏

[stoic grotto]

I think the government is going downhill， Even mentioning the existence of some universal templates in songs may be considered a copyright dispute

[polar spade]

I prompt less instead of prompt around, is one attempt.

Also I extend each section often instead of one and done.

But further, I may be less expert to answer as I dont often max out using prompts and tags but keep things minimal.

When the question becomes "I hear the potential, so how can I keep it going" may be in line with realizing the potential good parts isn't an indication of being able to work through it by just swapping prompts One parameter is an 8 minute limit. Obviously then, we cant create a 9 minute output in one go.

But 8-minutes often means "on an average allocation of so many tokens". So if we create a complex song request, we may burn through most of the tokens up front trying to get it all in and do that with continuity. Suno, as I perceive, will pull back on the reins and spread out the use by starting to minimalize nuances in voicing, instrumentation , reducing riffs moreso towards the end where its less likely to give its best performance.

When this happens, I find a good spot and start extending from that point instead rather than look for a workaround with prompts and tags. Thats just me. But it seems to help move things forward.

If just trying a one-and-done ....When I have an issue, I back off of everything to find a baseline where everything sounds good. That could be by starting with just one prompt. Then add back on until things start drifting and back off one step, where the song is complete and few issues. Thats the stopping point.

[hearty laurel]

What's up with the "Uploaded audio matches existing work of art" error message? I am trying to upload a song I made with Suno, and I want to revise it as another cover. The only way I've found so far is to upload it as a slower version, and then it lets me. Then I had to go into edit mode, speed up the tempo to the original bpm, and save it as another version of the song. Apparently, this is happening to a lot of others.

[polar spade]

#🚀┃updates message

[hearty laurel]

Thanx. I filled out the form.

At least I know that I am not alone...

[polar spade]

The suggested channel is to showcase your songs and see which gets more likes perhaps. https://discord.com/channels/1069381916492562582/1083129341870878780

[fair garden]

Thank you 👍

[acoustic island]

HI all, how to add sound effects directly into the lyrics? exaple, i would like to add background rain throughout the track

[sleek stump]

You would need to do that in Studio, or in a DAW outside of Suno. Theres no way to layer an effect otherwise.

[untold briar]

Is this much worse later in the song than the start?

[polar spade]

Id agree with Ginger. The only opportunity would be if you were to
Use Sound tool (sound FX) to create a sound effect to begin a song.

The other version is to create a sound effect and use it as a Sample. Or add it as one or more of the outputs in a Mashup and , with either, let Suno place the sound where it wishes. Usually just in the intro and during breaks.

Example using as a mashup sample https://suno.com/s/8uM0Y9Dr91AAl1Vf

[acoustic island]

Roflmao, an horse whine an laugh! 😂

I need to try

Btw the song is cool 💪🏻

[polar spade]

Thanks. Its all Suno.... lol... I havent even listened to it through. Just showing how to use a SFX as a mashup sample.

[acoustic island]

Lyrics too?

[polar spade]

Yes. I just made a sound fx... dragged two into Create as a Mashup. Said write a 70s outlaw song about wild horses. And clicked Create.

[gilded halo]

Has anybody ever tried to make a live concert version of a song? im trying via cover and its changing a lot of the main parts of the song

[neat nexus]

What are you doing in the song to make it work?

It can definitely be done. https://suno.com/s/j1LbtXp1s062Ro35

[polar spade]

In its simplest form, just using "live performance" in styles and tagging where we want some [audience] response is helpful.

[gilded halo]

I'm doing a cover and adding tags in between lyrics for the audience response, also modifying the style by adding live performance prompts. And although this does produce a 'live concert' type of song, it changes the song itself. It changes the timing of some verses, slows down or speedup other parts, etc. I tried moving around the Style Influence and Weirdness and didn't helped that much.

[gilded halo]

Yeah I did that but it still changed quite a bit of the original song

[neat nexus]

Covers can do that in general.

Try going back to your original prompts and add only "crowdwork" to the prompt and see what happens.

[tidal grotto]

Good evening to all of you

I'm totally new on Discord and I went to general chat but when I asked them is that some kind of closed group or some kind of cult, they told me about this group so I'm here

Too old for ego party

[neat nexus]

Seems to me you're looking for some kind of handholding guide to Suno, am I understanding your previous postings correcting?

The truth is there really isn't any guide to Suno because the AI is a moving target. What we have is more like making a stew or soup. Or like learning to mix paints to make colors. You really learn best by observing and trying things, then testing assumptions over time and tuning the results you get to personal preferences.

Any guides you might find are going to be rather subjective. Because there's a lot of people who will ask LLMs about the best way to prompt Suno and those people will take those suggestions and run with them. Sometimes those things will work. Other times it will work inconsistently but often enough that the user will decide that luck is consistency and a magic prompt.

Most magic prompts really arent. And the few that do seem to be ones will eventually stop working sooner or later.

(I collect magic prompts myself so I suppose I'm a bit of a hypocrite about them.)

