📚┃suno-school

[hollow hare]

It...does not sound like you are using Suno. https://discord.com/channels/1069381916492562582/1396887662186598460

[unborn star]

ye thats wild to me

maybe you post a picture of the side you try to get in

[robust anchor]

FFS, I am using Sunoai dot ai!

[unborn star]

As a paid user you have it like this:

[hollow hare]

That is not Suno but more like a service clone/hijacker. Basically uses Suno services for less quality but ripoff prices for the user.

Suno official site are https://suno.com and https://suno.ai

[unborn star]

You go for custom on top layer and you see lyrics box and style box

there you can put your style description

and in lyriks your lyrics + additional style options

you have 5000 Characters for Lyrics and 1000 for Style

[robust anchor]

This is the site I am using.

[unborn star]

that looks fake

at least for this discord server

[hollow hare]

Unfortunately, that is not the official site. If you want to continue using their service, that is your choice..
But paying for words? That is expensive

[unborn star]

wow

[robust anchor]

I cancelled the sub, I will use the official site, thanks guys.

[unborn star]

Good thing I didn't went to the wrong page

np =)

[hollow hare]

Here are the subscription prices for Suno. For comparison, it costs 10 credits min. to generate, which gives you two tracks

Simple Mode is limited to 200 characters but Custom Mode allows up to 3000 characters for lyrics and 200 characters for styles.

Oops, that is the yearly pricing. It is $0, $10, $30 USD respectively for Monthly subscription

[robust anchor]

That's fine. Thanks again and I'm looking forward to seeing what I can now really make with this tool!!!

[wild sigil]

Can I see a list of all the prompts that Suno knows or understands do we have documents like that?

[robust anchor]

The difference is amazing! I could get my full style words in and it's beautiful!

[unborn star]

not really but it understands a lot

[unborn star]

Good thing you joined the server otherwise you would have lost money. =)

[radiant breach]

Cover is broken right now, don't waste your credits

[left juniper]

How can I find my prompt for an older song i made?

[half dragon]

On the songs themselves?
Or you mean the prompts you saved?

[left juniper]

On the song where you only have 200 characters to use to describe the song

[half dragon]

Should be on the song - the old one you made.
EDIT: I realized you asked for the prompt you used in Simple mode, where you described your song, not Style prompt.
I don't think it's available after the song was made - instead you'd find the generated style prompt on the song itself.

[gusty hinge]

Hi! I have a PRO account, but many of my generations still produce the "this is a fake-up" watermark vocals. Credits are deducted, but I cannot use these songs. Can you please help?

[brisk flint]

There are several "wikis" you can have a look at like sunoaiwiki.com but nowadays also GPTs like Gemini and ChatGPT5 have a pretty decent understanding of how Suno works and can help you with achieving your goals by giving you prompt-suggestions. But best thing is, have a look and listen to tracks/songs of others and have a look at their prompts to learn how they did this and that. And if something doesn't work out, don't get frustrated and take a break.

[near flume]

click on the title of the song and you'll see the style prompt underneath the title on the full page.

[near flume]

Define "this is a fake-up" watermark vocals. What are you using for your vocal declaration?

[brisk flint]

I also wondered what this 'fake-up watermark vocal' is. I've created a lot of tracks with Suno, but I've never come across this term, nor do I have any vocals that would fit this description. On the other hand, I do have a lot of tracks where people are screaming their heads off 😉

[polar spade]

Typical cause for the accounts being flagged:

• API usage (either by you or through Third-party services, including non-official apps)
• Specific IP address (e.g. VPNs)
• Payment method (e.g. stolen card used)

If using the APP, here are a lot of scam apps on the app/playstore. Ensure you have the official apps by going to
https://suno.com/Android
https://suno.com/iOS

Due to people abusing Suno's internal API for their own gain, Suno have been detecting for any accounts that may be related to these third-party services (e.g. non-affiliated parties, non-official apps, IP address, etc.)

Looks like your account got flagged. Nothing to be alarmed about as there have been several false flags (mainly due to VPN).

Please fill out the following form and the Suno team will investigate your account to remove that false positive: https://sunomusic.typeform.com/to/zGsJSzsf

Be aware it may take some time to resolve as the investigation team is small and only work during Eastern USA business hours

[native minnow]

Hi everyone, I hope someone can help.
Maybe it's just me, but day by day, I'm getting more and more issues creating a clean/clear sound song.
I'm using v4.5, I don't like at all the new 4.5+ (for my taste, 4.5+ is too "complex")
It's hard to explain, but in every one of my songs, the voice is a bit metallic. Make sense?
And, now even more often, if the song "sounds" good at the beginning, always around the last minute start to lose quality.
I'm sure I'm doing something wrong, otherwise, everyone would be here to "raise up" this.
Can I kidnap someone a bit picky for help? Please
Please, I'm losing my enthusiasm and it killing me.

[half dragon]

Provide prompts and example songs might help others to be able to recognize the issue.

[native minnow]

Here is my latest song with this issue. For today I lost my good mood on it.
[LINK DELETED]
Pretty much in the entire song the voice is "metallic", try to listen from 2:45... the quality is terrible.

[polar spade]

More recent doesnt always mean best. Random is random and Suno is random. Try your options. Try a remaster in 4.0.

[native minnow]

100% agreed, honestly never tried 4.0, I joined in when 4.5 just came out and was good. I will give a try!

I wanted to hope with 4.5+ in "beta", the 4.5 became "stable" ehehehe

4.0 seems less noisy/disturbed, I will give more try for sure!

Not ok, but better

[polar spade]

4.5 and 4.5 are STILL good. But its made clear just one prompt change can throw an outputs fidelity in a totally different direction, and then the next output randomly do a 180. Let alone, each of our own attempt with multi-prompts, metatags and lyrics.

[polar spade]

Hah. Now you may be getting into "subjective" territory. 🙂 one users metallic is another users mellow.

[native minnow]

thats why im here to ask some suggestion, im sure i can improve it without blame suno

[native minnow]

100!

[polar spade]

I have a simple process. Buy 30,000 credits and click 100 times. One has to eventually be decent.

[native minnow]

im sure everyone can get me because kinda the the same situation: its pure frustration when u "throw" away 500 credits for the same song.
when u scream: YES finally... start awesome, perfect voice... and then the last minute is trash 😄

ahahahahahahahah

yup... just random. get it 😄

back to serious: there is some "prompt tips"?

some special word/keyword to use, like: High Quality.
I confess that Im trying everything

[native minnow]

More i'm here... more the suno's meme make more sense

[robust anchor]

Anyone selling any music for profit?

[raven turtle]

If you mean once you factor in the cost of Suno credits, mastering plugins etc, then distribution costs like Distrokid snd look at income from Streaming or Bandcamp sales… not a penny! 😂

[graceful condor]

WTH is Suno's Problem. I nearly finished a song. Needed to extend it now only this line left

[Final Chorus] I heed the call

[Outro: Fading piano melody, lingering synth pads] [End: Fade out]

And now it is saying I can't generate it because the Lyrics are copyright? WTH!!!! This is my own song. And I already created a couple of version of it with the same lyrics. And now only this last 4 words are left and now it is telling me it is copyright? This makes no freakin sense.

[near flume]

There's something going on with Suno, it's buggy as hell right now. Remix only populates the Style portion, can't get any vocals other than male no matter what you do. Is there an update in progress?

[graceful condor]

But at least you can literally cover Shake it off from Taylor Swift but if I wanna Extend my own lyrics with "I heed the call" It says it is "Copyright" this is dumb beyond understanding.

[half dragon]

Seems like all 4 triggers it - "heed the call" is fine - "I heed" is fine - "the call" is fine
Maybe spell I as "eye" or something sounds similar?

[woven hull]

Hello, I need your assistance. I have been attempting to create a dark, deeper female voice for a song for quite some time, but it is not working. Do you have any experience with this, perhaps a suitable prompt? Best regards, Chris

[brisk flint]

First of all, I would suggest using the style-prompt-windows and the lyrics-section to enter prompts in both sections: At gthe beginning of the lyrics-section you can describe the overall mood, instruments and the desired type of voice, later on before each verse / chorus you might repeat it.
Second, I would give it a shot with adjectives like 'deep, smoky, warm, gritty' and best try to avoid words like 'soulful' as this could sometimes lead to sort of an misinterpretation, i.e. Suno then adds some Soul to the whole song.

[woven hull]

Hello Tia, thank you for your reply. I will try it. Thank you.

[brisk flint]

I just tested it briefly, and your task is really very tricky, especially since I currently get almost the same voice regardless of prompt changes 😭
Only here did Suno briefly let a rather androgynous dark voice sing after the Eilish-alike intro, until later on it was clearly a man singing. Sorry, seems to be too difficult a task for me:
https://suno.com/s/LXPnsiKpzzysjWjy

[eternal sigil]

Anyone notice that all their covers are coming out the exact same duration with 4.5+

[humble verge]

uh.. what?

Also why does SUNO think female vocalists are male vocalists?

[humble verge]

Yep and also cuts the song when time is up 🙄

God SUNO is bugged beyond reason.

[magic ridge]

So… I’ve been playing around with 3.5 for weeks, just subscribed, and was excited to try 4.5+. But the singers basically SCREAM or BELT all the time! It doesn’t even fit my genre/style, yet they always start screaming during the chorus, lol. It even ignores the exception list. Is that a 4.5+ "feature"? Is 4.5 better in the sense that it doesn’t make the AI singers scream constantly? I’m just wondering if it makes sense to stop using 4.5+, even with its better quality, if other versions work more smoothly :/

[ebon sand]

yes it's very biased to do that, maybe you'll like 4.5 better but i don't like either of them

4.5 didn't scream as much so might be easier to prompt for you

[magic ridge]

I'll give it a try, thanks!

