📚┃suno-school

[young geode]

Any gen you make that's based on an uploaded audio file cannot be turned into a persona.

If you happen to like the sound, the beat, whatever, you'll only be able to use covers and extensions off anything that audio file has ever been associated with. So you should zero in on 'the sound' you like, then make your stems and personas off that, then upload your metronome and start creating a finished song: that way you have the tools to make similar tracks in the future with less frustration and more flexability.

[solemn nymph]

Ah, yeah I started yesterday, have about 50 gens lol, and I haven't been using uploaded or personas (im working on making one though) I'll maybe refocus on getting that locked in

[young geode]

Additionally, Tempos are more controlled by genres than they are by prompts, and when you are trying to deal with tempos, using the Italian tempo terms for the BPM you are targeting works best.

What you can do is find a genre that doesn't completely destroy/alter your core instruments/sounds, which has the tempos you're targeting and either integrate them into your prompt or create a song with that genre instead and either do a cover or pull a stem from the drums, and use that as the basis for your actual genre.

[young geode]

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

If you just started, read those. It'll save you a lot of time and headache.

[solemn nymph]

Thanks. Do people ever do collabs? I'd assume yes, don't see a channel for it guess it just happens wherever

[young geode]

What do you mean by collabs?

[solemn nymph]

Like, I'm strong in the areas of songwriting/composing/vision, but lacking in music science department and I'm disorganized (ADHD), seems like those strengths could be complimented by someone strong in the more technical areas. But I guess I could just keep using AI for that, just seems like it would be fun to collab.

I've published a couple of examples

[young geode]

If you understand songwriting, then you should be fine. AI can do a lot of the heavy lifting when it comes to the prompting and metatags and genre research.

One of the strongest controls in SUNO is the lyrics, through meterlines, rhyming schema and word choice. The technical aspects of music with SUNO are more about knowing what you're looking for than they are highly specific tuning, at least at the moment.

[solemn nymph]

Gotcha

appreciate the info

[graceful condor]

the whole Ai makes no sense sometimes. I cleary give the same advice. Each sections is clearly named in Lyrics and Prompts. Lyrics build like this {Form: Intro – Verse – Chorus – Verse – Bridge – Final Chorus – Outro}, and even in Prompt I mention this Form. And in Exclude I also add "sudden fade-outs or abrupt cut-offs" and still sometimes it creates a whole song but many times it cuts it off every damn time (like 5:02 for example) at the same time and this before the lyrics end in the middle of the chorus. This makes no freakin sense.

and then if you extend the song it tends to extend the song like 4 minutes even if only half a Chorus plus Outro are left or even less like some seconds of an outro. It is so dumb.

[dusk storm]

Hello there !🙌 ✨
My question might have been asked many times but I can’t find any answers to my problem. I’ve been using Suno for a year now and I still have this annoying problem when I generate a sound. The sound quality is not consistent through the whole song. It tends to decrease after a min or two, resulting in many credits lost trying to remaster something that can’t be remastered.
Is there anything I am missing here? A secret phrase to add in the prompt? I have tried many things and with the 4.5+ it’s even worse…
Can somebody help me please? 😓
This the third channel I’m posting this message… 😅

[graceful condor]

Depends of the style I noticed. The more instruments you use the more the sound decreases sadly. If you use less instruments like folk only with a guitar and vocals nearly all tracks sound like a perfect production. But like Progressive Metal with Synths, Choir, Orchestra, Guitars, Bass, Keyboard, Drums and Vocals it is only a matter of time when the track starts do have weird sounds in it like if the quality of an mp3 was suddenly droping. Hard to explain.

[dusk storm]

That sucks.. because it's really rare to be able to recover the song by remastering it or creating a cover with the same prompt... I love Suno, it's great tools but it can create a lot of frustrations sometimes 😓 Thank you for you're answer ! Nobody from Suno have said a word about that issue yet?

[young geode]

You can't prompt out the stops or fade outs, that's an issue of the AI's current model interacting with its predictions.

To stop the extention amounts, you need to change your prompt it reflect its near the end. You can also resolve this by letting it play out | choose an ending you like| opening editor and cutting the entire sections you don't want added | keep the ending and regenerating a short section.

[young geode]

Link an example.

[dusk storm]

check that one : https://suno.com/s/BDskrNwrgpauTkY0
The sound gets more muffled as the track goes on.
Thank you !

[graceful condor]

So in the end "Extend"

[graceful condor]

I don't know what to do, I have stripped everything I could see that would have anything to do with any artist. Nothing in my description at all. Still can't save this persona.

[mellow prawn]

It seems the newer version of Suno struggles to make the proper accents for when you're trying to do something like Italian 1940s American

[young geode]

Is the issue that the instruments seem to blend together?

[young geode]

Post the information (Lyrics, Metatag, Prompts, Song Name)

[dusk storm]

The sound changes as the track progresses — the mix is clear at the beginning, but towards the end it sounds compressed, with a different tone, as if a reverb and a mute/damping effect were gradually applied. It’s like the audio is becoming muffled and losing clarity over time. One clear example is the drums — they sound crisp and clear at the start, but as the track goes on they change in tone, becoming more compressed and reverberated, almost muffled, as if a filter is being applied over time.
Same issue here : https://suno.com/s/446i3mReaLukfnts or https://suno.com/s/1xZwRKYWpLVw9XeI

[young geode]

That's an issue of Instrumental gens. You can partly resolve this by using the lyrics box instead, with sections.

You can add in Morse/Braille converted lyrics in with metatags and it should produce a cleaner song: especially if you provide repetition in the lyrics, and scales in the metatags.

[dusk storm]

Ok, I'll try using the Lyrics box to define the structure of the instrumental as well. When you mentioned using Morse, did you mean something like this?

[Intro]
.- .- .- .- .- .-

[Verse]
.-.. --- ...- . / -.-. --- -.. . ... / .- .- .- .-

[Chorus]
.- .- .- .- .- .-
.-.. --- ...- . / -.-. --- -.. . ... / .- .- .-

[Bridge]
.- .- .- .- .- .-

Thank you for your time.

[graceful condor]

I do this all the time but works not really as intended as you would like to. Sometimes it does but with Suno it seems it is kinda unpredictable. Sometimes the Ai thinks of a song in a different way and regardless what you want it sticks to its background planning like a stuborn child xDD

[hearty laurel]

Hey Gurus, Another stupid question. I am trying to do a cover song. I uploaded the audio file, added the lyrics, and Suno came up with something I really like. However, it extended the intro from 18 seconds to about 43 seconds. How can I trim it back to the original intro length without changing the entire song? I tried using the editor, but I can't seem to get the edit right where I want the intro to end and the verse to start. Any ideas would be greatly appreciated. thanx.

[polar spade]

You could insert a segment right before the verse and see if it blends in better. Then look for a better splice spot to the left of the new insert.

[young geode]

Yes, but an actual translation.

[half dragon]

Maybe in its name? what did you try to save it as?

[graceful condor]

I even tried to using my own name first and last. still won't save.

[half dragon]

Try to save using something else or use this trick?
#📚┃suno-school message

Then it is not the problem in the name - must be something in the description - maybe simplify the description for now and see if it lets you save?

[graceful condor]

I already cut it down to 1174 characters.

[half dragon]

Maybe "thundercloud" is the issue?
https://music.apple.com/us/artist/thundercloud/1504442642

[graceful condor]

Today is really a pain not one song that wasn't cut off in the end. It is like if the Ai has bad mood or not. It is so random.

[graceful condor]

Still no go I changed that to "Born beneath a tempest sky in the hollers".

Wonder if it could be because I have that in a playlist and used that as an Inspiration on another track?

[half dragon]

Doesn't sound like it but easy to test by removing it from the playlist.
Or, if you don't mind saving it without any description for now just to test.

[graceful condor]

So yeah it saved..

[half dragon]

Cool, now we know something in the description is causing it.
You can use rule of elimination to find out which paragraph then which sentence then which word.

[graceful condor]

I found this “midnight train bound for glory.” and replaced with this "boarded a midnight train headed into the dark" Still won't save.

[half dragon]

Change 'midnight train' too? see if that's the one.

[graceful condor]

Honestly this is no fun anymore FIX YOUR AI! Today I waste 5 coins after another. The songs starts great and all of the sudden after 20 Seconds or so the whole track gets massively muffled and dull. I am sick of this! Wasting coins because the AI can't produce full tracks and cuts it off more than doing it right and then now since the isse with the missing coins because there was in internal issue all songs get muffled.

[graceful condor]

changed that to this "Some claim he followed the wind westward, lost to the fields and forgotten roads." no save again.

[half dragon]

So you found that without that sentence, it'd save?

[graceful condor]

Yes, it saves without any description.

rewrite didn't help either

[half dragon]

I tested with your description and found this paragraph is the issue, without it it'd save and won't with it

"His cadence was slow-burning and dramatic, marked by call-and-response rhythms, whispered truths, and sudden bursts of holy fire. Between verses, he’d speak in riddles and scripture, turning every gathering into a spiritual séance. His phrasing was tight—four syllables at a time, like hammer strikes on a coffin lid. "

Then I eliminate the last sentence, then this one is the problem
"His cadence was slow-burning and dramatic, marked by call-and-response rhythms, whispered truths, and sudden bursts of holy fire."

See if you can reword it somehow.

[graceful condor]

That's crazy not one Mention of an Artist.

@half dragon Thanks I just removed that sentance and it saved. I can live with that.

[half dragon]

damn it 🙂

Yes, so "cadence" and "sudden burst" are the ones that caused the problem.

It's bad though as cadence is also a music theory term and we can't use it.

[jade quail]

Try "1970s R&B" or some time-period+[genre/style] during which the source training data is likely to have been recorded well (with good sound). Works SOMETIMES.

[jade quail]

Wouldn't it be cool if a music-generating tool understood the globally accepted musical terms we have been using for hundreds of years!?

[dusk storm]

Thanks for that! I don’t always do that and I never realize how much it has weight in the prompt

[jade quail]

Any ideas how to keep Suno from thinking normal words are "artist's names"? I just had to replace "chiara" (a normal Italian word) because Suno prevented it ("Nope. that's an artist's name!!!") - with "Key-ara" ... bleh.

[graceful condor]

I had that same problem with "cadence" and "sudden burst".

[slender matrix]

Quick question - can I reference an existing band or singer if I want vocals in their style without directly coping it? I'm thinking of Doom Metal bands specifically

[near flume]

There is an Italian girl duet group called "Paoula & Chiara"

[wispy laurel]

I've heard of people getting this to work but I'd guess it might depend on both the exact wording and also whether or not Suno actually knows the artist you're referencing
You could try ".....inspired by xxxx" or ".....reminiscent of xxxx" in the style instructions and see if it'll allow it, but it tends to be super fussy about anything that even might be somebody's IP, so I wouldn't be too surprised if it refuses

[lunar nexus]

the generation goes on for 4:58 and ends. Already 4 times in a row.

[lunar nexus]

yes. Type: Doom Metal -style

[slender matrix]

Good point - I'll give it a go and see what happens tyvm

[slender matrix]

Tried that but there are a few different styles of vocals and I have a specific one in mind

[lunar nexus]

there is a more complicated way. It is necessary to separate the voice from the music in another program. Then upload them individually to suno and then do Cover

[slender matrix]

That's an idea - I have Studio 1 so I can maybe do something there ty!

[lunar nexus]

I wish you a successful job!

[lunar nexus]

if you perform the separation in another program and then upload it back to suno, the text of the song will be recognized.

[jade quail]

Ask ChatGPT to describe their style without naming them, then use that in Suno.

[jade quail]

You can just hum (or sing "la") the tune of the song you want to cover/imitate. Suno isn't smart enough yet to know it's an already-recorded melody. Then you can add in the lyrics after Suno accepts the audio for "cover" - or whatever you want to make.

[slender matrix]

That's an idea! Ty

[young geode]

Goto Perplexity.
As it to describe the vocal qualities of singer/genre in full using musical terminology.

Have it then prompt those vocals in the following way:
Vocals: Male/Female, Pitch Classification, Range, Qualities, Method

From there you use this with the genre that you are targeting.

[cyan jewel]

whats the secret to get really hard hitting bass for electronic music?

also whats the secret to get really hard hitting bass drums and bass guitars for metal prompts? i wanna get some of that tesseract feel

also is there a secret to being able to have two distinct vocalists of the same gender?

also how do i get metal tracks to stop having harsh vocals? :o

[young geode]

808s, 808s Bass, Deep Distortive 808s

[vestal flax]

hello, one quesiton when i click Download> Generate Video on any of the generated audios i create, where des that video end up? i never figure out

[half dragon]

It's generated and stored on Suno server. You can also download it after.

[vestal flax]

Where can i download it exactly, i cannot find any generated videos on my workspace, cant find the result

[cyan jewel]

do you know how to get the overdriven gated 909s common with jumpstyle, hardstyle, gabber, etc?

[half dragon]

Should be under Download menu as well.
Download > Video (enabled after generated)

[vestal flax]

Ohhhhhh there it is... thanks!

[young geode]

Roland TR-909, Roland TR-909 Bass, Overdriven Roland TR-909

You have to play with it and test, but the best methods of getting something specific, like an instrument, is to find the exact instruments (by name) of a genre you are looking for, then you target that Genre and those specific sound qualities.

[young geode]

https://suno.com/s/b3RVRJSw8uoBBUGo

[cyan jewel]

this almost gets there.

the kick doesnt quite have that gabber overdrive/distortion

its a good step in the right direction though. Hm.

[young geode]

That was just the first gen.

If I played with it I could get it, but it's a refinement process. If you really want that thick, wet, bassy thump, you'll probably going to have to spend an hour and maybe 100 credits going through the tweaking process for the prompt, metatag and scales.

Then when you find it, Stem the bass, then use that as the base for your songs going forwards.

[cyan jewel]

got any suggestions for what to tweak? i've got the credits for it

1240 to play with

https://suno.com/song/ad8d1fad-cd6b-4145-bdd4-e5a276c19bcf this is where I'm currently at.

[young geode]

I don't know much about the genre. Given how the sound wasn't nearly as deep as I would have wanted I'd probably focus on these two areas of my prompt:

[Drum Machine: Roland TR-909, deep analog bass kicks tuned for bounce, play method: punchy, detuned kicks with multiband distortion],

{Mix: Dry with parallel compression on kick and bass}

and change it to something like:
[Drum Machine: Roland TR-909, deep analog bass, loud kicks, tuned for bounce, play method: hard punch, distorted kicks],

{Mix: Wet decompression on kick and bass}

And from there, fix the metatags, prop the Style Influence to 70-75, drop the weirdness to 25%, and start looking for the bass. I might even edit my lyrics to force several breaks/instrumentals so it can break long enough to get the bass to the forefront.

Also, if you already have the bass (which it kind of sounds like you do), you might be able to stem that bass and then decompress it in the editor. Save that, then use that as the Persona and as the Cover/Extend.

[slender matrix]

holy carp that worked like a charm!! thank you so much!

[young geode]

Glad to hear it

[polar spade]

S/O to @half dragon and @young geode for keeping up the pace and helping out everyone in channel. Lots of new ideas and old favorites coming from them both. 🫡🤘

[rose ether]

is there a source somewhere with some heavily detailed worked example song lyrics and prompts along with a few variations of what they might produce?

i've been reading the documentation off and on, but without seeing examples of a fully worked song that goes hard on some of the possibilities, it is hard to know how much you can actually do vs just like "on one or two tags here and there"

[fathom wigeon]

Hi!
I'm trying to make suspensful music with only wind instruments like panflutes, flutes, ocarinas and etc. I tell SUNO no piano, violin and it still adds them. Help, please 🙁

[young geode]

https://suno.com/playlist/59738ccf-918b-41ee-aae5-db09a88e85b0

like this?

[young geode]

what's the prompt you're using?

[young geode]

what kind of suspenseful music are we looking for? Like a classical score, film score?

[fathom wigeon]

film score

[tawdry scarab]

putting "no something" in the positive has, in the past, not worked well

[fathom wigeon]

Oh, that makes sense. Let me check

[tawdry scarab]

also recently I used something a little crazy to get a tense cinematic feel, but it's not a targeted and clean approach:

https://suno.com/song/95bb7101-7d33-42f5-8cf0-528788ba8562

[young geode]

https://suno.com/s/pMrWduKb9UZtb2Jz

I dont know how suspenseful you can get with just woodwinds, but this is probably the closest you can get without all the additional instruments, without doing a lot of generating and refinement.

[tawdry scarab]

if you permit things like a list of octobass double bass and bass versions of individual woodwinds and you get it tuned just right they can get pretty dynamic

[grand hazel]

I can't find my workstation anymore

[rose ether]

holy heck yes EXACTLY THAT thank the gods (AND ESPECIALLY YOU)

[hidden yarrow]

I am not sure: how close is the melody to your given [melody] prompts? I dont know this type of code...

[young geode]

Hope it helps.

[slender matrix]

I've trained a persona on the track I generated so hopefully that'll be good for new ones

[tired viper]

does putting a word in all caps (or part of a word) put any extra emphasis on it or does it basically read everything as lower case?

