📚┃suno-school

[knotty stag]

Thanks, it actually has created lots of stems, but I was also trying to get the MIDI files, which did not materialise.

[pallid leaf]

I found the "new" editor, but it gets me to legacy editor after an error

[polar spade]

OK. Cool. Problem one , stems discovered in Library. Check.

So you only have Pro then, or Studio as well (Premium)?

[polar spade]

Ongoing issue from Friday. We may have to wait for that one.

[still frost]

hey, does anyone know how to make 128d audios for free? (ik its not a suno ai related question but still)

[echo hill]

Sup

[languid forum]

if i move one workspace to trash, will all songs in that workspace go in trash too?

[half dragon]

I'd think so. Try it -- you can always restore it back from trash bin.

[polar spade]

No. If you only delete the workspace, since there is only one file of each output, it just remains listed in your Library or playlist.

Youre not literally moving it from here to there, youre just adding a placemarker for the file when you put them in a workspace or Playlist.

You always have to trash the file itself for it to visually "disappear" from the Library.

[unreal sleet]

is it common that inspos can make vocals slip up? I've used the same inspo for like 600 credits and it's been slipping up at random parts every single gen, not exaggerating

[forest hemlock]

Not common since inspo tends to affect the texture more, but could be if there's something like that somewhere in it, for me it's usually more related to the prompt itself or the particular lyrics

Maybe it could happen if the inspo is drastically different from the style you want to create

[unreal sleet]

inspo is clean, using same style as the inspo

made several changes to the lyrics to be closer to the inspo rhyming schedule or what u should call it

[thin karma]

I recently started making beats my self i want to learn everything there is to know about music

[unreal sleet]

https://klipy.com/gifs/wtf-wth-19

[polar spade]

Thats a lifetime of learning. But the internet has enough info to fill a lifetime of learning as well. I would suggest, learn to at least play one instrument then. That will help.

[unreal sleet]

you started making beats or you started using AI?

Like did you just start out with a DAW and now u want to try learning about AI instead?

[quiet solar]

What kind of slip ups are you experiencing?

Inspo is usually quite broad and general - while accepting when it comes to Suno the rule "everything affects everything else" tends to apply, unless there's something really odd about the flow of the lyrics in the Inspo you're using, I wouldn't expect it to be causing problems like that. I'd have thought it's more likely to be Suno struggling in some way with the new lyrics.

[languid forum]

<@&1435720322207715419> <@&1372996242325508176> ahhh thanks you everoun

[unreal sleet]

Yeah like I said I've tried to change the lyrics where the words cant be pronounced, and even if it rhymes (which can be an issue if it doesnt) it transforms into inaudible jibberish. but now it even struggles to rhyme cost with lost so i am very confused. its just always somewhere, in every song i make, not following a pattern like always after the chorus or in a specific verse, it can literally be anywhere

remade the styles and brackets in the lyrics and no success

[sleek stump]

Are you trying to use Inspo to do lyric corrections? If so thats not going to have any degree of success.

[unreal sleet]

not at all, just want the inspo cause it sounds good

and it did what i wanted after a lot of trial and error with styles

[sleek stump]

If you have a track that you like from inspo, just do a cover of that with the correct lyrics and slider settings W: 10 S:50 A: 80 and leave the styles window blank.

[unreal sleet]

huh, im trying to make a new song, not another version

[sleek stump]

Same thing, if you get a new song that you like but it has messed up lyrics, cover that with the process I noted.

That process works for me 95%. Using remaster for that is 0% success for me.

Im just noting options you can try.

[quiet solar]

Cover, rather than remaster. xD

[sleek stump]

Not remaster.

I am assuming you are using v5...

[quiet solar]

Just to add, I use 15/50/80 for the same all-purpose "have another go at it" that Ginger described.

If you're using non-English lyrics, your mileage may vary a bit on corrections. Suno isn't as quite as strong in some other languages, I believe.

But as a general tip, if you run into odd inexplicable problems, it's worth trying to get to a version that doesn't have Persona or Inspo or anything else attached other than the audio feed from the song you're covering. Once Persona's established the voice, you shouldn't need to reapply it every time.

[shut marten]

So... I want to use the extension feature for a song but I noticed they changed it where you can't adjust the bar from the front just the back only... but my question is what if you have a situation that you want to keep the end but not the beginning of the song? I tried to move the bar but it wont let me... just allows to adjust the ending....

[gritty mulch]

I have found exactly one use for remaster: i often produce several covers of a song so that i can splice out the inevitable glitches/mispronunciations with a good segment from a different “take.” Sometimes the quality or tone does not match up. If I can get it close with EQs/volume remaster does go the last bit and make it sound more homogenous but you have to get it fairly close in sound

Oh and i guess sometimes its capable of ironing out a single glitch if its minor but theres also risks of it introducing more though i find that pretty rare. Always use subtle.

[sleek stump]

You can use cut and insert to do that for the front of the track.

[brisk nova]

Error for two days now, Pro, Edit, Open in Editor, click on three dots under wave form and this error message pops up: Message: Cannot read properties of null (reading 'showNav') Trashed cache, logged out/in. Help.

[sleek stump]

Have you tried the Legacy editor?

Also, try clearing your cookies, I have had corrupted cookies do really weird things in the past...

[brisk nova]

Don't like Legacy editor! Yes to cookies. Support said: Given the persistent 'showNav' error in the Editor across all songs on Chrome/Chromebook, and that both cache clearing and re-logging have not resolved it, this qualifies as a technical feature issue.

[brisk nova]

Thanks for the idea... any more are welcome. I'm having withdrawal.

[proud crow]

Same here.

[polar spade]

Drop by here often. Many good ideas to be found.

[azure crane]

The vocals and instrumental solo going nuts in the second verse all the way to the end. High long notes, screams, adlibs on speed, etc. When will this be fixed??? This is no small problem, no, it ruins the whole song.

[sleek stump]

YOu can try using vocal queues in your lyrics... i.e. [Chorus - soft] [Verse - whispered] ect, and call them out in the styles prompt.

[azure crane]

Well, I've already tried that but it doesn't help.

[sleek stump]

You are likely trying to gen a genre that Suno insists has all of that stuff...

[azure crane]

Definitely not.

[polar spade]

Before any models after 4.0 we already would use the Legacy editor that used Replace section which, without sliders, we trusted the defaults to clone the melody , instruments, etc as well as possible, but we could change prompts to help tone down thing, including vocal delivery.

Thats an option to run 4 replacements as a test and see if any feel usable.

Just a thought.

Of course if its not so good vocals from a point , all the way to the end, then doing extension with the same idea of adding "ambient" or "ethereal " about 6 prompts in among the older prompts helps to calm things down all the way to the end again -- might be a more extensive option. . Yes, we then deal with the idiosyncrasies of extends, but again, im just throwing out possible suggestions.

[quiet solar]

If you share an example of the song, so we can see the prompts and the output it'd be easier to give more specific suggestions.

Otherwise we can only give very general thoughts and say that if this is happening in a genre where you wouldn't expect screams and adlibs, then either Suno was having a very bad day or there's something in your prompting or the way you've presented the song that's steering Suno in that direction.

It's not normal behaviour for Suno to do that, but we'd need more info if you wanted help diagnosing and fixing.

[worthy phoenix]

Somebody know how to do a song with two different female voices changing the voice in all over the song?

[frozen dagger]

Look in https://discord.com/channels/1069381916492562582/1252267444773326848 for duets and multiple singers.

[quiet solar]

Getting two singers of the same gender isn't straightforward- though it helps if you can prompt for them being as different as possible (e.g. different vocal range, characteristics, singing style).

Switching over for different parts of the song is then a separate challenge. If you want to 'control' exactly who sings what, you'll almost certainly be looking at a mix of direct editing in a DAW or Studio to support what you can do with prompts - if you're relaxed about exactly how and where they change over, it should be possible to do through prompting alone.

Some additional info and advice here, if it helps: https://discord.com/channels/1069381916492562582/1466598465969590465

[worthy phoenix]

Thank you, gonna look in that

[rough depot]

all of a sudden they singing too fast https://suno.com/s/vH1VMJ20ltpCuU38

[quiet solar]

Hey again. 🙂

I'm tempted to refer you back to what I said before - the syllable length of lines has a big impact on the pacing of the song. The opening verse typically sets the rhythm for everything else that follows, and in this one you have lines of 13-15 syllables; which is a lot to cram in to an 8-line verse. Between that and the overall tempo being high, Suno has your singer going like a machinegun.

Contrast that with Verse 2 where you (accidentally?) gave it 6 lines where it would be expecting 8. When it gets to the last two lines, Suno drags them out over two lines each to maintain consistency with the overall song. And when it splits those 17 syllable and 12 syllable lines in half, it gives a much slower delivery.

If you want to slow the singer down, I'd suggest either;

• Specifying a slower tempo for the overall song. It's currently at about 165bpm - maybe put 120bpm into the style prompt, or 'moderato'. Alternatively, use Adjust Speed to forcibly decrease the existing song.

• Split your verse into two and spread each current line over two lines. That would give you a double verse before each pre-chorus, and should help send a signal you're looking for a song with a slower pace of singing that's more similar to the chorus than the current verse.

[rough depot]

Tried a new one https://suno.com/s/9zSaqsS3gwKqxslk

[cold zenith]

I'm in studio, and i'm trying to covers the vocal stem, i put in style ( Acapella, no instrumental, ) but it always puting instrumental. Does someone has a trick ?

[rough depot]

How the f DO I GET IT TO STOP YELLING THE SPOKEN WORD PART https://suno.com/s/iLF3JC2JqQD9hh5g

like srsly how stop yelling in spoken word parts https://suno.com/s/bEzHymTeFzeOrz4a

FFFS STOP YELLING!!!!!!! https://suno.com/s/N0VNvbAkxJIC4uCW

[polar spade]

Remove the exclamation points helps. An exclamation point , with Suno, says the whole sentence is important and said with energy.

[sleek stump]

dude...

[rough depot]

Literally put [Spoken Word][Narration][Deadpan][Calm][American Accent] in the tags above the paragraph

[rough depot]

energy shouldnt mean "sing it"

[sleek stump]

Ya thats flooding Suno with a lot of tags for one thing... Try this.. [Intro - soft spoken word]

[polar spade]

As you wish. But tell Suno that. 😄

[fathom basin]

exclamation marks literally mean to be excited. most people get louder when excited.

[sleek stump]

You are asking fopr power metal, which is full of raised voices and screaming...

[polar spade]

keep in mind, suno understands common phrases and colocations, so things like "making a difference in this world" will already be understood , often, to mean emphatic.

[fathom basin]

not sure if im asking in the right spot. but is there a way to download my whole library in one go or do i have to download them 1 by 1?

[sleek stump]

Nope, you need to do each individually

[rough depot]

https://youtu.be/Sx5BVLgRu_o?t=142

Literally, spoken word in symphonic power metal.

[rough depot]

emphatic doesn't mean start screaming.

[fathom basin]

https://tenor.com/view/why-banging-head-gif-15181510

you could try "emotionally" rather than words that typically mean raised voices

[rough depot]

"Some days you just can't get a damn brisket sandwich.. oh well, high adventure is just as good. Plus, I reckon I'm really making a difference in this new world., which, quite frankly, wasn't ready for this kind of speed.."

[polar spade]

You win my friend. Leave the exclamation points.

You're preaching to the choir. We just have to know where to make adjsutments and kow-tow to Suno. Thats the purpose of Suno-school. To pass on the knowledge.

[rough depot]

[Acoustic Guitar Solo]

[Spoken Word]
[Narration]
[Deadpan]
[Calm]
[American Accent]

"Some days you just can't get a damn brisket sandwich.. oh well, high adventure is just as good. Plus, I reckon I'm really making a difference in this new world., which, quite frankly, wasn't ready for this kind of speed.."

[Whistle Dixie]

Also i think i may need to add my own dixie somehow

[polar spade]

That looks cool. Try that one.

[rough depot]

Thats what i been trying

[sleek stump]

and it isnt working, is it?

[rough depot]

it also wont give me the acoustic guitar at the beginning and some times it skips the entire intro

correct, it isnt working.

[fathom basin]

for leasing with the acoustic guitar part. try adding "only"

it helped for mine for my fold fiddles

folk*

jesus i can't type today

[rough depot]

like so?

[Intro]
[Acoustic Guitar Solo]
[Acoustic Guitar Only]

[Spoken Word]
"Some days you just can't get a damn brisket sandwich.. oh well, high adventure is just as good. Plus, I reckon I'm really making a difference in this new world., which, quite frankly, wasn't ready for this kind of speed.."

[fathom basin]

right

[rough depot]

what i been doing has worked for EVERY character image song (eventually) except this guy's

[fathom basin]

its AI.

[polar spade]

Cut the tag back to [Spoken] and only one other choice... narration can be redundant, so try

[spoken] [calm] " some days . , ,"

or how you have it above could work as well

[fathom basin]

some models still just be wonkky

[sleek stump]

Also, Suno will not know what to do with [Whistle Dixie] and igjnore it.

[rough depot]

Heres some songs that did it right https://www.youtube.com/playlist?list=PLG5mHUq6GRRXjI_tB_F0-JCQk7HMyHLjg

christ why wont it give me the guitar intro

[Intro] [Acoustic Guitar Solo][Acoustic Guitar Only]

[Spoken][Calm] "Some days you just can't get a damn brisket sandwich.. oh well, high adventure is just as good. Plus, I reckon I'm really making a difference in this new world., which, quite frankly, wasn't ready for this kind of speed. "

GAH WHY

It keeps doing this weird thing

https://suno.com/s/YjSUjZn0fRWlpbYx

[fathom basin]

[Whistle the tune of "I wish i was in Dixie"] might work. its not copywritten anymore so it should be fine

[rough depot]

"...just as good. Plus" but without ANY pause between good and plus

[polar spade]

I mean... all i did was take out the exclamation point and combine the whole line as one quote. spoken calmly, no problem.

https://suno.com/s/uMN4c4IqLOasBg31

[fathom basin]

a period indicates a pause

[rough depot]

But its SKIPPING the pause thats the point

[fathom basin]

you don't have to use proper punctuation

oh

try a comma

[rough depot]

"It's saying good plus" instead of "good" [beat] "Plus"

Its ignoring the period as though it werent there

[sleek stump]

This literally gave me eactly what you are asking for...

[Intro Soft Spoken Word]
Some days you just can't get a damn brisket sandwich,
oh well...
High adventure is just as good....
Plus, I reckon I'm really making a difference in this new world., which, quite frankly wasn't ready for this kind of speed

[fathom basin]

if that doesn't work try a friggin semicolon lol

you're writing poetry, don't need accuracy, its about effect

[polar spade]

Mine had a big pause with the elipses..., before the word..... Plus.

So one and done isn't always Sunos forte. You might need to leave it as is and click a couple of generations as well.

[fathom basin]

how long of a pause are you looking for?

[rough depot]

Like a normal pause between sentences

[fathom basin]

pause for effect or just a separation of statement

yea, try comma or semicolon

[rough depot]

soft spoken word is still resulting in yelling

[polar spade]

Am I on "ignore" lol.

[fathom basin]

ellipsis will likely do a dramatic pause

[rough depot]

Its ignoring punctuation entirely

1. Its pausing too much between lines
2. And he's yelling again

https://suno.com/s/jfCfklk6wPi9cRMG

[fathom basin]

could also put the next sentence on the next line as awell

[rough depot]

And turning Rev it Up into 5 sillables

oh shit it finally did it https://suno.com/s/j8bjg6Gwzt7YqlZA

...but its still got too much whooaing.

[polar spade]

Check, check and check. 😄 I must be on ignore. Going to bed. Good night all.

[rough depot]

link?

[polar spade]

#📚┃suno-school message up there at 1:10am

good night folks.

[rough depot]

danke

[polar spade]

bitte

[polar spade]

What's the advantage to NOT having access to a list of the 172+ songs that include just the one Persona? Many times Ive rediscovered songs ive used a persona on that I forgot to put in a playlist. Im going to start focusing on adding to two playlists now, a regular one and one for each specific Persona. I guess just add them as I rediscover them. 😁

[south fog]

Is there a way to turn off that V5 learns from the genres you previously generated? I was experimenting a bit too much with reggae style music and when I try to do a different style it still has a reggae feeling or sometimes does only reggae when the genre I put in is something completely different… I really love the quality of V5 but I can’t really use it because of this issue… ||(I mostly cover songs I made irl to find out how they would sound as a different genre)||

[glad harness]

Is there a way to force Suno to create an additional verse section (and an additional chorus section) with the exact same style as other verses, but just with different lyrics?

[frozen dagger]

duplicate the section you want transformed in the editor (or studio) and then cover it. (syllables matching is best for less change; adjust sliders to stick to audio)

[frozen dagger]

according to the team, the only thing that effects a generation is what is in the style prompt, lyric prompt, and any audio influences added.
the training it does is on how you use it, and (i believe) trains on the outputs you make, as well as the uploads (since this are being used to create outputs).
i am skeptical sometimes myself. i create a new workspace and start fresh in there. (just in case)

[quiet solar]

v5 (indeed any model) doesn't carry memory of previous generations like that, except where you're bringing audio feeds over yourself (e.g. using Inspo or Persona).

If you're not bringing in audio influences, then perhaps try resetting your browser cache if you're on desktop. If none of the above fixes it, maybe raise a ticket with Suno Support.

[gloomy plover]

In Suno studio I want to take an existing stem with say, an electric guitar, and apply that EXACT sound/style to something I have hummed. Can anybody tell me how to do that?

[south fog]

Does creating a new workspace really reset V5‘s „memory“?

Also what’s this then?

[quiet solar]

@gloomy plover Suno and 'exact' don't typically go hand in hand, but the best way to do that in Studio would be to insert your humming just after the existing guitar track as a guitar stem and then cover the whole guitar section on a low weirdness and fairly high audio influence.

Maybe start with 15%W 70%A as a starting point, and play around with both as needed; more weirdness/less audio if Suno seems to be struggling to process then new section, less weirdness/more audio if you're getting the new part procesed but too much deviation from the original.

[quiet solar]

According to the devs themselves, there's no memory to reset.

Suno hasn't mentioned "Adaptive Creative Intelligence" since, but since it's talking about making suggestions, I presume it's related either to the Suno Chat beta, the cycling of suggested style ideas in Create, or possibly the magic wand.

Worth bearing in mind that was basically a marketing release rather than a technical feature list. Apparently each musical element remains consistent across multiple generations, which would raise a few eyebrows (or at least prompt a discussion on what is meant by 'consistent'). 😉

[frozen dagger]

This is an excellent question.
My best guess is in the tools like the magic wand that can create new prompts, or with auto lyrics, or other tools that come with using v5 with Studio.
Excellent question. Clarification on exactly what that means would be welcomed.
As far as the new workspace. The idea there is just that if it is looking at things in my Workspace, then making a new Workspace may mitigate whatever "memory" may or may not be happening... but as far as I understand it. There is not a "memory".
I can say it must know your genre types and styles that you use (most often) as it auto-generates those on your profile if you don't manually set it.

[quiet solar]

But whether it uses those genre types and styles to influence future creations is a very different question. To which the official answer is: no.

@south fog Can you confirm you're not using audio feeds like Inspo or Persona in the generations that are spontaneously going Reggae despite you not prompting for it? And if you're happy doing so, could you share an example song so we can see the prompt and output?

[frozen dagger]

If the prompt has "Caribbean" or "hand pans" or some ska reference.... I get a cookie.
You're in good hands with @quiet solar . Back to cards. GL.