[ancient haven]

Anyone have any tips or tricks to do when you want to prompt like a pause in a line or a way to dictate cadence?

[mellow coral]

[pause] or [break] works for me in the lyric box, but it can be finnicky. I think it's dependent on the genre. It tries to justify it in the correct style and doesn't always do it how you might expect.

[winged peak]

Yes, and it's just my feeling that the screaming has increased in recent weeks.
At least you can ‘overwrite’ the relevant verses with the editor. 😉
But yesterday I really jumped out of my chair because the screaming came out of nowhere from the speakers during one of my deep house tracks.

[past kernel]

In which browser does the son geditor beta work?

[half dragon]

Sorry, the what?

[past kernel]

de song editor beta

that fancy song editor, the legacy one works fine. the betaone works in not one browser

[half dragon]

Ah, the Editor? I use Chrome and it works great there.

[past kernel]

the new one? standard chrome?

[ancient haven]

WIll it possibly work well with Opera systems?

[half dragon]

Opera browser? haven't tried but it should.
Let me try quickly.

[past kernel]

vivaldi is, afaik, a opera based browser

[half dragon]

Works great!

[past kernel]

weird

for me it does not

[half dragon]

I'm on windows 11

[past kernel]

you and me both

it simply isnt starting the playback

[polar spade]

I used editor UI a lot yesterday and would get good replacements most on the first try, and alternates were just as good. Only a couple of slightly off splice points.

[ancient haven]

Awesome! Can't wait to get my hands on it

[half dragon]

Not yesterday but the day before - all except one song, it was good on first try on replacement.

[polar spade]

Hey wheres your playback button. Yea, hmmm, a display format issue for your monitor or browser?

[past kernel]

format in form of resolution of my tab?

[half dragon]

I see the buttons there - but maybe it didn't work?

[past kernel]

it doesn't change whatever i do. whe n i use floorp or any firefox based browser, iot works fopr like 3 minutes

than the tab dies

[half dragon]

Do you have any plugin or extensions installed?

Have you tried in private mode? like Chrome Incognito mode not sure what they call it on Opera

[polar spade]

Any conflict with using a remote keyboard, if you have one?

[past kernel]

i have non, i tested it in icognito - icognito has no extension installed

[polar spade]

So it stops about after that first orange chorus?

[past kernel]

it doestn start at all

[polar spade]

Ahhhh.. OK.

[half dragon]

But it plays a bit then stop when you tried Chrome and Firefox, right?
That's strange

[upper ice]

https://suno.com/song/28914c80-464e-4ccc-a87f-b7626a4167bc
current WIP. ai flubbs the prechorus, the first chorus, and then the very last line of the song. not having any luck fixing it with covers. seems to either completely remove sections of the song, or have the same mistakes. even at 100% audio influence it sometimes completely changes the intro to the song. any ideas would be appreciated!

[polar spade]

Practice using edit UI to fix those points. Its assumed many output have these glitchy points that often cant be magically prompted to work -- so Suno created an editor interface for these moments. Its almost easier to just manually edit through each one.

[upper ice]

i can try using the editor, but my experience with it so far has been far from good. any section i regenerate ends up not sounding anything like the original which is a little irritating.

[polar spade]

Ytou can run another generation and rearrange the lines a bit. This will force Suno to re-tokenize the syllables and often corrects the issues of it skipping words. It does that if it sees a rhyming scheme for instance and wants to keep that scheme throughout. For instance, it takes, FIGHT, tried to rhyme with Paralyze as well as NIGHT, so , since you have "you csnt..... " twice, it basically ignore SCREAM, to get a better metering. It likes to do that.

Try putting "no wrong" and "no right" on the prechorus on the same line, so it see its more as an appropriate rhyme.

Once doing those, it may actually flub something else, so it can be frustrating trying to change everything in one pass.

Interesting it has such a long 1:30 intro. Thats often a sign Suno is itching to hallucinate, as it is running out of "space" to fit in the lyrics.

Killer song, by the way. Best of luck.

[upper ice]

i'll try that, thanks for the tips. i actually really like the long intro so im trying to keep it when i do covers, but it like to cut it out a lot. haha

[polar spade]

Yea. The long intro worked well. Just saying thats unusual most times. And yes, Covers likes to cut down long intros and make a radio-edit version.

[ebon cairn]

what are the best ways to ensure that a new melody will follow the arrangement of an uploaded track in suno. Does creating the sections in editor then saving as new song work?

[blissful peak]

If your goal is to preserve the melody or structure of an uploaded clip, the Cover Song feature is one of the most effective tools.

[ebon cairn]

Yeah that's all I ever use. I'm composing my own songs primarily then using suno for that extra little bit but when I struggle with a melody for instance and I want suno to generate one to my track, I'll upload the lyrics with arrangement tags but it doesn't always get it right. Especially if it's a non traditional arrangement for instance

[fathom siren]

Anyone know the best possible settings for a wav/mp3 to upload so it stays as close to the source file as possible? I assumed using suno wav download settings would be best but song gets further and further away from original every time I download/upload (just testing).

[half dragon]

There is no such setting that I know of.
Suno only provides upload/download tool.

[faint crypt]

Create error
Create failed due to internal error. Please wait a few seconds.

What does this mean? Why can't i create?

[raven turtle]

And I'm just being lucky at the moment or is Suno having a quiet background update as the sound quality seems to have taken a big step up for me.

[polar spade]

The arrangement of an audio would still be followed rather closely, by using the audio slider to near 100% , or thats the intent. If you then include prompting in the lyrics or style prompt box to "follow" , enhance, enforce or divert from the audio, then it may go off on a tangent.

As i do more single instrument/single vocal uploads, thats going to be much easier a task, I would say. How well suno can regurgitate but enhance as well my upload as I include multiple elements is up to what it interprets from the sounds presented, I would guess.

[brisk flint]

It probably means, that you should wait a few moments and there was an internal error 😉 - We have learned that Suno likes to make changes to the live system. So, be patient, make yourself a cup of coffee and try again.

[humble verge]

what's the deal with SUNO beginning and/or ending every song with a click?

[toxic kraken]

Hey everyone! I'm having an issue with Suno - when I try to upload an audio clip as source material, the progress bar gets stuck at the very end of the "Uploading Clip" phase and never completes. Has anyone else encountered this same problem?

[woven hull]

Hello, how can I create a persona from another song, not from my own? I tried something and song radio is displayed.

[brisk flint]

As far as I know, the creator of the song would have to create the persona and then set it to “public” before you can access it.

[half dragon]

You can't.
You can only create one from your own song.
You can use Public Persona if someone decided to share.
Ask yourself this: Do you want anyone else create a Persona from your song?

[brisk flint]

I had that last month too, then I just took a break for two days and it went away. Seems to be a glitch every now and then.

[brisk flint]

I just created some new tracks, but I didn't notice that. Do you have a link to an example?

[woven hull]

I'm looking for a deep, rough, and smoky female voice, but Suno doesn't have that. There are only generic voices that don't suit me. I've tried a few things, including prompt, but it didn't work.

[remote zodiac]

Hi,
A lot of music, especially where there aren't many instruments, has a background noise like wind and also when you separate the instrumental from the vocals a lot of noise (wind) remains. Do you have a solution for this?

[hard beacon]

i think i got trolled by the 4.5+ Remaster button with 4.5 outputs 🤨

[sleek stump]

We all did

[polar spade]

If this is getting in the right direction

https://suno.com/s/n7ec51WRDinwfrI5

Then try working around the "cabaret" genre.

[sleek stump]

It appears that they just renamed the 4.5 to 4.5+ remaster

I compared outputs and they are pretty much the same from before the label change.

[hard beacon]

well.. the stereo wight is way different.. also the vocals in 4.5+ have more brilliance

if that is the real 4.5+ remaster, then it's not that good 😉

[sleek stump]

From what I am hearing, its still just 4.5.

[woven hull]

Hey, thanks for the tip. I listened to it. It's exactly what I wanted. I also have a song in the cabaret style: https://suno.com/song/d40056f2-e434-4ff2-aeb7-b10cde765626

[ripe galleon]

not really like prompt talk but

im just curious about how better versions of suno handle audio

i only have access to v3.5 cause im free tier

but i like, wonder what v4 or v4.5 or v4.5+ or watever would handle audios like this

poor recording of me playing the synthesizer

[polar spade]

OK. Thats great. Glad it was a helpful tip.

[lost holly]

Like this: https://suno.com/s/fMuyJmAZeMOsk0T0 but it depends what your going for.

[ripe galleon]

sometimes i feel like it just

takes the instrument and just makes a new song out of it

im playing a chord progression there, thought it takes inspo from it and stuff

[polar spade]

They must be making changes to the server. Carousels are a tip something is changing (or being repaired).

[ripe galleon]

but

thank you helios though

i appreciate your help

[lost holly]

I would argue 17 sec is to short for that

[ripe galleon]

let me send something..

i did remix the audio into a weird guitar song and it did kinda atleast keep some of it

its hit or miss for audios like this, isnt it?

[lost holly]

What do you want me to do with it? Just remaster, Cover?

[ripe galleon]

oh uhm

well truth be told: an extension

use "tension" as the prompt i think

thats what i used for generating my extensions of this audio

[lost holly]

So just extending with style tension? Nothing else?

[ripe galleon]

yeah

thats what i did

[lost holly]

To short to little info for suno, the result is just one of the 2 sec clips.