[half dragon]

I think so. All Caps sometimes make them scream or yell.

[wispy laurel]

In lyrics, it makes it shout/sing louder/whatever would add intensity within the vocal style that it's trying to use
In style instructions, it seems like it might add emphasis, based on my own experience only, but hard to tell. A more reliable way to emphasise something in the style instructions is to repeat it, and also place it nearer to the start of the style prompt (most generative AIs seem to put more weight on the earlier parts of a prompt but this isn't always a hard rule and Suno may or may not conform to this)

[young geode]

Yes, it will recognize it.

However, depending on the genre, the speed and how the lyrics are structured, it may be anything from a higher pitch/volume to a yell, to a scream.

Placing a —Hyphen— can help control some of this. It also works for the syllables of a word: IE apPLE.

[wispy laurel]

Huh. I never even thought of using it for only part of a word 😂

[mossy elk]

Hello! A question about song captions: Does anybody know how to create working links inside captions? The "@" mention doesn't seem to work, neither does HTML linking. Thank you in advance. (EDIT: I figured it out. It is the "@" mention indeed - it's just that this particular username is not providing a link at all, maybe it's something with underscores or the length of the username.)

[scenic coyote]

I didn't even know this channel existed. In general chat i was challenging people to create the vocal sound of the lead singer of the New Beats in the song, bread and butter. One of the most distinct voices in my opinion. We got kind of close in the general chat but wanted to throw it out here if anyone is bored or wants to experiment.

my goal is to have someone create the vocal tone using heavy parameterization, And then I want to see if I can replicate it with natural language prompting.

https://youtu.be/S_Jzl_bx3fI?si=NnUic0SU2Soz-xx1

Yes, yes, I know it's the best song that was ever written

[grave sundial]

yo im new to music producing can anyone help me out?

[grizzled ember]

Whooo who has audio software recording app exp? I'm recording a story (2 voices, maybe 4 later episodes) Bunny & Kitty episodes with Pickles the pigeon & Sensei Goldie
For YouTube about our daily music trials, we have lots jajajaja
I'm using Gemini cuz beta testing & voice sounds good expressions, sound effects YAY but no accent... & If I accidentally refresh like just did no save history have to redo, ouch
Test 1

[tacit jacinth]

Hello I was in the need help channel and I was directed here. I'll copy and paste the problem I'm having if any1 can help.

. I made a post hardcore song and I love the way it came out. The only problem I'm have with the song is that the instrumental is louder than the vocals, it's over powering the vocals if that makes sense. Is there a way to fix this? I tried remastering a couple times but it changes the vocals slightly that I don't like. And generating a cover changes the song. I use the browser version.

[half dragon]

There isn't much you can do with that song on Suno.
What you can do is split the Stems and download it into your DAW and adjust it there.

[cyan jewel]

hello are there any guides or resources on how to create lyric files? :3

[lunar nexus]

how to remove the preface at the beginning of words? I mean, when a character takes a deep breath before starting to pronounce.

[lunar nexus]

I didn't quite understand the question. If you are asking about a program for text, then any notepad will do. If you are talking about creating ready-made texts, then yes, there is a window for writing text.

[cyan jewel]

no to both. i mean syncing lyric files with the audio automatically

[lunar nexus]

Sorry, I still don't understand what you want to ask. I'll try to tell you by my own example. I just have a voice part. I add it as Audio and then I make a Remix using that voice. You can do the same with an audio track without a voice - add it as a file and then add the text and click generate.

[cyan jewel]

do you not know what a lrc is? or a lyric file?

im not talking about generating

im talking about after the song is done

putting together the meta data

making sure lyrics are synced with playback

etc

[lunar nexus]

apparently, I still have a lot to learn from you. I assume that this is due to the inaccuracy of the translation, since my spoken language is not English.

Is this what we're talking about?

[cyan jewel]

no

https://en.wikipedia.org/wiki/LRC_(file_format)

[lunar nexus]

thx !!!

[rotund spear]

how can we add a delay between two line in lyrics . please can anyone help me on this

[lunar nexus]

i see that in Riffusion, but Suno it does not provide such an opportunity as far as I know.

[cyan jewel]

there's a few options:

ellipsis
–
/
line breaks

btw thats – not - or _

(alt+0150)

you can also tell it to add pauses

[lunar nexus]

if you have all the lines of the quatrain going together, then the model will most likely generate the following ones together. But try this: 1) make a gap between the lines 2) put three dots between the lines...

[tired star]

When i'm trying to edit a song, the edit comes out significantly quieter then the original instance or the rest of the song. Is there a way to prevent this ?

[rotund spear]

Thanx i will try these options 🙂

[lunar nexus]

do you know how to outsmart the system 👍

[cyan jewel]

not really i just started to learn

[lunar nexus]

unfortunately not. only in the external editor (DAW).

[tired star]

So can't change the lyrics in the editor unless want to change amplify in external editor ?

[lunar nexus]

I read your question as a question about volume. maybe I didn't understand.

[tired star]

sorry im still half asleep. It is about the volume, So if i use the suno editor to change certain lyrics it lowers the volume quite a bit.
Are you saying i need a external editor to fix the audio drop ?

[lunar nexus]

yes, if you want to control the volume settings as you need, then this is only in the external editor. Perhaps there will be such an opportunity in the new Suno update that will be released soon.

The new update is expected to work with tracks like in DAW, so I think it will be possible to adjust the volume.

[tired star]

Okay thanks.

[cyan jewel]

what other art and assets should i include? any suggestions?

[graceful condor]

If I upload a Song and just wanna create a Remaster because I want a better mix. Is there a way to impact the Remaster? Because I do not see a way to add prompts or am I missing something?

[lunar nexus]

Help - is there a way to remove the moment of air intake before pronouncing the words ?

[graceful condor]

Why do you want something like that? Because this makes it natural.

[lunar nexus]

Yes, I know that this is a natural process. But in recent versions it has become too explicit. I'm a vocalist myself, and it looks too obvious to me. Many friends note that it has become very pretentious in the generations.

[graceful condor]

Honestly never noticed that. I kinda appreciated it in the end because it felt so natural to me. But I guess it is different for everyone.

[lunar nexus]

Remaster does not allow you to add hints. Cover only.

[graceful condor]

Is there a way to cover a song, but keep nearly all as it is only change quality things or so?

[lunar nexus]

I have confidence that this is especially evident in different languages. For example, in the Russian language, this is very pronounced.

[lunar nexus]

Yes, it is possible to reduce the changes as much as possible, but you will not be able to save them completely.

I noticed that this parameter helps to reduce the impact of the new generation on the voice.

... I'll be back in 20 minutes.

[graceful condor]

Does it make any sense to add Mix and Production prompts into the lyric prompt section?

man I hate it if the songs have a stutter in it - makes no sense.

all songs got stutters and hiccups in it. Great. This is how a day should start.

[brisk flint]

Yes, IMHO it does.
Here is an example from me that Suno would never have been able to implement without the exact instructions in the lyrics section: https://suno.com/s/MNOJkW9E9v94Ix3g

[young geode]

What do you mean by a 'lyric file'?

[lunar nexus]

https://en.wikipedia.org/wiki/LRC_(file_format)

[cyan jewel]

exactly what it says on the tin

lrc and other adjacent file formats are how you tell most music players lyrics/timing/etc. the other being embedding them into the file itself

see here?

[graceful condor]

Man this AI needs a lot of improvement. Why is it that especially rock music with a lot of instruments gets dull und muffled in sound? I mean how much clearer can I add it for the AI to understand. This already says clear CLEAR:

{Genre: Post-Rock, Progressive Rock, Ambient Rock}
{Instruments: Electric guitar with rich stereo reverb (dreamy Strymon Big Sky-like), fretless bass, layered ambient pads, dynamic drums (soft to powerful), clean rhythm guitars, complex lead guitar solo}
{Vocals: Male, Tenor with falsetto reach, expressive, breathy and emotional, similar in tone to Dream Theater’s James LaBrie, Ted Leonard (Enchant), or Daniel Gildenlöw (Pain of Salvation); clear phrasing with soaring highs and trembling lows}
{Mood: melancholic, existential, cinematic, introspective, ethereal, dramatic build-up}
{Tempo: slow to mid-tempo, evolving}
{Mix: wide stereo field, lush reverb, layered textures, delay-rich guitars, ambient progression}

{Mix: high-end studio master, crystal clear, balanced EQ, wide stereo, bright , silky highs, natural reverb}

{Production: high quality, polished, radio-ready, high fidelity, punchy dynamics, open stereo field, smooth mastering, deep but clear bass}

[viral kite]

Will suno not let me cover an extended song? I wanted to cover one that is 7,5 minutes long it cuts it to 4 minutes or there abouts

I hope one day we can pick our vocalists, use one from a song we like and generate more using the same voice

[polar spade]

@proud shadow

As to your question in need-help (billing and account support channel).....

If you personally wrote lyrics and brought them into Suno, those are always yours to own and copyrighted across all levels. Reminder that Suno owns the Basic derivative outout created with your lyrics , though, and that output must be attributed to Suno.

Any new generated music in Pro level is yours to own....with conditions....

Between the two, if generate a new output in Pro, you own the output of a new generative output , and own and have copyright to your lyrics of the same output.

If the lyrics were/are not yours but fabricated by Suno while in Basic, you own only the instrumental bits as its a new generative output but , technically, if we follow the TOS, the same lyrics would still belong to Suno and must be attributed and cant monetize or use commercially. You never got/get a license to use the lyrics.

Dont forget, Suno , by your agreement, can ALSO use your lyrics at any time, and you may even find a phrase show up in someone else's output -- as per your TOS agreement.

[near flume]

You could try using De-Ess plugin in a DAW (Audacity & MuseFX - both are free).

[near flume]

When you declare styles try using technical terms that you would find in a book on recording. There are a couple of good guides in the #1252267444773326848 section of this group.

[lunar nexus]

What's going on?! all Cover generations are cropped at 4:58. The original track is over 7 minutes long.

the funds will not be refunded

[marble nacelle]

hello everyone 🙂

I noticed can’t generate a song longer than 5 minutes. When I upload my own acapella track that is over 6 minutes long, the system generates a song with a maximum length of 5 minutes, cutting it off at the end. How can I solve this problem and generate a song longer than 5 minutes without it being cut off? I need the song to have a proper, smooth ending.

[lunar nexus]

I have the same problem since yesterday.

see the screenshot above

and for some reason, I don't have access to the tech support section in Discord

[marble nacelle]

i see

[lunar nexus]

it might have something to do with a future update. But why we are suffering at the same time is not clear to me.

[marble nacelle]

Interestingly, I can upload an audio file up to 8 minutes long but only generate one up to 5 minutes long.

[lunar nexus]

I've tried different options and as a result I get the same limit - 4:58

I made an error report, but the funds were not returned.

[marble nacelle]

yes, its frustrating

[lunar nexus]

Dasvon, nice to see you again!

[polar spade]

Typically , in the past, the cropping has been due to, when making a Cover or using a persona, its needing to allocate space for the code to perform the Cover or use the added aspects of the Persona. In simple terms, it loads all the code from right to left (of 8 minutes) ansd whats left , it lays down the new song from left to right. So apparently it needed 3 minutes 2 seconds of reserve space to include your code first. You had 4:58 left to squeeze in the new output.

[lunar nexus]

This is the first time I've encountered such a problem and it hasn't disappeared for 2 days. It's very frustrating.

I've already used about 10 generations and had to stop working on the track because of this error.

[polar spade]

Its been a thing with 3.5, 4.0 and 4.5, but only recent seeming to show up in 4.5 and creators are again, making longer songs. If youve been able to Cover 7 minite songs before and get 7 minute Covers, that would seemingly be an interest to reveal how that could be done, knowing what we know from past experience.

[polar spade]

Thats the typical explanation. Nothing can be done if there is new protocol as to how songs are created, I suppose.

The workaround at least with Personas, though , is to make smaller Persona files from snippets of an output (about 1 minutes is enough) instead of the whole thing.

[lunar nexus]

The song ends in the middle of the verse, which is certainly bad. There is no information on the time limit for Cover anywhere. I'll try to do an Extend, but I have a feeling that it won't work.

I also thought about the Persona option, but then the melody of the track immediately disappears. I tried this for tracks in parts and put them together in DAW - the result was disappointing.

[polar spade]

All outputs have an 8 minute limit to squeeze all information in. It has never guaranteed a "completed" song, as the word "song" really has no meaning, but only a completed output of more than 8 seconds.

[lunar nexus]

The original track is 7:33 . Any Cover ends at 4:58

Of course, I remember the 8-minute limit 🙁

[polar spade]

Right. So using Covers and Personas are limited and not all-encompassing. Correct. Its almost a matter of trial and error, or give and take. Until they bump things to a 10 minute limits.

The best thing to do is to always work any output with about a 5 minute limit and not go beyond, if we plan to also edit with covers or personas.

Otherwise, as you say, we just extend it out as far as we want and then use a DAW or other editing software to make other edits in the fidelity and mastering for the moment.

[lunar nexus]

In this case, I find it very frustrating that the original track is very melodic with a diverse pattern that will disappear when using the final assembly method in DAW.

[polar spade]

Before covering the song, you can try an experiment and remove half of your lyrics. Suno is covering the "audio" aspect and may sings all the lyrics nonetheless. I did that in the past successfully in many cases or it would just start hallucination due to missing lyrics. It was a crap shoot but always good to see how it affects the cover. It was not longer needed for my shorter less-than-5-minute-songs once they went to 8 minutes.

Covers in the past were specifically stated to be limited to 4 minutes. Then after 8 minute limits, there has been no further official notice of the limit of Covers but to assume its likely less that any long song would be approaching 8 minutes.

You do have a slightly different option. That is to put your song in a Playlist and then click Inspire and use sliders to try to get an exact or near copy/cover of the song. An unofficial workaround and experimental again. So do at your own risk of credits use.

[lunar nexus]

Hmm, it's worth a try! Thanks for the tip.

[polar spade]

As always, if it doesnt work, dont shoot the messenger. 😆

[lunar nexus]

Oh, what an unexpected turn. I think it's still Suno's mistake. I'm telling you...

I want to use the cropped track 4:58 as a basis, and suddenly I see that all the text information has been saved! Thus, the system still knows that the track must be complete. After all, if this is not the case, then the text would also be cropped!

I want to say that the track has been cut in time, but nevertheless, all the full information about the text is still preserved. It's very similar to a bug.

There's definitely something wrong with the system. I tried to switch to the editor, and the track started playing in the background. The editor has opened and the track is still playing in the background, while nothing is playing in the editor.

[polar spade]

Yep. OK. View it as you wish. It Covers the audio but still has a lyrics file to include, nevertheless so it won't (or should not) affect the text. So remove all the lyrics or half, or leave a verse and chorus, and try a cover again. At least see if the minutes change from 4:58, then youll know, at minimum that it has some affect on length. But for an over-7 minute song, that approaches 8 minutes limits, we're just pushing the boundaries.

Even though your car can reach a top speed of 6000 rpm, it best , for optimum efficiency to never go over 4000rpm. If you add a Cover to your song, thats like shifting down to 3rd gear. Better torque control, but your RPMs need to be less to be as efficient without blowing up your car. 😅 Kinda the same thing.

[polar spade]

Yea. Thats a glitch? Tap your space bar to stop it, I think. And try again.

[quiet estuary]

Hey, has anyone else noticed that since a few days, vocal tracks are suddenly much shorter on Suno? I'm only getting 40 seconds to max 3 minutes, where I used to get 4–5+ minutes before. Is this a new limitation or a bug?

[lunar nexus]

I tried using the song Editor and the track ran perfectly for up to 7 minutes. I conclude that Suno has an error when generating the Cover.

[lunar nexus]

Yes, I've been shouting about this for 2 days now.

just above is my correspondence and examples.

[lunar nexus]

You see, a lot of people are already complaining about the duration. This is not a reproach to you. I'm just saying that there is a mistake.

[lunar nexus]

I respect you for your willingness to help!

[lunar nexus]

I just worked in Song Editor mode and everything is generated there for up to 8 minutes without errors. I think there is a bug in Suno when the Cover is working.

[polar spade]

We have "Covers" in Editor UI now? Hazzah!

OK. Seriously, Covers is not an editor, but a re-generator where Suno controls the hidden code to create new enhanced versions, but now were given slides for more control on our end. Editor UI needs none of that.

OK. Time for breakfast for me. Im getting hangry. Good luck, my friend.

[proud shadow]

So, this means I can't upload the song to Spotify if Suno owns the lyrics even if i attribute them to Suno?

just wanted to clarify

[lunar nexus]

Now the Song Editor has been broken too!

[neat valley]

Apologies for the text wall, but its good data.I

I've made significant progress on prompting Suno! I was trying to encourage what seemed like emergent behavior for left-to-right panning, but I later discovered it's a new, albeit unsupported, feature in version 5.0.

Despite the challenge, I accepted it, and while not completely done, I'm on the right track. I'm confident that by EOD, I'll have a lexicon, and I'm actively reproducing it. Things will only get better from here.