[south fog]

Yea, i put in a song i made irl using Ableton. Then i made multiple Covers of it using suno in reggae style. After I’ve found the best result i wanted to try something different (i think it was melodic metalcore or some sort) but it still had that reggae feeling and a reggae guitar. Only difference was it was played a bit „heavier“ if you know what i mean.

And it did that after multiple tries to generate a different style

I’ve looked through my library and trash but can’t find the generated songs anymore that had this issue. I’ll try to do the same thing again and see if the issue still exists or not once i get home

[quiet solar]

Was it the Reggae cover that you covered into melodic metalcore? Or the original Ableton song?

Because if it was the Reggae version, that's 100% the source of the Reggae influence in what you subsequently generated.

Some very distinctive genres are a bit hard to totally switch for a different genre, since the melody, pacing and lyrical flow won't be well-suited. For example, a slow 60s love song crooner with short lines and lots of long sustains would be a nightmare to convert to a Nu Metal song that would typically have a high syllable-per-line count and very rhythmic lyrical flow. Suno will do what it can, but it'll be constrained by the source material you give it.

[south fog]

The Ableton song. I covered the ableton song into reggae and then i covered the ableton song into melodic metalcore.

[polar spade]

They were asking was the Ableton song reggae that you uploaded?

Or did you upload a "regular" song, covered and prompt for reggae, then deleted the reggae prompts (blanked the style prompt window) and tried another style and still got a bit of reggae?

Or did you go back to the original upload and cover it again in a different prompt and still got a bit of a reggae vibe?

[south fog]

no the ableton song was melodic dubstep. and i did the second thing you mentioned.

[polar spade]

Just clearing the style prompt and adding s new style, its possible (some feel) it may have some residual affect, but all depending on about 12 other factors. Best thing is to clear the Create page and start again. Repost lyrics, put in new styles, re-position your sliders and so on. Its all conjecture but with some anecdotal feel as well of "i swear it sounds like that last vibe" a bit, if you just change out only the prompts. Such a discussion shows up every now and then.

Of course, you should be able to , rather than just change out the prompts on the same Cover of the upload, clear the create panel first, then go back to the original upload in the Library and Cover it again, adding the prompts you want. That avoids number two scenario and should give a new output.

[south fog]

Oh no, i think you got me wrong. I meant the second thing as in the second "paragraph" of your previous message. I did not refresh the whole page but rather i cleared only the styles. I usually do that and it always worked but when i tried out v5 for the first time it did not work...

[polar spade]

Diving a bit deeper (sheesh im sounding like an Ai now) as @quiet solar Kanthric mentions, 1) there's never been any in situ demonstration on what Adaptive Creative Intelligence encompasses or how it is seen or affects out creation, so its unknown really.

And 2) the devs, also as @quiet solar mentions have emphatically noted that this residual feel should not really exist or occur.

[polar spade]

Then Id go further and do the third thing. Start from the upload in the Library instead.

[south fog]

hmm ok, ill try it later...

thank you guys

[polar spade]

The good news is suno is constantly updating, perfecting, making things better... the less good news is occassional with these improvement it might temporarily open a gate somewhere that affect the outputs.

People also get 10 minute outputs, 4 outputs instead of 2, and other little sunocorns occassionally.

Its sporadic and not contant and like the weather, wait a couple of hours, it should change back to normal You might have gotten a rip in the fabric of the sunoverse and reggae just crossed over.

Ultimately its all seen as a positive growth. So thats why we have many users stick around for years now.

[crimson silo]

https://youtu.be/bitwd5K1gro?si=W9b70ht40JbWIciv

how would i prompt this?

[polar spade]

Most times, its easiest to use an AI sourced song genre analysis software, jot down the styles it finds and start with those.

Or just outright ask an Ai chatbot. Grok threw out these:

ultra slowed phonk, brutal bass, 65 BPM, chopped and screwed, massive 808 distortion, dark villain phonk, 90s nintendo glitch effects, eerie pads, deep sub rumble, night drive, menacing atmosphere, extremely slow tempo, thick reverb, brutalism

I dont many will work out the exact brand and model synth or preset used, so the instrumentation may be more about research of the band.

[crimson silo]

thanks!

[crimson silo]

yeah the goal is to get something as close as possible, but maybe i will need to simplify it? Interesting results though s ofar

wow okay it looks like it might be worth getting more credits again xD

Decent outputs but I can't seem to figure out how to word it to get close to what i'm looking for hmm

[polar spade]

Well, one step closer is . . . well , one step closer. Always a good thing.

[pliant steppe]

I sure would like to use it

[polar spade]

You mean learn to use Suno? Its not too hard for just making a song about anything we can think of.

[sweet ivy]

Guess I did something wrong. I highlighted a piano stem in Studio, then clicked Cover. I didn't see any prompts come up for style or anything, it just starting generating two new piano tracks.. Tok quite a while to complete. I put piano on Solo, then clicked version 1 & 2.. Both had vocals & some strings them I was expecting, just a new piano track ?

[dusky stone]

hello guys does anyone have experience with "Enhance your music with
Mixea's intelligent mastering engine". Is it worth it (10$ each song?!?!). Thanks for sharing your experience

[sleek stump]

This is a known thing. Currently covering just an intrument will gen a full song like normal create. Our work around right now is to stem that and use the new part you were originally after..

[sweet ivy]

@sleek stump I guess I could insert the new piano, have a new version, extract the stems and be able to hear the piano track and see if I like it, what parts I could use. I could drop that new piano stem into my DAW, or I guess Suno Studio.

I asked Suno support and they said simply go to Song in Studio an select Keyboards, then hit Cover, that should generate 2 versions of just keyboards

[sweet ivy]

That worked, but never a dull moment with Suno. listening along and suddenly a different chord at the start of the bridge??? More Rhodes style piano than orginal. but I probably could cut ad paste parts it.

[lone nova]

I am not very experienced, I saw a video about someone want to insert a guitar solo:
he sing the melody directly into mic, use DAW to generate and select 1 guitar solo, get the guitar STEM again, cut and change volume from 0 to 100 and down to 0 using exp curve.
That video was for studio, which i do not have.

[sweet ivy]

@lone nova Appreciate the suggestion. I saw that video also. Good idea for a small section of a track. I think I will pull the entire piano stem in my DAW and cut and pasted what I want.

[lone nova]

How to Produce a Real Song in Suno AI (Don't Just Generate)
from AI With TechZnap

[dull hedge]

My favourite part is prompting music theory, and the AI actually applies the music theory in real time.
-especially the part when she sings the higher notes, and the lyrics reflect that. I don’t know how, but SUNO matched the motion perfectly. I don’t usually share songs but I had to for this one.
AND THE ENDING TOO when the bass comes in 😻

[sweet ivy]

@lone nova Yes, I watched this before and it's a good guide. Thanks

[lone nova]

My problem currently is I do not know what bad music is.
And posting in suno-showcase does not give me any constructive feedback. 🙁
On top of that, music is highly subjective... 🙁

[neat wolf]

Got it you want people to judge your music

We should have a workshop channel 🙂 for people to get objetive opinions

[lone nova]

Correct me if i am wrong:
The game construct in chat is to get other people to listen to my song, get the play number higher in the hope of some one really appreciate it in the early stages.
Triggering a trend or even to the leader board.
No one have incentive to make each other look back. songs get ignored all the time.

I do not want to play this game, I care about fixing my beginner mistakes.

[strong tangle]

typically people don't really listen to anyone outside of the niche cliques that exist. showcase is kind of a dumping ground with very few people actually taking the time to check out others. but i have insight for things to avoid. in the beginning stages. i wish someone gave that to me so now im helping people avoid those things.

if ya got a question feel free to pop it in. i'm not interested in the game aspect either.

[lone nova]

I made a EDM song with an interesting voice. but if I want to use that voice persona for a rock song, it just does not work well. Sounds like AI.

[hard beacon]

what's yor definition of "sounds like AI" ?

[strong tangle]

id have to take a look at it to see whats going on and how to remedy it. but putting the characteristics of vocal tone into it do help out to a degree like timbre, cadance and stuff like that.

[lone nova]

for reference:
this is the EDM song: https://suno.com/song/2d71b9f8-f9e1-48ce-9b92-e1f9a20b6522
persona https://suno.com/persona/941eb270-e5cb-4f6f-a9f4-a5469dc5f4bf
sounds like AI https://suno.com/song/6a74c1f4-2c72-4710-b553-7d118f6b2ec4
I regenerated this again
https://suno.com/s/EoJ5n1RzdQYxgDy5 still weird vocals

[hard beacon]

to be honest.. why using that voice as persona? that's the standard suno voice without any special difference 😉

but i know what you mean with that AI voice in the pop punk song

[strong tangle]

so there are a few things you can do here.

[hard beacon]

first of all.. i would remove "pitch-perfect" from the prompt.. and first of all try to create a cover with the same prompts and an audio influence up to 35%. And maybe try to create a "new" version with remaster

[strong tangle]

the first approach is one that i typically use whenever im generating a persona. what i do (it could be wrong or whatever, it works for me) i use a global meta directive for the vox initially. like this:

[Vocal: (gender) | (tone) | organic human delivery | live studio phrasing | breath-driven articulation | microtiming drift | push-pull groove | subtle pitch variance | expressive dynamics | natural vibrato | conversational phrasing | minimal pitch correction]

all the stuff are options that you want the vocal to have. about five will do at most after the gender.

"live studio" will often create a live track. if youre into that then by all means. but ensuring the prompt takes the intent having a nlp version of this along these lines will help hammer it home

something like this for your vox line in the style description:

Vocals feel like a real singer performing a live studio take. Timing drifts naturally against the groove with breath-driven phrasing, subtle pitch imperfections, and expressive dynamic changes, creating a raw human performance rather than quantized studio polish.

of course edit for gender and tone.

it may take a couple of generations since suno appears to register the prompt in the 2nd or 3rd iteration more effectively. which is why i typically will only use personas as a remaster work around and have this type of stuff in the prompt ready to go for new generations.

things in square brackes typically always go in the lyric sheet. ive seen people use this in style descriptions with some success but suno documentation states that it just reads meta directives as normal text kind of defeating the purpose but whatever works idk

also "vocal-robotic-quantization", works in the negatives if youre really struggling.something to thta effect

[hearty laurel]

Ok guys, I have one for you. When I go to extract stems from a song in Suno, I get Ver1 and Ver2. But the only full song I can download by itself is ver1. I hear that ver 2 is a higher quality version. Is there any way to download the ver 2 full song, without getting the stems and mixing them in my DAW to hear it??

[neat wolf]

I think there is a big difference between 1 and 2, but sometimes 1 is better than 2

I always thought it was more like a gatcha pull

they just give you two pulls 🙂

[lone nova]

I do not listen to standard suno voices, I wanted the early Rihanna vocal without mention the artist.
I got what I think is good enough with the first song, and trying to preserve it.

[hearty laurel]

If I look at the stems in ver 1 and then switch to ver 2, there are differences in the waveforms in some of the tracks.

[neat wolf]

correct They are different 100% its more than just quality its all sorts of differences

[hearty laurel]

Therein lies my question. Can I download the ver2 full song to hear the differences?

[strong tangle]

you can do research to get you kind of around the same ballpark as an inspiration in an llm but its not going to be one hundred percent which to me that's kind of the point. you arent tring to make a carbon copy of the artist or impersonate them. which is grounds for a lot of legal stuff that no one wants to deal with if you were to just upload their vocal track and dissect it. having an llm and knowing how to utilize it effectively for the purpose of getting that influence is really beneficial. then you can take it where you want after the fact.

[neat wolf]

Honestly, I have the same problem. Sunos algorithm is more or less a very simple popularity contest

If you publish a Song, only the people will get a notification, and even if they do, they may be following 1k more people so they may never see it

You won't get random people seeing it very often

its more or less a popularity setup, more followers the better

and that is a major pain.

So Here is my thoughts, If you need someone to sit down and listen.. just ask 🙂 I mean if i have time ill listen and give you honest thoughts how i feel about it we are all creators here not listeners, so its a little different than youtube or tikotok or whatever

[neat wolf]

Just to make sure i fully undestand you
When you are in studio making a stem, you are wanting to download the second stem

I thought it is in your workspace after generation you may have it filtered out?

[strong tangle]

the thing is the suno community is more or less niche driven. even in spaces like these there's very little you can do if you aren't in the preferred niche. being a grunge artist primarily and with no desire to entertain the whole anime thing or dubstep thing theres not much you can do outside of utilizing social media to get feedback.

but anyway hope that helped shed a little bit of light on the situation. im out.

[hearty laurel]

No I do not have the studio version. I have the pro version. When I go to extract stems, I choose all stems, and I get a choice of all ver 1 stems, and I can get ver 2 stems, but I can't get a ver2 full song where the original full version is at the top of the list.

[lone nova]

I posted the song above, I just want to correctly isolate a single vocal out of a multi vocal song, and use it for a different genre of music, in this case from EDM to rock.

[quiet solar]

New songs drop in showcase almost as fast as you could listen to them, and I suspect most people there are not seeking critical feedback. It's also a minefield for the person giving any, because you have no idea how it's going to be received.

It's also quite hard to know what feedback is most useful. Are you asking "how could I make this song appeal to a wide audience"? Or how to make the lyrics and music hit harder on the theme? Or Suno-specific suggestions on how you could achieve a certain effect? Or just "Do I like it/dislike it, and if so why?" Because how I'd respond to each of those would be very different. Without knowing what you're trying to achieve, it's not easy to know what feedback would help you.

Just posting up a song and asking for feedback is a huge effort asymmetry (time for you to post and ask; a few seconds. Time for someone to listen to the song, review the lyrics, type out constructive suggestions - several minutes at least), so you need to recognise you're asking for a favour and make it easier for people to help you. I'd suggest making the genre clear, as well as what you were trying to do with the song, what do you think is/isn't working well and what you want advice on.

Also, it's worth thinking about whose feedback you want and why, and whether you might want to politely reach out to them directly. For example, if you wanted thoughts on an EDM track then I'm definitely not your guy. But someone like Robin Clark, on the other hand...

[neat wolf]

Hey, I think what’s happening is Suno lets us hear Ver2 in stem extraction, but for the normal full-song download it still seems tied to Ver1. So unless there’s some secret goblin button hiding in your account, I don’t think Pro gives a straight “download full Ver2 mix” option.

If you want the actual Ver2 sound, I think the only real workaround is exporting the Ver2 stems and bouncing them back together in a DAW. Annoying? Very. Elegant? Not even slightly. But I’m pretty sure that’s the path.

[hearty laurel]

Thanx. That's what I thought, but I had to ask. Maybe they will add that to an upcoming version.

[quiet solar]

Both versions are attempts at retrospectively splitting the original song into it's respective stems. So in principle, v1 and v2's stems should both 'add up' to the same identical, original song. In practice it's not always quite that straightforward, though.

[hearty laurel]

Thanx for the reply. But if you switch back and forth between both versions, you can visually see differences in the waveforms on multiple tracks. Thanx again.

[lone nova]

pop punk female without vocal instruction always sounds like Avril Lavigne. That is kinda default behavior IMHO.
I am trying to avoid it

[quiet solar]

Yes, because it's retrospectively trying to break apart the sound and assign it to different stems. Suno takes two attempts using slightly different parameters, which helps avoid issues like "this particular stem split always puts some of the guitar in the synth stem, or puts some of the vocal effects in the guitar track, etc. etc."

[hearty laurel]

I was just hoping to hear what those differences were before downloading all the stems.

I appreciate all the help guys!

I still have a lot to learn.

[quiet solar]

You can preview individual stems. But previewing what the whole sounds like...yeah, not currently an option.

If stem split has done it's job correctly, previewing the stems as a whole in both versions should just sound like the original song in both cases. In practice, the split isn't always perfect and there will be slight-to-very-noticeable differences in the overall mix. I can see why previewing the mix as a whole could be useful - I might raise that as a feature suggestion with the Suno team.

[hearty laurel]

Sometimes when I do a mix with the stems, I wind up using most of ver 1 stems and some of ver 2 stems of the same tracks., because of the slight differences.

I like the idea of a feature suggestion for the team.

I think it would be helpful for some.

[sleek stump]

For all intents and purposes, the 2 stem versions are just 2 different runs at splitting the whole. The master track in the top IS the original track, it is not combined v1 of the stems. If you are a pro account, the only reason to split stems would be to download them into a local DAW. I always dload both versions and tend to use some from v1 and some from v2.

[dusky stone]

Guys is LUFT -14 that crucial before uploading to streaming services? I mean if I only do LUFT in Audition...is not what the streaming services is doing anyway? Thanks

Ah btw is there a good free DAW out there?

[strong tangle]

It's a good target to have in mind I think Spotify is -14 lufs I think apple and YouTube are a bit off by a db or two. Audacity is okay for a free daw if you're just wanting to set it to that loudness. Being able to do it in prompting helps but it's never promised.

[dusky stone]

only luft I can do it audition, is it ok, usefull?

what level should I keep as agolden rule suno studio master volume, like -2db

[quiet solar]

It's the target average loudness that most streaming services aim for to ensure songs are consistent in loudness without the user having to mess around with volume controls as songs change in a playlist. If your song is way out of whack above, the streamers will adjust your loudness down to that level.

What they won't necessarily do is adjust your loudness up to that level if it's lower, as they typically assume you know what you're doing if you deliberately go for a quieter volume, which can catch you off guard if you get into the bad habit of relying on them to fix it all for you.

As to whether that's where you want to master the loudness, that depends on your song. You might want to have the loudest part at -10, for example rather than a perfect -14 average across the board. Equally, you might want to avoid boosting loudness since you may not like how it sounds with that much compression in terms of dynamic range and sound quality.

There are also limits on decibel peaking too - not having peaks above -2db is a good general rule across a lot of streaming services.

[dusky stone]

so your suggestion is to

• keep the suno studio muster volume minus 2
• export the wav and do a -14 LUFT in audition anyway

[strong tangle]

So the golden rule (forgive me if you already know this) of all audio production in the phase is something called "Gain Staging" which means dialing back your track volume just enough to hear the transients and dynamics of any given track. You'll hear them when you slowly go from zero upward and then bring them all in the mix.

I assume suno studio works like this but I dont have it so I'm just trying to give production advice.

When you have them at a listenable level you can use a limiter to bring the volume up to that desired lufs target without losing too much transient information (the punch and all of that)

since suno stems are pre limited when you take them into a daw just make sure the song isn't getting squashed with too much compression or limiting.

Once you have a target your satisfied with streaming should be smooth imo.

[quiet solar]

If you wanted a simple workflow, then yes. That should help ensure that the song as you hear it when playing is how it will sound when played on streaming services.

Whether that's going to have your song sound as good as it could is a whole other question. Audio production and mastering is a skill in it's own right, and the advice from @strong tangle is a really good piece of info.

[dusky stone]

ok thanks both. What do you think of mastering engine (like Mixea's intelligent)?

do they really worth it the 80$/year

?

[strong tangle]

No it's not imo. You can do a lot better in a daw with dynamic eq which artificially "revives' transients sonically. Mixea just seems to add a high shelf boost to everything from what I've heard done to one of my tracks in the free trial. Getting a dynamic eq in a daw will give you much more control. It could be fantastic now but 80 bucks is steep if you're not getting a return on what youre releasing.

[dusky stone]

ok, thanks. in DAW you mean I should work on every track? or just bring voice and Instrumental stems?

I also tried to use EQ in Suno Studio, is it worth it?

And last, sorry all new to mee, what should I do to "daw with dynamic eq which artificially "revives' transients sonically"? I have audition and Premiere can I use them? Thanks!

[quiet solar]

Whether it's worth it is obviously subjective, but depending on your music production costs $80 could be substantial. I'd want to be comfortable it's giving you something worth that extra investment.