[ripe galleon]

oh thats.. weird

for me it generated something like this

maybe change to normal v4.5 or v4?

not sure

if well, you do manage to get something out of it with the other models

generate a full song out of it and ill like use my weird brain to compare the audio i generated and yours or something

[lost holly]

Sorry, only get 2-3 sec extensions...

[ripe galleon]

out of both models?

[lost holly]

Out of v4, v4.5 and v4.5+

[ripe galleon]

damn uhhhh uhh

is it too late to ask if another audio file of a more normalish song works or am i being too selfish

(i just dont like wasting someones time for nothing, sorry)

[lost holly]

Give me a moment, want to try one more thing

[ripe galleon]

alright go ahead

[torpid haven]

Can the length be prompted ?

[ripe galleon]

not sure..

[torpid haven]

And is Exclude just the negative prompt like with other diffusion models ?

[ripe galleon]

yeah think so

[lost holly]

A: https://suno.com/s/kkPNfVV2QWYzi3ex B: https://suno.com/s/hbrVZCUErTiX6LLg which one or none?

[ripe galleon]

put in exclude whatever you dont want in your song

[ripe galleon]

either one works

thank you

you could send me both of them as audio files

id have more samples lolo

[torpid haven]

Yeah that's how I'm using it but wasn't sure if it's maybe preprocessed because the example is for removing genres

[ripe galleon]

that im not sure about..

surely you can like exclude stuff like female vocals

or certain instruments

[lost holly]

MP3 or wav? For wav i would need a mail^^ feel free to pm me mail, not here

[ripe galleon]

by need a mail like..

email?

just wondering

[lost holly]

Aye, they each 35 mb, to big for discord

[ripe galleon]

OH jesus

i see

send me as an mp3 file then please

[fathom siren]

I meant what settings for the file (type, bitrate, etc..) I assume If I upload it in the format it prefers to store them in, it would suffer less changes during the upload process

[half dragon]

Ah okay. Suno don't specify the specific but always keep your audio at highest setting possible for quality.

[humble verge]

why can't i make a persona with an uploaded vocal stem?

[half dragon]

Maybe they prevent if one wants to upload audio of known famous artist then make a Pesona from it?

[humble verge]

it's a persona from another song i made

it works in one menu but not the others. It's weird.

[sleek stump]

Nope, Suno ignores any attempt at length.

[sleek stump]

Pretty much, yes.

[remote seal]

So if you guys want to make an album with a persona but want to avoid repeating the melody or similarity what do you guys do. And voice wise what techniques do you guys use to try to keep it close to the persona.

[spark brook]

did you ever get this falsetto to work?

i can only get a falsetto chorus, but seems inconsistent

https://suno.com/s/IlyoNG5JoVERwk0y

also this is just a test song so please ignore lyrics lol

[spark brook]

ok i think i have it figured out: https://suno.com/s/pGW5IxnDDL05PRCP

this is all falsetto

process is something like this: create falsetto chorus
delete everything except chorus
create persona from chorus only
re-add all lyrics
i extended up appending lyrics to "chorus" so the whole song are basically choruses
remix - change speed
0.9x, keep pitch

i ended up rewriting all parts of the verse that I initially used because it's possible the persona still had remnants of how it wanted to sing the initial verse

[wispy laurel]

Make a persona. Use that persona to make another song, with the structure and style prompt different enough that it can't just copy directly from the persona (or, use the persona to cover a song that's very different)
Objective is to create a second song that sounds similar enough to the persona that it sounds like the same artist. So now you have two songs, with some style elements in common but clearly not the same track.
Put those two songs into a playlist. Use that playlist as inspiration to generate a third song, again with a different structure to the existing ones. When you get a good one, that sounds distinct but still like the same artist, add it to the inspiration playlist.
In effect this works kind of like a persona made from several different songs at once, it's not exactly the same but the result is similar. Fairly quickly you should have a pretty recognisable style emerging, with common elements across different songs, and when you use it to create new ones it kind of mix-and-matches the different songs that went into it.
An advantage of this is that by deciding what does or doesn't get added to the inspiration playlist, you can 'steer' the style in a direction that you want, while still giving it room to "evolve" by generating new stuff
Drawback is, you can't currently use inspiration when covering an existing song. But in that case you can possibly work around it by using a persona based on whichever one of your current songs is closest to how you want the cover to sound
It's not quick or guaranteed to always work but this definitelydoes work to develop a distinct, recognisable style if you put some time (and probably quite a lot of credits) into it

[wispy laurel]

Alongside that, make sure that your style prompt explicitly describes whichever aspects of the existing style are more important
Like for instance, when I do this the main things I'm looking to put into the new song are distinct recognisable voice, and the combination of mostly classical instruments with rock rhythms and structures. So I have an inspiration playlist that I use but also make sure the style prompt includes "dark, melodic folk-rock - drums, piano, upright bass, violin, electric guitar" and "low, soft, sexy rap, with a cool, relaxed vocal style and a clean, smooth voice"
Because after experimenting a bit, it turns out that Suno interprets that as meaning what I want it to mean (the voice in fact isn't particularly clean or smooth and not really rap either but Suno thinks it is so it works out 😂 )
Personas or inspiration can help if you use them right but an equally important part is understanding how to manipulate the AI effectively

[ripe galleon]

if you put your voice in the intro of an uploaded song which has no vocals otherwise, would suno pick it up and if you add lyrics to the song, make your voice the singer?

[spark brook]

can you try it and let us know? (serious, i am curious)

[ripe galleon]

oh

ill try it

[ripe galleon]

could we DM?

i got somethin

[spark brook]

heya

[polar spade]

You mean upload a song containing your voice only the intro ? Or as an instrumentless vocal intro before any music begins as your upload?

Either way, this is a common way (since 4.0) for suno to "pick up your voice" when you then Cover the upload. Granted 4.5 follows our voice characteristics better (maybe I should say closer and more frequently) than predecessors rather than just give a Suno-stylized version.

[polar spade]

For just Suno tools (no DAW) In simplest form: create a song offline (not sure of your varied talents, but perfect singing and playing prowess is not quite needed). Record. Upload. Cover with a persona of your "artist" voice.

Or ....create or use a song in Suno. Cover with your artist persona.

Thats been the go-to since personas were introduced.

[ripe galleon]

instrumentless vocal

basically just a raw recording of a vocal

this also works for v3.5 suno

sometimes it probably just sees the recording as like a oneshot sample sometimes

[gusty pendant]

Hello to all the pros, Is suno using some kind of hidden equalizer when you use website/suno mobile app to play? I feel there is difference when I listen to the song on the web/app and when I download it in wav and play it directly with flat equalizer

[polar spade]

To note: Once they brought forward sliders, ite not easy to get that sweet spot sometimes. It seems different eith each vocalization presented. 4.0 had often hit a "cloned" version in a random moment, so its still good to keep in the mix of generations to try.

[ripe galleon]

forward sliders?

[polar spade]

Well, I meant they ( they developers) had their own idea of what the sweet spot for ALL users and genres, and so that was the default, before sliders.

Now with sliders made available (brought forward from behind the curtain), that default setting is gauged at the 50% mark for Audio and Style sliders and we get to change ourselves.

Since that point though, for me at least, the default has not given the same results as when sliders didnt exist. So I have different settings for different personas and genres and cloning my own voice ,written on a notepad to keep up with my own "default" settings.

[ripe galleon]

oh

i wish i had access to them lolo

[polar spade]

Lyrics, genres, styles, they will affect vocals including personas vocals. So things always need adjusting/tweaking.

[ripe galleon]

i see

[gusty pendant]

yep you allways have to tweek it more, in my case it usualy just dont follow the decibels of anything around it

[polar spade]

After awhile, with so many tools, its more difficult to answer "How do I.....?" As either there are 12 ways to do something, or even if one way works for me, its usually because I spent 3 pages of outputs tweaking it, so its not then a one-size-fits-all answer.

[ripe galleon]

man

[cyan jewel]

I'm running into a lot of issues where songs are 7:59 and cut off. What do I need to change to be able to keep a song close to 8 minutes and it actually end around there xd

[gusty pendant]

I think your best option is just use extend from the part that it is still good, what kind of songs you make so ti takes them 8 minutes?

[cyan jewel]

symphonic metal, doom/death, baroque, dark cabaret, gothic stuff in general, some anisong stuff like kajiura/kalafina/akiko shikata-inspired stuff runs pretty long

[gusty pendant]

Yeah your best bet is extend then

[cyan jewel]

https://suno.com/song/4fa85c13-fc6a-4414-974a-e45388bb4086 even stuff like this can run pretty heckin' long

stupid bot im 35 >:C

let me say a swear!

Where's the "i'm old let me say a swear occasionally" role >:C

[gusty pendant]

The picture on the song tho 😏

if you looking to extend it do it around 7:48 mark when the beat drops down a little

[radiant breach]

Cover feature is still broken......

[cold bramble]

Is anyone having an issue when you replace a section, the vocal sometimes is changed and the music gets louder? anyone know how to fix it? it completely ruins the song. this in v4.5+. thank you

[lime peak]

Any advice on inputing prompts regarding song length? Especially with audio sources that are only 1 minute long, suno sometimes hallucinate and creates a 7+ minute song with unnecessary stuff, when it's supposed to be an accurate cover

A few creative liberties here and there are not an issue, but then the song length shouldn't deviate too much from the original

[jagged narwhal]

😄 why it says to my original work?

[half dragon]

I uploaded an audio from an old Suno song I downloaded before and got that same error.
Someone told me to rename the file and that did the trick.