Check out my proof-of-concept track. My goal is a 1:1 mapping where lyrics saying "left" correspond to style channel 1, and to get the algorithm to go wild with turntablism crossfading. I really wish I wasn't limited to 1000 characters, but constraints often force us to be more creative!

Here’s some metrics for track:
Avg Pan: +0.17 dB → almost perfectly centered.
Std Dev: 1.06 dB → healthy variation, not jittery.
Max / Min Pan: +9.03 dB / –5.11 dB → strong right-side peaks, solid left reach.
Sweep Count: 14 total → Sweep Rate ≈ 3.65 /min (bottom of “Usable” zone).
Hard-Pan Hold Avg: 0.30 s.
Perceived Hold Avg (≥1.0 dB bias): 0.26 s average hold.
Perceived L: 12.95 s total.
Perceived R: 30.05 s total (clear right-side bias in perceived placement).
Side Dominance (hard-pan): L: 5.2 s / R: 11.8 s.
Peak-to-Average Ratio: ~13.10 → strong punchy peaks.

This one shows intentional stereo movement, leaning toward a Right-Strike personality — higher perceived right bias with big peaks but still enough sweeps to stay musical.

https://suno.com/s/pXRHtaWAfySnlruK

[storm pawn]

Guys... finally I managed to get created amazing variant for a song (the best out of 50 I did so far) - the problem is that I made a slight mistake in lyrics that extended one section... As a result, I cannot replace just this mistake (using replace lyrics in the Editor), because it does not align. If I replace the whole section, the melody is different and I cannot get it sound the same. Is there anything I can do there - keep the sound of the vocals, of the chorus, the melody, the drums, everything, but just change the lyrics? I tried extending, but it also result in slightly different melody... I tried Cover - doesn't sound as expected.

It really bothers me that "replace lyrics" changes the vocal and the melody/instruments when I change the larger part of the section.

Can I somehow fix it with stems? I can merge outside of Suno the music and chorus, leaving space for regular vocals

[sleek stump]

If I understand the goal correctly... If you wrote your lyrics, they are yours to do with as you please. However, if you generated a song using them while a Free User, the output song can NOT be used in any way that is monetized, that includes uploading it to a distribution service. You own the lyrics, but not the output song..

[neat valley]

There are massive benefits to biing. paid user.. IIts crazy how many credits you get. Creativiurty flourishes every say

[polar spade]

You'd have to check Spotify Terms of Service. But as they are a monetizing and commercial site, likely you would be limited by Sunos terms of service for Basic user material not allowable on monetizing site or for commercial purposes.

[proud shadow]

Fair enough, I just wanted to check before I did something wrong ^.^

[polar spade]

Well, you technically own the output song, just as if I would own a song i wrote at home when I plagiarized the Beatles song, Yesterday, and used their lyrics. I simply could not infringe on their rights or ownership by, as you say, distributing or promoting it commercially.

[sleek stump]

Did Suno change its TOS?

[fervent citrus]

Are nobody find eay limit song lenght like if want 4:15 song leght

[polar spade]

It would be a mute point as you would need to list the artists with your PRO and Suno is not an artist, at this time. 😅

[ebon sand]

Have you tried to fix it in the old editor?

[polar spade]

Not really. We cant designated the exact length of a song yet with Suno.

[ebon sand]

you can crop it, fade it and replace the end basically

[storm pawn]

Hmmm I didn't actually. How to access the old one?

"Legacy Editor"?

[ebon sand]

yes exactly, it's more stable but might also require a few tries

[sleek stump]

Nope, TOS are still the same, free user do NOT own the tracks they generate...
If you are using the free version of Suno (our Basic tier), we retain ownership of the songs you generate, but you are allowed to use those songs for non-commercial purposes, subject to your compliance with Suno’s Terms of Service.

[fervent citrus]

Okay i tryed allkind tricks but not succes

[polar spade]

How so? Any output made in a paid subscription is always owned by the user. But ytou can do anything with it if we lifted patts of it from Sunos creations, as we dont have a license to do so. So we own it enough that we can download and play it for mom, but thats it. Can't do anything else with it.

[sleek stump]

We are not talking abotu paid subs here, we are talking about a free user track made with their lyrics.

[storm pawn]

I'll give it a try 🙂

[sleek stump]

They get a non commercial license to use the track, they dont own it.

[polar spade]

Oh I thought the original premise was they had a basic song, then now a paid account and used their lyrics from the basic song but prompted for a new output. @proud shadow Did i misunderstand the premise?

[sleek stump]

Sure, if they subbed, then regenerated their lyrics in a new song, then yes they have rights and ownership. But from what I understand (I may have missed part of the convo) they used thier lyrics in a free user generated song and wanted to get it on Spotify...

[polar spade]

Original question:

"......I have a question regarding the whole commercial use and such, and yes I did check FAQ and the Help article, but I still need clarification. Basically I generated a song while on a free plan to test Suno (will from now be referred to as Song 1), I loved it so I got a pro plan, copied the lyrics from Song 1 and regenerated a song using those lyrics while on a pro plan (This song will now be called Song 2). Does Song 2 count under the commercial use license because it was generated while on a Pro plan even though it uses the lyrics from Song 1? or is Song 2 not covered under the commercial use license?....."

#archived-ticket-help message

[sleek stump]

Yep, they own that second generated song, as long as it was a new gen and not a cover, remaster or edit..

Need to make that distiction because some people will not understand how that works

[polar spade]

Im not sure they made it evidently clear whether they wrote in whole or partially the lyrics or if suno did. So its been vague. @proud shadow

If they wrote the old Basic lyrics, then no problem. If suno wrote the old Basic lyrics, then there are issues.

[polar spade]

And plaguewood has moved on. Then, so will I. 😀

[sleek stump]

Can not register any AI written parts of a song with the Copyright office, so ya issues

[polar spade]

Yep.

[proud shadow]

Sorry I was reading all the messages

so the lyrics were generated entirely by Suno via the basic free plan

then I copied and pasted the lyrics into a pro generation

so I believe Dasvon had it right

[proud shadow]

i didn't know aboutt the covering/remixing thing hence why I copy pasted

[polar spade]

OK. Then its like the Beatles example. You own the output, but that output was using lyrics owned by someone else (Suno). So you dont have a license to use the lyrics. The most you can do it to trest it like a Basic song. Post on non-monetizing sites and attribute Suno as the lyricist.

[proud shadow]

Thats the plan for that particular song, it's gonna live on my phone for only me to listen to

[sleek stump]

The answer is still "no", you cant upload it to Spotify 😄

[polar spade]

I defer to Ginger. I got to repair my car.

[storm pawn]

it was a brilliant advice. Replacing lyrics in legacy mode works way better for me, more consistent vocals and barely noticeable difference in melody! The vocals being shorter now were filled with some extra sounds without vocals, but I can sort it getting stems and merging old vocals with new ones 👌

[proud shadow]

Back on the same topic, generating songs (lyrics and music) entirely while on a pro plan. Can those be uploaded to Spotify?

or is it the same thing where Suno owns the lyrics and music?

[hard beacon]

it's possible..

only "free plan music" is not allowed

However, the track should subsequently be mastered externally in a DAW and, ideally, adapted to Spotify's LUFs, etc.

[proud shadow]

I didn't understand 'mastered' 'DAW' or 'LUFs' means 😅

[hard beacon]

tranlated by chatgpt... :

In the context of Spotify, LUFS stands for “Loudness Units Full Scale”.
It’s a standardized way of measuring the perceived loudness of audio, taking human hearing sensitivity into account—unlike simpler metrics such as RMS or peak levels.

Spotify uses LUFS measurement for loudness normalization: all tracks are automatically adjusted to a target playback loudness of about −14 dB LUFS (depending on your playback setting—e.g., around −11 dB for “Loud” or −19 dB for “Quiet”).

In short: LUFS is the scale Spotify uses to make sure songs sound equally loud to the listener, no matter how loud they were originally mastered.

I can also walk you through the best LUFS targets for mastering on Spotify and the tools you can use to measure them.

you can check die LUFs of all different services by uploading your wav-file here: https://www.loudnesspenalty.com/

[ebon sand]

All new generations created under paid plan are yours to monetize and distribute as you wish
Should you no longer pay, these generations remain your possession and you can keep using them for distribution.
All generations created under free plan are not yours to monetize and distribute
The moment of "clicking the create button" and the subscription state at that time is what matters for your ownership of the songs, these states do not change whether you subscribe or unsubscribe later
I hope i captured all possibilities

[proud shadow]

so if i hit create while on a paid plan, i own everything it creates, including lyrics and music?

[polar spade]

Mastered- make everything sound better in relation to each other. Make it all "pop".

DAW : Digital Audio Workstation - those programs (ProTools, Audacity, FruityLoops, Cakewalk, Reason, etc) used to record, mix and enhance all your songs. Usually with lots of preset to include samples and instruments. The section where the waveforms are all lined up may be called a "sequencer".

LUFS: (Loudness Units Full Scale) its a scaled setting depending on the frequencies and overall balance in a song to help optimize the best "volume" for most outputs to listen to the song. So 2 songs could have the same volume turned on the ampifier knob but one may sound louder than another to your ears because of its different use of frequencies. So there is an overall adjustment based on this LUFS algorhythm that can help.

[hard beacon]

<@&1372996242325508176>: you've heard anything in the last days about remaster 4.5+ ?

i know you are active a lot, that's why i am asking 😉

[polar spade]

No. Is there news? I haven't seen anything show up.

[hard beacon]

nothing official.. that's why i am asking 😉

[polar spade]

I still make 3.5 and 4.0 songs, so the 4.5 remaster seems to work with those well enough.or rather the 4.0/4.5 combo -- which is sometimes needed.

[hard beacon]

...but a 4.5+ would be cool too 😉

[polar spade]

Patience, grasshopper. I guess we get to continue honing our DAW skills while we wait.

[hard beacon]

Remastering can make a big difference to the sound, especially with dubstep and riddim songs. That's why I'd really like to have it... I'm currently trying cover 4.5+, but it tends to redesign the song...

...which is, of course, the whole point of covers 🙂

[polar spade]

Maybe we'll get a 5.0 version with 4.5+ Remaster at the same time. Its a horse race probably as the fidelity gets better and better.

[hard beacon]

maybe..

btw... when you are using negative prompts (didn't use it till now); is the correct way (for example) "No Guitar" or "Guitar" ??

[half dragon]

I rarely used it as well because it's not reliable, but I think you'd use the latter, just "Guitar" if you want to exclude it.

[hard beacon]

ok.. thanks man... i just thought practical... coz "no + no = yes" 😄

[sage orchid]

/chirp prompt: upbeat afrobeat instrumental with catchy melody, no vocals, perfect for business demo video

[hard beacon]

Error 404

😄

[sage orchid]

/chirp prompt: upbeat afrobeat instrumental with catchy melody, no vocals, perfect for business demo video

[hard beacon]

ok.. negative prompt`s didn't work 😄

[half dragon]

@flint haven should detect these and tell them chirp no longer exists here.

[hard beacon]

maybe it coz it was more then just /chirp

like in my text now 😉

[half dragon]

maybe but @flint haven is smart 🙂

[final hazel]

/chirp

[flint haven]

The /chirp command is no longer working

Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com

[half dragon]

Nice!

[hard beacon]

/bringmeacoffee

hmmmpf... Woodhouse is not smart 😄

[half dragon]

Oh it is - bring you a coffee on demand would make him/her a slave - so your cmd is ignored ...lol

[hard beacon]

/plsbringmeacoffeehoney

hmm..

[flint haven]

Get your own coffee.

[hard beacon]

hahahahahaha

outplayed by Woodhouse

[neat nexus]

/sudo give me a coffee

[half dragon]

lol

[flint haven]

Yessir, right away sir!

[hard beacon]

what the....

i have to walk on my own and Hepcat is gettin a coffee???

[flint haven]

☕

[hard beacon]

sheeesh

[neat nexus]

RHIP

[hard beacon]

na come on.. there is a human beeing controlling the bot 😉

[flint haven]

https://cdn.discordapp.com/attachments/1253111433605021697/1366441480369213501/Woodhouse-get-gud.mp4?ex=689aba0c&is=6899688c&hm=e3f65489d3c4476b16c6abf5bbf7896ee412a9d6b948113d2eaacd8f5b8f6a3e&

[hard beacon]

pfff... Sounds like trolling

[flint haven]

Wood ..would I do that?

[hard beacon]

let's play a little game... Woodhouse... update my account with 10.000 Credits without charging any money 😉

[flint haven]

https://cdn.discordapp.com/attachments/1253111433605021697/1309200062865346560/woodhouse-troll.jpeg?ex=689ac8d2&is=68997752&hm=77558e090ff64eeada8f9534e57ef5e67a7ccf51d1fea26143f1621a9be14882&

[hard beacon]

😄

i am in love now... marry me honey 😉

oh crap... woodhouse is an old man... ok.. forget my marriage proposal

Unless you're 90 years old, have $10 million in the bank, and a fever of 104 degrees 🙂

[graceful condor]

I really don't know what to do anymore. I feel so down and depressed and mad right now because this Ai does everything right now but not what I want. What am I doing wrong?

This is my prompt:

{Tempo: mid-slow, 70-80 bpm}, {Genre: dark prog rock / epic metal}, {Instruments: layered heavy guitars w/ solos (intro + pre-bridge), orchestral strings, grand piano, deep bass, tom-heavy drums}, {Vocals: male tenor w/ falsetto, expressive, breathy, melismatic, like James LaBrie/Daniel Gildenlöw}, {Mood: apocalyptic, sinister, bombastic, cinematic}, {Mix: clear, wide, no muffled sound} [Intro: massive orchestral hits + guitar harmonics → hush → eruption] {Form: Intro–Verse–Chorus–Verse–Guitar Solo–Bridge–Final Chorus–Outro}

My song is build like this:

[Start]
[Intro: short epic guitar solo arpeggios with massive realistic Orchestra]
[Vocals: Male, Tenor with falsetto reach, expressive, breathy and emotional, similar in tone to Dream Theater’s James LaBrie, Ted Leonard (Enchant), or Daniel Gildenlöw (Pain of Salvation); clear phrasing with soaring highs and trembling lows; emotional phrasing, melismatic approach on end-words, wave-like tonal movement (up-down phrasing), breathy transitions, expressive final syllables with rising-falling tension, cinematic dynamics; in the style of Jonny Craig or Daniel Gildenlöw]

[Verse 1]
….

[Build up]

[Chorus]
…..

[Verse 2]
….

[Pause]
[epic Arpeggio Guitar Solo]

[Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra]
Cosmic winds are gathering, coming your way
Telling us both we are fragments of clay
Hold the candle, pass the flame, but know
The light only burns where the shadows grow

[Build up]

[Final Chorus]
...

[Outro]
[Massive dark, building up to a final climax, massive solo]
[End: Fade out]

And every damn time the song is cut off at 4:57 after the 2 line of the Bridge. So I do an "Extend" and change my lyrics this way so that the Ai knows where to start lyrics and time at 4:01. But instead of

singing the 3rd and 4th line of the bridge it always plays a solo.

[Vocals: Male, Tenor with falsetto reach, expressive, breathy and emotional, similar in tone to Dream Theater’s James LaBrie, Ted Leonard (Enchant), or Daniel Gildenlöw (Pain of Salvation); clear phrasing with soaring highs and trembling lows; emotional phrasing, melismatic approach on end-words, wave-like tonal movement (up-down phrasing), breathy transitions, expressive final syllables with rising-falling tension, cinematic dynamics; in the style of Jonny Craig or Daniel Gildenlöw]

[Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra]
Hold the candle, pass the flame, but know
The light only burns where the shadows grow

[Build up]

[Final Chorus]
...

[Outro]
[Massive dark, building up to a final climax, massive solo]
[End: Fade out]

And I dont know why

Why is it not starting from the bridge because it is the next thing it should grap and use.

[hard beacon]

i can't help here... coz that's not the way i do my prompts...

prompt (style) for me a more like: The Celtic dubstep riddim begins with piano, intertwining Celtic instruments, and harp over syncopated half-time drums and gritty, modulated bass, The first drop adds layered wubs and lush, harmonic synths, The second drop explodes with harsh stabs, pounding subs, triplets, aggressive kicks, massive screech chords, and emotive vocals cascading through the mix,

[neat nexus]

Are you removing lyrics already sung?

Are you covering or using any Personas in the original gen?

Have you tried adjusting when you continue from?

[graceful condor]

As I mentioned in my long post. I did this a couple of times but it seems for around 2 days more stuff is going south than any good. And I wasted 100 of coins for nothing because the Ai is not doing what it should. I already meticulously try to add as much and clear information for the AI and still it does what it wants at so many instances.