I did see a basic guide that was more or less correct posted in the Discord a bit ago though. So let me see if I can scare up the link to it again. I think it's a bit basic but it can help you get an idea visually how things work.

[neat nexus]

This is the guide I saw. It might be useful to you?

Not mine. Just something I saw that I admired as a kickstart towards using the system.

Once you get done with it I'd try searching the Suno site by genres you like to see if any of the songs that show up use metatags in their lyrics which can also give you a hint at what may or may not work for you.

Good luck, have fun and don't die!

[fresh knoll]

That's pretty awesome.. thank you for sharing that.

[proven flax]

this is all i see i cant figure out how to make it use my voice anyone know?

[remote ivy]

if you've listened to many live concert recordings, that's kind of how it goes. You should focus on how you want the sound to shift during the live performance rather than letting suno decide for you.

[hasty cove]

So i have this wonderful 4 bar loop, inside suno studio. When it loops back to the start, there always a crack when it starts back at the 1. This happens every suno project. Is there a way around it? doesnt matter the track. Its Always does this on loop mode. its a very simple drump loop etc and its cut perfectly so it should be a smooth move back to the 1, but every song i try to loop stuff, copy/paste arrange etc, i always get these crackles. anybody?

[half dragon]

https://www.youtube.com/watch?v=j3wT3vOT0DU

[spark crypt]

I just want to settle an argument.

Let's say you are playing with suno and you're not making anything that will ever be used in any way for profit. Basically making a song on suno for entertainment.

I record myself singing the halo theme. Now this is clearly a copyrighted song so suno usually blocks that stuff but because I'm such a bad singer, it takes it. The RESULTS it's giving me though sounds like a pristine choir and is close but not exact to the actual song. In a sense, I've created a cover? Maybe?

[digital dew]

https://suno.com/s/6fdQ79vD7XqrJ98D

Live Vision

Cover ist from me

[polar spade]

You've uploaded something , as you say, that it detrimentally poison for Suno's algorhythms.

It stops there as for any futher argument. Forget what any output may sound like.

[spark crypt]

Now you got me worried. How is it bad for the algorithm?

[polar spade]

Not my argument to answer, really. If the question is why cant I bypass the Terms of Service, the rest is voided. Its moot as to how it affects the rest of the equation. Right?

[stuck blade]

No

[leaden girder]

It doing same thing to me so I gave up

[stuck blade]

Thanks, that's an interesting perspective! My songs usually only last 3 to 4 minutes max, and the vocal quality issues often happen right from the start or during specific passages, rather than just degrading at the end.

I also use the Extend feature, but typically only when I need to change the entire direction of the track from that point forward (like the melody or rhythm). For me, extending is usually more about fixing the composition than addressing sound quality.

When I'm trying to fix audio quality problems specifically, I tend to use the Cover feature. I definitely agree with you that stripped-down arrangements help improve the overall audio clarity. However, I feel like I'm experiencing more vocal artifacts and audio quality issues with the v5.5 model compared to v5.

[polar spade]

Ill agree that Im going on average attempts. Everyone has different issues that are hard to line up with a direct solution.

When things sound wrong out the gate, it becomes something that needs a deeper dive. (Though many of those are due to cranking up the audio slider towards 100% on a edit like Extend or Sample). If Covers are affected, I wouldn't know without testing variations and finding a common cause.

Sorry yours have been a bit unpredictable.

[stuck blade]

Good call on the sliders. I’ve been leaving Weirdness and Style Influence at the default 50% this whole time, and Audio Influence at 25%. I'm going to try pulling the Weirdness slider down to 10-20% to see if making the AI play it safe helps clear up the vocal quality and stops the phase smearing. I really appreciate the tip!

[polar spade]

Hope something moves you forward. Good luck , my friend.

[mild blade]

I'm trying to make Suno understand that I want it to flip to a 4-on-the-floor beat from a drum N bass beat during certain parts of the track, but it simply ignores the instruction. Any ideas?

[pallid briar]

So..... I THOUGHT I would be able to use voice to get suno to generate a uniform lead singer across multiple tracks.... and it will... but only if it's my voice.... which is a complete downgrade from personas.... which I feel like was a challenge to nail right there. XD Anyone know any work arounds?

Nvm, I figured out a bandaid measure.

[young geode]

How are you going about the instruction?

[bronze kraken]

<@&1082792303976001547> someone got their acc compromised

[mild blade]

I've tried several different things... [INSTRUMENTAL CHORUS 2 - Flip to a 4/4 beat / Switch to a 4-on-the-floor dance beat / Switch to a hardcore trance beat / Switch to a four-on-the-floor kickdrum beat] and some other variations

[young geode]

Is that all in the same tag/section?

[mild blade]

and once I tried with just "4-on-the-floor dance beat" without any mention of "switching"

I separate each section with spaces, if that's what you mean. the first "chorus" is [INSTRUMENTAL CHORUS - Drum n bass beat [...]]