[proper bluff]

Guys. Am I the only one or should suno fix their vocals, like everytime they sing it sounds so unrealistic

[jagged narwhal]

I changed but I guess it tryn to match from lyrics lol

[brisk flint]

What do you mean by unrealistic? The autotune effect? The screaming that's currently annoying us all? Or that the notes are sung for so long that no one on this planet has the lung capacity to do it? 😉

[proper bluff]

The auto tune effect

[hollow hare]

If you have your work registered to a system like YouTube Content ID, that is the reason. You may be able to get Support to upload it to your account if you can provide proof you own it #support-guide

Though personally I wouldn't

Because you give Suno and their partners free use over your song. Will they use it? Extremely unlikely but they still have that option

[proper bluff]

Guys not gonna lie, who wants veo 3 to be free

[half dragon]

If your lyric aren't yours then yeah...

[proper bluff]

When will v4 be free

[hollow hare]

I find being more descriptive about the voice tends to help. Not so much like alto, etc. But like "soothing, soft whisper [gender]", "angry loud [gender]", etc.

[jagged narwhal]

and I this not normal

[proper bluff]

lol

[brisk flint]

To be honest, I never have. After several thousand generations. I think that could be due to a) the genre and b) your prompt 😉

[proper bluff]

Sorry for laughing but. A song for 2 seconds? Maybe you typed in 1 word like pizza and then it stopped.

[proper bluff]

Did you type in 1 word?

[proper bluff]

I think it’s not an auto tune effect, but it’s unrealistic. Udio has a very good ai vocalisation. Suno has a bad ai vocalisation.

[brisk flint]

No, occasionally Suno has a glitch and generates songs that are between one and about 10-15 seconds long. In the best case scenario, subscribers are credited back immediately.

[jagged narwhal]

no lol it has 5x verse 2x chorus

cant create lol

[proper bluff]

It’s a glitch, did you do it on the editor option. Or did you just do the basic one

[brisk flint]

I can't confirm that either, and I've been using Suno since almost the beginning. But of course, it's subjective what sounds realistic and what sounds less realistic to each individual.

[proper bluff]

Me personally I don’t use editor because of these type of problems. The best thing you can do is use the basic option.

[proper bluff]

orrr you can switch accounts and try again

[polar spade]

With many users creating multiple free accounts just to keep getting "20" more 4.0 outputs, it would appear that its (4.0) still a viable and sought after version. I use it often. So it make sense to keep it at the (some sort of) paid level for quite some time.

[quasi steeple]

Can those of you who know classical music tell me how you think Suno did on this one?
https://suno.com/s/bpnS18P9480HYBia

[knotty stag]

Great music, sounds like a real classical piece from the past.

[undone nacelle]

I reported my song as it doesn't follow the lyrics but what is my next step to get it to follow the lyrics? I really like vocals and sound but I need it to follow the lyrics.

while playing my song it pronounces some of the words correctly then in the next verse it pronounces it wrong when it is spelt the same. I just want it to say it properly

[quasi steeple]

Thanks. Look at the "lyrics" and prompt - I was trying to get it to follow some highly specific classical structure, but I'm not enough of a "real" composer to know if it suceeded.

[torpid haven]

Maybe stem separation and mix the correct vocals together

[undone nacelle]

how to get to those options, I got this app today lol

[torpid haven]

Use the web application in the browser on your computer

It's 50 credits but there are also free open source stem sep apps for the oc

[wispy laurel]

The mobile app frankly sucks
Even if you're on mobile, the website is a WAY better option with a lot more control

[torpid haven]

https://ultimatevocalremover.com/

You can also use desktop view on your phone browser

[autumn rock]

Seriously you cant prompt "SATB choir" ?
The censorship stuff is stupid. SATB is a scoring of compositions for choirs like of four voice types: soprano, alto, tenor and bass. But SUno thinks its some copyrighted band or something...

[neat nexus]

SATB is an acronym. Unpack it.

Try prompting as Soprano, Alto, Tenor, and Bass Choir and see if that works for you.

But yeah, that is frustrating. I know because I ran into it too.

[neat nexus]

You could also try continuing from before the second verse and see if another attempt gets it correct. Or failing that when you prompt use phonetics to deliberatlely force the pronunciation you seek.

Other people gave good advice too though.

[cyan jewel]

is there a list of meta tags that demonstrably work and are not just placebo?

[half dragon]

Not officially but you can search for one user compiled list somewhere.

[cyan jewel]

i have a few lists but they have a whole buncha stuff that like, clearly don't work which was why i was curious if there was like, a list of verified working meta tags and not just placeboslop

sd has a lot of the same issue with quality and media tags too where ppl use tags that just straight up don't do anything and just result in token bloat

also has anyone really TESTED prose vs tag use for styles?

natural language has been EXTREMELY hit or miss for me in sd so im skepticle of its usefulness here at well.

[ebon sand]

you'd basically have to curate one for yourself which behave well with the style you create, my metatags likely won't work for you because i have a very different environment and song style for example

it's not like a switch you turn off and on but it gets interpreted into a coherent result

[cyan jewel]

so far i largely just haven't had much luck with any meta tags working

a bunch of songs i've genned with meta tags straight up ignore them

[ebon sand]

i mostly just use metatags to define who sings what and rarely a break, so [break] has been working well for me and [instrumental interlude], they have been pretty stable

if i want to force something specific i do that in the editor

[wispy laurel]

I use natural language almost exclusively, and I find that I can usually get it to do most things that way with a bit of fiddling and experimenting.
Once you figure out what the AI thinks that certain things mean, and how it's likely to interpret things, you can do a hell of a lot with natural language
What I cannot do is extremely fine, precise control over specific aspects. But from what I hear, it's hard to achieve that with tags too, depending on what you're trying to control, so that might be a function of the model itself rather than the language used

[neat nexus]

The problem is both the placebo effect and confirmation bias. I can give you a list that I use and have great success with. Consistently even. And you'll probably discover that they don't work for you.

The issue here is that Suno is not a wordprocessor for sound. It'a a generative AI. We can suggest things but ultimately it is Suno who decides what output its going to do. The best we can do is narrow in, use the tools (extend from an earlier time is way more powerful than most people give it credit for!) and nudge it closer and closer to what we're looking for.

[graceful condor]

"we can suggest things but ultimately it is Suno who decides what output its going to do." pretty much this; the user controls things like lyrics, settings, and styles/themes or they can reference a sample. But it's a delicate dance between what the AI thinks words mean, and what you wanted them to mean.

[graceful condor]

I have found that metatags exclusively as an intstrumental guide it far better than any initial prompts ever could

[graceful condor]

yea mobile app is straight up trash unfortunately 😭

[neat nexus]

The AI will get smarter over time. And we'll get better at prompting it.

But I tend to think that if they tighten it down too much you'll lose the magic that happens between the beats. This isn't anything official, just an observation at what happens when you adjust your sliders too hard in either direction. It will sometimes produce something good. But most of the time it makes something super short and not that great sounding.

That's why it's music. Because there's randomness to the mix.

[ebon sand]

it rarely does anything in the music what i want from it from metatags so i often am content with what it comes up with without my doing on that end haha

[graceful condor]

yea when it works with the metatags it's great when it doesn't...well then I have to go back to the drawing board.

[cyan jewel]

the thing with genai is you CAN get consistent results if you know which tags the model is ACTUALLY TRAINED ON

[dense moat]

I think Suno did a great job on rendering this piece and your did a great job conducting Suno to do so. Your prompt is very straight forward and uses a know classical music theory structure which is truly remarkable. Suno seems to not only understand classical music but can follow these sorts of prompts extremely well including key changes, invoking mood, chord progressions, and time signature changes you might experiment with these in your prompt to see what results you get.

[cyan jewel]

what meta tags are you using

i found out the hard way it ignores vocal technique entirely.

suno barely knows what a fry scream is, let alone a false chord, or a gutteral, or an inhaled vs an exhaled

it thinks alto/contralto are MALE voices. i have to generate the same song like 30 times to get it to stop generating male contraltos

even with female vocals selected and males excluded

in general getting different voices is hard

even when sung in different languages the female voice is very samey even when i do entirely different genres, lyrics, languages, etc

[ebon sand]

maybe persona or inspo helps

i don't give any vocal instructions because they are in the persona

[cyan jewel]

i need a lot of vocal techniques

and what i want might change dramatically

[ebon sand]

then a varied persona might help a lot

[cyan jewel]

how do i get a varied persona

and how do i get every vocalist not to just sound like christina vee?

and the male vocalist always sounds like Jørn Lande

which, Jørn Lande is an excellent vocalist, but i dont need literally every song to sound like Jørn Lande lol

[ebon sand]

create it, curate it, meka has a guide here
#1358122820273569914 message

[cyan jewel]

oh yeah i followed this guide

it did nooooot help in the slightest

most of these metatags straight up do NOT work

[ebon sand]

i didn't follow it but i have around 400 personas, just save when you like it and work from there

[cyan jewel]

i have like, 2, and i never use them because they just make the songs sound worse lol

[ebon sand]

i don't know what solution you want then when nothing works

[cyan jewel]

to find out what DOES work

obviously?

[ebon sand]

these are the things that work for us, if they don't work for you you might have to find it out and report back here when you do

[cyan jewel]

its why i kept asking if there were like actually verified working metatags because so far it just feels like placebo

[cyan jewel]

well whats working for you

can you provide examples?

[ebon sand]

creating personas and using them

[cyan jewel]

ok

can you provide an example?

[ebon sand]

what exactly do you want to hear?