If you want a fire-and-forget solution, you might want to take a look at Bandlab's AI mastering. It's not hugely sophisticated, but it has the advantage of being free if all you're looking for is a super-quick pass to polish things a little and move on. Mixea is probably objectively better, but I'm not sure I would personally say it justifies the extra cost if that's the route you're taking anyway.

If you already have a DAW with the tools for dynamic EQ work, and are willing to spend an hour or two learning the basics, I'd be very tempted to start there as that may give you a much better long-term payoff in terms of outputs and your own skills. In normal music production, you would 100% want to work on every single track to get the mix and mastering right.

With Suno? That's trickier. Since Suno's 'stems' aren't really actual stems, sometimes working with them at stem level when they've been artificially split out using Get Stems is very far from ideal, and it might be easier to work with the mix that Suno made for you and just focus on mastering typically cleaner vocal/instrument split. I've certainly done that a few times pre-Studio.

Studio opens up some other options, as you can work with individual stems so that they actually are true standalone stems with consistent and separate audio quality, which then opens up the opportunity to genuinely mix and master them individually.

If you don't have those tools, or don't want to learn the arcane mysteries of audio production (which is entirely understandable, by the way), then really your choice boils down to how much time/money do you want to spend to improve from your current position, just like every other step in the songwriting and recording process. xD

[dusky stone]

Hey first of all thanks a lot for your time and all detailed explanation!

I currently have only audition

and already learnt how to hard limiter and match loudnes

[pure panther]

Quote:

Studio opens up some other options, as you can work with individual stems so that they actually are true standalone stems with consistent and separate audio quality, which then opens up the opportunity to genuinely mix and master them individually.

How are you getting true standalone stems with consistent and separate audio quality? Please teach me the way

[dusky stone]

I would like to have a simple (say 2 EQ PRESET) to have the same outcome of Mixea

I actually have many songs and few time (it's not my job, but I like to do it a lot) so if it's possible to have a 4 steps workflow would be fine. @strong tangle was talking about working with "dynamic eq which artificially "revives' transients sonically. " how pls?

[quiet solar]

It's a more likely situation if you've used Studio like a traditional DAW, building up through a mix of actual recordings and samples.

Otherwise, if your starting point is a song made/covered completely by Suno, it typically involves (re)generating each stem separately, which will be pretty transformative in terms of what the song sounds like at the end of the process, and not always easy to work out what the final version will sound like as you go.

[pure panther]

Ahh, I see. So, if I were to get an awesome Suno song, then "sing" each part into Suno for each individual instrument, could I theoretically "convert" it to clean separate stems? How well does it stick to the recording? (Like humming the guitar stem for instance) Also, does it have a tendency to add random vocals when using recordings? I had major problems with that when trying to generate stand alone stems before.

[dapper stirrup]

is there a way to lock in the ending of a song line being wavy delivery? like knOoOoOoOow

[pure panther]

I'm using this with pretty good success on my current song:
I'm ready, but my heart is bleeding~—,
The squiggly, then the em dash, then a comma, then new line

[quiet solar]

I'm going to defer to @strong tangle on that one, if she has the time to respond. But you might want to eyeball a guide something like this one to get an idea of the main steps in the process. Apologies in advance for some of the jargon - it assumes the reader has some basic knowledge of the terms: https://www.gear4music.com/blog/how-to-master-a-song/

[dapper stirrup]

it's ok, i'm familiar with - and -- and ---, -.., all those variations. not familiar with a wavy delivery prompt though

[pure panther]

It's on the very top left corner of the keyboard, you have to hold shift to use the ~

[dapper stirrup]

ok ty, worth a shot.

[pallid leaf]

I wrote a song long time ago with FL, and now I uploaded it to Suno, which created awesome cover with amazing singer voice, but I can't Persona her as the song itself was based on uploaded audio? Is there any workaround?

[pure panther]

Also, the em dash is different than all the dashes you mentioned. You can just copy/paste it from here. Otherwise on pc, you have to find the alt num pad code for it.
—

[quiet solar]

Yes, you could. Or use the original stem as your starting point if it's not easy to sing.

It'll stick to the recording as well as Suno ever sticks to audio prompts. So, y'know, expect some variations and multiple attempts if you're looking to replicate it as closely as possible. If you're taking this path, it's a good idea to start with a song you like the general melody and arrangement of, but have absolutely zero attachment to the specifics. I wouldn't personally do it on a song you're 95% happy with to start with.

Sometimes Studio will generate other tracks in a stem where it shouldn't, but it's usually pretty good at not doing that these days.

[quiet solar]

In short, no.

You can try prompting for a similar singer in a separate song and create a persona from that, but you can't create personas from anything dervied from a direct audio upload, to prevent people trying to creatively voice clone existing artists.

[main hazel]

so i kept creating many songs and the - tab appears USELESS!!! (they never fixed that bug)

the ai keeps screaming each time refrain comes that my ears hurt, i tried everything to reduce it it still keeps screaming if not in the beginning of refrain then in the middle. any idea?

[dusky stone]

about Bandlab's AI mastering, is it safe to use? and does it only EQ or also hard limiter/LUFT -14? Thanks

[strong tangle]

dynamic eq adds a boost to specific bands when the band hits a specific threshold you you've set. so its reacting to the music. the transients in those bands (highs, mids, or lows) sound more prominent when their suppose to when they hit the peak you set. TDR Nova does this pretty well and its free i assume you can download it and take it into any daw. there is a baked in degree of compression and limiting with ai generated music so having the dynamic eq to accentuate the transients that are there seems like a good idea even if they aren't the best. a good rule of thumb for any boost is to only go up about 1.5-3db max. if you do that to every instrument then that will add to an overall lively sounding mix.

another good thing to keep in mind is to boost slightly when you are making a cut that way it provides balance and you are making up for removed harmonics. if you cut something at 3.6 khz by -1.5 db boost the frequencies that are in front of that 3.7-5khz. with mixing less is more.

[dusky stone]

do you confirm that in Band labs I only do EQ/compression and then I need to hard liiter and lUFT in audition? Moreover I read that is better to bring a suno song alreadi with -3db, correct? TY

[quiet solar]

Was there something specific you were concerned about in asking if it's 'safe'?

I don't know if it's specifically intended to optimise for streaming services - you'd have to check what it actually changes in terms of EQ and limiting afterwards.

[dusky stone]

about "safe" I mean it's seam so good to be free...so I was asking if there are hidden royalties or other

[quiet solar]

Oh, gotcha. No, it's a genuinely free service

[tame venture]

Add a bpm to the beginning, if it doesn't change, edit the bpm by just one, it sets it on a new seed and should refresh what your tryna do

[dusky stone]

thanks I just tried and it looks awesome.
So please help me double check my workflow:

1. -3db in Audition
2. Eq in Bandlab
3. Hard limiter in Audition
4. LUFT -14 in Audition

[quiet solar]

Seems reasonable to me.

[dusky stone]

thanks a lot for your time and help!

[main hazel]

bpm to the beginning?

[finite rain]

good evening, im fairly new to this, and i've made some amazing tracks, but if i can get any tips or tricks that would be great!!!

[polar spade]

I get what youre saying, but where would one add a bpm? At the top of the lyrics? And this will keep it from shouting, or just , as ytou say, give it a new seed to work with and maybe that will change its perpensity to shouting the vocals?

You said "by one." So if there were no BPM tagged in the beginning, then just any BPM will do?

[dense hare]

I just tried putting

[Vocals: (Mexican Spanish)]

in my genre input. Suno returned an error and automatically logged me out

[lone nova]

btw who are those lurkers with 0 msg but keep inv people to dm? like benj527

[peak lava]

hello, morning. i don't know its my error or not. Since 2 days ago, when i use my PERSONAs and i try to make a song with INSPO, SUNO makes a songs but not all, only a part. it not finishes the lyrics and cuts the song at 4 minutes or like this.

[pallid leaf]

I subscribed to Premium specifically to finish a song that needed some adjustments in Studio. After that, I remastered it from v4.5all to v5, but even with very subtle settings the singer’s voice changes completely — unfortunately for the worse. Should the prompt be left blank when remastering, or what would be the recommended workflow to improve the overall quality of the song while preserving the original singer? The track uses uploaded instrumental audio from an old Fruity Loops project of mine.

[tepid moon]

Does anyone know where the New chat Ai exists, I can’t see it, vids on youtube showing it. I am on the premier plan, but its not appeared

[polar spade]

If needed, you can block the user. Posting users names publicly in a channel is not suggested. If you believe its harassment, you can report.

[quiet solar]

I believe it's being tested at the moment with a small set of users. It's not widely available yet.

[quiet solar]

Remastering between versions can lead to significant changes - it is, after all, a totally different model. Some songs make the transition fairly smoothly, others don't.

I'd recommend trying a high audio influence cover and see if that works out better for you. Cover the song, and copy across the same lyrics and style prompt from the original, then try the following under advanced settings as a starting point:

Weirdness: 15%
Style influence: 25%
Audio influence: 80%

Just to add, v5 deliberately made a lot of changes to vocalist performances, so preserving the singer might be a little tricky. Expect that this won't be a one-and-done process, you'll have to give it a few tries and tweak some of the settings (and possibly the style prompt, depending on what you prompted for the singer in the first place) to find the sweet spot of a v5 version of your v4.5 vocalist.

[pallid leaf]

Yeah I don't even know why tried to remaster to newer model, as that trick worked surprisingly well 🙂 Thanks again!

[fading harbor]

When I do new persona+covers, the final output sounds way too much like the original audio. For example, if the original track had extreme pitch shifting, the cover inherits all those flaws in the tone. I don’t know if this happens everywhere, but everyone in my circle is having the same issue.

[forest hemlock]

I have this problem too when I go v5 to v5 in particular, but with v4.5+ to v5 and persona it works better for me

[neat wolf]

scammers usually. Suno is a creation tool, most of them will try to sell you something to market your music...

[neat wolf]

This is has to do with Audio influence, You have to push your style slider up high and AI lower if you want to limit the influence

[lone nova]

He try to convince me to join him for drop shipping with Shopify... I think it is multi level marketing or pyramid scheme.
I always check if said user posted anything on this server or not. Huge red flag, if they do not care about suno at all.
This is the 2nd time btw.

I rejoined this server yesterday, I think they have a new user detection or something, that is why u guys do not get any dms like this.

[simple spear]

I'm trying to add tracks to a 6 minutes song in Suno Studio. Is there a 4 minutes length limitation to the added tracks? Couldn't get one long track that last the whole song.

[neat wolf]

Newer models (v4.5 / v5): up to ~8 minutes total with extensions

[simple spear]

If you are talking about a new song, I know that. I'm talking about adding a track in Suno Studio to an existing song, made with v5.

[neat wolf]

Wont let me post a link
Suno AI Song Length by Plan (2026)
Quick Overview
Plan Max Single Generation Max Extended Length Monthly Generations
Free 2 minutes 4 minutes (with extend) 10 songs/day
Pro 4 minutes 8 minutes (with extend) Unlimited
Premier 4 minutes 12+ minutes (with extend) Unlimited + Priority

[strong tangle]

It says you can upload audio into suno studio from what I'm reading. Mp3 or wav

And from your library with this method:

From Your Library: Open the Library Panel within Studio to access songs or clips you have already created or uploaded. You can then click and drag these tracks onto the editing timeline.

[simple spear]

I'm a Premier Plan user, I should be able to create a track in Studio that extends to the whole song length.

[neat wolf]

Ah got it — you're talking about adding a track in Studio to an already extended v5 song, not generating a new song.

From what people have been seeing, Studio track generation still seems capped around ~4 minutes per generation, even if the song timeline is longer (6–8+ min) and you're on Premier. It doesn’t auto-chain the way the main Extend feature does.

So the usual workaround right now is generating two sections (ex: 0:00–4:00 and then 4:00–end) and placing them back-to-back on the track.

Would definitely be nice if Studio could just generate to the full selected timeline length though.

Sorry lol i was thinking you were trying to see if you can change your song into a podcast 🙂

[simple spear]

I tried that but, even with the same prompt, the two tracks came out a little bit different, as expected. For example, tried to add a rhythm guitar and I got two different sounds.

[lone nova]

Hi how is going guys?

I tried yesterday to put voice of a real older person's voice into Suno, it was bad:
can't use uploaded music as persona,
cover, extend, inspo, all make the original sound lose character,
I have to go back use 2024 AI called SeedVC to do that vocal, and use external tool to merge vocals and instrumental back.
Suno just do not allow that kind of process.

Is this a legal or technical problem? how do other people deal with this? buy Suno studio?
Thx for response in advance.

[fading harbor]

In fact, I have already set the audio influence to a very high level, which can cause the sound color of the cover to be too similar to the original audio tone. My style slider is in default mode because I haven't written a prompt at all. I just want to keep the arrangement and switch to a tone that I like.

[neat wolf]

High Audio Influence + Cover + Persona + blank style prompt is basically telling Suno: “stay very close to the source.”
So if the original has pitch-shifted or ugly vocal tone baked in, the cover often inherits it too.
Fix path:
lower Audio Influence, don’t raise it
add a real style prompt with the new vocal target
raise Style Strength so the new destination is clearer
use Exclude for the exact bad traits: pitch-shifted vocals, helium tone, formant artifacts, glitch vocals

If you want arrangement retention but a cleaner new tone, I’d start around 20-35% Audio Influence, not high.
Audio Influence is a reference pull, not a quality slider.

And yes, your instinct was right. “I left style blank and cranked Audio Influence but wanted a different tone” is exactly the kind of human move that causes the machine to smugly give you the same problem in a new hat.

[dusky stone]

Question about BandLab. I upload the .wav pre decease to -3db in audition. There's an input volume sliders in BL. Should I put it to zero? Btw it's not possible to input a precise digit so it drove me crazy kumping from -0,1 to + 0.1...😅 thanks

[signal spire]

Is there any way to stop SUNO adding that generic synth to tracks especially if you are trying to generate trance/synthwave - it always puts the same synth sound with delays

[polar spade]

Replaces ad Covers will act no differently in Studio than they do in the Edit UI.

So Covering or Replacing anything longer that 5 minutes will need some space allocation to "seed" the new output with old information from the original. Results could therefore be a bit short at times. So expect some shorter tracks as a result.

As some suggest, one idea may be to break the original up (perhaps duplicate first and drop them to different tracks) to try to Replace/Cover sections, then listen for whether they splice back or work smoothly to go for the same duration as the original.

[quiet solar]

If you're trying to keep it locked at -3db prior to limiting and loudness balancing, then yes - that would be the way to go. Something as minor as +/-0.1db is going to be unnoticeable if you're doing further work on it around loudness.

[dusky stone]

perfect, thanks a lot, frinds, as usual!

[strong tangle]

You can specify the type of synth sound that you want to an extent. It takes some research to know which sound you're looking for but much of AI needs specificity to realize the intent provided in the prompt. Much of my guitar gear has been translated into suno which is pretty crazy.

[fading harbor]

Appreciate,i will try this way.Thank you for your explanation.

[wintry saddle]

Yes new persona really needs a good vocal persona to be worth while Imo, try an inspiration set instead maybe? Or legacy persona....

[blazing zodiac]

Hello, I need some tips... If I want a word to extend along part of the next verse, is there any instruction I can do? I tried several workarounds but nothing works at all.

[neat wolf]

Not reliably, no. Suno is good at broad section structure, but it does not give precise note-length control for “hold this exact word through part of the next verse.” Its own editing docs are still pretty high-level: songs are generated as one combined audio file, and even lyric/voice changes are handled by remaking or extending sections rather than by note-by-note editing. Extend lets you choose where a new section begins, but it still does not promise exact sustain behavior

[blazing zodiac]

Oh, I will keep working around with it, I was doing it in some cases but was not provided the precise result... Maybe iterating with those "better" ones as inspiration could do the trick?

[neat wolf]

Yep, the almost-right takes can help as inspiration for the overall feel, but they usually won’t guarantee one exact held syllable. For that, Extend or Replace Section from just before the problem spot works better than regenerating the whole song.

[pure panther]

I've had some success getting this to happen using, "polyphonic vocals" in both the style and the metatags (bracketed instructions in the lyrics) It's not 100% but it does work.

I format the line like this

first lyric (lyric you want overlapped) 

All on one line

[blazing zodiac]

Yeah I use that, but don't work when the secondary voice is the primary on the next line

I found a small workaround that could work but ran out of credits so I can't get more iterations

[wise moss]

Is there a way to give the ending word in the intro to a song an echo type sound?

[pure panther]

Getting a polyphonic song, in my experience, is super hard, herder even than duets. Getting polyphonics to behave with a duet? You're chasing a unicorn, if you find it I would love to hear it! 😄

@blazing zodiac

[blazing zodiac]

You mean using two voices?

Not hard to do it

[pure panther]

You can specify echo in a metatag, like

[Intro | Vocal Effect: Echo]

May or may not work

[pure panther]

I mean getting polyphonc overlap with two voices in the specific order you want them to be.

[blazing zodiac]

Oh, overlapping voices, I can do it, already did it

The problem comes when that "background" voice have to sing the next phrase, instead of waiting more steps in compass it cuts it out

[wise moss]

Thanks apppreictate the tip 😯

[pure panther]

I edited it to specify vocals

[winter wind]

hey people does anyone have any experience with or examples for prompting game background music that doesn't take a lot of attention and stuff like that? I'm only getting "real music", if I try things like soft and quiet for style it just makes jazz that still has the typical song structure and stuff like solos and special sections that pull a lot of attention

[stuck junco]

Hey Litnit, could you point to an example of what you'd like to make? Or maybe use the search function through the platform, type "Game" or something similar, and see what comes up. Seeing what others come up with inspires me all the time. If you happen to hear a song that covers what you're looking for, you could check out the style they used and try it out.

here's an example that could be fun to try if you like the song it made: https://suno.com/s/tN44AhNQx29V0Am0

subtle loopable game background music, low-attention underscore, minimal melodic movement, no big solos, no dramatic drops, no verse chorus structure, no featured lead, repetitive and supportive, soft synthetic textures, gentle 16-bit inspired tones, unobtrusive, seamless ambient gameplay loop

[frozen dagger]

prompt for the era of gaming music. but if you have some game examples in mind I like just typing the game name into the Suno Style prompt and using the magic wand. It doesn't always hit the mark, but it's great for coming up with ideas for which words to use. A chatbot good for that too.
If you're looking for retro gaming music, then chiptune and stuff going to pop up a lot. Using that magic wand and chat bots can also help find words to hit the mark.
Another fun way is to use Suno's "Sounds" to create some sound samples with prompts too like "retro gaming music" (be more detailed when know what you're looking for specifically. Use that sound in the Samples feature to turn it into something fun and useable.

[winter wind]

just whatever any prompt that can help me make it more like background music

[winter wind]

that's really nice thx

[stuck junco]

Happy to help! Hope you have a great day!

[blazing zodiac]

BTW, is there anywhere to see what instructions SUNO knows regarding lyrics?

[blazing zodiac]

Do you mean on something like this?
https://suno.com/s/ZmVOrLW5u1j6zcM6?time=226

[pure panther]

It's a 🦄 😂 Very nice

[blazing zodiac]

lol, this is my first song ever with Suno, also without using Studio

But I have to say for that song, that was really special for me, I dedicated A LOT of work (almost 60 hours in total) tricking all aspects of iSuno to get the song as I wanted.

[pure panther]

Some of my best ones were in my first few days too, I had no vision so I let Suno do whatever. It gets a lot harder once you have a specific "vision" for the song.