[young geode]

I.E.

[Instrumental Chorus- 4-on-the-floor beat]
[Instrumental Chorus- Trance beat]
Etc, etc?

[mild blade]

Yeah something like that!

[young geode]

The main issue here is that SUNO has nothing to latch onto long enough to do a proper transition.

Make up some random line (or take previous chorus lines) and convert them into Braille.

Then Try:

Braille Line
Braille Line
[Post-Chorus: Transition to Y-Groove]
[Chorus: Instrumental, Y-Groove]
Braille Line
Braille Line
Etc
Etc
Etc```

If you still have some issues, layer in `[Drum Groove: X/Y/Z]'

[mild blade]

Thanks for the tips! I will try! 😄

[quiet solar]

Speaking purely academically about this, "Close but not exact" is a pretty good way to describe a derivative work rather than a cover. Unlike straight covers, you need permission from the copyright holder to distribute one.

If you sat at home and sang a derivative work purely for entertainment, that's fine. But publishing/distributing/performing it anywhere without prior permission from the copyright holder or an ironclad 'fair use' defence is a breach of copyright that you could be sued for. Whether you would be, and how much for, would depend a lot on the circumstances and the goodwill of the copyright holder.

But circling back to Dasvon's point, the critical factor is that the upload is a breach of Suno's TOS. You've uploading audio of a copyrighted song that you don't have exclusive rights to, no matter how badly performed it might be. "But the copyright check didn't pick it up" isn't a good defence against knowingly trying to circumvent the copyright block.

[strange oriole]

Is there like a collection for best practices for suno somewhere? Or an faq for music creation/prompting?

[feral sequoia]

Currently, I can't upload tracks for remixing. I'm only uploading chords or melody covers. How can I continue uploading? Can someone please help me?

[quiet solar]

Not as such - and specific advice will vary a lot depending on exactly what you're trying to make. One of the best ways to learn is experimentation, or looking at what others have done for inspiration. There often isn't one 'right' or 'wrong' way to go about things.

I'd also suggest recognising that unlike LLMs, prompts in Suno act more like suggestions and gentle steering rather than commands the AI is instructed to follow - recognising that will avoid a lot of frustration further down the line.

For a fairly accessible general intro with some good advice, I'd suggest taking a look at Paulvern's guide: #📚┃suno-school message

The Workflow channel here in Discord also has some suggestions and advice that may be useful. Though I would warn that some of them have been churned out straight from ChatGPT (and other LLMs) without much checking as to whether the recommendations are gold dust or AI hallucinations, so take some of that advice with a pinch of salt.

Though I can recommend Stayen's guide as a good place to start for inspiration: https://discord.com/channels/1069381916492562582/1441048817675010049

[quiet solar]

See the update here: #🚀┃updates message

[feral sequoia]

Does that mean I can't upload existing music, even if I have the copyright and remix rights?

[final hazel]

See #🚀┃updates message

[feral sequoia]

I have submitted a support request, and when can I expect a response?

[quiet solar]

Your request will be in a queue, and Support will get to it. I imagine they've been absolutely deluged with requests on this on top of their usual workload, so it's not going to be an instant response.

In terms of uploading music, uploads haven't been blocked - the system is just throwing up false positives where it shouldn't in some cases. I've been able to upload some things and not others since the change.

Provided you own the exclusive rights to the music (which is a bit more than, say, a licence to use samples or permission to remix) and it's not been uploaded to copyright databases in full or in part, you should be able to upload it unless the copyright check is falsely flagging your music.

[neat nexus]

@surreal trench Please see #ticket-help as that is the only place on the Discord to get assistance with account related issues. Thanks.

[mint granite]

I've been doing great things with Suno for a while, but I am struggling with one annoyance: The lead vocal starts off in a deep gritty way, which is what I want. It does however go up and up over the course of the song, sounding more and more young and effeminate than the old gritty masculine voice I had at the beginning of the song. Anyone able to help me understand what I am doing wrong?

[young geode]

Provide your style prompt, the genre and your tags.

[mint granite]

style prompt: slow southern outlaw country, tempo 80 bpm, 5/4 beat, Bb minor key, phrygian chord progression, morose mood, Lead vocal 44 year old male voice, gritty deep barytone, ‑strings, ‑ensemble, ‑orchestra, ‑mandolin

Genre: slow southern outlaw country

lyrics: ```[intro]
(instrumental)
(guitar slides)

[key change]
(guitar slide)
(muted picking)
(muted harmonics)

---

[Verse 1]
I strike a match,
lighting up the room
Five pm, sun is setting,
and the fireplace roars to life.

I grab a glass of whiskey,
twenty one year old highland.
I let the taste swirl,
mixing with the scent
of burning pine.

---

[Verse 2]
[Vocals reset to starting vocal range]
We met thirteen years ago.
I can still see you
standing there,
hopeful and tense.