[cyan jewel]

whatever you think is your best working product.

and explain why

[ebon sand]

hm i think my best one is my old attic persona because it pronounces old attic correctly

[cyan jewel]

https://suno.com/s/ZWF8GPpHsFAPYMGL

So here for example none of the meta tags do anything here

I didn't prompt for male vocals but got male vocals

i've fone a bunch of iterations of this song specifically to see if i can get the meta tags to do what i want them to but its 100% rng ngl

[ebon sand]

i only use very few metatags, 4 in total and sometimes none at all, i don't know if they work in such excess

[cyan jewel]

Post which ones you use and actually find success with

And why you use them

[ebon sand]

[break]
[instrumental interlude]
[hierophant]
[cultists]

oh and
[spoken]
[whisper]

that's all i use i think

break and instrumental interlude are selfexplanatory, hierophant and cultists are my vocalists and spoken and whisper should also be clear

[cyan jewel]

yeah spoken/whisper never work for me

i'cve tried a bunch of times but they do nothing

[ebon sand]

my persona has examples of that in it, but i remember it also works without it

[pseudo blade]

U tried ASMR?

[cyan jewel]

about to

nope

does nothing

[whispered]

[asmr]

https://suno.com/song/b0bf745a-d783-4043-b920-8bd5e3eb4bb9

[ebon sand]

do you also change your prompt to support that

[cyan jewel]

change it in the prompt?

which?

[ebon sand]

the styles

[cyan jewel]

no. how would it look like in styles

[ebon sand]

you describe how you want the song to sound like

[cyan jewel]

if i put [asmr] in styles how does it know to only use [asmr] for a specific passage?

well i dont want to put [asmr] in the style i only want it to be in specific passages like a bridge for example

how would i do that?

A gothic symphonic tapestry woven from the shadowed threads of dark cabaret, doom metal, and melodic death metal, Dueling female vocals echo through cathedralic orchestration, underscored by brooding string sections, minor-key piano, and baroque choirs, Cinematic swells rise against thunderous double bass drumming and towering electric guitar riffs, invoking haunted elegance and sorrowful grandeur, Drawing from modal folk mysticism with harp, kalimba, and layered invented-language vocals, and hypnotic string ostinatos laced with electronic undercurrents, the track breathes with myth and tension, Melancholic beauty merges with introspective heaviness, binding shifting time signatures, neoclassical textures, and emotional crescendos into one spectral, ritualistic suite, Ethereal yet apocalyptic, this soundscape summons forgotten cloisters, spectral circus tents, and moonlit ruins

what would i update here

[ebon sand]

i'd probably delete all and write it again with my own words, this sounds suspiciously like an AI tried to be fancy

[cyan jewel]

what would "your own words" look like

and yes this is auralith

it took my words and converted it to prose

[ebon sand]

Rhythmic Percussion, Acoustic, Strummed Guitar, Upright Bass, Dark Desert, fast-paced, Ritualistic, Sinister, Menacing, Hierophant, Male Cultists Chant, Call and Response Litany, Clean Sound
I basically just type like this, i think only v4.5 needed full sentences, haven't noticed v4.5+ needs prose as much

[cyan jewel]

ok let me be more specific

[cyan jewel]

how would you rewrite

A gothic symphonic tapestry woven from the shadowed threads of dark cabaret, doom metal, and melodic death metal, Dueling female vocals echo through cathedralic orchestration, underscored by brooding string sections, minor-key piano, and baroque choirs, Cinematic swells rise against thunderous double bass drumming and towering electric guitar riffs, invoking haunted elegance and sorrowful grandeur, Drawing from modal folk mysticism with harp, kalimba, and layered invented-language vocals, and hypnotic string ostinatos laced with electronic undercurrents, the track breathes with myth and tension, Melancholic beauty merges with introspective heaviness, binding shifting time signatures, neoclassical textures, and emotional crescendos into one spectral, ritualistic suite, Ethereal yet apocalyptic, this soundscape summons forgotten cloisters, spectral circus tents, and moonlit ruins

[ebon sand]

i cannot know what you're looking for, it's your song, i don't know what a gothic symphonic woven tapestry woven from the shadowed threads of dark carbaret is, i'd probably type "gothic metal" or whatever you want

[cyan jewel]

well its not just gothic metal i want

[ebon sand]

you can still add things until it sounds the way you want, i have bluegrass in my prompt because it adds a twangy guitar, no other reason

[cyan jewel]

facepalms

ok yeah im sorry for wasting your time. i think there's a communication barrier here. sorry.

[ebon sand]

i don't think there is, but no problem

[cyan jewel]

you're not understanding what im asking even when i repeatedly rephrase. so there's clearly a disconnect somewhere

[ebon sand]

you want me to rewrite your prompt

[cyan jewel]

either im not communicating clearly enough, or you're misinterpreting

well you said you wouldn't write it the way auralith did

so i posted my prompt

and asked you to show how you'd write it

which you didnt seem to understand.

[ebon sand]

yes i told you i'd start with gothic metal and then add whatever you want

i cant write your prompt

i don't want that music, you know what you want

[cyan jewel]

...you can you just dont want to

¯_(ツ)_/¯

like i said, sorry to bother.

back to the drawing board.

[ebon sand]

that's a rude accusation

but yes let's cease this

[cyan jewel]

its not rude or an accisation, it's just the truth.

whether you interpret that as rude or not is your problem.

[ebon sand]

I hope you find what you seek but please don't interact with me further

[honest compass]

Is gen or remixing not working at the minute?

[eternal sigil]

Every single one of my covers on 4.5+ has been approximately the same length (+/- 5 seconds) and cuts off abruptly for the past 3 days (600 credits).

Who do I talk to about that?

[flint haven]

https://discord.com/channels/1069381916492562582/1404828447695569037 @eternal sigil

[eternal sigil]

thanks @flint haven but this has just started appearing for me after a year of paid generations (idk probably 25,000 credits)... And also 4.5+ has been the best for me thus far at actually finishing the song.

Maybe it's just the nature of the lyrics

Still seems like something changed

The first 2 gens of this cover were 5:06 and 4:34 but every single one after (58) have been 4:40 to 4:45...

AND both songs in the gen are ALWAYS the same length now

with very little variety... before 3 days ago both gens were stylistically related, however, distinct (and with distinct durations)

[cyan jewel]

does anyone know how to get suno to do basslines that aren't just single note chugs? or guitar riffs that are a bit more... not just alt strummed single notes/powerchords??

[opal orbit]

I have had success by inputing things like [Dynamic guitar solo], [Electric guitar song motif], [melodic and powerful guitar melody]

[wispy laurel]

Keywords for this seem to be "melodic", "complex" or "sophisticated". Although like everything else, mileage may vary depending on how Suno interprets those terms in combination with whatever else you have there

[cyan jewel]

yeah cuz i wanna get like, djent/metalcore/melodeath/prog metal kinda riffs but its just been chugs and palm muted single chord or single notes like we're 2000s era power metal again

the goal is to have some sorta crunchy sparkly addictive guitar hook during versus/bridges

and then the chorus push the vocals to the front

i wanna brighten the outputs too but no luck

post processing in fls only does so much

also have any of you had luck with contralto vocals?

i keep getting male vocals instead of deep fem vox

[opal orbit]

You would need to take out the stems individually to brighten up the guitars, Suno doesn't always filter the fuzz well. With the guitars, do you have an example of what your getting vs. what you want?

[cyan jewel]

yeah

grabbing two examples: an output, and a reference

[cyan jewel]

oh as for stems how do you get them to not be absolutely nuked when separating stems?

the only stems that ever seem to come out well are drums and vox

[opal orbit]

lol no way I know of. Can't avoid getting them messed up a bit. I mix-n-match stems when I'm putting my songs back together, and some go through alot of compression and filtering

[cyan jewel]

the guitars always ALWAYS get butchered

i gotcha

[opal orbit]

Yeah, suno combines synth and guitar alot.

[cyan jewel]

https://suno.com/song/b1c4283a-3596-48fa-b7f5-892b645dfcb0

so here's what i currently have

https://www.youtube.com/watch?v=L_-pAcDpGQc
https://www.youtube.com/watch?v=JD_jzKoEOUE

some examples of what I'm tryna do

[opal orbit]

It sounds like in yours, it has its staple guitar motif, and then more or less just treats it as the 'beat' for the rest of the song (hence the limited chugs and notes).
In your examples, I'm hearing a lot of varying dynamic guitar riffs and melodies that play behind the lyrics, but carry their own rhythm that are (comparatively) different from the 'beat of the words'. Which is something Suno struggles with. Just to clarify, is that specifically what your having trouble recreating?

[autumn rock]

Orchestral has some great promise... took one of my kids songs and applied some music theory https://suno.com/s/tL6RFsv2G4QIrcRe

[cyan jewel]

orchestral and symphonic stuff works pretty well rn.

i've been getting some decent yuki kajiura-style tracks

[cyan jewel]

thats part of it. suno will do a banger opening riff then the rest of the song it just does powermetal style palm mutes as texture rather than like

any actual hooks, or riffs,

i had been struggling with tone

but perplexity MIGHT have solved that

im getting surprisingly good results from this prompt:

{Tempo: 170–200 bpm; T-Signature: 6/8 with double/half-time shifts; Rhythm: syncopated modern metal chugs & EDM grooves}, {Era: 2010s–2020s; Genre: Prog Metal, Melodic Power Metal, Electro-Metal; Subgenre: Modern Prog, Symphonic EDM Metal, Djent Pop}, {Instruments: [Guitar: Ibanez RG/ESP humbuckers, dual-tracked, mid-cut, Tube Screamer], [Bass: Ibanez BTB/Warwick, parallel distortion, EQ], [Drums: Superior Drummer 3 preset, Tama/Dynasty shells, punchy], [Keys: Serum2, Omnisphere, Sylenth1, Nexus, Supersaw, orchestral pads, trance arps], [Vocals: F/M, clean belt, harmonized, vocoder fx]}, {Mood: Energetic, dramatic, epic, uplifting}, {Mix: Wide guitars, parallel compression drums/bass, side-chained EDM keys, crisp vocals}, {Specifics: Gritty bass intro, chugging guitars, trance synth bridges, double-kick chorus, EDM arps breakdown, massive harmonized chorus, cinematic fadeout}.