[blazing zodiac]

Yeah, it was impossible for me to recreate that. I was trying to do something similar on some background voices for my second one, but it cuts out the voice

[rough depot]

okay I would like to MAYBE tone down the pitch on some of these, any way? https://suno.com/s/sVQvYq99T9lgLGUZ

[quiet solar]

Unicorns happen. Usually it requires tactical editing or steering with your own vocals, but if you can establish a clear chorus-like pattern which the backing singer leads on, then on occasion Suno allows the lead and backing vocalists to go their own separate ways:

https://suno.com/s/7GCb4PQY3K7bFhpY?time=189

[pure panther]

That's pretty cool

[quiet solar]

Not aware of any way to directly prompt for that behaviour, though. Suno doesn't really follow instructions at that level of detail. xD

[pure panther]

I was able to consistently get it to slip into that using the prompt "polyphonic vocals"
The final song I went with only had a little of it, but I was getting it pretty consistently.
https://suno.com/s/9m01NkmNnMOaDhsx?time=167

@quiet solar

[quiet solar]

A bit of divine intervention never hurts!

I'd agree that "polyphonic vocals" and "layered vocals" are good starting points for experimenting with something like this, but I've not found anything that produces the effect consistently enough to say that (x) is the way you can do it.

[trail bone]

Maiow?

[wintry jacinth]

Hey guys

When I type in the Exclude styles box, it creates the Style block like “-Synthwave” could I just write it like that in the Style Prompt

[stuck walrus]

guys, if I sub suno today and can I replicate the voice that I created 1 years ago using v3.5?

[wintry jacinth]

You could always download the song worse case but you should be able to make it into a persona

[stuck walrus]

yes that what i mean, the persona... I want to copy the persona from that song

[wintry jacinth]

You should have no problem with that

[quiet solar]

I'd advise against it, since the style box doesn't handle negative or conditional prompting well. -synthwave may well be read as "synthwave", which is the direct opposite of what you were hoping to achieve.

[wintry jacinth]

Thank you, I had a feeling but I am running low on credits so I am trying to be a surgeon with the prompts

[polar spade]

No. It wont read " -Synthwave" the same way to exclude it. Thus, the exclude box was born.

It was maybe a refinement needed when users started asking more than the "perfect" song, but one more structured to the individual, whereas negative prompting was more a baked in design for artwork AI from the beginning. Each company has their own design.

A negative prompt in the style prompts will still ignore the negative and include the style or ignore the style altogether, from last Ive seen.

[wintry jacinth]

Heard. Thank you

[quiet solar]

Yes, although worth noting pre-v5 personas don't always translate across to v5 neatly, because the vocal model is very different.

So you might have to play around with songs in v4.5 or other interim models before you get to the point where you have a v5 persona that sounds like your v3.5 persona which you can apply to future v5 songs.

[sweet ivy]

Seems like there is something weird every day. I took the last and tag, of the piano track and hit replace. The replacement was actually playing the melody of the song, over the chord foundation. The part I replaced did not do that ? Did it think it was now an instrumental of the song, rather than a keyboard part

last verse and the tag

Maybe, I took too big section?

[blazing zodiac]

Looking my today iterations. I seen I recreated it doing my second song with a collaboration with my first song Persona (Ithayi) so I got a real "Ren (my second Persona of this song) feat. Ithayi:
https://suno.com/s/YEs5ALIoBJvDO8oV?time=100

I'm a unicorn maker!

[queen linden]

trying a Mashup of two songs in Latin ,,,,,, I should be so in bed

welp, we wont be doing that again LOL liv e learn

[fading hollow]

Has anyone been able to separate saxophones from Dark/Doom Jazz? Strictly specifying it in both the prompt and the negative doesn't seem to have much effect

Nevermind, I got around it by avoiding any use of the word "jazz", and describing it as a cinematic/dark ambient thing with just the instruments I wanted

[dapper parrot]

I have a problem with making Suno sing in Hazaragi and Dari. It is spoken in Afghanistan but is a dialect of Persian. Suno is heavily trained in Persian. It has a thick Persian accent. Does anyone have any ideas ?

[south fog]

Is there a "correct way" to style prompting in V5? Some people do short words (e.g.: melodic metalcore, powerful guitarchords, atmospheric) and some to whole sentances (e.g. Atmospheric melodic metalcore track opening with airy synth pads behind haunting clean guitar melodies. Halftime drums anchor the verses, building tension with syncopated rhythms. Choruses erupt with layered, powerful guitar chords and soaring lead lines, maintaining an epic, lush texture throughout.)

Also is it correct that V5 ignores negative prompts in the more options tab? i generated a song with the style "atmospheric melodic metalcore, halftime drums, powerful guitarchords" and the negative prompts "double kick, growling, screaming". It doesnt do growling and screaming but it does 90% of the song double kick rhythms...

[polar spade]

"double kick" might not mean anything. Though it probably works in prompts but that may be since its often paired with genre that utilize it anyway. Try putting in EXCLUDE BOX those genres which double-kicks are often used: death metal, thrash, black metal, power metal, prog rock, hardcore. By doing that, Suno recognized a common style these all have -- double kick percussion. And while at it, might help excluding power chords, which you seem to want excluded.

[south fog]

Ah makes sense thanks. Oh and btw, if you remember our previous conversation i figured out that the memory thing may actually be just a bug. Because no matter what i do it doesnt keep the style i previously generated when entering a completely different style - which is good!

[polar spade]

I hardly remember what I had for breakfast. So if it worked for you. Huzzah!! 🎉 That's what matters.

[queen loom]

when requesting an extend and prompting details it's best to allow about a 3 second window? or set it to exactly where you want it to extend? it always seems to repeat the first few seconds twice otherwise

[frozen dagger]

the first half of prompt style was more for v4 (still applicable in v5), but v5 understands the language better so using natural sentences became the move. it's context, whatever works.

[quiet solar]

v5 can respond absolutely fine to both approaches.

Some people feel the natural language approach is more effective at grouping things together, e.g. specifying something you want in the chorus but not necessarily as a consistent factor across the whole song, which requires a bit more formal structuring if you do it by individual words. But ultimately both approaches boil down what you've put in to keywords and then look them up as influences on what Suno creates.

[strong tangle]

What is Max mode?

Please someone explain this to me.

[quiet solar]

As far as I'm aware, it's something that Suno toyed around with as a possible idea (assign more credits, get more computing time on generations). It was never released publicly as far as I know.

However, it prompted various "...did you know there's this one secret trick that Suno doesn't want you to know!" type voodoo prompts as to how it could allegedly be secretly activated and do amazing things. And through the power of randomness, some people found they got a really cool generation when they did it, and so the voodoo spread.

Personally, I always turn around three times and throw salt over my shoulder before clicking Create. I've found that to be a surefire way to get awesome sounding tracks out of Suno every single time.

[strong tangle]

As far as ai goes it needs specificity across pretty much every model and something like this that has no bearing on production quality or is attributed to any aspect of vocal or instrumentation seems like a really far fetched idea. And having a meta directive in a style description alone makes no sense. Since it's not geared toward understanding that as far as the documentation goes. If it does that'd be nice.

[quiet solar]

Indeed.

However, I can see the comforting attraction of the idea that there's a magic switch to flip that would make Suno always create amazing things, which doesn't involve any actual thinking or work on the user's behalf. 😉

[main hazel]

if i created a similar voice to an artist but it still sounds different than the original artist, i still can use that hybrid voice in my songs?

[hard beacon]

how did you create the similar voice? with Suno?

[main hazel]

inspiration tab

[hard beacon]

ok.. but the tracks in the playlist (for inspo) where made with suno.. right?

btw: don't know if you are able to create a playlist with uploads instead of suno outputs 😉

that's why i am asking that question

[main hazel]

yeah

[hard beacon]

ok.. so you've uploaded non-suno-created songs and you want to use the "original voice" ?

[main hazel]

no, not the original

re prompted

[quiet solar]

Are you the artist behind the non-suno created song?

[hard beacon]

ok.. i am not sure if i understand it right, so maybe the following message can be missunderstood 🙂 Take care about uploads from existing songs (Copyright). The Inspo function is more about sound. If you wanna use a good suno output voice you can create a persona out of it

[main hazel]

yeah after the output - persona

is what i mean

[hard beacon]

so... not inspo.. we are talking about persona.. right?

[main hazel]

yeah

[hard beacon]

be right back... Toilet-Time 😄

[hybrid copper]

I have a theory that Max Mode DOES work, but not exactly for the reasons we think it does. It forces you to write your prompt in a way that appeals to a machine rather than a human. Sort of like telling an art generator to draw a beautiful masterpiece realistic cinematic photogenic etc.

[hybrid copper]

It also seems to depend on the song you're going for. Seems to work well on my synthwave-fusion and rock songs, does more harm than good on bluegrass/old-time/folk.

[quiet solar]

@main hazel Can I just check with you - the songs you used behind the Inspo. These were non-Suno songs, uploaded into Suno and used as Inspo to (amongst other things) influence the vocalist.

Are you the artist behind those non-Suno songs, or in possession of the rights to create derivative works from them?

[hard beacon]

Ok @main hazel : we are talking about persona.. so what is your exact question?

[main hazel]

no, a friend perfromed and sent it to me

[main hazel]

yeah

[quiet solar]

If it resembles your friend's voice enough that it's recognisable, then definitely get their permission to use the Persona in your work going forward. That's the simplest way to avoid any future awkwardness.

[main hazel]

coz i just told a friend even if my voice sounded 70% similar to urs we still wouldnt have the same one

[ruby thunder]

is anyone elses personas not working, like I'll upload my own music I made then put it on suno and for some reason its all greyed out and I cant use it

[main hazel]

yeah u cant

[ruby thunder]

brotherr

[quiet solar]

Yes, but in copyright/intellectual property terms, you've used Suno to create a derivative work of their original composition. The Persona would continue to be a derivative work going forwards when you use it.

That's not a problem at all if they've given you permission to do that. But if they haven't, you're skating into potentially difficult territory. Quite honestly, even if the Persona only sounds 40% like their voice, I'd still suggest checking that they're happy with what you've done and how you plan to use it.

If they're a friend, there's literally no reason to be secretive about this unless you know yourself that you're doing something you shouldn't.

[urban lance]

Hey all I've got a 4/4 track that has a 3 beat pick-up bar before bar 1 in Suno Studio. When I try to import this track into my DAW that 3 beat pickup should be bar 0 from a compositional perspective, otherwise the logic of all my SD3 midi drum tracks are offset. Any advice?

[wintry jacinth]

Guys I am burning through credits I have no clue what i am doing wrong
I am feeding in an Audio Seed that is a simple pattern of high cut piano playing 8th notes as the 'leading' rythym. But its ambient and open. Then piano comes, in still open. It should do that for atleast 16 measures, but the generation is instantly adding in drums. Appreciate any help

[strong tangle]

Is this in suno studio?

[wintry jacinth]

I do not have studio. I create the rough sketch in Ableton. With minmal voicing. Grand Piano, Ambient Foley, and Bass. I have tried adding in just a kick from the drums later in the track to see if I can make it act as a 'placeholder' for when drums should come in but nothing works

This is my style prompt: dark cinematic score, neoclassical grand piano warm analog synths, overdriven analog bass prophet5 brass gritty analog saturation ‑Pop, ‑House, ‑Synthwave, ‑EDM, ‑Four-On-The-Floor, ‑Drums, ‑Hi-Hat

[strong tangle]

There are ways you can use just the piano as the overall influence and dictate when things enter with meta tags. Suno knows bars with instrumental stuff mostly but if I saw your entire prompt I could try and provide some assistance. I do this for my stuff a lot and it helps navigate it to what I want.

Or do this with an llm

[Task: analyze this suno custom window prompt (input 1) and style description (input 2). Revise the prompt to ensure the piano plays 16 measures without drums]

[Sub-task: explain what was causing the prompt to drift from the desired intent and the reason for the revisions]

[Input 1: <paste prompt>]

[Input 2: <paste style description>]

[wintry jacinth]

What do you mean by custom window prompt vs style prompt

[strong tangle]

The custom window prompt is where you put the lyrics with all of them meta tags. It functions like a prompt to me at least. The style prompt gives it more concrete direction.

[wintry jacinth]

I struggle with consistency for purely instrumental tracks especially when I feed it audio seeds. You do this for instrumental only tracks too?

But I do have an LLM assistant to help me prompting. I have tried many differnt angles it still wants to add in the basic drums even for a track that doesnt have any, and I put it in negative style?

[small bridge]

Is there a way through prompts to make sure a song doesn't end up too long, like a way to put in a prefered song length? has anyone played with that? I keep getting like 6m+ songs my lyrics aren't super long, I wanna do like around 3:20-3:50 max

[rough depot]

How do i stop it from 1) singing the spoken word elements, and 2) singing SO DAMN SLOW? https://suno.com/s/J5LpXa8VgIDYk6Iq

[lone nova]

you define the bar length of each section, include the intro and the ending.

[rough depot]

wait you can define bar length? how?

IT. KEEPS. DOING. THIS. https://suno.com/s/7HVoeSe7x0sgq5d4 These lines are WAY too long for singing this goddamn slow

[wintry jacinth]

Try wrapping it in paranthesis

[lone nova]

use bracket :
[INTRO : 2 bars]
[VERSE 1 : 4 bars ]

[rough depot]

[Verse 1 | 4 bars]
The blades of the killers are gleaming and cold
The shadows descend on the kingdom of old
The fourth princess is cornered, her magic is spent
The murderous stalkers are fully intent
She falls to the ground, and she prays for her life
As the assassins prepare for the strike of the knife
But suddenly thunder rips open the sky
A rift in the heavens makes the enemy cry
Like thus?

[lone nova]

yes or even [00:30 Verse]

also for intro and outro write some thing about how it should end,
you can also ask the AI lyric assistant to do it for you, add those tags ( orange big magic stick button)

[lone nova]

maybe 4 bars is not enough for this amount of text...

[quiet solar]

It's very difficult to get Suno to completely avoid drums, since percussion is so universally common it tends to get picked up in the reference material almost whatever your prompts are, and adding it to the negative styles isn't enough to outweigh that.

You may be better off trying things like adding 'acoustic' to your style prompt, although it's by no means a magic bullet. Often the easiest way is to extract the entire stem set and remove the drums and percussion tracks afterwards.

[rough depot]

I was trying what was suggested. how many bars should it be then? It was REALLY stretching words out DESPITE the length of the sentences being so long. it was helll bent on making the song slow

[lone nova]

more like 8 bars,
or just do like this:
[Verse 1: Fast and Driving]
The blades of the killers are gleaming and cold
The shadows descend on the kingdom of old
The fourth princess is cornered, her magic is spent
The murderous stalkers are fully intent
She falls to the ground, and she prays for her life
As the assassins prepare for the strike of the knife
But suddenly thunder rips open the sky
A rift in the heavens makes the enemy cry

[wintry jacinth]

I don't have Studio, and I can't justify spending more money when its so inconsistent like this. Its not that i don't wany any drums in the track, I just want a more ambient Intro, like the audio seed that I am covering has. Which is common for cinematic scores. Think Hans Zimmer

[rough depot]

need to edit it heavily too because gemini loves to make chorus too many lines long.

Wrote this chorus wholesale

[Chorus | 8 bars]
This is the legend of Red Lightning!
A Band of heroes with a power so frightening!
The outlander comes with an alarming speed,
To be a hero for a world in need!

[quiet solar]

You don't need Studio to extract stems, that's possible on Pro. You just need your own DAW or audio editor to stitch the stems back together afterwards. Ableton should be perfectly capable of removing the entire stem, or letting you tactically remove certain pieces.

If you don't want to approach it by editing and want to do it purely by prompting, be aware this is known to be difficult to achieve. However, in addition to adding 'acoustic' to style, you could try something like tagging sections as 'soft', e.g. [Intro - soft instruments] which can help weight towards no percussion or percussion light sections, although that will depend on genre.

"‑Pop, ‑House, ‑Synthwave, ‑EDM, ‑Four-On-The-Floor, ‑Drums, ‑Hi-Hat"

Just to check, I'm assuming that part of your prompt is taken from Exclude Styles? If you've actually got that in the main style prompt, remove it - Suno doesn't handle negative prompting well in the Style box.

[rough depot]

see i already have the "Ending" song. Now I am working on the "Opening"

[lone nova]

you can also tell gemini to give you 8 bar, short version or what ever.

[rough depot]

I asked for 10-10-10-10 syllables

of course it still gave me more than 4 lines a verse

aaand it keeps hallucinating and leaniing into cliches - I'm surprised the line "Be face to face" hasn't cropped up yet

[quiet solar]

Neon and Static are only ever one click away. xD

[rough depot]

what are those?

[quiet solar]

Very common words that AI likes to include when generating lyrics

[wintry jacinth]

Appreciate the help.

I misunderstood what you meant, yea I did try the extract stems before but its far from perfect and credits are tight. What do you mean by 'stitch' together though. I have 0 problem with the editing side of it, I always edit in Ableton afterwards anyways for mixing.

"Just to check, I'm assuming that part of your prompt is taken from Exclude Styles"

yes its not in the main Style box.

I have also tried to add in the lyrics box
[Ambient]
[No Drums]
[No Percussion]

That didn't help.

[rough depot]

https://suno.com/s/EDzXks0Np7b9fEAs nope that didnt work

[wintry jacinth]

I also raised the audio seed (the track has no drums) to 95%, do you know what its doing there?

[quiet solar]

Yeah, negative or conditional prompting very rarely works. It's more likely Suno will skip over the 'no' part and read "drums" instead.

By stitching together, all I mean is that if you extract full stems you'll have multiple individual MP3s or WAVs and you'll need a tool to put them back together again for editing and exporting. In your case, Ableton.

Audio influence determines how closely the song will try to stick to the sound it hears in the audio reference you've given it. However, that doesn't mean it won't improvise new things - the style prompt also has weighting, as does inherent weirdness.

[wintry jacinth]

mmm. yeah i don't know if its my specific style of music but it has a lot of synth, reverb, and ambience. Which i think causes the stem spltting to break or atleast drastically alter the actual individual channels. I have tried twice (100 credits :/) and it chops up the actual melodies played or just crazy artifacts show up

It would be cool if it could split as midi.

[lone nova]

too long, cut some text for real, no metal song with this amount of text can be short.
spoken and outro...

[rough depot]

cut horizontal, or cut vertical? be specific

The length of the song doesnt bother me, its the fact that despite the LINES being long, Suno insists on dragging out multisyllable words at a snail's pace

It was doing to longer words what most Super Bowl singers do to "O'er the land of the free".

Its like why on gods green earth would you drag out long words like that?

[quiet solar]

Bear in mind that Suno is trying to strike a balance between the BPM of the song and the syllable count when it spreads the lyrics out over bars.

The BPM of the song is a very fast c.200 or so and currently, each is line is fitting to 4 bars. If that was to be condensed into two bars, your vocalist would be going like a machinegun.

I'd suggest either increasing the syllable count per line so that the individual words are dragged out less, or alternative dropping the BPM so that having one line fit to two bars is a more likely anchor point for Suno to reach for.

Symphonic metal quite often has vocals running at half-time compared to a frenetic instrumental track, so I didn't consider the balance of this song unusual - even if it's not what you were looking for.

[rough depot]

How do i set the bpm?

how do i make it more "dire"? as in, a sense of urgency. This version of the first verse is too....happy? https://suno.com/s/tEyZ6a4nLtxE94w6

[quiet solar]

Yeah, if you're looking to stem split, you'd be better off prompting for minimal production effects and as few instruments as possible so that the splitting process is as clean as you can get it, then putting the reverb and other effects in yourself in Ableton.

Again, more tools available in Studio that help with that, but appreciate we have to work with what we have!