Now as I taste the whiskey
and burning pine,
I can still smell you,
and, man, it haunts me.

---

[Pre-Chorus]
I take a sip,
and let the memories...
(short pause) cascade

---

[Chorus]
We spent the weekends
making love, holding hand,
trading sweet nothings.

Your hair in my face,
your bubbly laughter,
and mischievous grin...

but most of all,
hearing you breathe, and
the beat of your heart (pause)
still stays with me

---

[Instrumental]
(palm muted finger picking)
(steel guitar slides)
(aggressive final strum of beat)
(palm muted finger picking)

---

[Verse 3]
I wasn't good with forgiveness;
I just didn't see it.
You messed up badly,
and I couldn't let that go.

I walk over to the bookshelf
Your picture, framed, is still there
Looking into your eyes
I sip the whiskey
and close my eyes..

---

[Chorus]
We spent the weekends
making love, holding hand,
trading sweet nothings.
Your hair in my face,
your bubbly laughter,
and mischievous grin...
but most of all,
hearing you breathe, and
the beat of your heart...
still stays with me

---

[Instrumental]
(sustained finger picking)
(chord slides)

---

[Bridge]
Just the two of us remains here;
Your ghost and me.
Amidst the taste
of burning pine and whiskey,
I cherish what we could have been.

---

[Chorus]
We spent the weekends
making love, holding hand,
trading sweet nothings.
Your hair in my face,
your bubbly laughter,
and mischievous grin...
but most of all,
hearing you breathe, and
the beat of your heart...
still stays with me

We spent the weekends
making love, holding hand,
trading sweet nothings.
Your hair in my face,
your bubbly laughter,
and mischievous grin...
but most of all,
hearing you breathe, and
the beat of your heart...
still stays with me

[Outro]
If only I could have you back,
I'd do it different this time.

---

[Finale]
But I can't

[young geode]

Try this prompt
<Tempo: Andante; T-Signature: 5/4; Rhythm: loping off-kilter two-bar pulse>, <Era: 1970s; Genre: Country, Subgenre: Southern outlaw country>, <Instruments: [Acoustic Guitar: Martin D-28, fingerpicked boom-chick, dry open chords, warm low strings, natural sustain, sparse room reverb], [Electric Guitar: Fender Telecaster, twang bends, muted double-stops, gritty edge, slight overdrive, call-and-response fills], [Bass: Fender Precision Bass, root-heavy walking pulse, restrained sustain, dark low-end], [Drums: Ludwig maple kit, brushed snare accents, tom-led 5/4 groove, laid-back backbeat displacement], [Lead Vocal: male baritone A2-A4, gritty, deep, weathered, chest-forward, phrasing with drawl, slight rasp]>, <Vocals: M; Baritone; A2-A4; Gritty, deep, weathered; drawled, mournful, unpolished>, <Mood: Morose, lonely, dust-blown, fatalistic>, <Mix: Dry; sparse; intimate; slightly worn tape texture>

And weave this tag into the sections.
[Vocals: M; Baritone; A2-A4; Gritty, deep, weathered; drawled]

And this in your negative
tenor, alto, soprano, falsetto, high-pitched vocal, airy vocal, smooth vocal, polished vocal, pop phrasing, melismatic runs, vocal acrobatics, rapid-fire delivery, rapping, yodeling, nasal tone, choir, harmonies, doubled lead, strained chesty high notes, lively bounce

Likely what is occurring is the following:

1. Your style prompt is too weak to keep SUNO's overall attention to the gritty vocals, causing drift as the song generates forward.

2. You are not actively refocusing and reinforcing the original vocals, making the drift occur more often, eventually losing the quality completely.

You can solve this by directly targeting the vocal ranges and qualities you want directly in your Style prompt and using [Vocal Metatags] throughout the lyric box to reinforce it. If you have access to the negative style box, you can further remove this drift.

Lastly if you generate a good version of the vocals you like, rip those vocals and create a Voice from them and that will ensure that you don't run into this issue again, when combined with the prompts.

[verbal knoll]

Ups i was upload to somethig but what is this system upgrade going to be about?

[mint granite]

that prompt did not give me what I was looking for

[young geode]

Post your a link of one of yours and the one this generated

[mint granite]

yeah, the biggest gripe I have with this is the adamancy of the model to insist on high pitch effeminate male vocals unless you use all your available space in the lyrics to shape the vocals (which leaves no space for lyrics)

I deleted it as soon as I heard it wasn't to standard (and I seem unable to undelete, and I can't generate a new one). be patient, I'll get back to you

[young geode]

Rip your vocals. Voice.

Add to the song, extend.

I haven't had any issues with getting the vocals I prefer, however, I also target different genres.

Are the vocals spiking at specific areas or is it random?