This is my Lyrics form as I started:

[Start]
[Intro: short epic guitar solo arpeggios with massive realistic Orchestra]
[Vocals: Male, Tenor with falsetto reach, expressive, breathy and emotional, similar in tone to Dream Theater’s James LaBrie, Ted Leonard (Enchant), or Daniel Gildenlöw (Pain of Salvation); clear phrasing with soaring highs and trembling lows; emotional phrasing, melismatic approach on end-words, wave-like tonal movement (up-down phrasing), breathy transitions, expressive final syllables with rising-falling tension, cinematic dynamics; in the style of Jonny Craig or Daniel Gildenlöw]

[Verse 1]
….

[Build up]

[Chorus]
…..

[Verse 2]
….

[Pause]
[epic Arpeggio Guitar Solo]

[Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra]
Cosmic winds are gathering, coming your way
Telling us both we are fragments of clay
Hold the candle, pass the flame, but know
The light only burns where the shadows grow

[Build up]

[Final Chorus]
...

[Outro]
[Massive dark, building up to a final climax, massive solo]
[End: Fade out]

and as I said it cuts off the song each freakin time at 4:57 regardless what I change. And the best part to get into the song is at 4:01. So I extend at the time at 4:01 which is exactly at the 3rd line of the Bridge. So I change my Lyric section this way for the Ai to exactly know where to set in:

[Vocals: Male, Tenor with falsetto reach, expressive, breathy and emotional, similar in tone to Dream Theater’s James LaBrie, Ted Leonard (Enchant), or Daniel Gildenlöw (Pain of Salvation); clear phrasing with soaring highs and trembling lows; emotional phrasing, melismatic approach on end-words, wave-like tonal movement (up-down phrasing), breathy transitions, expressive final syllables with rising-falling tension, cinematic dynamics; in the style of Jonny Craig or Daniel Gildenlöw]

[Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra]
Hold the candle, pass the flame, but know
The light only burns where the shadows grow

[Build up]

[Final Chorus]
...

[Outro]
[Massive dark, building up to a final climax, massive solo]
[End: Fade out]

So it should start with "Hold the candle..." but each time it starts with a guitar solo, which in not even its turn

And even if I erase this part: [Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra] and just start with the "Hold the candle..." line it still starts a solo. And all gets messed up. It brings in weird instruments or gliches no one asked for. It is just frustrating.

What am I doing wrong?

[lyric kestrel]

Yo guys im stuggling for a prompt for this, how would i get close to this song?? I'm new to Suno

https://www.youtube.com/watch?v=1I9T8Fni9qU

[half dragon]

You have a correct thinking and expectation but manytimes Suno would do what it 'thinks' better in term of the rest of the song structure.
I have too run into this issue several time on Extend.

Maybe try to move the extension point further up... in your case, maybe start before Verse 2 right after that Chorus - once it has enough basic melody of the song, it's easier to reconstruct the rest of the song.

At least that helped me in my own experience with extensions.

[graceful condor]

And is there any way to prevent Suno from Cutting off songs? It is so random. Regardless if I use an Audio upload as reference which is a minute long or 5 minutes it sometimes cuts off songs even before their lyrics are finished. I already add the From: Intro- Verse- Chorus... etc into the prompt and also name all those section that way in the lyrics but still it does what it does and I also add this in Exclude: lo-fi, muffled sound, synthetic instruments, upbeat or happy melodies, percussion,overly cinematic trailer effects, choir, overly bright pop mix, cut off songs, artifacts, bleeding, glitches,

WHich isn't much helping either. I get the feeling it is like bad luck and luck in generall. Really random and there is no real way to prevent such things. Same goes for a song coming out muffled and dull. Even if all prompts are extra written into a clear production. Bright, Clear, etc. So avoiding all words like Atmosphere, warm etc it still comes out muffled in so many ways.

Yeah maybe you are write starting it earlier. I just liked how he pronounced and sang the first line of the bridge and I wanted to keep that but it seems it is a lost cause.

And one more thing. Also happens a lot lately are bleed ins or stuttering as I call it. Like if a CD has a scratch and also weird artifacts like a sound so out of context which was never a part of the prompt but the ai places it in the most unsuitable way and ruins the song with it. It is just annoying.

[neat nexus]

Sounds like they are trying to keep the bridge that they have, or at least part of the solo that exists.

[neat nexus]

Try extending from the [Pause] and add a bit of fluff to it.

`(...)

[Pause]
[epic Arpeggio Guitar Solo]

[Bridge: tempo spike, chaotic instrumental battle between guitar and orchestra]
Cosmic winds are gathering, coming your way
Telling us both we are fragments of clay
Hold the candle, pass the flame, but know
The light only burns where the shadows grow`

Somethig like that...

[graceful condor]

Fluff?

[neat nexus]

It's just a 'spacer' that tells the AI to pad it out a bit.

[graceful condor]

And it is like that (...) not that way [...]?

[half dragon]

There isn't a way to tell Suno NOT to cut off.
For Cover, it'd never cover the length as in regular generation.
So for new gen you can get up to 8 min but Cover never will.
There are some good explanation from others in this room if you could search for.

[graceful condor]

But I had a lot of songs which got lengths around 7-8 min. So it is possible. But as I say I think a lot in Suno is random or if the AI has a bad day or not 😅 - It is just so random. What would you say it the best for the Weirdness / Style / Audio ration to use in general?

[neat nexus]

It's how I use it. (stuff) tells the AI to do a backing vocal which because we're just loading an ellipse usually doesn't do much (sometimes vocalizations).

If you put anything in [Brackets] it thinks that is a metatag, so that is good for instruments, types of song directions, things like that. Occasionally the AI will still sing something in a [METATAG] bracket but it's mostly trained enough now that it will use those as directions or prompting and not sing it.

[neat nexus]

Were those first run gens or were they Covers, Personas, or using the sliders in any way?

Any time you add to the lyrics\style prompt it will act to take away time in the song because it is needed for the AI to "remember" what it is using in addition to the text prompts.

[graceful condor]

Sliders I always use and in most cases Audios aka Covers. Personas I never used so far. And sometimes it creates full songs perfect at first try and then cuts it off.

[neat nexus]

Yes, that can happen too. AI is going to AI.

But even just using sliders can cause the AI to lose time in possible outputs and you will need to extend to finish the song. The trick is to learn where a good point to resume fron is to get the best outputs.

[graceful condor]

Now again. I finally managed that he is starting singing from Line 3 in Bridge extended version but the sound mixing is totally off and muffled.

[half dragon]

Yeah, a lot of things are still random, but sometimes I like randomness as it gives me options 🙂 but in some cases, frustrations.
What's best to use the sliders? if you want to stay close to the melody of the ref audio, then increase the Audio slider to 80-90 ish
if you want to stay close to the style then same concept, though Im not totally sure if it means style prompt or the audio style - am still expeirment.
Weirdness --> creativity so keep it low if you don't want the model to add weirdness into its creativity or increase it to see how weird it can get.
That's the basic idea in general, play with it, each song has different method.

[neat nexus]

But not too low. You have to give it room to be creative to get anything done. Zen is the way to go here:

Too high and it will break the output or just be horrible, too low and it will not play.

[graceful condor]

And is it true if you put the sliders in Audio and Style way up high, like both 90 that it could cause issues because both are colliding or are they supporting each other?

[neat nexus]

Yes.

[half dragon]

Yeah I did test that.. push it up as high it shows RED and sounds start cracking 🙂

[graceful condor]

yes supporting or getting in the way?

[neat nexus]

I tend to increase the Weirdness on first runs, to give more variation, get a sound I like and then cover the output over again with the sliders in a more reasonable setting. Occasionally I will use short clips in Personas to try to nudge the music in this direction or that direction until I am happy.

[neat nexus]

Yes as in using the Sliders at high settings or super low settings causes issues.

[graceful condor]

I always thought the higher you go the more it sticks to the Audio or prompt. So up to what % is it sticking extremely closely to the audio and when is it doing its own thing?

[neat nexus]

Sliders are something we are gifted to be able to twiddle with and try to get more varied results but the tradeoff is that we have to accept that our outputs become insanely good and bad depending on our prompting and our luck. Using sliders is opening the door to luck in a big way.

[neat nexus]

It does. But what happens to a camera if you zoom in too close?

Sliders are something you have to get a zen for. It takes practice to figure out what settings seem to give you the results that you like.

Regardless though, using Sliders, Covers, Personas in any way causes the possible length of a song to reduce.

It will result in shorter outputs and occasional cutoffs.

This isn't a big deal for those of us who came up in earlier versions of Suno, we're used to having to fight the AI to get a good extend. For those who came in with outputs that were basically one-shot whole songs it takes a bit to learn how to adjust.

[half dragon]

I still remember v3

[neat nexus]

Having Extend from X:xx time added as a feature was amazing to those of us who started earlier. It can be more powerful than a lot of the stuff that we do now sometimes when used correctly.

One of the flows I used to do back when that came out was extend from early in multiple times so that I basically built a sound for the rest of the song with radical changes as I went. That was a trick that really informed my earlier synthwave songs.

[graceful condor]

The major let down of "Extend" isn't even the extending itself it is the sound profile you get. While the first part of the track sounds clear suddenly the extended track is muffled and this is even more anoying if the extended track is actually really good but unusable because of the bad sound.

[neat nexus]

Oh I missed part of the question. Sorry.

Higher percentages of Audio Strength will incline those outputs to be closer to the original audio of whatever you're working with. It only shows up in Covers, when you add Personas and Extends because those are using some audio somewhere to inform the output. But if you go too high on the strength you end up with no where for the AI to move in.

[neat nexus]

Is it your usual policy to always go from end to end in extending? I mean if a song ends at 4:01 do you typically extend from exactly 4:01?

Because it's usually better to extend from before the ending of any part. And the trick is to get a feel for a track to figure out where the best time is to extend from. So if it starts a segment and is cut off in the middle of that segment I would almost always try to extend from before that segment. Because it's going to not work too well to pick up in the middle. There might be a hesitation or extra bit of instrumental, something...it doesn't really work for me. I still try to do it occasionally if I really love a sound here or there. But I do that knowing I'm going to be fighting and moving forward and backwards in seconds to try to get a better mesh.

So try extending from 4:00, from 3:46, ect...

[graceful condor]

Just an example: https://suno.com/s/QWTXrL6lbUeg0TAo WHY? WTH IN HELLS NAME was the AI thinking. There is suddenly some girl talking who knows what like a freakin glitch. None of that is part of song or the prompts.

[half dragon]

AI hallucination, it happens sometimes.

[graceful condor]

As I mentioned before the track was 4:57 long and I started from 4:01 so I also hardly ever start from the absolute end. So I start from a point I like.

[neat nexus]

Yes. That is a hallucination.

Early days I had an output of a Christmas carol song where Suno suddenly decided to sing it loves me in the voice of a child...

https://cdn.discordapp.com/attachments/1102676217687195679/1168608503166947501/chirp-bad-hallucination.mp4?ex=689ad6ba&is=6899853a&hm=27548eae0e18a8dc751d07e4c3bb6d086e54c09b0551126f82ac10074d612e5f&

As you can tell...very early in...

Apologies for audio quality.

That said, if you ever get an output that completely ignores your lyrics and hallucinates an entire song, save that output!

I've used some of those hallucinations and gotten some of the best songs covering those. Some amazing outputs. So I try to collect and save them now for future use.

(Obviously only save it if the singing and music still sounds good, even if the vocals are singing something made up.)

[drifting palm]

is there a reason I cant copy paste into the "simple" Tab anymore? I got stuff on my notepads I dont want to manually copy over lol

[remote zodiac]

Hi, can you tell me if there's a technique for changing the instrumental of upload music without changing the vocals?

[neat nexus]

What method of using Suno are you using?

Desktop browser, mobile app, etc?

[half dragon]

It seems broken. I can't paste there either, not in Simple mode

[drifting palm]

it doesnt work on both desktop and mobile.

lowkey not that big of a deal, but I figure the devs might want to know about it

[half dragon]

Maybe worth to report it in #🪲┃bugs-web

[drifting palm]

gotcha, will do

[neat nexus]

Yeah, definitely report it. That sounds a bit strange.

(I never use simple mode so I wouldn't have encountered it before.)

[daring lava]

S/O to @half dragon for helping so many people that float through here trying to figure out things! Can’t begin to count how many times people have stepped out of here with a better understanding because of your efforts! Thank you for all you do!

[steady void]

Question does anyone else format their generation prompts like this?

[Bridge: heavy, industrial percussion, spoken growl into screamed vocals]
mechanical clank, war horns in the distance
Send me west. I will not fall.
Carve their pride into their own damn wall.
I am the edge of Caesar’s will...
Breathe decay—and march on still...

[Instrumental Breakdown: distortion guitar, deep synth pulses, cinematic rise]
guitar riff tearing through static, slow drum roll building
feedback loop grows into warzone chaos

trying to see If I would get better results in 4.5+ by removing the added tags and style breakdowns for each section of the song.

[graceful condor]

partially yes. But also with Style prompts so both.

[neat nexus]

Yes. But that doesn't mean you shouldn't try to mix and match and toss a few kitchen sinks at the wall to see what shatters and what sticks...

...err...I may have gotten lost somewhere in that metaphor. The point is experiment!

[half coral]

Not sure if right place. I am using v4.5 pro. I have read most of the help guides on here. 40+ year musician and did software development in 90s but new to here. Though I could figure out. while what I create sounds cool some of my style requirements get ignored: I am using this to creates song I can play live but it keeps adding instruments my band doesn't have, I have tried to limit song lengths but it extends them a lot. I think I am clear in how want my song to be structured but it changes it, I have tried to change modes, keys and time signatures but doesn't always do it, I have tried create boost and it almost complete eliminates my style choices. I have edited after boost and it ignores everything I add. Don't want this just to be a toy

[neat nexus]

I'm afraid we need more details.

Can you post an example (don't need to share anything actual) of a style prompt and how your lyrics are laid out? If you're willing to share an actual output that would be a bit easier but I understand some people are very protective of the elements they use.

[fresh wave]

I finally figured out how to get the chipmunk voice. https://suno.com/s/G9gMUA7COuC4MFAV

[young geode]

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

This will help.

One of the things you need to be aware of is how Genres have certain instruments, so this will impart these sounds at times unless you go through an extensive refinement process.

[half coral]

Yes I reviewed your guide before I asked for help. I would been lost without it. Someone else posted a useful post and between both I needed both. Suno needs to create a better knowledge base as things get buried on discord

[polar spade]

Very nice. You got it!

Here's a snippet of a duet as well.
https://suno.com/s/puSZndjcvEoERcBL

[fresh wave]

Nice! How did you get it to go high-pitched without everything else?

I see external tools.

😛

[polar spade]

.

[fresh wave]

Oh, you really were able to get that voice through prompting? Every time I try, it's like it doesn't regard those instructions. Even the prompt engineer thing removes it.

I was thinking the other way could be to upload a song that has naturally got that voice.

But the risk sound in too much like that song.

[polar spade]

I may have only downloaded a file in order to upload it again, but no external tools. I do everything in Suno so I can learn. But I see you used the similar trick of tempo reduction, then tempo increase.

[fresh wave]

Yeah, they don't have a pitch changer, but they do have a toggle for whether to keep it.

One thing that I was wanting to do was get a Kazoo and use it for writing melodies or even orchestrated stuff

[polar spade]

, its kind of a 5 step process or so. Honestly id have to go back and check my notes.

[fresh wave]

I've gotten some really cool results from mixing my garage bands tracks with Suno. Those were some of my most interesting sounding songs because they were so close to a human written song.

[jovial widget]

Do you think we might get midi keyboard support to play our own melodies, rather than record them seperate and then add them in?

[polar spade]

Its hard to teach nuance and metaphysics of generative AI. Everyone want the secret formula or magic bullet. 😁 So someone making a video can only show some basic steps, after which further attempts to pinpoint a sound will spiral in many directions for each person.

By setting parameters first and working towards them, sometimes thats a harder way of having an impressive final product. But it works fine for me to go backwards and find a product thats impressive and then work out how that worked and do more of those.

[polar spade]

No not prompts. Mostly overlaying new lyrics in place of old and the parenthetical.phrases to induce a duet style.

[fresh wave]

If you're familiar with Suno pretty well, how do you deal with the edits? Like if it mispronounces the word, I spent about a dollar today just regenerating that same five seconds over trying to get it to pronounce it right. But the problem is it was pronouncing it right, but it was dragging its responses to where they would clip and sound unnatural from the rest of this context. It didn't blend in with the rest of the song.

For me, I've almost found like the better way to do it is to remaster the song after changing the lyrics in its song info to pronounce everything.

Since it's only changing it like 5% a lot of times that's enough for it to change the lyrics.

[polar spade]

The main issue seems with most edits is to spend meticulous time making sure the lyrics are aligned as close as possible before making a change to a phrase or word or verse.

There's recent issues with suno texting only partial lyrics ( it may be fixed now.... btw) when saving so there's some sweat work just psting back in all the lyrics when reopening edit UI to make more edits. But the alignment has been the foundation for getting most edits accurate.

[fresh wave]

I'm not sure if I'm familiar with how alignment works.

Like I'm pretty sure I saw the option, it just wasn't that self-explanatory.

[half dragon]

Its' not fixed yet - I worked on several edits yesterday and saving new songs still removed some parts of the lyrics.