[wintry jacinth]

I feel you. I am grateful for this tool, and I know its only going to get better from here. Just sometimes feels like gambling with little control on the outcome. AT TIMES.

[quiet solar]

Been on the journey since v3 - I know exactly what you mean. It's a fantastic tool, but at times it's like working with the most stubborn session musician known to mankind. xD

[wintry jacinth]

Exactly... lol and sometimes I try something random to change it up, and it nails it

Do you mind listening to the audio seed to see if something about it would cause the drums?

[quiet solar]

Add it to the style prompt (e.g. 160bpm) and v5 will generally respect it as long as it's within the bounds of what's normal for the genre.

For vocal tone, you need to prompt for that separately. Try adding something like "Angry, rebellious male vocalist" to the style prompt.

In terms of musical tone, that's a bit less straightforward. I hear what you mean in that it's more...epic and anthemic than you were going for. I'm guessing you're using an LLM behind this, so maybe ask it for suggestions for style prompts for a darker, more threatning atmosphere.

[quiet solar]

Sure, share a link

[rough depot]

I was feeding it my manuscript and asking it to write a symphonic power metal song about the opening of the book and how its the start of a new legend.

I've already had to edit it manually becuase i didnt like the results.

[quiet solar]

Yeah, Suno will need more of a steer than that if you want a particular tone and feel to the song itself. Symphonic power metal is just too broad.

[rough depot]

(For one thing it hallucinated a car crash...ffs my character DIDNT DIE!)

Hmm I'd say more like Dawn of Victory than Emerald Sword if that makes sense,

Gemini absolutely COOKED with my Valyrian song though. And Suno did it justice.

Goes hard. Very hard. https://suno.com/s/Y6l3R8BpwCViooLq

[quiet solar]

"There are no drums or bass"

Take that line out, it's doing you no favours. Suno doesn't handle negative prompts well, so when it boils it down to keywords it'll more likely read "drums" and "bass". Other than that, there's nothing wrong with the prompt - I don't know why it's reaching so enthusiastically for the synths, but I don't know what the original audio prompt that it's referencing sounded like.

[lone nova]

i started with 16 bar from gemini, and
use suno editor to add other instructions.
https://suno.com/s/HjOHxH10ihBPWbet
only define voice manually:

result in 3min 30 sec song.
It is a hit and miss sometimes

[wintry jacinth]

This is the original audio I uploaed from Ableton

Thats the auto generated summary

[quiet solar]

Ohh, right. Gotcha.

[rough depot]

This needs work hmm

[Pre-Chorus 1]
But suddenly rift forms in the starry sky,
A chariot of steel falls and assassins all die,
Four crushed beneath it, two trampled in mud
The chariot stopping in dirt and in blood

Ok happy with the first 2 lines...gotta fix them last two.

[quiet solar]

Yeah, there's nothing in there that sounds like it would be suggesting drums from an audio perspective. There's a very clear rhythm to it, but it's clearly musical notes rather than a percussion effect.

[polar spade]

I m just dropping by. Is this a new "need ideas for lyrics" post or something to do with anything posted above it?

[rough depot]

its the same song

[polar spade]

Oh OK. I have not read anything above scroll. So ill just go sit on my stool in the corner and lurk. 🙂

[wintry jacinth]

Thats what im sayiiinng. It's all good. Ill move on to another track for now. I was worried it was the E-Piano 8th note foley, but I want that effect in there.

appreciate you brother

got a sub from me 😄

[lone nova]

Why do u have 2 intro? 1 spoken and 1 instrumental? 😄 and outro.
You are lucky it is only 5:40
12 sections bro

[quiet solar]

@rough depot Is what you're going for something with a more obviously driving and pacy rhythm in the lyrics themselves?

If so, and you're open to reworking them, maybe try prompting your LLM to produce something in Anapastic Meter and see if that helps.

[rough depot]

i can try it yeah

[quiet solar]

Happy to help. Well, try to help at least!

[wintry jacinth]

You helped my sanity by saying its not just me

[rough depot]

just tried that, it butchered the name REd Lightning. ugh

"Lightning so red" Christ almighty, ITS A PROPER NAME!

So how many syllables SHOULD my chorus be for best results? and lines for that matter

[Chorus]
Hear you now the legend of our heroes, Red Lightning!
They strike hard and fast at a speed so frightening!
Of the outlander and his red Cadillac,
Fighting evil and never turning back!

[quiet solar]

I'd be stunned if an LLM produced something lyrically flawless right off the bat.

The aim was to see if having a more deliberately structured meter pattern as a starting point- especially one known for having a sence of pace and momentum - would help Suno translate that across musically when you create something. It might not work, or might not work the way you want, but I thought it might be worth a try to see if it helps you get closer to something stable to iterate from, as I feel a little bad watching you doing major rewrites on the lyrics over and over again.

[rough depot]

I'll try ana-whatever on the next track, just use it right off the bat.

Er, what does "Style Influence" Actually do?

[polar spade]

🎶
Suno glitchin on the verses (It wasn't me)
Suno bangin' like its cursin' (It wasn't me)
I even heard it speed the chorus (It wasn't me)
Neon lights again to bore us (It wasn't me)
Suno clicked and she sputtered (It wasn't me)
To much gibberish she uttered (It wasn't me)
Heard the screams gettin' louder (It wasn't me)
Suno, girl, take a powder! 🎶

[lone nova]

increase the style influence to 80 to 90% also helps. otherwise LLM ignore stuff your instruction

[polar spade]

Sorry. Shaggy song came to mind. I just had to sing it.

[wintry jacinth]

Masterpiece.

[lone nova]

also you can use the pro editor to cut away stuff you do not like and regenerate to connect parts.
reduce a 6 min song to 3min

[quiet solar]

I'm afraid that's a bit like asking which musical notes should the chorus include for best results. There isn't a single right or wrong answer.

Generally speaking, the more syllables per line the faster the vocal delivery will be. But too many syllables for the available time to deliver the line, and it'll steer towards doing half-time lyrics. BPM dictates the amount of 'time' available - the lower the BPM the more time available, but the less pacy the overall song.

Suno tends to play most nicely with verses and choruses being 4 or 8 lines long, but more often than not it'll handle 6 lines fairly well too. The key is not changing that between the same sections (e.g. don't have one verse be 4 lines long, and the next be 6).

[rough depot]

Man i wish suno could do weird al style parody. i been wanting to do "despacito but its all california city names"

Ojaiii.
FRESNO!
Duarte!
Oh
OH-Oceano
Si-mi Valley Solano Beach San Jose~

[Something[ San-
Ramon [something] San-
-ta Rosa, Clara, [something] Indio…
Mendocino

[quiet solar]

Style influence means Suno will put a greater weight on what you've specifically asked for in the style prompt when generating a song. That might sound like a universally good thing - why wouldn't you want Suno to listen to exactly what you prompted. However, it will lock Suno in more closely to exactly what you've asked for and may limit it's creative freedom to pursue something that 'sounds good' according to its underlying training.

If you're going with very simple, broad prompts as you have been so far then I'd leave the Style Influence at default.

[rough depot]

is it expecting 4 per verse or summat?

[wintry jacinth]

yah I learned this the hard way when i tried making 'Synthwave' music, but without the cliche Gated tom fills. Impossible if using that tag

[quiet solar]

No, it'll work with whatever you give it. But 4 and 8 are a very common format for 4/4 songs, so it'll find it easier to match it up to more things in its training library when looking for inspiration.

[rough depot]

Whats weirdness do then?

[quiet solar]

Weirdness is a catch-all for how strictly Suno should stick to what looks 'normal' in the genre references vs deliberately breaking them in the pursuit of something more creative and unexpected.

In principle, a Weirdness 0% song should be a very generic and predictable song from smack in the mainstream of that genre. A weirdness 100% song will often pay no attention to your verse and chorus structure because, hey, prioritise weirdness.

Some weirdness is good, because Suno uses that creative space for adapting what you give it. I'd leave it on default unless there's something very specific you're looking to achieve. And if you do move it, stay out of the red zones of under 15% or over 85%.

[rough depot]

I'll save chaos for...hmmm the ballroom scene song

I had to.

An outworlder and his friends in a crimson Cadillac,
Fighting forces of darkness, and cutting them no slack!

Better, THIS vocals is what i am after https://suno.com/s/Tq61IqNtfgNniVMu

[dim zinc]

Spricht hier jemand Deutsch?

[rough depot]

Unfortunately, its still struggling with the first 2 lines of my chorus

Hear you now the legend of our heroes, Red Lightning!
They strike hard and fast at a speed so frightening!

how can i make it less clonky?

[quiet solar]

What do you feel the problem is in the first two lines compared to the last two?

They're all being delivered at the same pace with a scattering of sustained notes.

[rough depot]

are they though?

[quiet solar]

Yes, musically speaking. 4 bars for each line.

[rough depot]

despite being the same syllables it adds uneccesary pauses feels inconsistent

actually more like....its the fact they screamingthe second line

[lone nova]

I managed to use local AI (seedVC) to add an elder AI voice(50Year) into suno, a rap song.
what do you guys think of the result? what can be improved?
https://suno.com/song/22d8f135-b738-47fe-ad47-76138b8ea536

[rough depot]

hmm i want to use a wizardly narration personally

[quiet solar]

If you're asking why it doesn't flow as neatly as, for example, the pre-chorus, then that's down to syllable count and structure.

The first two Pre-Chorus lines split neatly into lines that are 5 syllables (pause) 6 syllables, then the last two are 6 syllables (pause) 5 syllables. Really nice mirrored structure.

The Chorus is comparatively very different - there's no clear meter pattern between the lines, and the syllable lengths are 13-11-14-13.

[rough depot]

i figured it was something like that. so i gotta fix that

add blazing to line 2?

They strike hard and fast at a blazing speed so frightening?

oh ok so 5(pause 6) and 6(pause 5) is something i should shoot for?

[quiet solar]

It's worth remembering the pauses aren't unnecessary - Suno is fitting each line to the same amount of time. But if there's no obvious place to put the pause, it'll sustain a note or insert a pause where it thinks makes sense. And if there's no clear meter pattern then expect it to wander around a bit.

Take pre-chorus 2 as an example of a clear structure:

The princess arises, the danger has passed
(da-dum-da, da-dum-da / da-dum-da, da-dum)

The stranger is awestruck, the die has been cast
(da-dum-da, da-dum-da / da-dum-da, da-dum)

It's using a form of Anapastic Meter, it's clear where the stressing is on different words, and the structure is consistent from one line to the next.

[polar spade]

Some words can be removed as the subjects and action are inferred to be the same folks.

Hear now the legend. Our heroes, Red Lightning!
Strike hard 'n fast. Speeds so frightening!

Depending, of course, that may sound more chunky to you. At least its an option to give Suno more breathing room to sing it.

Is Red Lightning more than one hero? you have heroes as plural?

[quiet solar]

This isn't to say you should aim for that specifically, but that the structure is clear and rhythmical on the ear.

[rough depot]

seems like some of this can be fixed with just a comma, too

I managed a 6/6 split on a 12

An outlander and friends, in a red cadillac,
Take the fight to evil, and cutting it no slack!

[rough depot]

Its a "Unit". An adventuring party.

So its multiple heroes, as a collective.

[polar spade]

OK. just ckecking.

[rough depot]

Victor Kirkland, the man from another world. Together with Princess Illiana. Soon joined by Malcolm Kavian, and Illiana's sister Sylfaena.

[quiet solar]

If it helps, an Anapastic meter version of the chorus would look something like this:

Red Lightning! The heroes we sing of today,
Strike hard with a speed that will blow you away!
A Cadillac crimson, the outworlder steers,
Defeating the darkness and conquering fears!

[rough depot]

ok so whats the BEST way to add a female vocal spoken word part to the beginning?

I think i should just make a new song thats all spoken word and thread it together in Audacity with how often it just straight SINGS the spoken intro

[quiet solar]

Bear in mind if you're trying to introduce a sense of rhythm to the lyrics, it's not just about the syllable count but about how the words flow from one line to the next.

You can try...

[Spoken word - Female]
Your lyrics here

...as a starting point

Spoken word sections are a little bit more hit and miss to prompt for. And introducing male and female voices together will also tend to increase the number of attempts you'll need to make.

It can help to add something to the style prompt, but you'd probably want it to be descriptive - so something like: "The song begins with a spoken introduction by a sassy, youthful female". Though obviously replace the description with what you're actually after.

[rough depot]

there a way to prevent it from pausing too long between verse lines?

[quiet solar]

Not prevent as such, no. The main way to avoid it is not having a line that's "too short".

Your opening verse sets the tone and pace for the ones that follow. Say for example that it's all 12 syllable lines. If your next verse is mostly 12s, but has an 11 or a 13, then Suno will probably be able to condense or stretch without it being too obvious. If you include an 8, all bets are off - you're going to get a gap.

[rough depot]

gah

I hate this couplet Gemini gave me. I need to take a shower and think of something better

A magical gift so the two can converse
The legend begins in the heart of the curse

So begins a friendship, spanning worlds apart,
Hie they to Anaura, 5 syls of somethin

something about the engine start?

oh wait

With earnest gratitude, she thanks him from the heart,
So begins a friendship, spanning worlds apart

[quiet solar]

That first new line will probably break the flow. The implied stressing from the two lines before would have it sung as:

so BEgins a FRIENDship, spaNING worlds aPART

Might be worth considering something with a cleaner meter flow, like:

"And so starts a friendship, that spans across worlds"
And SO starts a FRIENDship, that SPANS across WORLDS

[rough depot]

ok but what rhymes with worlds

[quiet solar]

It was an example rather than a suggestion. Give us a sec, lemme think what you could do with 'apart'

[rough depot]

thanks. meanwhile i am gonna gen it see how it stresses

[quiet solar]

Maybe something like:

She holds out her hand and thanks him from the heart
A new friendship formed from two worlds far apart

Not 100% sure about it. Apart wasn't the easiest rhyme to work with naturally either, it turns out. xD

[rough depot]

I'll try it

closer. https://suno.com/s/yn7hZR2kE1IBGA6h

[pallid condor]

I need help.

I used to write stuff all the time, but now I keep having issue. I written and generated a lot of metal songs, but now when I do it... and for the last few months... EVERY TIME... I have a great powerful opening, it gets to the lyrics and all the music stops, gets real gentle an calm, and does like ballad clean vocal singing. It's almost comical at this point.

I feel I've tried everything... I'm thinking SUNO v5 isn't good at metal.

[neat nexus]

How are you formatting your lyrics?

[pallid condor]

Like [VERSE] [BRIDGE] [CHORUS]

[neat nexus]

Any other metatags?

[pallid condor]

No... just [PRE-CHORUS] [SOLO]. etc..

stuff like that

[neat nexus]

If you don't mind sharing what is your style prompt like?

[pallid condor]

Generally pretty simple... but I've gotten very detailed. Like: Technical Metal, 80s Thrash

Like this was super slow at the lyrics and just weird: The song launches with rapid-fire thrash riffs and double-kick drumming, layered by sharp, staccato palm-mutes and soaring lead harmonies. Verses drive with tight groove metal chugs, transitioning to a classic heavy metal-inspired melodic pre-chorus. Each chorus explodes with gang vocals, high-octane shredding, and face-melting solos. The bridge features syncopated breakdowns and intricate time changes before returning to a galloping finish, blending relentless energy with complex precision.

[neat nexus]

OK...let me tinker a bit...

Oh wait... what are your slider positions?

[pallid condor]

50

50

default

I even have tried exclude styles: clean vocals... didn't listen. 🙂

I'll try a style at 100 just to see.

[neat nexus]

Feel free to play with it but I was going to recommend a default if you had them set to extremes.

[pallid condor]

Ah, yeah, didn't really change much. Still doing it.

Tinker a bit if you don't mind. Thank you.

[neat nexus]

Testing with the prompt and a few seconds in ReMi:

https://suno.com/s/qj0qxfickCnSgVX2
https://suno.com/s/1Ye0w7saILb7PSF5

Have you tried without the Personas?

[rough depot]

We have a winner https://suno.com/s/QMFiJAtWLzeoODXd

[rough depot]

christ. why doesnt it obey male vocal female vocal

https://suno.com/s/CIk1S06yfLj01pcY

am i doing something wrong?

[median night]

Nah it's just duets are a struggle with Suno

[rough depot]

gah'

the lyrics came out real good too. captured my main couple perfectly.

[median night]

Could try.. doing all male. Then all female. And getting them as close as possible. Them stemming. And putting them into 1 song where you need em

Timings are a bit tedious tho

[rough depot]

what is stemming?

[median night]

Are you free or pro?

[rough depot]

pro.

no clue what stemming is though

[median night]

I believe pro can stem

The get stem/midi button

Then download those. And since you're pro. You'll have to use audacity or a daw

Since you don't have access to studio

It's a lot of work. But if your really into the song then it might be worth it

Unless you're more into just genning the music. Then only advice I can give is try like ..
[Style: Male / female duet]
At the top

[rough depot]

but what IS it ? ugh

[median night]

Stems are the different tracks on a song

[pure panther]

It's an option you can choose on the song, the three dot menu

It gives you separate audio tracks of instruments and vocals (vocals, backing vocals, woodwind guitar drums bass...)

[rough depot]

grr OBEY. THE. TAGS!

[polar spade]

In music production, we record each instrument, each vocal, all on separate tracks.

So those tracks are layered in the charted/graphed/timeline area called the sequencer of your Diggital Audio Workstation --, the computer visual version of what was just a 16 track ribbon of tape in the past.

To remove those tracks though is taking each away from the whole song. When doing that the tracks are now called stems. Technically, most stems are groupings of their related parts, so drum stems might contain a few tracks of all drums, guitar stems of all guitars and so on. So in the end, a stem can be one instrument or grouped instruments.

BTW: to extract them out you might here someone say they've "bounced" the stems from the master .

The term is a misnomer for AI songs as the original tracks never existed, but the idea becomes the same. Suno creates tracks by reverse-engineering of what would have been the individual instruments and vocals. So we borrow from the real world and call it "stemming" or breaking the song out into its stems.

To remove only the lead vocal, then you have just the instrumental PLUS background vocals , which youll often hear called the karaoke track, for obvious reasons , or karaoke stem.

Three individual tracks then that suno makes are called stems. Youll recognize the difference colored wave forms stacked up on screen and each one labels a different instrument. When playing them all at the same time, they should replicate the original song perfectly.

[rough depot]

ic....so how would i like...do the same song twice, once male and once female and have the instruments be the same?

jesus. first verse LITERALLY has [Verse 2: Male]
and it still comes out female

this is what i got https://suno.com/s/YyicJvvxbZfzE0l2

god damn it, perfect instrumental opening....but still woman's voice in the first verse. and switches abrutp

crhst[l

[polar spade]

Well, if we can ask Suno to replace or clone the male vocal track with a female voice instead, that would be enough. Certain outside third-party apps do just that.

This is the theory, not the application.

Suno does not do cloning well. It can replace, but it may not stick on the first replacement attempt.

[rough depot]

it keeps alternating them MID VERSE!!!

Victor calling Illiana "my Eagle" makes zero sense.

for obvious reasons.

It only makes sens the other way around.

[polar spade]

We can often hear a male literally glide into a female voice or female glide into a child's voice. Its often because they start singing in high ranges that triggers a more female voice to be singing. Suno cant stay locked into the male needing to sing falsetto or having a 4 octave range, , for instance, so it just turns into a female.

[rough depot]

i see....so if i put falsetto into the negative section....

whats the girl version of a falsetto?

like where a woman sings low AF

[polar spade]

No. Just to answer that question, we want a male to go into falsetto. Not turn into a female voice.