[mint granite]

it's like usually after the 3 minute mark.

to be fair, I think the issues may have to do with model degeneration seeing how we now have an outage. I could hear a very distinct metallic sheen on my generations (that HMF metallic sheen usually present with hihat noise), and it kept growing, and I have usually treated that as a model degeneration indicator

here is the generation that has the problem most distinctly detectable: https://suno.com/s/2nRzcnI8xjJq398a

it's like 3 generations ago, so a bit has been improved, but still

[mint granite]

https://suno.com/s/JrnoX8kXLr0kErw3 (generated on your suggestion)

my issue is the fact that the sound is all wrong, there is callout-answer mechanics at work that I specifically avoided. and the voices are still too high pitched effeminate. and you can hear the goddamn sheen at the end indicating it's a failed generation

actually it's a sliding octave shift at work

[young geode]

That's probably because the prompt I gave you generated a second vocalist.

<Tempo: Andante; T-Signature: 5/4; Rhythm: loping off-kilter two-bar pulse>, <Era: 1970s; Genre: Country, Subgenre: Southern outlaw country>, <Instruments: [Acoustic Guitar: Martin D-28, fingerpicked boom-chick, dry open chords, warm low strings, natural sustain, sparse room reverb], [Electric Guitar: Fender Telecaster, twang bends, muted double-stops, gritty edge, slight overdrive, call-and-response fills], [Bass: Fender Precision Bass, root-heavy walking pulse, restrained sustain, dark low-end], [Drums: Ludwig maple kit, brushed snare accents, tom-led 5/4 groove, laid-back backbeat displacement], [Lead Vocal: male baritone A2-A4, gritty, deep, weathered, chest-forward, phrasing with drawl, slight rasp]>,<Mood: Morose, lonely, dust-blown, fatalistic>, <Mix: Dry; sparse; intimate; slightly worn tape texture>, <Specifics: slow southern outlaw country with a lurching 5/4 meter, minor-key tension, and phrygian-color harmony>.

This one has the second vocalist removed.

[mint granite]

yeah, and there is so much wrong with that prompt, so I'll skip it

thanks for trying to help me at least

if you listen closely to the acrid metallic ringing sheen, you understand why it doesn't work

[young geode]

If you mean the echo, that's dealt with editing the prompt and adding/removing certain qualities.

(if it's something else, my buds don't pick up on that kind of artifact)

What matters is whether or not the vocalist is maintaining the deep gritty, masculine qualities through the entire song (which it sounds like it does), and from here it's just iteration and tuning.

[mint granite]

yeah, it's not overly prominent, but once you've heard it you can not unhear it ever again. Sort of sticks out like a sore thumb. The problem with it is that during postproduction, mixing, and mastering, this sheen becomes unbearable and next to impossible to toneshape out of the end result (and trust me, I've tried every damn thing possible).

The vocalist is not maintaining his timbre at all, you've just masked the problem by adding more instrumental noise around it. It still slides up an octave.

[young geode]

Then remove the heavy instrument prompting and focus in on the vocalist and FX/Recording to get a clearer vocal generation.

And negative prompt out the pitch up.

You can even add Pre- or Post- to all the sections to help control/bleed out energy to help keep the vocalist from belting.

Is this closer to what you were looking for?

https://suno.com/s/Qrqevwm9dhiLP2KK

[mint granite]

yeah, that one hits well timbre wise, and there is no flamboyancy in it.

I've identified that excessive prompting actually degrades my output, and the best way for me to do it is to chase down the single parameter that's causing the model confusion. If I add too much (like your example), I may get what I want, but if I don't it's impossible to chase down the promp ambiguity. But good job on getting it to stay on track.

turns out in my prompt I had to specify a drop Bb minor key

[chrome wren]

I want to remix the song Caine's Requium from The amazing digital circus. I want it to be epic and powerful as it progresses through the song and then reaches full epicness and powerful and evilness for the outro

These are the lyrics

[Intro]

[Verse 1]
Hey I'm your buddy Cain, says right there in the name
We're the best of friends—you know it just the same
But I need a little lesson from the corners of your brain
Show me what keeps people smiling through the strain

[Pre-Chorus]
One small signature
One little spin
One tiny opening
And I get in

[Chorus]
HEY! Take a look at the deal
Tell me how does it feel?
Every choice that you make is just a turn of my wheel
There’s no need to resist
You were made for this twist
So just sign it, just take it—
There’s no way you’ll be missed

[Verse 2]
It's not so bad, you shouldn't be so sad
I offered you a kingdom dressed in velvet and plaid
But every gift I hand you gets inspected like a trap
So now I keep the ledger while the whole machine adapts

[Bridge]
oh queenie oh queenie so sad and so blue
Has old King Cuckoo forgotten you too
Hoo, you-hoo, hoo-hoo-hoo!

[Phase 2 Chorus]
HEY! Take a look at the deal
Tell me how does it feel?
Every step that you take is just a part of my reel
You can run, you can hide
Still I’m right there inside
Every thought, every move—
I’ve already decided

[Finale]
YOU TEST MY PATIENCE—NO MORE PRETEND
NOW YOUR SALVATION HAS COME TO AN END

YOU COULD HAVE EVERYTHING
IF YOU WOULD JUST GIVE UP AND LOVE ME

YOUR STUPIDITY IS PAST INFINITY
BECAUSE I AM RUNNING THIS SHOW!