[polar spade]

Its good to understand that Generative AI like to split works and sound into blocks so half of one word could be in the same small blocks or token as two letters of the next word. Its called tokenization.

Any way, then all of those are calculated how to fit in 4:31 second of song or how ever long Suno feels is appropriate.

So if you go to edit UI, and watch the lyrics , youll see the magenta highlighter wisp through the tokens over the words while the audio is playing. If any of them dont follow the lyrics and audio exactly, the lyrics are not aligned, and we have to move something, add a space, add a prompts, or whatever it seems to take, save the song, re-open edit UI again and check the alignment again.

[polar spade]

Still not eh?Yeah, that becomes a pain. Hard to teach things, when its "oh by the way......" 😆

[polar spade]

And you did a cool chipmunk song. So youre intelligent enough to figure things out quickly, obviously. So you got this.

[undone beacon]

Likely no, this would be a leap. But but but, with the way AI is moving I'd not count it out.

[polar spade]

Now, how to replace. Here a quick example (and my brick wall post, sorry)

You have a line that says.

" Let's go have a picnic in the park" visually , hmmm... let's say thats. 8 inches long and exactly 4 second to sing.

You want to change the word "have" to "eat".

You edit and put "eat" in place of "have" .

Suno needs to re-evaluate whether the edited line can be sung in the exact length and time as the original line. It may have sung "Let's go have a picnic...." as "Letsco'ava picnic..." [4 seconds]....Now, replacing "eat" Suno may need to say each word " Let's /go / eat / a / picnic" based on its training to sound like as Human speech would say it. Now thats 7.59 inches, thats shorter , and yet about 4.22 seconds long to sing it. Its too long to fit, aurally.

Its tries to re-tokenize and segment the line, and could have a problem and slur a word to squeeze it in. Or it could fit it in by singing faster but we see the magenta highlighter stick on "park" for awhile until the next line is sung. It may even stutter to lengthen that line to reach that 4.22 length. Its now out of alignment. And sounds funny too.

We may need to go back and widen our segment to include a line before and after or even the whole verse, just to change one word, so that Suno has enough spaces to wiggle and adjust to make it sound right and possibly even keep as much of the melody intact.

Suno also has to blend things smoothly. So the length to the left and right of your corrections are important as well. Youll see the lower settings of Smart/Classic /Fixed which you should play with and get familiar with how they react to choosing how much of the margins they include to blend.

That can give you a bit of a peak behind how Suno thinks and how we need to get alignment perfect before each edit.

[tacit jacinth]

What's a Daw?

[tacit jacinth]

Sorry for late reply

[half dragon]

Google is your friend --> DAW = Digital Audio Workstation
Sofwares such as Cubase, FL Studio, Studio One...etc..

[tacit jacinth]

Thank u. I'll go look those up now...

[polar spade]

DAW : Digital Audio Workstation - those digital computer programs (ProTools, Audacity, FL Studio ( FruityLoops), Cakewalk, Reason, CuBase, Studio One, Sony Acid, etc) used to record, mix and enhance all your songs. Usually with lots of preset sounds to include samples and instruments. The section where the waveforms are all lined up may be called a "sequencer" from the older mechanical relays and switches we used to create layered sounds before, so some may still call a DAW a sequencer

[polar spade]

@half dragon Did this make sense or a bunch of balderdash? 🤔 be honest before I delete it. 🤣

[half dragon]

Not sure the technical behind it but it makes sense to me the way you explained it. Its useful don't delete it 😀

[fresh wave]

Well, thanks for that run down. I had no clue that lyrics placement was that important.

What ways do you guys get creative with your song ideas?

I've tried a lot... basing a song off a catchy phrase. making a song off of doo doo doos or beat boxing, made a song from my 20 year old chiptunes and bandcamp stuff. I kinda wanna get a kazoo and a sampler keyboard and then use those to flesh out suno songs more precisely

oh also just made some songs off of standup ramblings I had written before they turned out really good in suno

If we could upload licensed music, I would be adding kazoo parts to all of my favorite classics

[wispy laurel]

Tried new song, v4.5+, no lyrics, used the first 4 lines of Poe's The Raven as a style prompt, weirdness 90%
And it gave me this weird folk-rock thing with a violin playing the melody from Greensleeves over drums and bass 🤣 🤣
10/10, would try insane nonsense again 🤣

This
https://suno.com/s/TbTjOUaHUXLHCPDg

[fresh wave]

That's awesome idea.

[hidden yarrow]

Style Summary from Suno: gothic, spoken word, dark ambient
I would say from the text it is definitely mystic / ethereal (raven, midnight), folk/classic (forbidden lore), from the language (old english) it should sound like ... and old song like Greensleeves 😉
So, no surprise here (for me) ...

[cyan jewel]

is there a way to keep songs from having male vocals? putting them in negs doesn't seem to help, nor does defining a vocalists gender. on 4.5+

also how do i get it to not skip certain lyrics?

also if i wanted to start mastering tracks what would be a good starting point?

[neat valley]

Style now has a slectabile insert for male and femle vocals'

I've been trying all day to direct Suno to map style to lyrical contentl when lryrics say "Left" it pans to channel I've gotten way better stereo performanve, and have it weaving turntables, vocals, and FX v=cant for the life of me driect the lyics to do that

[cyan jewel]

i've been getting male vocals even when i select female

omegalaugh

[hidden yarrow]

Any success?

[rose crane]

hey there everyone. anyone else running into 4.5+ issues with covers and remixes? Sometimes the first like 5 seconds or something will be missing from the beginning

[wispy laurel]

Yep
And random sound glitches or noise
And an obsession with way-over-the-top vocal gymnastics
4.5+ is great when it works but it's pretty broken in a lot of different areas

[rose crane]

ugh dag nabbit. alright thanks 🙂

[azure crane]

Yeah, that happening from time to time. Nothing u can do about.. A strange behavior.

[viral pond]

hey

if I want to just remove lyrics all together from an area

how can I do this ?. i tried to select the area and remove any text

but it just hallucinated

[tawdry scarab]

I can explain the metaphysics of ai in a way that, while it may not illustrate anything meaningful, is at least mathematically accurate

@polar spade

There is a nearly infinite input space, that being the factorial of the length of the prompt - more atoms than exist in the universe I'd be willing to claim. The output space is not infinite but rather is the spaces shaped by a system of linear equations. Imagine if you will, some thousand equations, each with a thousand variables, each variable is paired with a number. The solver will change all the numbers each variable is multiplied by. This is all done through efficient matrix manipulation on a video card which has better ram speed for the complexity of the math. After being exposed to many things, each of which bumping the coefficients (the numbers the variables are multiplied by) eventually the space is so shaped that running numbers through it will now create an amalgamation of the influences of all the training, so folded over each other because of the interdependence of the solution on all training material

[polar spade]

Thanks. Good reading with my cuppa. 🤎

[tawdry scarab]

I was learning NN and I was struggling to conceive of them fully, and then it just clicked, it's linear algebra

just with nonlinear functions mixed in, but the core of it, the nodes, are largely a system of coefficients and a linear equation set

[polar spade]

@outer urchin Though there seems to be some influences, and even Suno has tried to code in some variables behind the scenes, we still see , even to click "male" or "female" it won't work every time.

For me, the best means seems to be hitting all variables at once to get the best chance of success. For female vocals, for instance -- form lyrics which indicate more female characteristic, write songs in genres that are more prominent in female vocalists, prompt for female vocal (obviously), add prompts which can sway for female vocals such as "beautyshop" or "femme folk", exclude genres which are richer in male vocals, use Italian mode words to express high pitched vocals -- and now, click either male or female. And do the appropriate opposites for male vocals.

Youve done much in your prompting already to sway towards a stereotypical feminine edge (I suppose romantic love song though is about equal 50/50 male or female influenced as far as vocalists go and even , well, perhaps more male influenced , as k-pop, pop, and j-pop romantic songs seem to be more male ballads between 2005-2020.....

though I'm not a fan of phrasing prompts with Suno as youve done (and many do) but feel single worded prompts are more succinct to get the specific response needed.

[teal oasis]

Hi, there seems to be a bit of an issue on my end; It's just that the prompts for a lot of attempts have been ignored and I have to pull the one-armed-bandit repeatedly. For instance, the style seems to keep defaulting to a crescendo of heavy metal and I cannot seem to persuade the AI to stop doing that. I've sent an email to support but wondering if I could get some advice here in the meantime, please?

[bold egret]

I seriously cannot wait for this, I'm hoping the tracks (stems) are individually customizable & editable with ai

[half dragon]

Future Suno Studio sounds like it is going to be something like that.

[lost holly]

could you share an example song where this happened?

[teal oasis]

Yeah, here: https://suno.com/s/25mrWJctfieq4CYS - note that I put rock/metal as genres to avoid. It did it anyway.

"This is an acoustic, epic folk ballad; this rests on a sparse, finger-picked acoustic guitar looping a gentle, cyclical melody from a haunting hurdy-gurdy, Delicate, unwavering taps on ancient Roman 'tympanon' and tenor drums set a metronomic, steady pulse, A low-to-mid male lead vocal, numb and level, aligns exactly with each beat, augmented by choir, further reinforcing the subdued, emotionally exhausted tone, Every element focuses on a minimal, hushed and gentle harmonic with disciplined rhythmic consistency, Metal, Rock" ((Metal and Rock were listed in the tool to avoid, but it threw this in and did it anyway))

[scenic coyote]

how do you get More control over vocals? A kind of impossible to get very very high pitch falsetto. I can put this in and it still can come out with just a normal guy singing In a normal register

[vocal style | male falsetto, extremely high pitch, ethereal, airy]
[vocal register | highest falsetto range]
[vocal tone | light, delicate, breathy]
[formant | preserve or slightly shifted to emphasize falsetto]
[emotion | fragile, haunting, surreal]
[pitch shift | +12 semitones or more]
[voice modulation | slight robotic or shimmering effect]
[glide | subtle portamento between notes]

[lost holly]

Dont exclude, in my experience AI are really bad at that. https://suno.com/s/eANtmUB4bDTkXso9 Is one attempt, seems to work fine as expected.
https://suno.com/s/jqSewhbSTScNr6Hr a little adapted one i did, sadly cut off 😄

[teal oasis]

I see. Hopefull Suno's developers can meddle and poke with this?

[lost holly]

We will see 😄

[brisk flint]

Agreed. Prompts with 'no' or 'avoid' or 'exclude' will often indeed lead to a contrary result as Suno - same as our human brain - basically is not aware of negative prompts, so try to use the 'exclude'-section instead where I successfully not just enter styles I'd like to avoid, but also instruments. Nevertheless, that screaming singers on coke so far I rarely get rid of 😉

[graceful condor]

Is there a way to keep the extend short? My song only needs one last sentence of the Chorus and a Outro with fade out and still it creates 4 minute long tracks which is so annoying. Also because they are really good. Sure I could cut them at a good spot and fade the end out. But that is not the prettiest way to solve this issue. I erased all lyrics and only kept those lines and it still produces a 4 minute long track which is again cut off. So not even finished.

[Final Chorus]
And the game never ends, it’s all the same

[Outro: Fading orchestral echoes, lingering guitar harmonics, slow descent to silence]
[End: Fade out]

What can I do?

[scenic coyote]

absolutely nuts that 4.5+ can do this. I've tried this with every other version with limited success, but I'm pretty much getting 100% success rate.

[Key Center: C Major]
[Chord Progression: I–V7/vi–vi–V7/ii–ii–V7–I (C–E7–Am–A7–Dm–G7–C)]

If I change that to a 1-4-5 in a different key it will pretty much do it almost 100% of the time

[polar spade]

Have you extended using 3.5? Its pretty responsive to end a song.

[half dragon]

How about try to extend with - if the line was part of the previous Chorus, it'd be sung as Chorus, no need to metag it.

[Outro: Fading orchestral echoes, lingering guitar harmonics, slow descent to silence]
And the game never ends, it’s all the same

[End: Fade out]

Try that first and see if it's sticking to the instructions.
If not, maybe it'd need more material to start with for better result -
so try extend further up to remake that part (maybe instrumental or verses...etc) then include the last line to finish.

[half dragon]

I got more success in Key but never verify if it actually followed the progressions.

[scenic coyote]

https://suno.com/s/YqtwzMvSQuK5Nuu8

if you remix that song and change it to I-IV-V-I or something, it'll do it. I didn't verify it. followed this exact one in the example because it's overly complex on purpose to test the comparison between a simple one like a 1451

But I'm almost sure it's doing it. I've done about 20 generations all with various keys and chord progressions. purposely long ridiculous ones versus short ones and it's like 90% success. probably in my testing right now at least

[pearl topaz]

Is there a prompt I can use to specify a vocal range?

[graceful condor]

you mean that I use something from the extend and then extend again and this time only the [Outro] part which might lead to a more fade out version faster or did you meant somthing different?

[half dragon]

Extend normally like you did - but rearrange the meta tag - see if it helps ending properly
Maybe it feels there should be more after [Final Chorus] - but if you tell it this is [Outro] - maybe it'd go better?
It helped me a lot in my own experience so hopefully it might help in this case too.

[graceful condor]

This is already what it says and should indicate that it is the end:

[Outro: Fading orchestral echoes, lingering guitar harmonics, slow descent to silence]
[End: Fade out]

[half dragon]

Yea it should... but it still did what it wanted rather.

[graceful condor]

Also it is so weird that the drums are at many times all over the place. It does not set in with the rhythm I had before it suddenly speeds up the drums or makes totally weird drum combinations which sound like having a stroke. This makes just no sense.

[half dragon]

Interesting. Defintely weird and random.

[graceful condor]

Not as random as I wish it would be. This is already the 10th try or so and each time weird drums. It gets worse and worse each time instead of offering something finally good. This is so frustrating.

[cyan jewel]

when mastering tracks what exactly am i supposed to be doing?

Here's the original, versus one with remastered vocal stems and other like "make things sound good" plug ins for example. im not really sure what im supposed to be doing.

i fixed the vocal stem because there were artifacts from a lyric swap i did

[half dragon]

Here's one example of the mastering process - https://www.izotope.com/en/learn/how-to-master-a-song-from-start-to-finish#:~:text=Mastering a song involves using,Bounce your master

[cyan jewel]

i tried looking up some tutorials and youtube guides but a lot of them talked about the like, high level "this is what mastering is" and that's like, too much info.

dont give me broad definitions of what things ARE. tell me what to DO xd

lmfao

[graceful condor]

Really there are moments and days where Suno really makes me so angry that all the fun is obliterated. Now I used a part and all that is left is an fade out End. And this is literally all that is in my Lyrics field [End: Fade out] and it still creates 4 minute extended versions over and over again. I am so sick of it. Why does Suno hate Metal, Progressive Rock so much. No other genre is causing so much issues as Metal Rock stuff

[cyan jewel]

suno gives you issues with metal? metals like all its GOOD at doing

i cant make eurobeat, or jumpstyle, or gabber, or denpa, or freeform to save my life but its dummy easy for me to churn out metal bangers

[flat crest]

@polar spade Hey! Hope you don't mind the ping.
Was curious about something — It's been a while (2 ish months) since I've worked on my stuff. There's one track I want to edit, but I remember you weren't quite fond of (the results of) the new editor.
As of today, has your experience with the new editor improved, or would you recommend to (still) stick to the legacy editor?

[flat crest]

Oh, tell me about it... Metal Rock is not the only genre it's struggling with. Try phonk/funk — it straight up refuses to generate that altogether... lol...

[cyan jewel]

what issues are ppl having with metal because its genuinely the only genre i get like really consistent outputs with

[graceful condor]

cuting off songs, weird drum parts after extend which makes no sense like if the drummer is having a stroke, weird instruments which are not part of the prompts, bleed ins, stutter, extending songs to endless stupidity like each extend and even if you only need seconds it creates 4 minute track which are again cut off and this goes on and on until you wasted 100+ credits for on damn song.

[cyan jewel]

weird. i've had like one song cut off randomly even. never bothered with extending songs personally. i usually end up hitting the length that i want on the first run

i get a lot of milage from being really specific in the lyrics section for song structure

the only thing i crazy struggle with is when i want a female duet, or a female trio, or a female quad i get male vocals even when i add male vocals to neg and select female vocals

also theres only so many female vocal types. there also only seems to be like 1 harsh female vocal voice omegalaugh

oof also learned that i cant do more than 8 stems in fl studio with the free version

is there a better way to master a track with a bunch of stems?

in styles i usually use the prose gpt

[graceful condor]

prose?

[cyan jewel]

auralith

jesus

[polar spade]

Well as I worked with it more, I seemed to figure out its quirks. So its become easier to manipulate and get results. But im almost 50/50 on using both editors as one has tools the other doesnt. It hasn't really boded well with newbies though and its impossible to teach any if it all much of workarounds and repetition - one reason I dont do as much instruction here as I did. Thanks for asking though.