But to answer the issue, usually its about Suno being random. Thats all.

The rule of thumb is with any place that shows a pause or separation such as quotations or a line space, Suno tends to switch gender from whatever was last, no matter what ytou have noted in the metatag.

[rough depot]

thats stupid,. ok how do i avoid that?

oh my christ how long was weirdness up to 80 and i didnt realize it?

[polar spade]

We work with "stupid" mostly .

Find what looks like the spot what may have triggered suno to switch, even though we had [male], to a female and work out how to tweak that line to make it go male instead.

For instance, throwing in an extra line of

[Female] Mmm

[Male] "But this is so wrong"

Just to have her sing anything for a split second or even in harmony, so it goes back to a male voice. This is just one example to apply to one situation. Not a catch all for every example.

[rough depot]

how do i make the vocals more lovesong-like anyway? Less Through the Fire and the Flames more Starfire if taht makes sense

trying this now : symphonic power metal , love ballad, fantasy, operatic, romantic

ffs symphonic power metal isnt all screams

[polar spade]

[Male] [contemplative] "if only I had thought"
[Female] romantic " Its ok. We'll get through this".

Try just prompting through it. Definitely no exclamation points. Tag the whole verse with [romantic], or [introspective], [calm] ..... even a [gasp] might slow suno down.

Em dashes " -- " can indicate a point of pausing to think as well.

[Male] " This is -- so -- i dont know -- wrong"

[polar spade]

I mean, theres always some good results coming from covering a pumped up version with just the prompt... ummmm... let's say... romantic.... or ballad.... and probably set Audio Slide to about 60% and Styles to hmmm.... 50%... [ or A-40% and S-20%..] and see how it affects the whole song.

More than one way to skin a cat.

[rough depot]

i just really really hate the deliberate disobedience is all. literally says male voice. on the tag. first line? woman singer.

[polar spade]

Put "male vocal" as you're first or second prompt in the style prompts as well. This tends to Prime the pump to trigger the [male] tag you have.

[rough depot]

symphonic rock, male vocal, love ballad, fantasy, operatic, romantic, electric guitar, violin, flute, keyboard, ‑falsetto

[polar spade]

It might help to include "cinematic arc", dynamic pivot" , emotional journey" ... maybe "dramatic shifts" its all a guess,, not knowing every detail or lyrics , other prompts and how many iterations already into this one output, but I think you get the idea, we cant just tell Suno "symphonic metal" without also nudging a bit more of the style we want, in my opinion.

[rough depot]

yeah i have -falsetto and he still high pitched

[polar spade]

We were talking falsetto as it relates to suddenly switching to a female voice . If he goes into his upper head voice, thats just a talented male singer. So Suno did its job and its still male. 😄

To just have him sing all lower register notes. I would not know exactly where to go in one big swoop but back to trying to calm him down with "contemplative" and "introspective" , "dirge" , "sad" type prompts. Before that part. You have symphonic, you have operatic , Suno is just giving you what youve asked for.

[rough depot]

dirge and love song dont go together

[sleek stump]

I have the strong urge... Suno isnt a word processor for Music... You cant demand anything of it, you can only ask it to give you things. It may or not, and often its not. 😉

[rough depot]

gonna try this now

[Verse 1 | Male Voice]
|[Male Voice]I fell from the stars to a world I didn't know
[Male Voice]A stranger of steel in the moonlight's heavy glow
[Male Voice]But there in the shadows, fighting for your life
[Male Voice]I found my true purpose in the middle of the strife
[Male Voice]Now we ride together, with the engine and the brand
[Male Voice]Forever beside you, my dove of the elvish land.

Brute force it

[polar spade]

As you wish. But we are jjust trying to add a dash of salt-peter to the recipe. So if its 90% symphonic and 5% dirge,, then we may get, a calmer symphonic vibe.

Im just throwing out prompts that might calm it down without changing the vibe or sound of your output.

Remember your prompts are weighted for about the first 7 prompts listed. The first prompt is anchoring most of what the whole song is going to sound like. And thats going to be "symphonic " it seems unless you re-juggle them and try a new combination or add a few off the path prompts. Heck you can put k-pop on the end of your prompts and it might calm the vibes down withou it sounding anything like k-pop.

OK. Move operatic down a few notches and put symphonic rock in third slot. Do that at least. See what changes.

[rough depot]

ic...ok so what should my string look like?

i got: symphonic rock, love ballad, fantasy, operatic, romantic, electric guitar, violin, flute, keyboard, dynamic shift, cinematic arc

[polar spade]

"Should be"? Well let's stick with just "what can we try?" Try rearranging whatever you got then. That will always change the vibe.

Sso how about: male vocal, love ballad, fantasy, symphonic rock, romantic, electric guitar, violin, flute, keyboard

Just move symphonic down a few slots. I i took operatic out but leave in if you like, somewhere behind romantic.

[rough depot]

but its NOT just one male vocal.

[polar spade]

Its giving you separate male voices? If so, that the operatic prompt kicking in.

If it is not giving you more than one voice and you want it to, then I dont have a good 2am quick answer on that. Maybe a 10am one, id think of one. 😊

[rough depot]

jesus christ thats not waht i meant

I want verse 1 male. verse 2 female. Pre chorus both voices

its not even OBEYING the first tag half the time, the rest of the time it switches mid-verse

[Male Voice]
I fell from the stars to a world I didn't know
A stranger of steel in the moonlight's heavy glow
But there in the shadows, fighting for your life
I found my true purpose in the middle of the strife
Now we ride together, with the engine and the brand
[Male Voice]Forever beside you, my dove of the elvish land.

[Female Voice]
I thought I was lost when the darkness closed me in
But thunder awoke and it drowned the battle's din
A chariot forged in a realm I've never seen
And you stepped right out like a king upon the green
You shattered the wicked, a force they couldn't stand
My darling eagle, taking hold of my hand.

[Pre-Chorus: Both Voices]
Through the magic and the metal, we are bound by the fight
Two souls colliding in the fire of the night
We started as strangers under alien skies

[polar spade]

OK.

Im not sure putting [male vocal] on each line will get any better results. Ive never seen that to force one gender. Its signaling to Suno for a change or at least a change and then to ignore that change and continue singing male.

You have male verse as your first prompt, you dont really need [male vocal] as an initial tag.

[polar spade]

If you can tolerate an interlude.........try to indicate a change by putting s break between verses. Obviously, this can then be edited out fairly easy if wanted.

Now we ride together, with the engine and the brand
Forever beside you, my dove of the elvish land.

[Interlude - violin]

[Female Voice]
I thought I was lost when the darkness closed me in
But thunder awoke and it drowned the battle's din
...

[rough depot]

the other problem is, even when it properly starts in male voice, it often changes DURING th esame verse

[polar spade]

Not having any periods until the end is good, but try adding quotation marks only to the beginning line and end line point. Just one pair of quotations indicating all one voice .

[rough depot]

I try thst tomorrow

Thx

[polar spade]

Schools out 🤘 peace and good night all.

[half dragon]

Welcome to the same challenge many others have encountered when making duets!
It's known to be very difficult. You can't rely on just prompting.
There are tricks if you searched on this room, there are plenty posts about it.
One way is to use Extend, making one section at a time. @quiet solar has a good write up https://discord.com/channels/1069381916492562582/1466598465969590465

[acoustic egret]

hey guys how do i batch export all my suno songs ive created?

[quiet solar]

There isn't a way to do that - they have to be downloaded individually.

[quiet solar]

Being a model that uses creative randomness, Suno can't really be "commanded" like that.

It's better to think of your style prompts and tags in the lyrics as being suggestions that you're making, which Suno may or may not take into account when it generates the song.

Duets can be quite tricky anyway. Duets involving the two singers harmonising are even tougher, and duets where you need specific singers to perform specific lines are incredibly challenging to do through prompting alone.

This new song you're trying to make is going to be so, so much harder to do than what you've been working on so far.

Realistically, the way around this is using a combination of prompting and audio editing. Studio can help a lot with this, though I've used Audacity to put together duets as well. It does require a bit of a mindset shift though, as you need to start seeing everything that Suno makes as potential samples for a final song, so you'll need to start taking into account BPM and key signatures and get familiar with doing high audio influence covers, or creative use of Extend.

[slate atlas]

yo did anyone figure out good way for edm tracks to remain compliant to bpm yet

[lone nova]

Say if you have 2 generated songs with the same prompt, identical bpm, both have elements you like, and you want to combine both.
Would you use audacity,FL studio to put them together? then upload to suno and use that upload for cover, remaster or what ever?

[lone nova]

I assume you did generate EDM with exact bpm in style prompt.
and you measured the result outside of suno and find out it is not the BPM you specified?

[slate atlas]

regret to inform that i failed to insert the bpm into the prompt. measured bpm can sometimes be very close, however the render of the track has some sort of meta-sway in tempo which is model inherent and cannot easily be bpm-corrected

[lone nova]

if you have a specific track, that is not copy righted, you can upload it and use it in the "cover" option and in the sliders, you can make that audio have more weight like 80 to 90 %, and there for have the same bpm.
It is really specific for every use case.

[rough depot]

creative randomness is an extremely frustrating design choice. whats the point in tagging if its just going to disobey them?

[slate atlas]

its an alien intelligence, we do what we can

[quiet solar]

Yeah, if they're the same (or virtually the same) BPM and the same key signature, then you can make a Frakenstein's monster version by picking one track to start with and pasting the bits you like from the other track in where relevant. Same with using your own vocals, if you want to give a clear steer to the singer to do something in a specific way.

Then upload it and cover with a high-but-not-too-high audio influence and it should smooth out the rough edges and convert everything into a fairly consistent overall song.

Alternatively, if they're really close, you can potentially skip the "upload and recover" stage in Suno and put the final version together in Audacity/FL Studio.

[quiet solar]

It's a pretty powerful design choice if you want Suno to create original songs without you having to programme in every note, or use specific pre-existing samples like a traditional DAW.

But the downside of it being a probability-based model is that everything is probability based, which is very difficult to combine with the kind of instructional "do exactly as I say, every time I say it" commands you were hoping for.

For the record, all LLMs and other music generating AIs are also probability-based. The problem you're running into is one you have to learn to work with and work around as needed. 🙂

[rough depot]

"Male voice vs female voice" should NOT be subject to randomness.

[slate atlas]

its kinda like trading though, you dont assume the single trade will be a winner, you assume out of many trades there will be some that win and some that dont

[polar spade]

Tenor sax and bass sax sould not be subject to randomness either but presently all AI just hear "sax". Things are getting better each day.

[quiet solar]

@rough depot But it is, and will be for the foreseeable future.

There are workarounds - I put together a guide for how to prompt as effectively as we know so far (which is not 100% effective) and how to use workarounds and editing techniques to get to the end result you want, because duet questions come up so frequently: https://discord.com/channels/1069381916492562582/1466598465969590465

Short answer is that they're a lot of work.

[rough depot]

OBEDIENCE is the only legitimate design choice.

[quiet solar]

Then don't use AI

I'm being slightly tongue in cheek there, obviously. But if you're looking for strict adherence to a creative vision, you really need to be speaking to session musicians who you can describe that vision to and who can respond to detailed, specific feedback. Or perform it yourself.

Music AI works really well on a much more general level, but specificity is not it's strong point and probably won't be for some time. You can always come back to the song when v12 (or whatever) releases, if you don't want to engage in editing and workarounds today.

[polar spade]

One of the more obedient maneuvers we can use is to use extensions from the get-go. Or at least start a full song with the plan in the beginning to create every verse in an extension. In one aspect thats how we created 10 minute epic sagas when songs lengths were limited to 2 minutes. Then the nuances of prompting becomes more obvious, I personally think.

So baby steps have worked best for me. Let things around our target go wrong as long as the one thing we focus on finally goes right.

[slate atlas]

frontier claude code model hallucinates all the time. is what it is afaik

[rough depot]

Its literally one instruction. ONE.

A SIMPLE instruction to comply with.

[polar spade]

Listening to Crazy Train while having my omelet. Can't get better than this.

[quiet solar]

But it's not an instruction, it's a prompt. It's a suggestion. That's how the models fundamentally work.

You can influence the creative engine, and you can learn to get better and more effective at influencing it, and it will typically try to respect what you're asking for as far as it's capabilities allow.

If it's easier, just tell yourself that Suno doesn't have the functionality for multiple voices and duets yet. Because that's not a million miles away from the truth.

[rough depot]

Suggestion my foot

[quiet solar]

I can only help describe Suno as it is, in the hope that helps you understand what you're up against and decide how you want to proceed. I can't wave a magic wand and convert it into what you might like it to be.

Would be cool if I could, though.

[lone nova]

That is what I did, suno can not replicate voice, retain voice so people can say after the edit, that is the same guy from RL.
any post edit of the voice inside suno sounds like a different person.
The "replace lyric" function is not that great for longer sound bite. I tried to replace "change the day" to "stop decay", I have to turn down the volume to hide it.

That is why after upload I only repair instrumental, that is also why I use seedVC for "vocal identity".

[polar spade]

Males tenors can sing in frequencies above some female contraltos, so now we have to assume male and female wont be enough, right. Some males singers such as Prince are small (petite) and can have quite similar female characteristics because of that. Some females are larger with longer vocal chords and so can be mistaken for males. That's now three different meters for AI to decipher. Suno has scanned them all.

Prompting for certain genres , certain styles, more common traits like "giggling" is likely found more by females in all songs over the past 100 years than males, tenor are likely found as a constant in barbershop quartet than as often in glitter rock.

We start learning the tweaks that can nudge Suno towards these pockets that help get a male or female vocal.

Now leaving the café with Petty's Free Fallin' playing. Nice. 🤘

[quiet solar]

I sort-of agree. Suno can replicate the voices close enough on high audio influence that if you're not editing them side by side mid-line, then generally you wouldn't know it wasn't strictly speaking the exact same voice.

And if you cover the whole song, then the lead singer's voice should be consistent throughout the cover (unless you've got the audio influence too high, and it pulls in the original audio in place of the now-slightly-different singer).

[polar spade]

We literally cant hire a vocalist to sing either as a male or female. We have to look around for certain characteristics, body shapes, and talent and hire that one and still prompt them to sing the way we want.

Some females do a great boy child's voice in animation voiceovers.

The real worlds not perfect. AI just scans and reflects the real world.

Having a child genius wont make them obedient. Ya gotta beat it into them. haha.... ok.. I mean you gotta guide them in the right direction.

[lone nova]

It is bad...
I use lyric replacement, there is no slider for influence, I assume it is hidden at 90%.
It should analyze the clip before and after to determine the voice.
https://suno.com/s/kb7vCUtSQSIKnPpS
go to 0:35 you can clearly hear the difference(from "change the day" to "stop decay", and I generated 10x for this result also lowered db -5.

[sleek stump]

Many of us use Cover now to do lyric corrections...

[lone nova]

well I do too, it is cheaper and easier, if there is more than 2 spot to correct. credit calc: 4*x vs 10
But my problem is, if i do cover, the voice becomes a younger version of my original, barely recognizable.
Too perfect is also a problem, I want to retain inperfection some times.

[sleek stump]

I have found slider settings that prevent vocal drift, and theres a term you can prompt for human-like timing... "subtle rubato".

[lone nova]

which slider? for cover?
audio influence to 90 and remove any vocal style? or better?

[sleek stump]

W; 10 S;50 (default) A:90, remove everything in Styles prompt window.

[lone nova]

let me try

no lyric? or same lyric?

[sleek stump]

It may take a few rolls, I normally get a 95% fixed version within 3-5 gens. Such is teh nature of generative AI

Leave the lyrics and make any corrections.

This is what has been working for me.

[lone nova]

after testing, I think it works with suno AI voice better than imperfect voices from RL.
too perfect again, lose identity.
let me do 99% audio influence...
same problem.

[pallid condor]

I don't use a persona... maybe it reacts to lyrics??

[sweet ivy]

Basic question, but can't find it. How do extract the stems from Studio. Asked support at Studio, who said to double click in the song, but that didn't do anything but move the time marker

[half dragon]

You select the clip on the timeline --> look on the right panel (Clip), there should be a button say "Extract Stems"

[sweet ivy]

Thanks I see it now. Suno Support, [e-mail] doesn't help much, sometimes

[sweet ivy]

Found it, down at the bottom

[rough depot]

It keeps spending so much time on filler that it skips the outro https://suno.com/s/SnmBN2cORWknJnym

[high snow]

Wha does this message mean?

[half dragon]

It simply means, you can plugin in a mic or other input and record your own voice, instruments..etc.. straight into the Studio timeline.

[wintry jacinth]

Oh wise ones I seek knowledge

Can someone help explain what the Persona of an instrumental track does, does it only copy the style that was used for that track? or does it look at the track itself

[high snow]

At, ty.

[quick wedge]

In Suno, a Persona created from an instrumental track does not just copy the general style. It analyzes the actual track itself to extract musical characteristics such as tempo, rhythm patterns, instrumentation, energy levels, and overall arrangement. These elements are used to build a profile of the track’s musical “vibe.”

[wintry jacinth]

Gotcha, meanwhile the Voice Persona is strictly for the vocal?

[quick wedge]

Yes, in Suno, a Voice Persona is specifically used to guide the vocal characteristics, meaning it focuses on how the singer sounds (tone, delivery, range, and vocal texture) rather than the instrumental part of the song.

A helpful thing to know is that Voice Personas and instrumental Personas can work together. The instrumental Persona influences the music’s vibe, structure, and instrumentation, while the Voice Persona influences the singer’s sound and performance style. When combined, they can make different songs feel like they come from the same “artist” voice over different types of music.

[wintry jacinth]

So is this a matter of Remixing a song that you used one of the Personas? I dont see how to add them togther.

[quick wedge]

Personas are pre-trained vocal styles. Remixing tries to match the original audio, so Suno doesn’t let you override that with a Persona voice it would break the purpose of Remix.

You can try stitching them together in Studio that might be something you can experiment with.

[polar spade]

You said you asked Studio? The studio chat room? Or feedback. I wasnt able to find your post about it.

[rough depot]

isolating an outro https://suno.com/s/y2eu1PYllBGtHJt8

oh COME ON. it LITERALLY reversed them

[half dragon]

If you rolled enough dices - you'd get lucky

[quick wharf]

Hey Guys, I like making covers of songs that I like with the instruments or voice I like most. What tools is recommended to bypass the "Uploaded audio matches existing work of art." or lyrics ? Maybe a opensource tool?

[quiet solar]

If you're uploading songs that you don't own the rights to and didn't make yourself. Well, short answer is you shouldn't be doing that - it's a breach of Suno's TOS (and a major copyright infringement risk)

If you're being hit by copyright blocking on your own work, that's something to reach out to Suno Support about.

[half dragon]

TOS says you must own the material you upload; so there isn't really a work around.
If the audio you uploaded indeed yours - then you should contact [e-mail] to resolve the false warning.

[slate atlas]

tf

[half dragon]

Just click Create 9 times that'd generate you 18 songs at the same time.

[rough depot]

[Intro | 4 bars]

[bridge | 4 bars]
"Under the twin moons we shared our first kiss,
The signs were there, how could we resist?"

[Outro]

[Solo Female Soprano | 4 bars]
"O darling party leader of Red Lightning,
My brave eagle from another world"

[Solo Male Deep American Baritone | 4 bars]
"O sweet party healer of Red Lightning,
my elvish dove with emerald eyes"
How is this generating a 5 minute song?

and WHY is there a pause between leader and "of red lightning"

[half dragon]

You can suggest the model but I don't think you can reliably determine request for the length of sections or the song.