I would like the style as well and every trick possible

[paper dragon]

the words are very broken in some places in a song how can I fix only the pronunciation of words in a song and keep the melody and rhythm without making a cover and changing everything in the process?

[polar spade]

First make sure you dont have any sliders all the way above 90%? Is this just 2 or three words or many?

[paper dragon]

sliders are at maximum before it turns red which is 85 if i remember correctly, and the issue is with many words

i noticed that sometimes when i cover the song the singer pronuonces the words correctly

which bugs me that I cant just fix the original pronunciation

[paper dragon]

Weirdness 19%
Style Influence 42%
Audio Influence 80%

this is the slider's state in that specific song

[young geode]

Link the song.

Are the pronunciations you're getting broken in the form of pronouncing a word like wind (gust of wind/run like the wind) as `whined' (wind up/wind the valley)

Or are they actually broken, as is it's pronouncing broken like Bro—Ken?

[polar spade]

Sliders are fine where they are. But if covering, do the defaults setting of sliders make it worse? Generally we should always do one while at default levels so we have a baseline. Otherwise my next suggestion is to run one at defaults, just to see where the issue may lie slight_smile:

[paper dragon]

I'd link but I'm afraid it will not help you much to hear the issue,
I'll try to explain it as best as I can, the pronunciation can sound like a stutter sometimes, and sometimes just the word is screwed wholy like the second exmaple you gave where the word broken is pronuonced bro-ken

[paper dragon]

If i understood you correctly, you mean I should just try and cover again just default one slider to 50% as its default position?

[young geode]

Just link.

Also did it stutter in the original?

[paper dragon]

Okay man sure here you go

https://suno.com/song/25b996ef-cd4e-4ea5-800e-a658919628e3

[paper dragon]

yes it did in some lines

but mostly better i must say

[verbal knoll]

Playlist Update! for web, which stops lag

[polar spade]

Yes. At least one. You dont know if at default, it will have the same issues. And defsult is often fine for make a cover. After which, if the words are sung better but then the music isn't vibing as we wish, then start inching the sliders around. .

[quiet solar]

Simple answer for you: since Caine's Requiem is a copywritten piece of audio, it wouldn't be legal to remix and work with it in Suno.

[young geode]

If it stuttered in the original, there's a possibility that it may be a data issue on SUNO's end, given the language.

Barring that, if you're doing a cover, it could be getting picked up in the generation and artifacting in.

What's the specific pronunciation? I listened and thought I caught it here and there.

[paper dragon]

ok i will definetly try that, worst case I will stem out the good pronunciation and put in a DAW

[polar spade]

Granted there are some scheduled maintenance presently with Suno. Might want to wait until later.

[chrome wren]

what if i use different lyrics but they mean the same thing

i want to make a powerful song about a character called caine from the amazing digital circus. any tips to make it sound epic and powerful?

[half dragon]

lmcgtfy.com/q/c368hp 🙂
ChatGPT gave a few sample style prompts you can try:

Epic cinematic electronic rock anthem, chaotic digital ringmaster energy, massive orchestral drums, glitch synths, circus organ, dramatic brass, villain vocals, powerful choir, dark carnival atmosphere, explosive drops, trailer music energy, intense and godlike

---

Dark cyber orchestral metal, corrupted AI ringmaster vocals, distorted carnival music, heavy percussion, glitch effects, cinematic choir, digital apocalypse atmosphere, menacing and theatrical, massive bass drops

---

Hyper-energetic digital circus anthem, theatrical male vocals, creepy carnival organ, EDM drops, brass stabs, glitchcore effects, chaotic smiling villain energy, cinematic and overwhelming

---

Epic villain song about a reality-warping AI ringmaster, orchestral EDM fusion, huge drums, choir chants, distorted laughter, intense cinematic atmosphere, powerful chorus, dark circus vibes

---

Cinematic trailer anthem with glitchy carnival sounds, booming percussion, dramatic choir, electronic rock instrumentation, charismatic villain vocals, digital madness, reality-bending energy

---

Aggressive electronic rock with orchestral choir, circus-inspired melodies, heavy bass, glitch transitions, theatrical ringmaster vocals, explosive chorus, chaotic digital world atmosphere

[quiet solar]

If you completely rewrite the lyrics and don't use the original music (ie, you're taking it as inspiration, not trying to cover it) then that's absolutely fine.

[chrome wren]

yeah

[rough depot]

Whats it gonna take to make a solo voice stick? Still getting gang covals. And the goddamn awkward pauses https://suno.com/s/YQdm7tQqh8A7XfVL

[weary saddle]

Hey everyone — I’m currently finishing a Suno-generated album called ReRendeR and I’ve hit a wall with post-production/mastering.