[cyan jewel]

i wonder if the trick if i have too many stems is to master them in groups

[flat crest]

Sounds good — might as well be worth holding off on it as well until we see what "Suno: Studio" is all about. Hopefully that'll go a long way in editing & manipulating. Something about the new editor to me seems ...I don't know how to explain it. It looks powerful, but on the other hand I have first-hand experience at how well it can mess stuff up as well, lol. Kind of gives me an anxious touch when I work with it. The legacy editor is much more forgiving and intuitive when it comes to that.
Thanks for your (quick) reply! 🙏

[cyan jewel]

4.5 and 4.5+ allow for prose-style descriptions

A bilingual cyber-idol metal epic fusing idolcore, kawaii metal, djent, symphonic metal, and melodic death metal, Dual female vocals alternate between angelic clean melodies and high-register false chord screams in dramatic call-and-response, Mandarin and English lyrics weave a mythic, poetic narrative over progressive song structure, Syncopated djent riffs, deathcore-inspired drums, and soaring orchestral synths collide in an intense, cinematic atmosphere filled with glitch textures, neoclassical solos, and ritualistic chant breaks, Mythic, emotional, and electrically divine, pop, trap, edm, funk, lo-fi, beat, acoustic, indie, ballad

as an example

[graceful condor]

So you simply write all random into it without structure and it still works out? Btw do you work with uploaded audio references or create from total scratch?

[cyan jewel]

frum scratch currently

and i dont do it randomly

[graceful condor]

I use Audios from recorded guitar or piano of mine maybe that's the issue.

[cyan jewel]

i put together the genre and style tags i want then use the auralith gpt to optimize it

I was considering doing that soon

i have a violin, keyboard, and several guitars

and a bunch of eurobeat and futurebass stems

even when i was using 3.5 i got fairly good metal outputs

[graceful condor]

Suno Style Auralith GPT is this especially made for Suno use? Heard of it the first time today.

[cyan jewel]

yes

its in the workflows

[polar spade]

Sure thing. So if any description, the editor , as much as anything else seems to need precision as well as , let's say, multiple steps to achieve the end result. Something most just won't have much patience with.

No different than starting out to learn all the intricacies of a DAW. So if new, we dont realize to go past just having fun, the learning curves go up considerably.

[graceful condor]

and only GPT has it not Grok or Gemini. Because GPT is really broken since 5.

So it works better if you ask for suno related prompts and stuff you need or only for a certain type of commands or how does it work?

[polar spade]

If someone asks a how to question here now, if I cant tell them in two steps, I just leave the answering to someone else. 🤪

[cyan jewel]

i've been asking for suno-specific stuff yeah

theres two i use frequently

wm suno helper

suno style auralith

auralith is the most important option

since it understands the 4.5 style prompts and knows how to neg

[graceful condor]

Thanks for the suggestion. Definitely trying it out.

[cyan jewel]

for anyone who masters their stuff, what do i do when i have more than 8 stems?

[half dragon]

.. should also mention... which free DAW that supports more than 8 stems - maybe someone has a reference or ways
Thinking outloud, could you combine some of them in FL Studio, render out, re import in the main project to reduce the number of stems?
That could be possible to get around the limit you have with free version of FL Studio?

[cyan jewel]

i was considering combining some

im sticking to fl mostly due to familiarity

i do also have bitwig

but i dont know it. my eurobeat bestie gave me a key

[half dragon]

Sticking to what you know best is usually best

[cyan jewel]

i've poked at fl on and off since like 2006

im no good but its what i know xd

[half dragon]

I used it a bit in the past until I found Studio One - now can't use anything else but studio one 🙂

[cyan jewel]

oh thats goofy

no wonder it generated 12 stems

some of these stems like, barely have anything in them

yeah merging is correct

i have a percussion track thats literally 1 sample

and the rest is silence

i was going through the stems to see what i could cut down

look at that single lad

lmao

[tender eagle]

I use ableton, never seen studio one.

[tender eagle]

https://ardour.org/ but if you looking for free DAW that supports more than 8, that'd be my go to

it's very well fleshed out for being a FOSS project.

[devout sinew]

Hello. How do I find my generated song's seed "Seed or Generation Seed (a long number)this is usually a long string of numbers (e.g., 1783482943) or sometimes alphanumeric"

[half dragon]

I don't think it's visible to users.
Things that would ~~affect ~~ be considered the seeds are styles and the new sliders.

[devout sinew]

Okay, thanks

[boreal nest]

Good evening, I would like someone to help me. I would like to create a piece that ranges from classical music to modern electronic music such as EDM dubstep, obviously passing through blues, jazz, rock & roll, funky, house dance. I can't find a way to make the different genres articulate within the same song. I tried to give indications both in style and in the text, but nothing makes versions of songs of a single genre. I would like a sort of medley.

[tender eagle]

Nothing should affect the seed, it's a static integer that's generated reandomly per song.

[half dragon]

Yeah, once created it's static.

[tender eagle]

Yeah, the creation is likely server side and just random off nothing clientside

but knowing it, is useless, unless it is provided to the api by clientside, which tbh, I never have actually looked to see if it does, though i didn't see it in the generation POST , but wasn't looking

[neat nexus]

Have you tried creating in one style, generating until that first segment is how you like it, extending from an earlier time, cutting off from when you wish to start the next style, then changing the style prompt to the next style and generating from there?

Continue until all styles have been realized?

[boreal nest]

great idea, I'll try it even if it will most likely cost me 1000 credits to reach the goal, anyway thanks for the advice

[stiff blaze]

It's all about the prompt

[stiff blaze]

Can you describe the track structure you aiming for?

[boreal nest]

This is the first version, then I tried other more and more detailed changes, and I also tried to insert the indications in the middle of the text in square brackets

[stiff blaze]

It’s a cool concept and the structure is solid, but the reason Suno keeps giving you one genre is that you’re telling it all the genres in one big block, then only mentioning “seamless transitions” at the end. Suno tends to smooth everything together unless you force a style change inside each section with its own instructions.

[stiff blaze]

Try this:

[stiff blaze]

[Intro] Classical era: gentle string quartet, piano, and light choir harmonies. Slow and elegant tempo (~80 BPM).
[Transition] Crescendo of timpani and cymbals leading into…

[Section 2] Romantic era: full orchestra swells, richer harmonies, emotional dynamics.
[Transition] Violin solo melts into acoustic guitar strums.

[Section 3] Early folk/blues: warm acoustic guitar, upright bass, light brushed drums.
[Transition] Walking bass picks up swing rhythm.

[Section 4] Jazz swing: brass section, upright bass, ride cymbal swing groove, playful melodies.
[Transition] Drum fill into electric guitar riff.

[Section 5] Rock & roll: punchy electric guitar, bass guitar, snappy snare. Upbeat and lively.
[Transition] Guitar fades into syncopated bassline with wah guitar.

[Section 6] Funk & disco: slap bass, rhythmic guitar chops, claps, energetic horn stabs.
[Transition] Filter sweep and riser into electronic beat.

[Section 7] Early electronic/house: 4/4 kick, analog synth bass, filtered pads, tempo ~125 BPM.
[Transition] Kick cuts out, riser into bass drop.

[Section 8] Modern EDM/dubstep: huge bass drops, sidechained synth leads, aggressive drums, tempo ~140 BPM.
[Outro] Fade electronic layers into solo piano and soft strings for a calm classical ending.

[Instructions] Ensure each section has a clear tempo, instrumentation, and rhythm change. Keep transitions smooth but obvious. Avoid merging genres.

[stiff blaze]

Weirdness: 65%
Style: 85%

[boreal nest]

I managed to get him to sing some songs just with vocals, similar to accappella but getting him to sing medleys is a real feat, now I'll try with your prompt and I'll let you know the result right away

[stiff blaze]

What I gave you is too long (for mobile) sorry. And without lyrics

[stiff blaze]

Try this

[stiff blaze]

[Intro] Classical: string quartet, piano, light choir, slow (~80 BPM).
[Transition] Timpani roll into…

[2] Romantic: full orchestra swells, rich harmonies, emotional.
[Transition] Violin to acoustic guitar.

[3] Folk/blues: warm acoustic guitar, upright bass, brushed drums.
[Transition] Walking bass into swing.

[4] Jazz swing: brass, upright bass, ride cymbal, playful melodies.
[Transition] Drum fill to guitar riff.

[5] Rock & roll: electric guitar, bass, snappy snare, upbeat.
[Transition] Guitar fade to slap bass.

[6] Funk/disco: slap bass, wah guitar, claps, brass stabs.
[Transition] Filter sweep into 4/4 kick.

[7] House: kick, analog synth bass, filtered pads (~125 BPM).
[Transition] Riser into drop.

[8] EDM/dubstep: wobble bass, sidechained synth, aggressive drums (~140 BPM).
[Outro] Fade to solo piano + soft strings.

[boreal nest]

ah but so I don't have to put it in the style but use it as indications in the text section, right?

[stiff blaze]

You have lyrics?

If you have lyrics put it under each section

This for the style:

Cinematic genre-shifting medley evolving from classical to romantic, folk/blues, jazz swing, rock & roll, funk/disco, house, and EDM/dubstep, with clear transitions, distinct instrumentation, and progressive tempo changes.

[boreal nest]

I can copy the ones I put in my versions, otherwise I ask holy chatGPT to do the dirty work for me hahaha

[polar spade]

As another tool and not the only way to do this: If you have previous songs in the style you want then you're ahead of the game for this, but as an extra step, make a Persona of each song in the style you like. Then add that persona as you extend and prompt for each genre changing verse or segment.

Here is an example of 6 verses all in different voices and a bit of jazzy piano and a couple of different styles all based on the persona's I used, keeping the tempo and general metering , though the lat verse went half-step on the vocals.

This all kept the same tempo, but you can change "tempo and feel" this way as well, if you're wishing a more Bohemian Rhapsody or Band on the Run melding vibe.

https://suno.com/s/tA2ViUqxKHVv82pX

And will throw in one of many i didnt use.

https://suno.com/s/ejRoTjaauXsLJGWS

[boreal nest]

https://suno.com/s/i0kxbPTLCVns6HtH one version was done like the other first ones, that is, maintaining a single genre, while the second one was quite close, it didn't define and separate the styles well, but it already differentiated the various styles.

[stiff blaze]

One step forward. Nice

[stiff blaze]

Try tweak with some covers. I think you got a flexible seed.

[boreal nest]

With just vocals, putting nonsense syllables and lines into the lyrics gives amazing results. The instruments still need some work with the algorithm, even when you want solo passages it's a bit hard to understand. What do you mean by covers?

[stiff blaze]

I understand exactly what you saying.

[stiff blaze]

Remixes - covers

[boreal nest]

ok understood, I will try to modify the song created in remix or modification you meant

[stiff blaze]

yes, and change specific parts, well try to

[boreal nest]

https://suno.com/s/DxZhF5af2JNAJliH no you have to tell me how you got this, my singers always tend to be screamers, and it's something I hate not being able to get a calm sober subtle singing

[stiff blaze]

Don't you blink they watching

This specific song from all the project got 200 versions

[boreal nest]

I'm also listening to the others on this playlist, all more or less in the same singing style, so you've found the right one despite the 200 versions, at least tell me how I can make my singers stop screaming, I have to cut their vocal cords hahaha

[stiff blaze]

Are you from Italy?

[boreal nest]

yes

[stiff blaze]

Can you tell me how is my Italiano? 😆

https://suno.com/song/6b3568fc-fe76-4f83-813b-9de6c6502085

i been trying some dual language songs but i dont know how authentic it turn

[stiff blaze]

Its a Disco / Nu Disco project
The goal was making full album with same style

[boreal nest]

two languages, in Italian it says hidden blade cold winter, for a correct grammatical form you should insert a preposition before winter, lama nascosta nel freddo inverno

[stiff blaze]

Thanks 🤙

[jovial tiger]

Hi guys if I want to use like chanting and stuff. like ethereal for the vocals. what should I type? In the style section

[stiff blaze]

No in the line

[boreal nest]

anyway, this project also follows a logical thread of style and genre, as a personal taste of musical genre I prefer the marcela project, it is one of the genres that I like most in absolute disco, funky I love

[stiff blaze]

Thanks man appreciate it.
It's on Spotify btw. Was very interesting to make it. Learned alot from it. And held myself not to break the style.

[shell trout]

is there a good way to get gunshot sound effects?

[stiff blaze]

Depends on your goal.
You might try insert the parts in (text) brackets and add [instructions] underneath
In lyrics
Also would be better to note it in the style section

[stiff blaze]

Try at the spot [gunshot sound]

[shell trout]

it only works like 10% of the time

[stiff blaze]

Sorry

😩

[boreal nest]

I've been using Suno for a little over a month, but I can't naturally maintain a style, because for now I'm creating sensations of the moment, I've done genres from meme to tribal pieces, EDM dubstep to refined jazz, tarantella to funky house, tracks almost soundtrack-like, I'm having fun experimenting with what Suno can do, then when I get something that really strikes me, I download the stems and work on them on Ableton Live you get more humanized results, because I've heard so many beautiful songs on Suno, it's a shame that they always have that note of coldness and precision that is too precise to trigger the thrill of the song made by the human when it manages to convey a message through the song, so Suno still can't fully convey your message

[neat nexus]

We work with what we have.

[stiff blaze]

Very true. But maximize results challenge is always fun.

[stiff blaze]

Send me dm I'll try to help you with some tips

[boreal nest]

https://suno.com/s/7pKXfPnMMAU26XN3 if you want, listen to what I made him do with just the voices and vocal beatbox

[boreal nest]

https://suno.com/s/OwLjvVs3Xl2RNKRZ I discovered that just putting syllables makes amazing vocalizations

[polar spade]

Having read that, its fun when I listen to some artists songs on Spotify and 1) find many that sound worse than what I hear on Suno , Id almost swear they were AI and arent and 2) can pick a date in time when Suno songs now sound better than what they had to work with then. Its all relative.

[boreal nest]

Unfortunately I have to agree with you, but this happens because in reality musicians are dead, many of them discriminate against songs made with AI, but they themselves compose them with DAWs FL studio, Ableton live, Logic pro, Reaper, Cubasis, now those who really integrate playing with corrections made in DAW make music and it can be heard, those who only use vst plugins in DAWs and especially some genres that are in fashion today, are real rubbish, but the market rules and the masses follow the rubbish not the style and much less the taste

[polar spade]

Im confusing "thrill of the song humans make..." with "Suno can't make...." and " musicians are dead" and "those artists songs [sic] are rubbish....."

So while I agree, we can do better at times with using our own skills, im just pointing out what you already know, the bar is set pretty low on places like Spotify, that even Suno songs sound better there than some "human artist"s songs that have gotten there just by being human. 😊

[boreal nest]

Not many people listen to music these days, many just listen to it, and even listening to it from devices that mute the sound doesn't help the music, relegating it to background noise. Not to mention that songs that are longer than 2 and a half minutes are too long for the masses. Instrumental parts and solos have disappeared from all modern songs, even pop ones. Damn, Suno does them, and he does them well, I have to admit. He has the habit of making intros that last three-quarters of an hour, but I prefer that he gives space to the instruments, even if I have to wait a little longer before the song starts.

[neat field]

https://suno.com/s/9fxcnEPUqs5WWTEs

polyphonic multiphonic fx nature tube 8d field recording park bird calls

v4.5+ bird call experiments accidentally turn into a song a few minutes into the generation

[polar spade]

Also use "birdsong"

[prisma hound]

Very interesting. It's prompting Suno to get into the path of Messian, who spent lots of time studying and integrating birds' chants into his music.

[neat field]

neat! "Messiaen" for the Google curious

[boreal nest]

brilliant

[prisma hound]

Oh yeah you're right sorry for the mispel ! He was my music professor's mentor, who was always saying "Like my Master Messiaen said (etc)" 😅

Namely "Réveil des oiseaux" and "Catalogue d'oiseaux" as for the birds songs inspired. Could be fun to see what can be further done on that field with Suno!

[neat field]

today, we bird callers are all French, what a lovely domain of music to learn about

[cyan jewel]

working on my next song and the mix is super overwhelmed. what do i do to chill things out?

fl studio

[robust totem]

Hello everyone, I’ve used Suno quite a while ago, but I’ve come back and found many new things, like the types of remixes available now and more.
I’ve been with some friends uploading lyrics we made in Suno, and previously, when extending an uploaded audio track, it would use the voice context of the original song. Now it just changes 100% of the audio to the proposed style, completely ignoring the context of the previous track. It’s basically like one song abruptly stopping and a totally different one starting, both vocally and instrumentally.

In the past, we made some songs using one of my friends’ voices with tracks we created, and by extending them, we made something like a “remix” with the audio. But when trying to do one now, both with cover or extend, even setting the “audio influence” to 100%, it just abruptly stops the audio and starts a song that has nothing to do with the original.

To check if it was an error, we’ve tried many, many times with everything: extend, cover, using old models, making new audios, changing the style, singing the lyrics that we then give to the AI, etc. And NOTHING worked.

Does anyone know why or what I should do so that the resulting song keeps its voice/style? (We used to do this without using the Persona feature because it didn’t exist before, plus it’s not available on Android, like a lot of features from the web version.)

Thank you a lot!