This was obtained by having perplexity analyze specific sonic qualities i was chasing

the bass of Tesseract's Altere'd State abum

The drums of Tesseract, Dynazty, 

the keyboards of Metalite's A Virtual World and Amaranthe's Transhuman```

oh lol even with the typos

[wispy laurel]

Add "[melodic electric guitar hooks]" into your lyrics at the relevant sections and see if that does anything
I find that, or some close variation, usually will give me what I think you're trying to do. Although like everything, Suno gonna Suno so there are never any guarantees

[opal orbit]

Nice. I've done that with ChatGPT when i'm trying to figure out a specific sound.

I think the main issue Suno is having with your song is that it wants to keep it in the same motif theme it starts off with. Reintroducing a new melody conflicts with whatever rhythm it sets for your lyrics, and it naturally will mute it down to overvoid overpowering it because they overlap.

Suno won't naturally introduce conflicting instrumental motifs and its hard to force.

I would try doing section replacements in the editor to get variety and depth you want, (don't worry about it perfectly connecting yet). Then cover what you end up with.
Suno should stay true to the songs varying motifs and you can get more actual hooks and riffs. (You can also do this by piecing together a monstrosity in your DAW then letting Suno make it sound good with the cover feature)
Also try covering it as just an instrumental to minimize the risk of it muting out your guitars in place of vocals.

You can always add vocals later once you've gotten the sound you want.

[autumn rock]

Just to be clear, I don't use "instrumental" switch > I use the lyrics [instrumental] and I get promising results treating the lyrics for [ instructional ] orchestral composition. Either way i import into Logic and get to work.

[woven hull]

Hello, thanks again for the tip you gave me. This voice would have been perfect for me. I tried everything to incorporate the voice into my song, but it didn't work; it always ended up being a high-pitched voice. Do you have any tips on how I can do this to make this deep voice persona? The song "Moulen Rouge" is an older song of mine. I would be happy if you could help me

[manic bay]

Hey, thanks for sharing that with me. I get what you mean it’s tricky when the voice keeps coming out high-pitched instead of deep. I can definitely help you figure it out. Sometimes it’s about pitch control or layering techniques. If you’d like, I can show you a couple ways to shape the voice so it matches that deeper persona you’re aiming for. And I’d love to hear more about your song ‘Moulen Rouge’ too it sounds interesting

[inland thistle]

so in exclude styles, do i just type "vocals" to prevent singing or do i have to actually use "-vocals" or "no vocals"?

[tulip jay]

hej, newbie here, is there a way to force SUNO to produce 7/4 time signatures? 🙂

[inland thistle]

yep!

[tulip jay]

i tried adding lines like "[TIME SIGNATURE: 7/4 (grouping 4+3)]" with no luck

[inland thistle]

hmm trying to find a reference for you

[tulip jay]

🥹

[inland thistle]

here we go - i was trying to find Dragnier's version for you which was better - but this sort of gives you an idea - oh i can't link suno songs in here almost fell for the same trap again 🙁 check general-suno mate where they dont seem to have a problem if you link a track - i will tag you

[tulip jay]

I've checked out tracks you linked, cool! but not 7/4 ;D thanks a lot for your effort tho 🙂

i had this vision of creating a house tune but with kind of Venetian Snares feel to it 😉

[inland thistle]

not 7/4 specifically it does 1/3rd but the prompt mechanics are there. you just need to turn it into 7/4 with a little effort

something like

  [Time Signature: 7/4 — Heptapulse Recursion 2.4]

or

  [808 Sub Pulse Drift — 4:7 Polyrhythmic Anchor 2.1]

etc something similar

with the grouping, you can try specifying what you want to group

eg: [Rotating Groupings: (A) 3+2+2 → (B) 2+3+2 → (C) 2+2+3 → (D) 4+3 → (E) 5+2] or something like that

[tulip jay]

the structure of prompts looks awesome and im already tweaking it to see what i can get from it 😉

[inland thistle]

a specific ghost note or instrument or synth lead or 808 etc

[tulip jay]

just my lazyass mind had this vision of getting plain answers right here right now ;p 😉

[inland thistle]

don't thank me, thank Dragnier (The Cybard Project)... He is a legit Suno Prompting Demiurge

[tulip jay]

of whom I wouldnt have had idea if you didnt mention him 🙂

[inland thistle]

also, took me forever to realise you can use variables and loops and such so you can define things and then later modify them by say incrementing or stacking etc

Suno understands programming 😄

which makes it easier if you want to have different rhythm loops/sections like a loopstation/sequencer

[polar spade]

Not sure much i can help with, no.

But only to say, it helped more as I added "grunge, deathmetal, and creaky" to my prompts. Vocal fry used to work decently with 3.5, But I could have found others prompts as well, like Gregorian Chant, or Mbube, and still might have gotten some help with those as well. As I said I ran through about three pages of quick prompting to come up with some thing close. Its always like that to me to find a good recipe that fits.

Here were the most recent before sending you the "54 second one that sounded decent.

Oh oh wait.... and when I got something I liked , I then changed tempo speed to about 87% to lower the pitch. Dont forget, you can do that.

[lyric kestrel]

hi guys! i'm wondering if anyone knows a prompt for this song

https://youtu.be/5A7g3edXdEU?si=fhkDdE9sg5b9870V

[storm pawn]

Have you guys managed to create a song in a similar tone but with slightly different instrumentals between verses? I'm trying to make a hip hop song with latin style/vibe (reggaeton) in the first verse, and then after chorus to switch to japan/oriental style, but it keeps making the whole song oriental, regardless if I mention the styles in Style or in Lyrics (as instrumental before verses, or next to verses).

[sleek cobalt]

About a week I was able to create songs with "french R", (uvulvar R, back throat R), as this is my dialect from Bergen, Norway. But now, whatever I do I get rolling R's. Has there been a change? Is there anything I can do to get the "french R" back?

[hoary marsh]

So I have a question on making a Cover. I am making a cover of my own song. I wanted to clean up my prompts and add a small new section. The issue I am seeing is that, while the original song was already 5:30 long, every cover returns a 4:58 version. Each and every time. Any idea what is wrong, or what I'm doing wrong?? I am on v4.5+

[half dragon]

You're not doing it wrong - It's been known Cover would not cover the entire length- you'd have to Extend to finish it.

[hoary marsh]

Ug, so then it's broken by all accounts and unusable. I get that extending it is a stop gap, but at its core, it should cover it completely. Thanks for the info. I'll look for a different solution.

[half dragon]

It's been like that for that feature - somewhere on here there was an explanation.
It might look like it's broken and it may change or improve but it is what it is with the current system.

[wispy laurel]

It's not unusable, it just requires one extra step
The AI has a limited amount of 'space' available to work in for each song. If you're using a cover, or a persona, or extending an existing song, that data occupies some of the space, meaning it can't generate as much new information in a single step. Any cover or any song based on a full-song-length persona will be capped at just under 5 minutes but, Extend works perfectly fine to generate the rest of it
I have 9-minute songs using multiple extensions, I've heard of people going over 20 minutes. You just can't do it all in one step.

[livid trout]

personas seem to have a tendazncy to start the song with the same melodic intro that was used in the song they were created from. how do you tell the algorythm not to do that ?

[hard beacon]

that's why i am not using the persona function 🙂 But maybe i am just to dumb to use it correctly

[livid trout]

in case the persone tends to use the instrumentation and melodic similarities, then I wish you could specify to only use the voice

[hard beacon]

i am trying to use the "re-use function" at the moment... looks like the better way to use vocal styles 😉

[ripe galleon]

can suno do asmr

like not as in a song

[wispy laurel]

I start the style instruction with

[Use the persona's voice and vocal style]
[Ignore melodic elements from the persona]

And this seems to work
This may or may not be confirmation bias, I don't know if the AI actually really recognises/understands those instructions but I do find that using them seems to improve the odds of getting the same voice and a different melody

A more involved approach is, don't make a persona using a whole song if all you want is specific aspects. Crop the song down to just the section that most clearly represents what you're trying to use, and if the voice is the only thing you want to replicate then use the vocal stem of the song rather than the whole thing. You can use a short segment of a song to create a persona that works perfectly fine, retains the things that you most want to keep and doesn't have much melody or an entire song structure attached to it

[hard beacon]

and i have good results by using the cover function but with "style influence" at 0% 🙂

[polar spade]

If it was just a week ago, Id say there hasn't been a change. Suno is not consistent with prompts. But are you using a persona made of a vocal with that quality. Id think, that would be more reliable or consistent.

[polar spade]

Thats sound a fun experiment. Upload a bunch of ASMR type of noises and see what it emulates. You may need to throw in lots of. xplosions exclusions to avoid any music at all.

[manic bay]

That makes sense. Starting with just the voice and style is a smart move it keeps it focused. Cropping the song down to only the part you need sounds like a good trick too, I might try that. I can help out with persona building if you want, sometimes just a clean vocal stem makes things way more consistent.