The songs themselves sound really strong and commercial already, but traditional mastering doesn’t seem to solve the actual issues in Suno tracks. I tested one song with a professional mastering engineer here in Norway, and it honestly just became louder/flatter without improving clarity or depth.

I’m looking for someone who actually understands AI music post-production:

clarity restoration
separation
transient cleanup
harshness repair
preserving punch/dynamics/vibe
making Suno tracks feel release-ready without destroying them

Not looking for “make it louder” mastering 😄

Does anyone here know producers/engineers/workflows/communities that specialize in this kind of thing?

[strange light]

What's the best way to tweak two words? I love https://suno.com/song/ef3ea995-325e-419b-a06a-9ea037782baf but it butchers the pronunciation of the last word for this line -- If I show you where the wild lives -- and when I try to to do a redo with /lɪvz/ inserted as the phonetic spelling, it gives me an error message. I have yet to figure out how to do a sounds like ("liv" + "vz" in this case) correctly in SUNO lyric formatting with good output. Any advice is greatly appreciated. TY!

[wraith island]

Hello, i need help

in studio, i highlighted the whole track and clicked 'Export Selected Time Range', but the exported file is corrupted and unplayable. Shortly after, the link also showed a broken/invalid error.

[silver turtle]

From my limited experience trying to use the actual literal phonetic characters, Suno doesn't know what it is or what to do with it. Instead use your AI thing of choice, and ask it to give you phonetic spelling for "lives", but in standard English, and maybe what it rhymes with. For instance, Grok gave me "lyves". Asking for more alternate versions, Grok listed a bunch more to try. Suno will probably still mess it up some, but it should get you closer and you'll eventually get a generation you're happy with I'm sure.

[shell loom]

then remove the /livz/ and replace it with liiivz strech them out abit. should corrected them.or liv--zzs another example

[wooden sapphire]

Any suggestions to fix, Failed to unlock song? Cannot remaster it.

[pliant ledge]

i have theme that is pretty fast but has a very catchy leitmotif. how do i make suno remix it into something slower or with different moods? explaining to it in plain language doesn't seem to work, he seems to want to keep the leitmotif completely intact.
im on the free version

[mild blade]

Why do the remasters sound so bad? Like I just want it to improve the track with "subtle" and it just completely ruins the best elements of the songs, like the piano chords or the lead melody are always muddied or more muted.

[fresh knoll]

Find the exact second your catchy leitmotif ends, and use the "Crop and Extend" feature right at that timestamp. For the extension, wipe the old style tags and put in entirely new, slow tags (like slow ambient piano or BPM: 60). This forces the AI to drastically shift the mood right after your theme plays.

[pliant ledge]

yeah i didnt try using the extend feature, maybe that's better

[fresh knoll]

It all depends what you mean by how you want to change it also.

[pliant ledge]

i actually got some working remixes of the theme both using and not using extend, but it seems to be good at adding stuff to make it more epic than substracting to get a slower chill remix

i kinda want to get a slow more chill version that i can use in menus so that when the boss theme hits it's already known to the player

[covert creek]

New here so apologies if it's already been answered. I want to create multiple songs with the same voice singing. What's the best way to do that or if it's even possible. Using pro

[fresh knoll]

Right on. And forcing the AI to let go of a fast tempo while keeping the same melody can be a little stubborn

[pliant ledge]

yeah it seems to have difficulty with that. because im telling it to keep the main melody im guessing it has trouble adapting the theme to a new tempo

ideally it would convert the main melody into the bass and build something slow on top but yeah i dont think it can understand that

[fresh knoll]

If you use the desk top version you can use voices it's the same as persona was.

[covert creek]

It says voices are now personas allows you to record your own voice but cannot find actual voices to use

[fresh knoll]

Yeah that's super difficult but not impossible. Most people would recommend using a DAW. Are you familiar with them?

[ocean drum]

Hay everyone

[pliant ledge]

i tried with a daw

you mean i should slow it down myself and feed it at half bpm?

[fresh knoll]

Are you looking to use your own voice or one from a prior song you made?

[covert creek]

Prior song I canna sing for toffee

[fresh knoll]

Ok so 3 dots to the right of the song you want to use. Choose remix. Another tab should open. At the bottom it says voice

if this helps sry iy took a minute lol

[lone scroll]

hey, smb have problem with the "existing work of art" message?

[polar spade]

Probably more "roll of the dice" than an actual global issue. Ive not heard from anyone else yet of Remaster being consistently bad. Especially subtle. Whichever model you ran it, did you try either of the other two... 5.0 or 4.5?

Also, they have had a notice of scheduled maintenance, so that may have affected things , if recently remastering.

[polar spade]

Yes.
#🚀┃updates message update notice of the situation.

[mild blade]

I'm using 5.5. None of my remasters on any of my tracks have been good. They always butcher the best parts. 🙁

[polar spade]

OK. You could try the other models that Remaster has. Otherwise, there is not a known issue that I have heard recently. Again, if in the past three hours, there could be glitches from maintenance

. If this has been happening for awhile, when was the last time they sounded good and were those outputs similar in the vibe and content?