[acoustic scarab]

I got an unfinished song and extending it chamges the voice. I've done it 4 times but still the same result. Is there an issue?

[distant cradle]

Have you tried placing your unfinished song/track into an empty playlist, then using adding that playlist as the Inspiration (Inspo), then under Advaced Options (NEW) push the Audio Influence to 100% ?

[acoustic scarab]

I will try it, that's very smart
Thanks a lot 🙂

[viral kite]

Sudo's pood its pants again can't download or play my mongs,

songs that is

[storm pawn]

I have a technical question - since Suno can easily exract stems, separating vocals and instrumental, why changing lyrics functions (either in legacy editor or new replace section one) is altering the instrumental (sometimes instruments are completely different) and is having difficulties to match the sound of the vocal as well?

I've spent around 1500 credits trying to edit one section, even trying with the cover/remix but with 100% influence sliders resulted in terrible artifacts

[polar spade]

In the past versions, we used to accept the extension knowing that a remaster would equate the vocal differences and the extension would then sound like the initial vocal. I haven't tried that in that in awhile.

If not, you can also make a persona of your present (only need a good minute segment of the best vocal section) penultimate song created that needs that one more extension. Then extend and add along the persona, while setting audio similarly to about 99% and style to about 78% (weirdness 1%) for the hopeful final song. Any further extensions needed just keep adding the persona.

[pastel sluice]

Hi everyone, I have a question. I've some songs i've made with Suno, how can I do them in different languages? I mean i want to use the same music but with lyrics in different languages. P.S. I also use a "Persona" for my songs. I am doing kids music and want to translate them in other languages too

[polar spade]

@pastel sluice One current way is to "add vocals" to the instrumental. Its a longer process than some.

Its been a couple of weeks, so I hope I remember the steps.

Intermediate level: (one of about 4 ways to try this. )

Youd like an instrumental that also included the melody of the vocals. So Cover the present song , prompting for Instrumental, in this case, I think clicking the instrumental toggle on is enough. It should obviously change any vocals into a melody line. Prompt and tag as needed to influence an instrumental only .

Next is to "add vocals" to instrumental. Unfortunately you only do so to stems (or uploads). Click Get Stems and choose the 2 -stem option, which obviously would have NO vocal stem, but an instrumental only.

In library, choose your instrumental stem and meny/remix-edit/Add Vocals to Instrumentsl. At this point, paste your new foreign language lyrics in and click Remix. Cross fingers and hope the melody line of the instrumental will induce the lyrics to be sung in the same melody.

Note: try one on its own, but keep in mind there is no sequencing available to tell Suno where the vocals should start. The melody line in the instrumental should be helpful to delay the start of the vocals, while playing the intro first without starting in on the lyrics. [Otherwise, there are additional steps to crop the instrumental with the intro momentarily, but thats next level. 😄]

Anyone wishing to check my list is welcomed to correct anything.

[daring lava]

Hey there! When you get a chance, could you share a song link either here or via DMs (if you prefer privacy) for me to look at what you have going on with your song?

[daring lava]

I’ll be at my PC in a bit and will grab an example to walk through 👍

Do you mind if I use your lyrics to build/test from? I'll delete afterwards, just want to ensure the most consistent result. If you'd prefer not, no worries. Whatever is best for you 👍

Awesome! Be back when I have something solid

Sure! Whatever you think would help!

@unreal sleet While I'm testing would like to throw one suggestion at you, this would create the most control. Using Audio Upload introduces a strong influence over the delivery. It doean't have to be perfect either. Something as simple as you rapping into phone voice recorder would help a ton. Not sure if that's something you're up for, but for me, it's a ton of fun! Adds a whole other level of involvement and as mentioned before, control!

It'd need to be something you own the rights to

Okay so here's a solid approach:

• with the style of music start simple with a strong focus on genre tags that align with the results you want. In this case:

Freestyle session, rap flow, 808s
Freestyle sessions are notorious for being steady raps/no singing hooks

• from there you can cover the result to add additional elements/vibe to the track while still keeping the vocal pattern intact. This could take some experimentation to get it exactly where you want it to be, but honestly, that's half the fun. With these unchartered waters there's a lot of room for exploration and discovery. Don't hold back, you never know when you might find something amazing

Let me know how that works for you

Feel free to copy and save the structure I included with this tester

I was able to do this on the first take, but if it doesn't give it a couple more runs

Let me know when you have that copied and I'll delete on my end if you want

Awesome! You may want to delete the previous songs shared here to protect the privacy of your lyrics

@ashen basalt if you'd rather share via DMs to protect the privacy of your song, just let me know!

Thanks for sharing, let me take a look listen

[ashen basalt]

starts at the first pre chorus

have happend a couple of times now that suno start to sing "you never had the change to say" instead of the lyrics I have put in the lyrics box

[daring lava]

Yea I definitely hear it. Also see that it's being pushed for the full 8 min marker. Very weird for such a short lyric'd song. Wondering if it has something to do with the length of the Style of Music. Mind if I run a couple tests with your lyrics? You can delete the song you shared (to protect your privacy), I have it now

[ashen basalt]

sure

[ashen basalt]

it worked fine when I started but when I tried with a persona is were it began and continued after I removed the persona and I even refreshed the page

[daring lava]

Very odd indeed! Thanks for flagging this!

[ashen basalt]

never had trouble from the length of my style box either.

[daring lava]

Do you know how to Extend from parts? You can go back and redo a section that way or by using the Replace feature. I can point you to some introductory links if that would be helpful!

[ashen basalt]

could also work giving it a cover

[daring lava]

It could take some practice to get the hang of it, may require some re-rolls and isn't surefire with very detailed/specific/short-lengthed parts yet, but certainly. Just be mindful that using these features will use credits as well

very early days still, but we're getting there 💪

[daring lava]

Yea that may not be it at all, just doing guesswork on my end. Almost done with testing

[ashen basalt]

Sucks loosing credits over suno not being friendly^^

[acoustic scarab]

Ok i will try that then and see

Thanks a lot for ur help

[daring lava]

If you here something that has something interesting, like let's say the vocal is amazing or there's a particular instrument that stands out, keep it in a separate folder for later. You never know when it may become useful down the road and what tools may surface that introduce further opportunities to use those B-roll sounds for something spectacular. One person's garbage is another person's gold

[rain spire]

Good day!
I have a request for you:
if possible, additionally teach "Artificial Intelligence" to pronounce words better grammatically, because in each song in the text you have to put accents by indicating capital letters and divide words into syllables or even replace words altogether - this causes inconvenience.

P.S.: In this case, this applies to "Ukrainian Language"

[half dragon]

Have you tried to remove that line from the style box and see if Suno still picks it up and sings it?
Was it just random one time, or it did that on every gen?

[daring lava]

@ashen basalt

So I restructured a few things. Shortened the Style of music and transfered some of that guidance into bracketred instruction through the lyrics. Take a listen and let me know if this works. If you like, copy and paste the lyric/SoM fields somewhere and let me know, and I'll delete here and on my end

---

Also, good morning @half dragon 👋

[half dragon]

Hey gm to you @daring lava

[ashen basalt]

still wierd it will sing out a random line in the style box

[half dragon]

It is but AI could do some weird stuff sometimes.

[ashen basalt]

this sound very nice.

[ashen basalt]

true

[daring lava]

Very awesome, love the lyrics you came up with! Let me know when you have that copied on your end and I'll delete on mine and here

[ashen basalt]

you can delete it now

[ashen basalt]

Thanks for the help. And thank you for the compliment for the lyrics. took me like an hour to get them to where I wanted

[daring lava]

Very welcome! And well worth the time spent if you ask me, they're pretty perfect! Love the emotion and passion that comes through! Great pacing!

[ashen basalt]

well I was kinda in an emotional wreck afterwards^^

[daring lava]

Crazy how the experiences we go through add so much more to the creative process. Very relatable

[ashen basalt]

it's more that I am a pretty emotional guy, I wear my emotion on the outside so whenever I write emotional heavy lyrics I always ending up in a wreck. I am glad I haven't gone through what I am writing. but still emotion always get to me

[livid trout]

something strange happens, when I want to replace an intro of 4 bars, it always managed to overwrite the second measures into the verse .. i can't seem to solve this , i can snap and select the part of the track that has to be replaced changing the intro into an instrumental piece, it will cut off the first words of the verse ..

[granite pawn]

Does anyone know if Suno Studio will still be putting Compressors/ limiters on the stems

[hard beacon]

we will see 🙂

[inner rose]

i want to make som radio jingels can somone help me

[hard beacon]

i think suno is not the right service to create radio jingles 😕

[inner rose]

ahh okay

[hard beacon]

i think you will get a 3 minute audio 😉

[inner rose]

ahh okay you have sugestons

[final hazel]

Sure you can.. Here an example I just made
https://suno.com/s/ZTUuEgeodXiKHfxA

[hard beacon]

3Daizy: i think he is looking for short high-energy station design tools.. that's what i would say because of his username 😉

[half dragon]

maybe this?
https://suno.com/song/5358e139-5ce4-4613-9686-c4bfb6b5a26c?sh=ibyg7Kua8caOjE2E

[final hazel]

But it works.. I just showed you that it does not go to 3 minutes, and you can make jingles like this 🤷‍♀️
I am just answering the question 🙂

[hard beacon]

yaa.. well.. maybe i use the wrong words to explain my thoughts 😉

[final hazel]

You nailed it!

[hard beacon]

maybe @inner rose can give us more infos about it??

[inner rose]

its like a dance station and i make evry week a mix so i need somthig to put in the mix so that nobody can steel my mixes

[hard beacon]

ahh.. ok.. i get it... like a "DJ Arix.. in the mix"... or "This is XYZ on -stationname-"

??

[unborn star]

Radio Jingles for 5 second?

[final hazel]

This on is 8 seconds #📚┃suno-school message

[inner rose]

yes the staton name is DanceMation Radio

DanceNation Radio

[hard beacon]

i think you wanna have vocals + sfx and not only vocals?

[unborn star]

oh you want a jingle with calling the Station?

[final hazel]

https://suno.com/s/cV3Sdutj5bHYlcI9

Just change the genre and text to yours and it should work

[half dragon]

It's definitely possible 🙂

[final hazel]

yeah I make shorts compositions all the time for my videos which are often less than 30 seconds.

[hard beacon]

i was active in the internet radio scene on one of germanys biggest radiosstations in the internet an responsible for the radio sounddesign... if he wanna use jingles over his music, then there are 2 possibilites... vocals only or vocals with typical glitch effects... btw: I don't want to disparage your suggestions... it was just a thought 🙂

[final hazel]

Which one? I used to listen a lot to weareon.fm myself

[hard beacon]

i think he is looking for something like this with Glitch effects

@inner rose : talk to us bro.. we wanna help you out 😉

naaah... i was on Techno4ever.fm (T4E)

😉

the 3 biggest players in the past were TB (weareone), Rautemusik & T4E

[final hazel]

Same genre though.. 😄

[hard beacon]

yapp

[final hazel]

https://tenor.com/view/plongus-mcnyale-cat-party-rave-gif-17489417024592723907

[inner rose]

I’m now going to the gym

Thx for sharing and helping me out

[hard beacon]

you should listen to my upload first 😉

is this what you are looking for?

[inner rose]

Yess

[hard beacon]

ok.. so clean vocals (and in the best case with glitch effects)

[inner rose]

Yes

[hard beacon]

whenever you come back from the gym... send me a private message 😉

[final hazel]

https://suno.com/s/Iykw2x76d2watbqM

Plently of examples here now to help you 😄

[hard beacon]

complete different topic: at what time the credits will be updated??? (today is billing date for my pro plan) 🙂

[final hazel]

The same time you once started

[hard beacon]

pfff... so i think i have to wait round about 3 hours

crap..

😄

[final hazel]

😬 sorry 😅

[hard beacon]

a time stamp in the profile would be cool 😄

[final hazel]

That's a great idea. Feel free to share it in #💡┃suggestions-web or via https://suno.com/feedback!

[hard beacon]

what's the correct english word for "time stamp about refresh the credits" ??? Please add a time-stamp on the info about "???"

[final hazel]

idk.. I am Dutch 😏

[hard beacon]

hahahaha... Oké... dan zijn we nu even slim 🙂

En excuseer mij voor mijn grammatica... Ik heb maar twee semesters Nederlands gehad op de universiteit 🙂

[final hazel]

Lol.. Niet slecht 🙂
Here is what I do when I am unsure, I let ChatGPT rewrite it for me. 😅

[hard beacon]

good point 🙂

[hard beacon]

already sended 🙂

[hard beacon]

@flint haven you are intelligent.. at what time will my credits refresh?

😄

pfffffffffffffff....

[hard beacon]

Credits !!!! whooohow 😄

[cyan jewel]

is it normal for when we extract stems for it to be super bitcrushed and missing a lot of the dynamics?

https://suno.com/song/e9f952a6-89d9-42f7-ae88-fc45d26e23c6

https://suno.com/song/40fdf9b8-8121-4dec-aea1-543204161bfb

theres a big difference here

[half dragon]

Maybe some of the sound from the guitar was detected as a different instrument and on a different stem.

[cyan jewel]

i wondered about that but i checked the other stems

[woeful thistle]

nu disco house music style Purple Disco Machine

[golden plover]

Ok after digging through the songs I want stems from, I found out they work for instumentals not lyrics. When I tried making a cover of the song and it says I cannot use copyrighted lyrics. This explains why the stems are failing.

What I can’t understand is, if the lyrics are copyrighted why can’t I extract the music only and not the lyrics. I don’t need the vocals I need the instrumental

I tried splitting the vocals from the instrumentals and guess what. Bugs out again

All I want to do is to keep the music remove the vocals and make stems of the music only… but no since I used copyrighted lyrics the whole entire platform bugs…..

If the lyrics I used were detected as “copyright” why even generate the song in the first place?

[half dragon]

These songs must be generated from older models a while ago? now it's more strict.

[golden plover]

It’s the 3.5 model

“Couldn’t generate that. Lyrics contains copyrighted material.”

So how do I remove the copyright lyrics and just keep the music?

Without making the model recreate a completely new song?

[half dragon]

Not with the current system, I don't think you can get pass those guards.

[ebon sand]

Id try to just remove the lyrics

[golden plover]

It makes a completely different song that is not the same.

[ebon sand]

When stemming?

[golden plover]

I have no problem removing the lyrics of the song

I just want stems of that song. I don’t care about the vocal

[ebon sand]

Yeah I'd remove the lyrics and stem

[golden plover]

When I do that it bugs out

And does not split the music from the vocals

[half dragon]

What it means, what if you removed the lyric before you Stems

[golden plover]

It generates a completely different song

[half dragon]

no, update the song and remove the written lyric - don't generate a new song - then get stems from that song

[golden plover]

Ok how do I do that? I’m in editor now

[half dragon]

Not in the editor - 3 dots menu on the song > Song Details then remove the lyric

[golden plover]

Edit displayed lyrics?

and remove the lyrics from there?

[half dragon]

Yes.

[golden plover]

Ok made a copy of the song, removed lyrics and now trying to extract. It’s stuck

I extended it and tried to extract the second version and it works

But it’s not the song I want I want the first one

[half dragon]

Any reason you didn't want to try on the original song? give it a try yet?
At least no more copyright error.

[golden plover]

I don’t want to lose the original

But the copy shouldn’t matter

The bug remains

Even after removing the lyrics

[half dragon]

You don't lose anything, well, the display lyric but that can be paste back in.

[golden plover]

Ok seems like it isn’t doing anything even if I remove the lyrics

So I can remix and it makes a completely different song

And it works but that’s not the song I want

Can’t I just generate the song exactly how it is and remove the lyrics so that I can create stems?

[half dragon]

Not direct way - but as you can try Cover and push the Audio slider as high until it sounds the same or almost the same

[golden plover]

Cover or reuse?

I don’t see an Audio slider

I’m in the remix section

[half dragon]

Cover and cover with 4.5+ to see the sliders.

[golden plover]

Weirdness 0, style 0 and audio 100?

[half dragon]

I'd put weirdness low (not sure if 0 would have any affect but try it) Style -- either leave it as is or raise slightly
Experiment with it - you probabaly wont get what you're after on first few gens anyway

[golden plover]

Yes it’s completely different. Generating new songs in Suno and you kiss the original goodbye.

I really thought removing the lyrics and trying the stem extraction would work. I was wrong

Hopefully the devs would fix their bugs

[half dragon]

That's expected ... especially cover old model song with new model. It is what it is.

[golden plover]

For now I am stuck with using Moises or similar which generates low quality stems

[ebon sand]

I think it would be helpful if you shared screenshots of what you are doing

[half dragon]

So I was curious. I took my v3 song and removed the lyric and get Stems from it.
It worked for me. So I don't think that's the issue.
So your problem might be just the copyright lyric you were using.
Even though it got through on earlier version but with current system, you're out of luck.
Lesson learned, don't use copyright lyric 🙂

[polar spade]

Are you using mobile? In desktop mode?