[lone nova]

https://www.youtube.com/watch?v=I6dPdLoAZEU

throat singing could be nice as FX for any song I think

[quiet solar]

Because in the same way most songs try to avoid locking syllables to the beat (as it can sound very formal or robotic), Suno learns from that and usually reaches for a more natural phrasing and offsetting. That inevitably means a variation in gaps and pauses.

Specific details like that are hard to prompt to include or avoid, but you can nudge Suno in a certain direction. If you have a song with a mid-line pause, then if you cover it and add a little more to that line you may be able to push Suno to use that space for the bit you've added. Or it may approach the line completely differently.

If you're looking for more precise control over exactly how a line is performed, it's worth noting Suno is much better at replicating audio than interpreting exactly what you had in mind from prompting, so you can sing it as you want it to be performed and then cover that using your style prompt.

It won't always replicate your phrasing perfectly, but if you really to want to specify exactly how something sounds then you're already looking at an editing job - running a few attempts to get a vocal sample for that line which is close to what you want and then editing it in is probably going to take less time than trying to generate it randomly by luck from prompts alone.

[sonic folio]

Is it possible to get vocal like a certain song that is "Logical" from (s)cooter

[lone nova]

male or female?

[quiet solar]

Possibly.

That kind of vocal sound is usually created by speeding up the playback of a voice recording and allowing pitch to rise as a result, which is how you get that sort-of "chipmunk" effect. You might be able to try prompting for it, but it's fairly unusual, so I don't know that Suno will have a lot of reference material to go by - and it will probably try to sing it naturally at that pitch with a lot of production effects, so it may not sound quite the same.

It might be worth trying speeding up a vocal recording of your own, including that as an audio prompt and seeing what Suno makes of it.

[rough depot]

Its the melodic outro i don't need it to go to the beat.

[velvet granite]

Does anyone know if its possible , when using songs as inspiration, to tell suno what part of the song uou want to use? Like you do with a sample maybe?

[sonic folio]

What ever works 😅

[half dragon]

I don't think it's possible. No.

[lone nova]

good enough?
https://suno.com/s/tSq8P4dZWnTDcClB

the goal is to make a persona of that vocal, so I just uploaded a remix, covered it to sing a different song,
retain the vocal, cut the male out, use only female part to finally be able to generate a persona.
keep the audio influence high.

main vocals: Pitched-up sped-up Melodic Vocals, use squeaky, helium-like, smurf effect, Chipmunk effect,

[pure panther]

You could cut it either in studio or an outside app like Audcity, then use the section. I don't know how well it would work, but if I were trying to do what you're doing, that's what I would try.

[lone nova]

works very well, i think the pro editor can also cut for "free", not stem extraction, but cut the male out and keep female works.

[brave gull]

perhaps you can crop the part you want and use that as inspiration? idk if that would work

ah i didnt see that this was just mentioned above whoops

[rough depot]

Frigggin 2 power outages today

[rough depot]

ok so HOW do i force it to treat "O sweet party healer of red lightning" as a single uninterrupted sentence?? That pause is AN-NOY-YING

[wintry jacinth]

Does it happen to any of you guys where the singer will pronounce a word wrong randomly

[rough depot]

Is there a tag i can use for "dont bother with the 6/5 meter"?

[rough depot]

Yes

[wintry jacinth]

Ahhh and the rest of the track was good too but he noticeably butchered the word lol

[rough depot]

its also doing "Under the twin[EXCESSIVEPAUSE] moons we shared our first kiss".

[quiet solar]

Suno can't be 'forced' to do anything.

If you have a sentence that repeatedly gets broken up with pauses in the wrong places and you don't want it to be, then either it's not long enough for the time it occupies (ie, there's room for a pause), or Suno is struggling to parse the rhythm of the sentence.

To try to fix it the first way around, add more to the line.
To try to fix it the second way around, use commas or other punctuation to give a stronger steer as to what the flow of the sentence should look like.

[sleek stump]

It happens and in every language. if you have that happen you need to use phonetic spelling to get word pronunciation correct.

[wintry jacinth]

Also I am having a recurring issue here:
[Chorus 3: Hopeful Delivery]
(FADE-ING) fading Where the silence. settles in.
(HOLD-ING) holding To a thread. I've worn.
(SEE-YING) seeing Past the night behind these eyes.

EVERY time no matter how I try it when he comes in to sing the chours he always skips the word fading. The choir also skips it but they sing HOLDING, and SEEING.

He just starts at 'Where'

[wintry jacinth]

the word is 'day' lol, he gets it right sometimes but he sometimes says 'bay'

[quiet solar]

Have you tried something like:

[Chorus 3: Hopeful Delivery]
(FADE-ING)
Fading where the silence settles in.
(HOLD-ING)
Holding to a thread I've worn.
(SEE-YING)
Seeing past the night behind these eyes.

Suno is sensitive to punctuation, capitalisation and structure and might find the above easier to parse.

[wintry jacinth]

ill give that a shot

[rough depot]

trying it with 7 syls a line

[Outro: Melodic and affectionate]
[Solo Male Deep American Baritone]
O my sweet party healer,
Elvish dove with em'rald eyes!

[Solo female]
O darling party leader,
My brave eagle from afar!
see if it does what i want

[cosmic birch]

Am I able to search for my own songs by my own personas without having to make it all public? Specifically through the web interface.

[pure panther]

I don't think so.

[cosmic birch]

Ah, well. I thought as much. Thank you!

[wintry jacinth]

That structure actually makes it skip every word in parenthesis :/

[rough depot]

Am I hallucinating....or did it finally get it right??? https://suno.com/s/xzMaaOVrixb0Q0xA

So now i just find the gen of the full song i like best, and stick the vocal stems on it right?

[rough depot]

is there a prompt or tag that can reduce the chance of screaming?

[rough depot]

This one is perfect....shame its just this tiny section of song and not a full one https://suno.com/s/r3m65LMBAnep9SEi

How do i make it not high pitched? https://suno.com/s/iTynbT7BYywRGFWa

[pine sandal]

https://suno.com/s/sOYWrGxXgdxJjpin

[quiet solar]

Did you only test it the once? Worked first time for me:

https://suno.com/s/vG0YlZZ1LIhp7ZTq?time=34

Usual caveat of "your genre may vary dramatically", of course.

[wintry jacinth]

I wonder if it’s because my meta tags are confusing it

That ‘chorus’ is different wording and it essentially is an extension from Chorus 2

It goes [Chorus 1]
4 lines

[Chorus 2]
3 lines with a [Pre Chorus]? Rising beat
Into this chorus 3

[quiet solar]

It'd be easier if you shared a link to the song itself - feel free to DM if for any reason you don't want to share it widely

[wintry jacinth]

Yeah this one.. this is the first complete song I wrote

So it’s special 🥹

Thing is that vibe of (chorus) line, (chorus) line, (chorus) line. Happens fine in chorus 1 and 2, but 2 ends in measure 3 and it catches it nicely, does a small little break and leads into this chorus

[quiet solar]

Ah, okay...that knocks out the first thing I was about to suggest.

So what you're saying is that Chorus 3 consistently breaks, even though it's arranged exactly the same as Chorus 1?

[wintry jacinth]

Correct.

My main suspicion is the transition from chorus 2 to 3? Is it a chorus still? lol is it the pre chorus.

Since chorus 1 is 4 lines. (Chorus) singer 3 times, then singer sings the 4 line resolving nicely.

Chorus 2 purposefully doesn’t resolve the 4 th line as instruments rise up for the final chorus

[quiet solar]

Then yeah, that's unusual. Suno is pretty good at replicating choruses as a general rule.

Can't really diagnose much further with the info available - as an outside shot, maybe take off the "3" from Chorus 3 in case that's steering Suno that you want it to do something differently from previous choruses.

An alternative would be to use an audio editor or DAW to literally cut Chorus 1 and paste it over Chorus 3, then reimport to Suno and cover it to smooth out the join. With an audio guide to follow, Suno is less likely to improvise. Although this may not work so well if your backing singers are singing at the same time as the lead comes in.

Best of luck

[wintry jacinth]

Chorus 3 is different words from the rest it’s the format of chorus to singer that’s the same

I’ll just share a link lol

[quiet solar]

PS. The transition you mention shouldn't be a problem in itself. I've used unfinished choruses leading into a second chorus a couple of times, and Suno generally gets what you're trying to do.

Oh, right...yeah, having totally new words for a fairly complex setup probably isn't helping. Do they match syllables and rhythm with the original chorus?

[wintry jacinth]

Ok cool. What did you have in between

Like bridge, pre chorus?

[quiet solar]

Neither, just ran them on straight from one chorus to another. It'll be easier if you just see it - links are to the start of the double chorus at the end.

First: https://suno.com/s/gcoaIGNUXaPxfZSm?time=121
Second: https://suno.com/s/cfRkhrPecuiWnOhP?time=100

In a twist of irony, that second one is the first proper song I put together in Suno, when I was very much testing out what it could and couldn't do.

If the problem just so happens to appear when you've completely changed the words in the chorus, my first instinct would be to scan for whether you need to make changes there for Suno to be able to fit your new words against the existing rhythm and meter.

[wintry jacinth]

❤️ massive nostalgia with that style. But I notice that you naturally lead it over with the lyrics, mine is probably to instant?

I sent the link in DM don’t know if you saw the structure yet

[quiet solar]

Just saw, one sec

[lost kindle]

I'm trying to remaster a song I made but it keeps modifying my lyrics to gibberish. Am i doing something wrong?

[main hazel]

@polar spade

need ur pro help with a promt

[polar spade]

@main hazel You can always post the question. And leave it open for anyone to review. What's the question?

[main hazel]

theres a song i been struggling to find a proper promt for but need it badly to create similar styles

[main hazel]

the rythm etc - could send u the youtube link if need, but at beginning (intro) is violin which i dont look for

yes i used those back then but would get completly different songs lol

maybe i was doin somethin wrong lol

https://suno.com/song/6a705c9e-5668-4e8a-8c85-0aada5538433

this

was rly good

but says deleted remixes

[main hazel]

cant do much with this "oom pa, bass kick drum" promt rite now

[polar spade]

I put it as last on the playlist now. Check the playlist again. Amd i got it by mashing some songs I think. Ytou might remix a couple snd try to mash teo together? Im not sure what will work.

[main hazel]

playlist says this still for me

[polar spade]

Thats not a playlist. Those three songs, though, are ALL in the playlist . You're trying to click on remix list in that one song? Just go to the playlist. https://suno.com/playlist/1bd13140-6f18-42d2-baf8-6a0fafa1a0a1

Songs I typically remix from are not themselves made public. But yes, I stuck them in the Laiko Waltz playlist.

[main hazel]

which of these 6 did u mix for the Inspired by Russian Laiko Waltz-

[polar spade]

Oh. Got it. Hold on. In the Playlist I put them in order of what seemed to work best. But I could not say how I did them. I assume i did an inspo of the playlist itself. So it would be the first three.

[polar spade]

So it would seem "Laiko waltz" was a driving prompt in all three. You can maybe "remix" my versions. Stick them in a playlist , and inspo them for yourself. Thats a thought. It might help?

[main hazel]

problem with walz is i think the song i sent seems rather some chanson, but if i type that in i never get such out

for some reason

[main hazel]

https://www.youtube.com/watch?v=MNxuPZ0dUMU

maybe that helps

1:16, 2:02

[polar spade]

Whats your question? How to get something exactly like this?

Try

[introduction | viola] ? Thats a good start. Or violin or even cello.

If you lead your prompts with viola, it may pepper it throughout the songs. So try different things.

Its also a bit "orchestral" as it has about 6 strings at once in some places.

Still it could be difficult in one generation or be very simple and give you something nice the first try.

Yea, just stick "viola" or "violin" near the front of your prompts, and perhaps even stick "orchestral" about 8th prompt towards the end.

Then tag in the lyrics [viola] or [cello] as an interlude or by themselves to help Suno pick up on the vibe

[main hazel]

so i d have to add those extra [promts] in lyrics tab itself rite? not just promts in promts tab

[polar spade]

The style prompts help give Suno permission to contain certain vibes, but putting in the lyrics helps trigger both where or emphasize the insistence on having them in the song, yes.

[polar spade]

Sometimes Suno doesnt listen to the "where" bit as much as it at least sticks a viola or cello somewhere in the piece.

[main hazel]

Because I was even avoiding violin typing -

to exclude

[polar spade]

At this point, i think you know what youre doing, its just a matter of testing different patterns and see which works. Thats all I do often.

[polar spade]

As always, all prompts work in tandem with each other. Granted , "waltz" as a first prompt usually gives a waltz. But the other prompts may have enough influence to move the needle and not get a waltz on a few. Who really knows. But if youre never getting a waltz, then there is the idea of removing prompts until a waltz shows up.

[sand lodge]

Hello, I need your help. Could you please tell me how to contact customer service or an administrator? My account has been stolen, and I am currently unable to remotely force a logout from his mobile device.

[polar spade]

You can post your request in https://discord.com/channels/1069381916492562582/1479483412250366054 to begin.

But always then make effort to contact Support by email as soon as possible. https://discord.com/channels/1069381916492562582/1373016771103166575

[wispy venture]

Hi everyone! Have v5 generations improved since yesterday? Or are generations still slow/sluggish, and some of them still getting stuck?
Any improvements generation wise (drunken vocals, shimmering, bad "regressive" audio quality reminiscing older v4 v4.5 models)?

Feels like some stuff is being deployed/tested behind the curtains

This has been happening for me (and many others since a couple of days ago)

[polar spade]

Well, as is said, if you're standing still, you're going backwards. They're constantly progressing, so glitches are built into the expectations. Then its balanced by having something nice on the return trip. Cost averaging keeps going in an upward direction.

[main hazel]

any clue why the persona voice always changes after it sung first minute? i think thats a bug

happens with any persona i made

[main hazel]

gemini says: The fact that the voice "breaks out" after about a minute is a known problem with Suno, especially when using the Persona feature. The model often loses its connection to the original reference clip as soon as the composition becomes more complex or a new song section (like a bridge or chorus) begins.

[rough depot]

why. wont. you. stay. consistent? https://suno.com/s/r3m65LMBAnep9SEi

It nails

In the web of death five archmages caught,
Androbas of whom I know naught.
Tarian my master, who met a dreadful end,
Wise Merlinda who taught my friend
But then slows down on
Mihoshi, the sun of the Eastborn flame,
Gandore the very land calls your name.

can some one explain this to me. why cant it friggin keep up the same cadence?

[polar spade]

@rough depot Have You worked with the edit tools, New Edit UI and the Legacy version?

If we go into them and watch the magenta highlighter go over the words as it sings, we can watch like a karaoke and see how it at least is thinking of the cadence. it might hang on some words, or speed through others.

Take a minute and check them out. In your menu click Edit/Open in Editor and the New Edit UI will open. Or in the Editor top left corner it says Legacy Editor in grey letters. Click to go to a simpler and easier editor.

Im not giving you a miracle tool. But just showing you a spot so you can simply see how Suno thinks when it sings. It has to groups letters [tokens] together called "tokenizing" and has so many it can use in an output. Well, a token is not a syllable, its more like the characters or sounds it can fit in a certain amount of time and space.

[rough depot]

no clue what those are, no

[polar spade]

oops... sorry . see above.

I"m trying to make a quick video to get there. Hold on. Take some time next week to explore all the tools, they'll help you in many ways when things need editing.

[rough depot]

literallyput growling in the neg section still get growling and screamo

it seems like it cant decide how many syllables it expects?

[surreal basin]

Anybody who ahs the trick to stop the " maria carey" mode when using female singer

[polar spade]

Yes. in the line "Do you want to eat" Suno might see that as 3 or 4 syllables. Because its also speaking in native English intonations, so most of us may sing, "jwanna eat" which could be 3 tokens or it could be a ballad and sing out every word "Dooo yoooo want.... toooo eeeat. " which might be about 8 to 10 tokens

it decides what it will sing based on everything we give it ... lyrics, prompts, estimated tempo , emotion, etc.

[rough depot]

this is semi passable except for the incantation https://suno.com/s/u4loAxkbwNYxeQvJ

The way it stresses "you and I". The final "ultimate massive heal" is good. But. Then it does weird stuff with the final chorus and outro.

nails "The black order will know true elvish wrath"

Like damn son that's Susan Calloway "helpless in our cries" level.

[polar spade]

Sorry. I wasnt able to get the screen vid recorder to work properly.

anyway, click into the edit and open in editor... go explore that tool a bit to get some idea of how it works.

At times, we can use the editor to add a space, type some words together like doyouwanna eat, or elongate word and sentences, add commas, line spacing, and so on to just a line or a verse and try to force a difference in the tempo and cadence of the lines, rather than just trying the whole song over and over again.

[rough depot]

https://suno.com/s/Po1BdvVqEmu5BY8w

ok hold on

[polar spade]

This would be a next week project you should do to take a couple of hours to at least explore it, check youtube for demonstrations and so on.

It has a learning curve on some things such as "Replacing" words , lines and sections.

[rough depot]

got it.

the part where she starts singing about Sylfaena's eyes should briefly get hopeful. i dont even know the musical t erm for such a shift so i dont know what tags to use

[polar spade]

I wont be helping at the moment. Got some priorities for today needing done.

[rough depot]

jesus christ https://suno.com/s/lCAny5I8zLg21PsU Suno ALMOST HAD IT but then it made the friggin incantation a chorus of men!!!!

[limpid belfry]

Can anybody tell me how you can extend a song, in which the original sound is preserved and extended. I have trouble in the sense that A) it makes a mess of sound, lyrics and structure... and B) it adds a different sound quality and intensity... alsof drops certain instruments. When i extend i always add the same genre input as the original (i am very specific with my genre input - maxing out the characters in detailed description), i keep weirdness and style influence at the same level (but for audio influence is aim for 60-70%).

[hoary sandal]

[INSTRUMENTATION: Electric-Violin, Processed-String-Quartet, Mechanical-Pulsing, Guitar-Slides, Bass-Guitar, Double-Bass]

[MOOD: Cinematic-Dissonance, Hybrid-Acoustic-Electric-Resonance]

[MIXING: Dark, Cinematic, Tactile, Velvet-Distortion, Warm-Mids, Deep-Stereo-Space, Ritual-Like]

[THEME: Orchestral-Rebirthed-Machinery, Grief-laden]

[INFLUENCES: Rap, Rap-Metal, Melodic-Metalcore, Alt-Metal, Grunge, Symphonic-Metal, Nu-Metal, Dark-Ambient]```

If I wanted to add sounds of wind and hammers to this and make this more like a Ballad without losing its core foundations, how would I go about it?

Nevermind...don't need it....Suno generated me something that tear jerked me

[polar spade]

Been there.

[hoary sandal]

https://suno.com/s/hlC81k5CS5s4XOUR

[polar spade]

Righteous. Good job.

[wintry jacinth]

I also struggle with this, there is this guide though: https://discord.com/channels/1069381916492562582/1468637403647447062'

[quiet solar]

@rough depot Just going to tag you on the above as well - I think you might also find sangderenard's explainers helpful:

https://discord.com/channels/1069381916492562582/1365192916687126650
https://discord.com/channels/1069381916492562582/1468637403647447062

It won't give you a copy-paste answer to the problems and challenges you're running into, but it might help explain why such an answer doesn't exist in the first place, and a potential approach to resolve it that may be less painful than constantly recreating the songs from scratch in pursuit of perfection.

[rough depot]

So how do i get it to stop pausing at weird times during spoken word?

Upon a summer some sixteen years before the outworlder's arrival my best friend Sylfayna and I teleported to the Grand University of Magic. " It keeps pausing after Sylfaena. https://suno.com/s/KYq7EXZU21053i5a

[quiet solar]

@rough depot As I've said before, if it's something that happens consistently, there's a likely a problem with the flow or structure around the line that you need to resolve.