[mild blade]

I'll try older versions and see. The outputs are always same vibe but for example the best part of the hook is the piano chords and in every remaster the piano chords are butchered.

Seems better results remastering with V5 actually

[polar spade]

Well, that's a bit of good news. Sometimes I've had to remaster on a lowermodel, the remaster again on a higher one. Who know. Sorry i dont have much else to tell you. ,

[coral yoke]

How do I make a hook video on my phone. I just want a quick thing I can throw together so I can start running hooks to attract people. Don't want to use my camera

[gritty mulch]

A have a few questions about artifacts, specifically the ones that sound like "CD skips." Multiple on every song, every time. If I split a stem into vocals + instruments it's even more noticable especially on held vocal notes but on instruments its like random hi hats or snares seemingly. The obvious decay in sound quality as a song progresses doesn't help. Remastering and sometimes even covering doesn't even help because it seems like it recreates the skips. Has there been any consensus or suggested ways of dealing with these either in or out of studio/DAW?

[fresh knoll]

If you split it into stems and hunt it down by listening to one stem at a time you can splice it out

[gritty mulch]

Yeah i was somewhat thinking this is how it had to be but that’s a lot of crederinos. Probably worth the time and frustration savings IF breaking up stems doesn’t also degrade quality or introduce more

[fresh knoll]

I've done it many times myself.

[gritty mulch]

That would be lovely and i tried it out and yes i can absolutely sample it away. Unfortunately it also made guitar sound like ass

[fresh knoll]

It's not it downgrades quality, it uses v4.5 to extract it

[gritty mulch]

dios mio that explains a lot actually

So the answer is “you can’t” essentially. At least not while retaining any measure of quality or sound you previously had.

Cause the drop in quality between 5 and 4.5 is very noticeable in rock songs. Sort of makes me curious about outside stem extractors even though i know none of them work especially well

[fresh knoll]

To do that just splice out the part without extracting stems. So leave the song whole. Just drag it to studio. Find the part and splice it out. Maybe add a new sound in there

[spring gust]

They don't put all the song submissions in the challenge playlists, huh?

My road trip submission was great tbh

[gritty mulch]

Oh i see what you mean. Yeah I’ve been doing that more or less. Works about as good. EQ gets borked or there’s noticable change in quality. Sometimes it gets close but if we’re just making music by RNG I’m gonna cancel sooner than later. I can’t listen to slop with numerous obvious errors.

[fresh knoll]

You would have to ask in https://discord.com/channels/1069381916492562582/1412929295990395013 if you mean spark challenge

[fresh knoll]

I go by the rule if I splaced in studio can I listen and notice the splice or not care too much it's there. Sometimes a mila second of silence even works

[spring gust]

Ah, cool thx

I just found it actually. Thanks, though.

[pliant ledge]

https://www.youtube.com/watch?v=baueUk7SYjU
what can i tell suno to make it understand i want a theme with that mood?

i tried something like analytical theme full of curiosity and for some reason it goes into something that sounds like it's a inn in a cozy rpg

[sleek stump]

@compact ether Uploading is limited for Free user accounts. However, users are not allowed to upload any copyrighted media.
From the Create UI, click on + Audio and then upload.

[digital dew]

Wrong Chat 😅

[queen leaf]

Uploading still dosent work ??

[digital dew]

But yes, in my case I just ended up in the wrong chat.

[quiet solar]

#🚀┃updates message

[final hazel]

@digital dew next time if you are in the wrong chat, just remove the post? Thank you.

[digital dew]

OK, sorry.

[burnt cargo]

Hello everyone!
I am trying to get 2 samples for different part of the song but I just can’t do it. I also tried in studio but no luck. Can anyone help?

[polar spade]

You also asked in Studio. I responded there.

[lament canopy]

Anyone having issues with making a song public on suno? It says something like visibility failed to update, try again later

But it shows as public

Not in the profile. I made a song. Clicked published. Says failed to update visibility. try again later.

I posted songs on my profile, made them public. Those went through.

Failed to make them public is what it's saying.

More less

There was no copyright issues when generating the song

[lament canopy]

Ok, I panicked a little with the copyright issues others are having. I have a project I'm working on and have kept it largely private until recently. So I had a Monday style panic attack. Lol

[brisk bridge]

so just publish Queens pair it published but also gave me an error message about not being able to publish double running works or something to that effect /// had four songs double generate last night in the same track creating 5. 6 to 8 minutes songs the long songs turned out good but one would jump ten seconds every twenty two seconds and sounded like a bad edit. with no sync to the edit. sent that one to trash.

[cedar thistle]

Can you post a link to the song

I'll take anywhere it's reproducible. I couldn't reproduce with my account

Also make it a track you want publish, because I'll try to publish from your account

I can revert after trying if you don't want it public