[polar spade]

You may want to tag to whom you're response is? So they will get pinged.

[manic bay]

Ah got it, thanks for the tip! I’ll make sure to tag next time so the right person gets the ping 👍

[digital kraken]

Hello everyone.
Do you know if it's possible to add radio-like sounds, like Manu Chao's songs? A kind of sound that isn't an instrument, used as a background loop?

[polar spade]

Did you mean the vocals sounding like theyre coming over a radio, or the background sounding like a radio is playing?

[manic bay]

Yeah, I can help you out with that. You can layer in radio-style sounds or ambient loops under the track, kind of like Manu Chao does. It’s just about finding or creating the right samples and blending them in so they feel natural in the mix

[polar spade]

Technically, it would be possible as Alina mentions. As per Suno TOS, though, we cant upload samples, for the most part to add into the mix.

[digital kraken]

Yes, I saw that you can't add a sample (which is a shame). That's why I thought about mentioning a South American radio sound in the styles prompt, like adding a guitar or drums.

[manic bay]

Yeah true, samples aren’t supported directly. But you can still get that radio vibe by working it into the style prompt, like you said. I can help you shape it so it feels like that South American radio texture layered in the background without needing an actual sample.

[digital kraken]

With pleasure. I'm curious to try this.

[manic bay]

Hit me up

[manic bay]

With pleasure 🙌 I’d be glad to help you out with this. I’m also curious to try it and see how close we can get to that Manu Chao radio vibe.

[granite pawn]

Is Suno down

[opal carbon]

it was working a few mins ago for me but maybe it changed

[prime solar]

hi everyone, anyone knows if an api will be available? there are some options but not very reliable

[wet minnow]

no sorry. at the moment there is no api and i have not heard of any release date yet.

[half dragon]

It was also advised NOT to use any sites that offer Suno API - it'll get your account banned.

[flint haven]

Beware: there is NO official Suno API

🚨 Suno does not have a public API. Any claims otherwise are unauthorized.

[quasi steeple]

where can I listen to his stuff / see his prompts

[inland thistle]

https://suno.com/@dragnier

[ripe galleon]

explosions??

[ripe galleon]

is it possible to do a formant shift effect like this with suno?

[polar spade]

Sorry. Exclusions.

[ripe galleon]

oh

[woven hull]

Hi thank you

[blissful pumice]

Any advice on entering style prompts? Should there be bracketed prompts entered with the lyrics?

[wispy laurel]

In the Lyrics field, as a minimum use bracketed tags to define the song structure

[Intro][Verse 1][chorus][verse 2][Chorus][bridge][chorus][outro] before each section

If you don't know what you want the structure to be, leave these out and let the AI figure it out
You can give more details in each section if you want to

[Intro]
[Instrumental, drums and piano]

[Verse 1]
[Soft rock rhythm]
[Melodic acoustic guitar hooks]
[Soft, sad vocals]

[Instrumental]
[Sad saxophone]

But you don't need to do that, it just gives a little more control if you already have a clear sense of what you want to do

In the Style instruction, opinions vary as to whether you should use natural language, bracketed tags, or a combination. I personally find that I get the best result using mostly natural language, with just a couple of brackets at the very beginning

[Low, soft, sexy rap]
[50% hip hop 50% jazz]
[Piano, drums, bass, acoustic guitar]
And then the rest is just natural-language descriptive text

But different people have different approaches to this, and it's hard to say what works best. Natural language relies on you being able to clearly describe the sound that you want (Suno may misunderstand so you sometimes need to figure out ways of saying it that you might think are saying the same hing but the AI interprets differently). Tags rely on the assumption that Suno actually understands and recognises what that particular tag means
Things entered at the start of the style prompt tend to take priority over things entered later (which is why I put my tags right at the start) but like everything else here, this isn't 100% guaranteed

Honestly though, the best way to learn how to prompt well is to experiment for yourself, try out different things, see what works, what doesn't work, what did something but not what you thought it'd do

[balmy mortar]

Hey, quick question — is there any way to export or access MIDI files from songs generated in Suno AI? I’d like to get accurate chord progressions, riffs, and tabs from my tracks so musicians can actually play along, but the approximations aren’t cutting it. Does Suno currently allow MIDI export, or is there a workaround (like stems or a hidden feature in Custom Mode) that makes this possible?

[half dragon]

No Midi but you can get STems.

[wispy laurel]

Can't directly export as midi, you'd need to split the stems, download and use separate software to extract what you need from those

[balmy mortar]

Where do I get these stems from?

[half dragon]

If you're a subscriber, go to each song, 3 dot menu > Get Stems

[balmy mortar]

Okay cool thanks for the help!

[half dragon]

Suno Studio is coming and I heard you can get Midi.

[balmy mortar]

That’s dope!

[half dragon]

Sure. But keep in mind, it's not free. It'll cost you.
2 Stems option is cheaper, one instrumental and one vocal.

[balmy mortar]

So I just clicked on the 3 dots it doesn’t show that option on the mobile app on an iPhone

[half dragon]

Maybe not available on Mobile. Use PC browser.

[balmy mortar]

Okay

[wooden canopy]

I paid for $30 plan after I let go ten dollar plan havent received my credits yet

Sent proof of payment in email

[half dragon]

For account and subscription issue, after sending email to support, best option is to wait.
Follow up on #archived-ticket-help or #support-guide

[wooden canopy]

OK Ty hun appreciate it

OK

[ornate island]

Hey, all, I’m curious How you all are dealing with the situation, where when u sing phrases in, Suno is taking my voice literally and sprinkling horrible sounding acappella and beatboxing into the music.

[wooden canopy]

I get it too not sure why

[ornate island]

Yeah, I thought it would be obvious that I don’t want beatboxing in my songs. And of course, when I say “no beatboxing”, Suno gives me double the amount of beatboxing. Lol.

[hollow hare]

When you do that, the AI sees "no" and "beatboxing" separately. So it will give you beatboxing and whatever it's pattern learned with the word 'no'.

Hence why there is the exclude style prompt
https://help.suno.com/en/articles/3161921

[soft rain]

I like making songs from time periods that sound authentic. Especially 70s and 80s. It's hit or miss but sometimes the recording sounds authentic. Other times it doesn't sound like an old recording. Is there a specific prompt for that?

[hollow hare]

Try to be more specific in the subgenre of what you want to improve your odds. I personally use https://rateyourmusic.com/genres/ for that
I also use things like "1970s, 70s," as part of the style to help try to get the style I want

You can also ask a LLM bot to try to get the style/subgenre/instruments you want by referencing song titles, just verify its answer because they can lie to you with confidence.
I like using the LLMs hosted by DuckDuckGo: https://duck.ai

[wispy laurel]

What Lumine said, but also "analogue equipment" or "lo-fi recording" if that's appropriate to the era you're aiming for

[polar spade]

How did you "let go" of a plan. If you update midstream, it should automatically auto-draft the premium payment on the next refresh date.

[cyan jewel]

quick q, i finished merging my track and its 11+ minutes. I want to now upload it to Suno to post it/share it. Are we not able to upload tracks longer than 8m?

[ornate island]

I understand this— but that’s not helping me get the beatboxing out of my song.

The “hum a song” feature should let you use your mouth to make a song!!

But check out what Suno is doing here, it’s dumbbbb:

I’m humming and singing:

https://suno.com/song/4640db4f-98b5-42ce-8535-9063c1b644f3

the result:

https://suno.com/song/c4684312-ca58-4d85-b134-0b115f2e013c

If I simply excluded “acappella” and “beatboxing”, would that achieve what I want?

[hollow hare]

What style are you trying to get and what are you trying to make it sound like? I can try to help you with that.

You can try to Cover your song with the style you want, like this (used a trance style):
https://suno.com/s/oLZ9bK9nC3AMLwUE

[manic bay]

Suno doesn’t export MIDI yet, only audio. Best workaround is using stems + a MIDI converter. They said MIDI will come with Suno Studio soon.

[ornate island]

So if I hum a song and add the beat I have in mind (specifying the style and instrumentation I want), I can then cover it to zap out remnants of vocal crud leftover?

[hollow hare]

You can use Get Stems to split the vocals out, probably cleaner that way, if your vocal is interfering with the Cover

[stoic vine]

found the prompt that allows you to ascend https://suno.com/s/PjcyAD3TsWfLWtSz

https://suno.com/s/m96RyCdqg7MP65gk

Weirdness 100%, Style Influence 100%```
[null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object][null][void][object object]

https://suno.com/s/fTzoEbHIPTprfest

01000001 00100000 01110011 01101101 01100001 01101100 01101100 00100000 01101101 01100001 01101110 00100000 01101001 01101110 00100000 01100001 00100000 01101100 01100001 01110010 01100111 01100101 00100000 01110110 01101111 01101001 01100100 00100000 01100110 01100001 01101100 01101100 01101001 01101110 01100111 00100000 01100110 01101111 01110010 01100101 01110110 01100101 01110010

https://suno.com/s/PkI5ugpPpRx7Uu2a

something beautiful about how they start

[raven turtle]

With Suno Studio invites apparently being issued can anyone report if the fidelity has improved at all? Is it based on 4.5+ or is the sound quality closer to CD 16bit yet?

[vale obsidian]

So if anybody saw any of my other recent posts here's an example in nearly real time.

Damn near same runtime and issue as every OTHER generation

FML .... Apparently Murphy's law is strong with me today.