Suno is a music generation AI. Although it can handle spoken word, it's not intended to be a prose narrator. It'll be aiming to fit the spoken word around the pacing of the music in the background.

It's also not an LLM - it won't understand contextual directions like "only pause on periods". And tempting though it is to rage against the machine, instructing it in colourful language not to pause won't have any effect at all - that's not how prompts work (see the links I shared with you above).

In terms of fixing this specific problem, you can use punctuation or phonetics to influence the flow and sound of words. Or, in this case, a combination of the two. As a test, I fixed it by:

• Changing the line structure slightly
• Added a preceding comma after "Upon a summer"
• Handled Sylfaena phonetically, linking words immediately after with dashes (that tends to nudge towards a consistent flow across the whole)
• Cleaning out prompts that were unlikely to have any effect

https://suno.com/s/FLzoUdhJWBFcwazz

[rough depot]

You cant tell me that even in a song you would normally pause after two words https://suno.com/s/8ZiF9AxHuQhHAg8S

[sharp falcon]

Here's something you probably shouldn't overlook. This has been super effective for me. On my last 15 tracks, I've been using neurological hijacking to draw people in to increase listener retention and it's becomming very effective.

Several different large language models (LLMs) are used, each with their own specializations.
One method uses the "Viral Vitality" formula: Total_V = (Video * 8.5 + Audio * 8.5) * (Sync_Bonus x 0.2) to solve the retention problem. Layering four LLMs uses each one's specialty. Claude is used for high-authenticity lyrics, Gemini for frame-by-frame visual analysis, ChatGPT-S for high-level orchestration, and Grok-AX for the heavy mathematical formulas.

The result gives a very unique music presentation with storytelling that draws people in and makes tracks that trigger human emotions. I'm also using micro-jumpcuts to trigger a physical "danger signal" in the listener's brain. Check out some of my tracks before you dismiss this method. Let me know what you think of the tracks. https://suno.com/@awarmfuzzy_com?page=songs

[quiet solar]

Okay, that took a couple of listens to understand what it was doing, but it's definitely phrasing it against the bars of the music. The pattern is essentially 1 bar [spoken] followed by 1 bar [no speaking] with each line getting 4 bars worth of time to be performed over. With a slight offset to exactly when the speaking starts (eg, it's not always on the first beat of each bar, sometimes it starts in the bar before).

Ultimately, that's causing each line to be split into two pieces. And in the first line, that split comes after "sixteen years". It does make logical sense, even if it doesn't make 'human' sense.

The 1 bar spoken / 1 bar pause isn't an arrangement I've seen before, and obviously isn't what you were looking for. I'd file it under "these things happen..." and look for a different generation to use as a better starting point for further refining and tweaking.

Bear in mind Suno has no idea how you want the song to be performed. It'll just try creating different arrangements each time you start again. Because you're aiming for a very specific end result, you may find it a lot less infuriating to pick an attempt that's in roughly the right place, then focus on how you can adapt and tweak that by using Cover or Extend to get the end result you're after. Trying to brute force Suno into performing specific lyrics in a specific way using prompts is typically a recipe for frustration.

Alternatively, pick up your microphone and record an audio guide for how you want Suno to deliver the lines, then create the song using that as an audio input. That should be a great help in steering Suno to the way you want it performed.

[slate atlas]

guys i found a good way to get nice bpm. you just have to listen for the rendered swaying tempo and when you find it during bpm matching, just artificially prolong the bar till it sounds right, probably by fractions of a beat, then when beat render recovers in the track, put back to exact e.g.

note the second yellow marker is slightly after the downbeat

[plain star]

It seems I've been hacked? Explanation: I paid $10 for subscriptions for two weeks, but they've tried to withdraw money from me more than 5 times already. I have enough credits! How do I get rid of this? If there was money in my account, they would have withdrawn about $60 in a week. Who do I write to about this problem?

[plain star]

Hlp

The problem itself might be that I need to cancel the subscription?

[polar spade]

Official Support is through email which can start with https://discord.com/channels/1069381916492562582/1479483412250366054

SUNO SUPPORT HELP - Note email review and responses are handled predominantly Monday 8 am EST through Friday each week. Emails are in a heavy queue at this time, but the only source of support for account issues, so patience is required and requested.

For support assistance, please contact the correct team by email:

Billing and subscription issues: [e-mail]
General support and all other concerns: [e-mail]

If its technical, a question about something glitchy or a noticeable bug - feel free to report in the 🐛 bug channel https://discord.com/channels/1069381916492562582/1082775596767117312 or give feedback at suno.com/feedback.

If you need help understanding who to contact and what information to include so the process goes more smoothly, feel free to check the guidance provided in the ⁠support-guide channel or start a private thread in ⁠ticket-help https://discord.com/channels/1069381916492562582/1479483412250366054 for assistance.

Please note that Discord is not a secondary support channel and will not speed up existing support requests. Support is handled in a queue, and requests are worked in the order they are received.

https://discord.com/channels/1069381916492562582/1373016771103166575

[polar spade]

And as Kanthric says, Suno generates all at once, so the music is balance with the lyrics, and the lyrics with the music is you try a one-and-done generation.

As Kanthric also suggests, if we can record a spoken recitation first, then add music to it, then the music will be directed by the oral spoken word, its pauses and intonations, more closely than the other way around.

[polar spade]

Hey. Sylfayna... thats like a southern Belgium name isnt it?

[rough depot]

Its supposed to be Sylfaena. Sylvan plus fae plus na. But suno butchers the pronunciation. Third princess of Anaura, mage.

[polar spade]

cool. Yea I think Sylfaen and Sylvaine are more popular names from the southern Belgium / France area. I know a Sylvaine

[rough depot]

Cool

I did a song for her too actually

[azure crane]

How do I get Suno to play in the correct octave on the piano? When I upload a piano piece, it always plays an octave up.

[rough depot]

https://youtu.be/Q27beVNXrqU?si=pzWWnw0yUyA9LHz5

This woman is the POV of the Spellscourge song and corresponding chapter. Nenewyn Andaure.

[polar spade]

All I could suggest is to make sure the recording is as clear possible. My instruments uploads are mostly 90% in the same key and pitch as the upload. And usually correct on the initial outputs and changes on subsequent outputs.

Since an upload, I guess syles and audio sliders about 85% both or leave them as defaults.

[polar spade]

You think there are any high notes or counterpoint type playing or contrasts of anything an octave higher as well that its picking up -- or even overtones from the reverb in the room? Speaking of reverb, maybe keeping it as dry possible as well might help.

[azure crane]

No, it’s only my digital piano. I will try to change the sliders then.

[polar spade]

And by the way, an official hello and good morning my friend. You and I have been here with Suno since about the same time. Always good to see you. Enjoy the day!

[azure crane]

Ooh, Such a warm interaction. Same to you — and thank you for always being there to help us 🙏.

[sleek stump]

You can also add what key it's in, in the styles prompt, which can help Suno stay on key.

[sleek stump]

@stone vortex Can you share the link to a track this is happening on?

[primal knoll]

Is there a way to "reset" a song from a mashup perspective? I'm getting errors about there being too many layers or too many past mashups for a couple songs I want to do a new mashup for. Remastering doesn't seem to work. I'm hoping to find a way to get Suno to just see the songs as a new song

[stone vortex]

Sure
~3 weeks ago - no vocal https://suno.com/s/1ZYQHURhIdOPYIFp
~2 weeks ago - no vocal
https://suno.com/s/BzAxL7iBd95cLQzm
~week ago - no vocal
https://suno.com/s/btSgaRE1yOGn1kBh

now - vocal
https://suno.com/s/bcHdM3g016TrL0Ux
https://suno.com/s/Du5vZhb5uaSvse5G

Of course, there is always chance to get hallucinations, but now percentage of full instrumental / vocal hallucinating results is flipped

I'm not too good at prompts, and don't calling myself as professional of suno, just create something for myself and trying to figure it out, why i'm start getting so much hallucinations

[sleek stump]

Looking at this now...

Ok, I opened up your prompt quite a bit, Suno 4.5 and newer models really benefit from more descriptive prompting. Look at these and take note of the prompt, empty lyrics window and teh slider settings.
https://suno.com/s/zqKGPsYiOdgIzzag
https://suno.com/s/fGhiDvvjt8j2IR4p
@stone vortex

[stone vortex]

Thanks a lot

[sleek stump]

You can adjust the prompt in any way you need, but note that the "first" thing there is that its an instrumental.

[lone nova]

why did you remove the word "fully" from "fully instrumental"?
maybe add -vocals as negative prompt, also -beatbox

[stone vortex]

Just experimenting how it work with different versions of instrumental tag. i've also tried "only instrumental"

[languid forum]

does persona creation not work anmore?

[lone nova]

https://suno.com/song/b442426b-0c3a-4242-96e3-f5afcffb45c8
https://suno.com/song/8d3cf017-17c2-4887-8769-5da77a08000e
works so far for me with my edit on your prompt
maybe increase the style influence from 50 to 70+
I hear a bit beatbox in my version at 0:55, with style at 50

[lone nova]

I think you can not create persona from uploaded music, only from generation. or after multiple edit of the upload.

It works for me, but the effect goes away after 1:30. I also need a solution for that.
https://suno.com/s/xGOx69khJ1yelW6P
here my chipmunk persona was clear at start 0:22, but at the end 2:25 it is gone.

[languid forum]

i just want to make a persona from a song made in v5.0 but i cannot make it into a persona

[urban sandal]

Anyone know any good tricks to exclude vocals??? Any time I use a prompt including the word "Soul" I get nonsense vocals contaminating what would otherwise be a great track.... The negative prompting just straight up does not work.

[lone nova]

url for the song? are you sure you did not use any upload in the process to create/cover that song?
if you go to create->persona->create new persona-> can you see the song u want to make persona? is it greyed out? with a message down there, saying the error message?
or if you try to create that persona, the system just do not react, give your error mesages?

[rough depot]

Is this structure gonna work? I am low on creds and dont wanna brute force it atm

[Verse 2]
Down vaulted halls with crimson glass where poison traps ascend,
We broke the cultists' heavy guard and fought them to the end.
With armor black and glowing swords they thought to seal our doom,
But shadow strikes and longbow shots confined them to the tomb.

[Pre-Chorus 2]
The widget-shaped inner sanctum calls beyond the iron door,
Fayd sits upon a black throne, time to even up the score!

gimme a minute to fix the first line too actually. I hate gemini some times

(verse 4 line 4 imma also change to be aboutteam tactics) point is, do you think these lines will cause problems in the app?

Also this

[Outro]
Three rapid shots rang out to end the dark wizard's spell.
"They fear me," Victor whispered as the lifeless body fell
Fayd lies dead in a pool of blood and disgrace,
Now time for our heroes to loot this place!
The Mastermind is not here, but we'll bring the pain,
Hear our battle cry: Red Lightning Strikes again!

this one i am even less sure about the syllablry

[rough depot]

Changing line 1 of pre-chorus to

We entered the black sanctum, beyond the great iron door,

Ok so its a mismatch right? one is trochaic the other iambic?

i might need to change some of the "the"s into "they", so it sounds like Rhapsody of Fire,

[lone nova]

it is fine if you have difference, u need to learn the meaning of those words 😄

[rough depot]

as an homage

I suppose I do. But do you think Suno will have an issue with these lines?

[lone nova]

I think it is fine.

[rough depot]

I meant in terms of meter and stuff

[lone nova]

just read your sentence loud with rythme, it is not that hard 🙂 have some confidence

[shadow plaza]

Can somone help me with my project...

Im trying to do 2 male Duets.

I tried give them names and descriptions of their unique voices

Also in lyrics box i stated who sing what part

But still... Their voice sound very similar and on chorus section i want them so sing together in duet but that never even happen

[rough depot]

yeah definetely got a clunker...

Soon the Black Order will feel the pain,
They will cry Red Lightning Strikes a-gain!
workin on it

Until they cry?

no the vowel preceeding the R makes it hard. Scream maybe?

[rough depot]

https://suno.com/s/MKyG3bqFy3EPk3im

What neg prompt do i need to eliminate the whining? I can't understand the lyrics

[fathom lichen]

I need a help for having two voices in different moods of singing (like one rap and the other singing) because I love to make my AI songs to look like the Linkin Park songs.

[rough depot]

Where its singjng a story at you metal style

[rough depot]

nvm got it i think.

[quiet solar]

Try making one a tenor and the other a bass-baritone and see if that splits them sufficiently apart.

PS. It sounds like you're doing a lot of the right things, but be aware that controlled duets are very challenging in Suno.

[solid briar]

is it possible to make a voice robotic and dark?
it seams to be hard to control the voice in SUNO.

[quiet solar]

You might need to clarify what you mean by "work".

Verse 2 has a very stable iambic meter and syllable length. It's more common in poetry than music, so may steer Suno to perform it more theatrically or more conversationally, so if either are what you're looking for then great. It should fit fairly neatly into lines provided the BPM isn't too high. Anything below 140bpm should probably be okay. Anything higher and there's a fair risk it'll split the lines in two (not good in this case) or stretch them out over twice the space with gaps and sustains.

Pre-Chorus 2 is close to iambic, but the opening doesn't scan thanks to "widget shaped" which adds an extra syllable. For it to sound natural in iambic meter, the stress should fall on the INN of "inner sanctum", so the 4 syllables of "the widget-shaped" need to be shrunk to 3. If you're looking to follow that with a 14 syllable iambic second line, black sounds odd as a stressed syllable.

A 'true' iambic version would be something like:

the DARKest INNner SANCtum CALLS beYOND the IRon DOOR,
where FAYD enTHRONED aWAITS, it's TIME to EVEn UP the SCORE!

Whether you want it performed in iambic flow is a stylistic decision that really only you can answer as the creator. And probably only by testing it a few times and seeing what Suno makes of it.

[full field]

have u tried just putting that in? dark robotic voice

[rough depot]

ah ic ic

[frozen dagger]

I like to use "Sounds" feature to create some voices. Then I duplicate the small sample in editor/studio, then Cover it to create audio with the voice sample in the track, and then make a Persona out of it.

In "Sounds" put something like:
Dark Robotic Voice, "What you want it to say"

Example: https://suno.com/persona/cbb9c4db-21a8-41dc-85e4-3d17efef3dfe

[quiet solar]

Continuing from my previous post, the second line is iambic and the first line is almost iambic, but thrown off by the word 'dark'. The flow would be restored if you replaced it with a two syllable word where you can stress the first syllable naturally (e.g. evil).

The rest of the outro looks like it was originally in Anapastic meter but got edited away from it?

"fayd lies DEAD in a POOL [missing unstressed syllable] of BLOOD and disGRACE"
"now [missing unstressed] TIME for our HERoes to LOOT [missing unstressed syllable] this PLACE"
"the masterMIND is not HERE [syllable pause] but WE'LL bring the PAIN"
"hear our BATtle cry RED lightning STRIKES [syllable pause] aGAIN"

I'm not sure about masterMIND being the best way to stress that word. Probably worth switching it out for something else where it's natural to stress the second syllable, e.g. the dicTATor's not HERE but we'll STILL bring the PAIN would convert the line to Anapastic meter.

The final line doesn't scan as Anapastic meter at all, so it'll likely be performed differently to the previous lines, with a more freeform flow. Which isn't a problem if that's what you want.

Also worth noting that even if you feed Suno a poetic meter, it will still likely vary the delivery. If you were hoping to use iambic or anapastic meter to prevent mid-line pauses, there's no guarantee it would work. These are all lyrical/songwriting decisions that will influence how the end output might be interpreted by Suno rather than control it.

[rough depot]

i am removing the mastermind line actually.

also i have made edits to that whole thing

[Outro: High tempo, triumphant, swagger, heroic]
The evil Fayd lies deceased in a puddle of disgrace
Our heroes know time has come to thoroughly loot this place
Soon the whole damn Black Order will feel holy epic pain
Then they’ll scream like sissies Red Lightning Strikes a gain!

(getting it to pronounce the word "again" "incorrectly" for poetic purposes is hard)

[quiet solar]

You can switch the pronunciation if you write it as again (ahgehn) vs a-gain / a'gain (ah-gain).

[rough depot]

I'll try that next (after I attempt the silly song I have planned)

[quiet solar]

What you did with "a gain" should also work.

[rough depot]

50/50 yeah

[quiet solar]

50/50 is sometimes the best we can get.

[rough depot]

https://suno.com/s/YxJHnRy46Bj6lNTV

[quiet solar]

@rough depot Meter isn't the be all and end all in songwriting, but it's useful to have an awareness of it. Even if only to help with scanning lines to see if they're likely to flow naturally or not, depending on whether you want them to or not.

You might find something like this useful as an introductory guide - just skip past the desperate attempts to flog you e-books you don't need: https://lyricassistant.com/how-to-write-lyrics-about-meter/

[rough depot]

Can i prevent the long list from slowing down?

[quiet solar]

The long list?

[rough depot]

From my other link

[quiet solar]

Yeah, I don't know which part you're referring to

Do you mean the bridge?

[rough depot]

Yes

The hardware store style bridge

A funny song to follow a heavy one

(Its cute 200 year old elf princess singing)

[quiet solar]

Yeah, that's a meter problem.

The part in bold is all fine. 16 syllables fit perfectly to the beats of 2 bars at double time, with that consistent "da-da-da-da, da-da-da-da" pace. The first line is perfectly set to that, the following lines have 15 syllables, and Suno is handling that with a space for one half beat at the end of the line. Still sounds very rhythmical and deliberate.

The last three lines break the flow. Suno can see exactly how to handle the first four, then breaks the rhythm starting on line 5 because it needs to prepare to fit what you've given it into the space available.

**Iron hinges, brazen brackets, battered shields and leather jackets!
Pewter mugs and silver spoons and tapestries of crescent moons! [beat]
Goblin cheese and salted meats and dusty wool from dingy sheets! [beat]
Tallow candles, rusty nails, and water from the wooden pails! [beat]

Iron skillets, kettle drums, the pastry tray of stale crumbs [beat]**
Broken arrows, bowstrings, spare subligars, barrels, random paper scraps,
Copper piping, braziers, tongs, musical instruments, Pickaxes and iron chains and buckets barrels, caliipers, empty tins,
loos bricks and every single thing we cross

If you wanted to fix the last three lines, you're looking for groups of four short syllables, e.g.

"I-ron hin-ges / bra-zen brack-ets / batt-ered shields and / leath-er jack-ets"

To pick out example problems:

Bro-ken arr-ows (fine) / bow-strings spare sub (problem)
Mu-si-cal in-stru-ments (problem)

[rough depot]

Ahhh...

Ok imma adjust that

[quiet solar]

Bear in mind you've also got internal rhyme going on in that first section too:

Brackets / Jackets
Spoons / Moons
Meats / Sheets
Nails / Pails

If you want it to 'sound right' you'll need to replicate that in the final four too, so that the 7th and 8th syllables rhyme with the 14th and 15th, or 15th and 16th depending on the length of the line.

Avoid having lines longer than 16 syllables or shorter than 15 syllables like the plague if you want to keep that really rhythmic flow going.

[rough depot]

Im also experimenting with making the vocals less hardcore...hmmm unless that makes it funnier?

[acoustic egret]

I've got suno pro, but when I export to stems I don't see the "MIDI" option

[quiet solar]

You won't straight away.

Assuming you're doing it from the Library, you need to export the stems first. That should take you to the correct window - if not, go back to the original song, click Get Stems/Midi again and choose "See Extracted Stems".

That should get you here, where you can select MIDI from the download options:

[acoustic egret]

I don't see those 2 options