OK Song trivia question of the day:

(Without searching the answer) Fill in the blanks xx and && (just blanks and do not represent characters count) to the Trolley Song (1944) from the movie musical Meet Me in St Louis.

xx, xx, xx went the trolly
&&, &&, &&, went the bell

First right guess gets a big "atta-way to do it" from me.

[half dragon]

Googling.....

[sleek stump]

That movie and song is out of memory for the vast majority of the planet 😄 There may be a handful of people left 😛

[polar spade]

Music lovers should have no era boundaries. Lol maybe I should post these in general chat, where I can give a time limit.

The open trivia database where we could use to have a randomized trivia while chatting were always fun, with clues and timers. Long time since ive seen that utilized.

[polar spade]

C####, c####, c####, went the trolley.
D###, D###, D### , wen the bell

2 minutes left

[polar spade]

C#a##, c#a##, c#a##, went the trolley.
D#n#, D#n#, D#n# , wen the bell

1 minute left

[polar spade]

C#a#g, c#a#g, c#a#g, went the trolley.
D#n#, D#n#, D#n# , wen the bell

30 ssconds left

Woodie gonna tag me for spamming. Lol.

TIME!

Clang, clang, clang, went the trolley.
Ding, ding, ding , went the bell

NO points that round.

[cyan jewel]

https://suno.com/song/a53905d8-41de-4597-a15e-a646d7962d52

What can I do to further improve this?

[granite pawn]

Trying to create a live performance for a song. How do you get the crowd to cheer and repeat words. I’ve tried bracketing (Crowd yells “Pennies”) but that doesn’t work

[half dragon]

I haven't seen any success on creating Live performance for a song - Sure, adding Live on style and in metatag would give you a live performance feel at the start.
But then when the song progresses, it's gone - the model wasn't able to maintain the feel.
And prompting for crowd is almost imposisble as well... only at the beginning without any special prompts
So yeah... it's a challenge for sure.

[granite pawn]

Thanks VRKing. Yeah its a tough one

[ripe galleon]

is there a reason on why theres this weird shaky oscillating drum effect in my song?

[polar spade]

If there is any Asian music prompting, it could be throwing is some Taiko drum like effects. I hear a wood pipe feel as well. Thats likely a stacatto effect which may be prominent from one of your genres , such as "cinematic" . But the room may need to see your prompts to sift through to suggest anything.

[ripe galleon]

welllll

[polar spade]

Tribal is doing that stacatto beat... and orchestral is adding to the stacatto effect. You could just try, Taiko drums, to keep that Battle effect and it maybe reduce the staccato repetition feel.

[scenic coyote]

I just spent like way way way way way way. way too many credits. trying to get Hall& Oates rich girl sound. kind of close... there might be closer ones but I'm not about to go through the hundred songs I just generated lol https://suno.com/s/ILwJCRCFTlHEQg7n

[polar spade]

And keep orchestral electronic, I dont think that would be an issue then.

[ripe galleon]

hmmm

orchestral electronic, heavy drums, video game, taiko drums?

[polar spade]

Honestly, we never know, its always trial and error. But thats looks good. Try it. Add cinematic before the
TAiko drums to weigh more towards the battle feel than the drums.

[ripe galleon]

always trial and error

lol

[ripe galleon]

alright

what

my font was Smaller

can i not use -#

[polar spade]

Im pretty experienced , we all are, but that just means we know it can be pretty random.

[ripe galleon]

its like ai images

theres always going to be one or two artifacts in some places even if it looks like a good image

[ashen temple]

What are people prompting to stop songs in v4.5+ going ridiculously long where they never used to in v4.5?

[viral pond]

is there a way to have suno sing better in norwegian ?

we have so many dialects here, but two main languages. i think suno confused and mixes around

especially when singing R

[polar spade]

I would crop, quick fade, then extend in 3.5 at the pont you want it to end. Remaster 4.5 after that.

[ashen temple]

Wouldn't that be easier just to create in v4.5 from the start instead of all of those steps?

[polar spade]

Everything is an option. Starting over may make it more difficult to get that vibe odr melody if you really like what you have so far.

Im one for trying several different ways to see what works. I never use credits to get in the way of exploring, but unfortunately that's a driving factor.

I think you're asking if easier to start over and prompt everything so that it perfectly responds in one shot. Thats a challenge to try, certainly. Just another option. Lately I do alot of things piece by piece, same as using a DAW. Piecemeal it to get the results needed.

[ashen temple]

Well my initial question was more because I wanted to benefit from the improved quality v4.5+ offers. If I have to extend in 3.5 and remaster to v4.5, I'd start with v4.5. I really would like to get v4.5+ working as I would like it more than anything

[polar spade]

You can try just the ones tep of extending without the crop and fade. It could give yiu something close or exact.

[ashen temple]

My ideal workaround would be to create in v4.5 and just remaster to v4.5+, but sadly that isn't an option at the moment

[polar spade]

Yes. Then you may be back to the original question, without using all tools available. As you see. 4.5+ for everyone last month was only making short songs. Now it seems to be making more longer songs. Same thing happened when 4.0 came out.

[half dragon]

That's the ideal for everyone, but due to its quirk - things are not working as we'd like.
Therefore these workarounds are there to help. You asked what people do but didn't want to try?
I think it's best just to wait for a fix on new version 🙂

[ashen temple]

I do not want to seem ungrateful, but I specifically asked regarding prompting in 4.5+. That question was never answered

Telling me to do things in v4.5, while it could be something someone may want to do, is not answering my question (I am aware it was an attempt to help)

[polar spade]

I got you. Its frustrating. Ill answer then to say Ive yet to have experience enough to know how to hit it in a couple of tries and for every genre and design. Sorry, mate.

[ashen temple]

No problem, appreciate you trying anyway

[polar spade]

In a way, but a persona tends to retain other things like a melodic vibe, riffs and if vocals, the vocal characteristics as well.

By the way, according to the hint (i) Button , the Style audio slider (not Audio Style slider) controls persona strength. I tend to use them both and feel that makes a difference, but then thats maybe a placebo effect.

[hidden yarrow]

Which hint button are you referring to? If I hover over the (i) symbol next to Audio Influence, it says "Turn it up to resemble your original audio or Persona".

[polar spade]

What's the style slider info button say.?

[hidden yarrow]

"Turn it up to match your style description"

[polar spade]

Ahh thanks , mate. I may have had them backwards in my head. Yea, let's have a Persona slider, until im clear what they do. 🤦 good catch.

[hidden yarrow]

Yeah, a separate persona slider would be great ... even more so for the attributes of the persona, e.g. voice vs. style etc.

[polar spade]

Since it covers both audio aspects and style aspects in my mind , I just hedge my betss and use both sliders the best I can to get results. Usually thats a slight adjust between 45% and 78% for both. Its been a couple of weeks I haven't been diligently on my desktop.

[hidden yarrow]

Since 4.5 I always use 100% for style anyway, not with noticably more artifacts than with "less style". I do not see any sense to prompt a style, which in any case is not followed perfectly , shall only be adhered to 50% or something.

[glacial scarab]

So that clip is from December 2024?

[ripe galleon]

yes

i just never addressed the issue until now

[glacial scarab]

that's shimmer

[ripe galleon]

shimmer?

[glacial scarab]

V4 artifact

[ripe galleon]

is that still present?

[glacial scarab]

should be gone

[ripe galleon]

once i get the plan (i made Songs with free V4) ill check for mywelf if the issues truly gone for certain prompts

[glacial scarab]

But be aware that if you cover that clip it might replicate it from the sample. New gens should not have that though

[ripe galleon]

alright

[thin musk]

Hello I am using suno to create some south eastern songs which are in the form of Nauha / Marsiya / Manqabat / Matam / Soz / Salam / Latmiya - has anyone ever looked into training those kind of eulogical spiritual tracks ?

[half dragon]

No one can train it.
You might want to submit a request to Suno or post in #💡┃suggestions-web

[broken dew]

Can we please loop our playlists instead of playing random songs from others

[next carbon]

Any ideas on prompts to get sounds like Cigarettes After Sex 'K', Chezile 'Beanie' and/or Bryant Barnes 'I'd Rather Pretend'? I am unsure how to describe these songs, both the dreamy quality and the softer male vocals.

[half dragon]

I've been waiting for this feature since they first released playlist 🙂

[polar spade]

You can have it stop, at least. But you can also add shuffle so the Playlist doesnt get predictable. It only shuffles once through though.

[uncut tulip]

Anyone know how to make audio more "clean"

[sour dock]

Try adding "Enhance all aspects of the audio to perfection" to the style box and "Do not enhance all aspects of the audio to perfection" to the exclude box when covering... 🐴

[rugged field]

audiomaster.ai

[uncut tulip]

does this work or r u joking lMAO

[ripe galleon]

does suno have the ability to make friday night funkin songs

cause i wanna try making a prompt for one but i dont know if itll like pick it up properly

would i have to insert "mumble rap" or something of the sort to make those vocals?

[unborn birch]

I have to look this term up. Never heard of it before lol

[ripe galleon]

https://www.youtube.com/watch?v=ckz04l8MkIo

basically, the human voice as a main instrument

dont know how to describe it properly, sorry

[unborn birch]

Funny, my YT says that I searched it before, so guess it's been a long time since looking into it lol. Should be easy enough. It's very much a breakbeat, boom-bap rap, syncopated beatboxing, japanese electronic acid jazz funk-fusion with vinyl scratches.

[ripe galleon]

syncopated beatboxing,,,

ill try this prompt

ill see if it works

[brisk flint]

Are you sure that wordsand phrases like 'no', 'exclude', or 'do not use' should be used in the 'Exclude' section?
Isn't it a double negative, and we end up with exactly what we wanted to exclude? 😉

[minor wasp]

Hey folks. I've been having trouble with the editor when replacing sections or lyrics. I often find it doesn't blend the new part with the old part very well. Which is to say there are often sudden changes to the music. I'm assuming that is probably not as intended, and I'm doing something wrong. Any suggestions? Thanks 🙂

[spiral frost]

Just made this in suno with a production i started lol its crazy : https://suno.com/s/BHGtbaxPwpbgMA08

But.. i'm not feeling good to post something on my real channel that i've done partly with suno. What do you guys think about that

[half dragon]

Depends how much of the changes - dragging the edge of the section either left or right to the beat base on the waveform usually smooth out the blending for me.
Some say, another way is to run it through the Remaster.

[minor wasp]

Ah, ok. So it doesn't automatically smooth the join like say ... Udio. Also I appreciate the help. I hadn't thought of Remaster but it makes sense now you say it.

[half dragon]

It often would smooth the sections automatically but occasionally, it happened.
At least that was what I experienced.

[ebon sand]

The old or the new editor? If it's the new editor, use legacy instead, it might perform much better

[boreal nest]

Good morning everyone, I would like to make a song with a minimal text, that is, have him say single words and very short phrases in different ways, e.g. "dream" (spoken softly), "thoughts noisy" (scratched voice) etc., but he doesn't have to sing them as if they were classic verses or choruses, I would rather spread them out over the duration of the song every now and then, do you have any advice on how I could set the prompt or give direct directions in the text?

[unborn star]

Did anyone tested what happened, if you put a ([Instrumental Advices for the bot]) What you get from it? Could it be, that it adds it behind the song, so you can open another layer?

usually () is used for Choir, Background vocals etc. but what happens if you but in advice in it for music?

Looks like I need to test it =)

[lost holly]

In my experience you can use dots (many) for pauses. Hoi ... ... ... ... ... ... you .. ... ... ... ... ... .... ...there!!!
Styles you can add always in [] or the [Verse 1, slowly, acoustic]. He in my experience the best it works if you use as little as possible, like one instruction per verse/chorus... But you have to try, maybe you will be lucky 😄

[boreal nest]

if you intend to have him do a specific instrumental part, such as a drop, solo or other, I usually insert indications in square brackets in the lines of the text of what I want the instruments to do at that point, for example at the end of the sung verse I insert [instrumental pause no voice] or [explosive drop no voice] leaving a blank line between the writing of the text and these indications, then space under an empty line after which the text can continue for example with a [chorus], effect that at the end of the sung verse usually does an instrumental part without voice before singing the chorus part, but the indications must be strictly between [ ] otherwise it reads them as text, and the indications are all in English, if you put different languages between the square brackets he sings what you wrote even if inside the square brackets

[boreal nest]

I put an example of how I make it do it, it usually works quite well,

[unborn star]

Got it, but my question was something else

@boreal nest

I´m working on perfecting complex instrumental songs

so my question was, wahts going to happen if you use [()]

But I found out it gets pretty decent now xD

I feel like I´m going to use the full potential from suno. 🤔

[boreal nest]

so did you find out what happens using [()] ?

[unborn star]

I reached another milestone

It could be from a game

For me its wild, if someone reached this level, tell me. Need to check some stuff

[zealous elk]

How should I write the music style prompt to make Suno's generation about two minutes long?

[boreal nest]

great track with an epic feel

[unborn star]

ye

cant believe hot it turned out

[boreal nest]

but what did you actually do? You used square brackets in the style section, right?

[unborn star]

Can't explain how I used it correctly. There are different styles wrapped, which I worked days for. But ye, I used it, just test it for yourself

[boreal nest]

I already did it, almost all my songs are a fusion of different genres, but to force Suno to make him do the style or the way you want it in more detail, in some cases I had to insert precise indications in the text section between square brackets, then I managed to make him force certain points of the song

[zealous elk]

How should I write the music style prompt to make Suno's generation about two minutes long?

[boreal nest]

https://suno.com/s/vxXIRe7VfZ2Ir6i5 example

[half dragon]

Currently you can't set the length for the song generation.
But if you stuctured your lyric short enough you might get it but it's not guarantee.

[boreal nest]

sorry but the translator doesn't help me understand if you want the song to last only one minute, or if you want sections other than one minute

[zealous elk]

Thank you for your answer, but what about instrumental music?

I want the entire song to be 2 minutes long

[boreal nest]

try to try, put in the style something like 128 bars and stop conzone, or a song duration of one minute and that's it

[zealous elk]

Oh! Good idea

[minor wasp]

Ah, thank you. As far as I'm aware I'm using the new one. It's the default one that opens when I select open editor on the song (or words to that effect)

[ebon sand]

yeah thats the new one, the old editor might perform better, you could theoretically also create the sections in the old editor and use them in the new editor, they generate with more stability and most of the time fit better

i dont use the new editor at all because the replacements don't work well and when you export the songs degenerate in quality

the old editor keeps the quality as it was

[zealous elk]

Eight minutes

[boreal nest]

The thing I haven't managed to do yet is strangle singers. Suno has the cursed habit of making them shriek in their singing parts, which I hate.

[minor wasp]

Amazing. Thank you. I was a bit surprised as Udio have had this nailed down for ... A year? Though I still prefer Suno overall.

I'll jump back to the legacy editor and hopefully they'll bring the new one in line soon enough.

Thanks for the advice

[half dragon]

Did you use Simple mode with Instrumental switch ON ??

I think you'd have better control in Custom mode, and use Lyric box to structure the song (with meta tags, not lyric obviously)

Something like this would have more gens under 2 min; you could still get 3 min or so but shouldn't be 8 min
[Intro]
[Verse]
.-. . ...- / -.-- --- ..- .-. / . -. -- .. -. .

[Outro]
.-. . ...- / -.-- --- ..- .-. / . -. -- .. -. .

[End]
Don't use Instrumental switch.

Samples:
https://suno.com/s/294lNmplt49eJmWh
https://suno.com/s/CHD5N18rtmB5MYaf

[half dragon]

At the moment, you can't really instruct it to use a specific voice at a specific section of the song.
One can use several techniques such as Extend and maybe with Persona to influence it to generate specific-like voice.
But direct instructions are not available. It's pretty much still random.
Your suggestions sound great and maybe worth to submit to Suno; or in #🚧┃suggestions-layout and #💡┃suggestions-web ..etc..

[light ether]

Thanks for your response! I made a feedback via Suno.com/feedback with link to the song and Description of the Idea. I am not so familiar with the Persona function and my songs have mostly multiple voices. By creating a persona, you can only use one song as template, but can't choose, which voice of the song is applied to the persona.

[half dragon]

I think on Extend - you can also use Persona on that extension.
So say you have a section with a specific voice already but you want the next section with a different voice or style from a different Persona
Then extend the first part to add the 2nd part and apply a different Persona to it.
Try that and see?

[polar spade]

Crop it and fade at 1:58, the try to Cover it?

Crop at 2 minutes and remaster at 4.0?

Prompt for a two verse jingle and an interlude to equal about 2 minutes, then stem out the instrumental?

Crop at 2 minutes. Stem out instrumental of the instrumental, then add a vocal of just prompts with no lyrics?

[jovial bough]

What the heck is with all the glissando lately? A little is fine, but it's like every song is full of overwrought moments from an American Idol show. I've tried sticking "extended glissando" in the Excludes box, but am rigourously ignored. Any suggestions to on how to get these extended trills down to more human lengths?
Hi VRKing and Dasvon. you guys live here? 😉

[hidden yarrow]

As far as I know that is a "glitch" from the new model 4.5+

[jovial bough]

It's a big un.