[shut ferry]

I created an Playlist to use the +Inspo feature, but whenever I do EVERY song generated is now 4:58 in length regardless of if the song is done or not. No song in the playlist is 4:58.

[shut ferry]

The songs in the playlist are mostly 5-7 minutes. The other problem with the +Inspo seems to be the same problem with Persona, every song has the same intro melody and the vocals progress the same no mater what I do the Style or any meta tags. Just going to ignore +Inspo just like I do the Persona until they fix it.

[wispy laurel]

A handy thing that you can do with Inspiration is, don't use a playlist of actual songs. Crop the best bits of individual stems from songs, so that you have a bunch of short clips of guitar hooks, basslines, drumbeats, random bits of vocal that you liked. Then make those into a playlist instead, and the AI now has to mix-and-match because it can't make a whole song out of any one of them.
Like everything else in Suno, this works when the stars are right and the gods are smiling, but it does work

[shut ferry]

ok, I will try this, thanks

[mellow hound]

How can i suno to put vocals my istrumental song without changing the beat?

[void jolt]

Not sure if it's supposed to be asked here but is there a way to make the instrumentals not feel so... overloaded? So many times if not always you hear two seperate drum tracks ( on v4.5 ) or any other instrumentals overlap with each other and it just feels overloaded

[polar spade]

The idea behind clicking edit/remix , then Add Vocals, is to add lyrics to the instrumental without changing the vibe of the instrumental. Has that not worked out?

[mellow hound]

Well i want to add Brazilian lyrics and i don't speak Brazilian

[polar spade]

Yes? And whats the connection between Braz. lyrics and adding vocals to instrumentals?

[mellow hound]

Well where do i get Brazilian lyrics

[hard beacon]

chatgpt ? Google Translator with written vocals? 1000 ways 😉

be creative.. that's what music is all about 😉

[polar spade]

Keep in mind its a style of Portugeuse, just like American is a style of English.

[mellow hound]

Yes i have the lyrics (chat gpt) but i can't read it

[hard beacon]

well... that's a problem with languages you don't speak 😄 😄

[polar spade]

There is a Brazilian Portuguese channel in here, which those that also speak English might work with you to help with lyrics. https://discord.com/channels/1069381916492562582/1154779408989687848

[humble verge]

am i the only one who is getting extremely bad audio quality when remastering? It's like nine out of ten are just straight for the trash bin.

[hard beacon]

remaster 4.5+ is 4.5 .. don't ask me why.. lookz like a bug (or feature) 😄

[polar spade]

The favour for most over the past couple months have been to remaster in 4.0 first, then 4.5. Comments have been positive. So its worth seeing what comes of it. May get 6 of 10 for the trash, but 40% is good odds.

[humble verge]

i tend to get a lot of shimmer with 4.0

I gotta say, in general the audio quality seems to have gone down hill.

background noise, artifacts

when remastering i mean

[polar spade]

I get artifacts when stem splitting vocals..... but then I cover it and the artifacts are gone. Might try taking that extra step and see what 4.5 does to 4.0 shimmering. Just do two gens then, see if two sound decent. If not, tick that as a "no", then.

[humble verge]

regardless of the track i am remastering, the remaster will start with a loud CLICK and a hiss and end on clicks too

[polar spade]

Perfect. Then its easy to crop out the click and hiss as they are on the ends.

[humble verge]

not when the click happens right when the instrument starts

it tends to happen when i remaster a remaster

i just did seven takes and every single one added a noise at the start of the song

[polar spade]

Darned. Thats definitely more a bug or feedback issue then. Thought we had it. Just a domino effect, eh, of issues. Sorry its been an uphill climb.

Dont want to keep riding a bike with a flat tire. I get it. Any thoughts around it? I guess my ideas won't help.

[humble verge]

The click thing is fairly common at the beginning and end of songs, but usually it's just a split second and before the song ends/after the song finishes. But this is more than that and it's consistent.

Here's the instrumental I am remastering:

https://suno.com/s/8802woLagqbC5eIE

Here's the bug:

https://suno.com/s/DjDoQBmQtdRtDtWm

[polar spade]

Cool. Ill listen. I had the click problem a month ago and for two weeks I had crackle in all playbacks. But not in the downloads.

Ill see what yours sound like.

[humble verge]

i've been getting this bug a lot over the past week or two

[polar spade]

Thats an "explosion" kind of noise I had heard before. Yea, it all sounds poor.

Ill play around with it and see what comes of it.

[humble verge]

i did seven remasters in a row and they all had it. Even the 4.0 ones

[polar spade]

Was there an issue with the first one? It says Remaster.

[ionic wind]

Any tips to prevent the vocalist from yelling at the top of their lungs during the chorus (and constantly thereafter) in rock songs?

It makes everything sound like 80s power ballads

[humble verge]

no the original is fine, except for a hiss throughout the track - which is why I remastered it in the first place.

the hiss is actually brushed drums, but SUNO just makes it sound like noise

so I did a remix where I removed "brushed drums" from the prompt, then remastered the remix when i was satisfied

[polar spade]

But the crash at the beginning wasnt intended right. I did NOT hear a "click" so much as sound like dropping pennies on a drum head.

[humble verge]

no it's not intended and it's in all the remasters

[polar spade]

Sent you a DM.

[cyan jewel]

can someone help me figure out why i cant embed a video file i made?

[hollow hare]

If you mean why your suno link does not embed a video, assuming that the channel (like #💬┃general-chat), the likely cause is that you do not have a video generated (limited to subscribers AFAIK). I am assuming you have a subscription for the rest of the discussion.

When you publish a song, Suno will automatically invoke the video generation.

If you are sharing a link to a song that has never been set to public, you will need to manually generate the video by going to Download > Generate Video.

Once the video is generated, the video should be embedded. If you already posted the pre-video link on Discord, you will need to clear your Discord Cache to see the post-video changes

[cyan jewel]

no im trying to add a video to my song

it says upload failed

:Upload failed
Please try again:

[sleek stump]

Is the video 10 seconds or less and is it a mp4. is it a 720x240?

[cyan jewel]

9 seconds, mp4. the upload thing the video has to be at least 720px tall. was that written incorrectly?

[sleek stump]

Those are all the reqs, has it failed for a long time now?.

[cyan jewel]

it fails every time i try

[sleek stump]

Try a different song, if that works then there may be an issue with that song, if it fails again, ther emay be a larger issue happening.

[cyan jewel]

ok i had to compress the crap out of the video.

yeah i think the websites wrong about the video requirements. the website says "at least 720x" but its really "no bigger than 720x" lol

[sleek stump]

As far as I knew the size was 720h by 240w, that size always works for me.

[wispy laurel]

Yeah this is super-annoying, Suno just loves to yell 😂

I find I can somewhat avoid it by tagging both in the lyrics and the style instruction
In lyrics, instead of just [chorus] I use something like
[Soft chorus]
[Calm, relaxed, soft vocals]

And somewhere near the start of the style instruction, something like
[Even, consistent vocal tone and register throughout all sections]

This does not completely or reliably solve the problem every time, but it does seem to at least improve the odds. Mileage may vary though depending on the style of the song, it's very difficult sometimes to override genre conventions so if Suno thinks "this is the kind of song that has shouting in it", it can be hard to work around it.

[sullen jay]

hey everyone, I hope I am in the right channel to ask about this - I have created a song in the sort of nu-disco, synthpop, indie electronic direction, and it seems to have completely skipped making a lead melody. It always just has some background meldoy, but never a main melody accompanying the singer in the chorus. I find this quite frustrating and wondered, if anyone might know a way to deal with this in remixing. I hope it's possible to maybe add one after the song is done and is being remixed, but it seems to be a pure style shift, never anything major. Any ideas? Thanks.

[graceful condor]

Has anyone found a viable way to make good acapellas

[polar spade]

Its not often we want a "unison" performance throughout the song. Mainly in the chorus parts of musicals or anthemic rock songs, Id say. So, Suno likely has trained on less of those.

If it were me, I would

1. create the song
2. Split out the vocals to get a melody (sans instruments)
3. Cover the vocals as an instrumental (specifically as an instrument, if wanted) so you now have a melodic instrumental.
4. Choose "add vocals" to, well, add lyrics and vocals to the melodic instrumental.

[sullen jay]

Nice idea, thank you

[polar spade]

Ive done similar where I want a musical feel, but cant say my goal was to have them in unison throughout the song.

I also have done in jazz and new age, just to have a melodic 2 minute intro, then add vocals to sing tacked to the end. Some worked solely using Suno, others just easier editing in a DAW.

[neat prism]

Bon après des milliers de crédits ....Suno ne clone jamais ta propre voix pour faire une chanson....

[graceful condor]

All good figured it out

[polar spade]

Cool. Ill remove my comment.

[sullen jay]

Oh, I now get what you mean. I don't specifically want the melody to be the same as the vocals, I just wish for there to be a main melody in the chorus. The tracks I sometimes get have no melody in the foreground, just the background chord progression, which is present through the rest of the track too so the chorus doesn't really strike as a chorus that much

[polar spade]

" Well after thousands of credits....Suno never clone your own voice to make a song...."

Je ne pense pas que Suno ait jamais prétendu faire ça, n'est-ce pas ? I dont think Suno ever purports to do that, clone ones voice, do they?

[polar spade]

SUNO is teasing about a tool called Suno Studio (?). If it functions like a DAW (as the promo photo looks like) and we can kind of cut and paste stems around, all these other techniques may become moot. We'll see.

[sullen jay]

I'm looking forward to it 🙂

[ripe galleon]

is there a way to make a song sound worse in suno

i want to extend this and make it sound like Donkey

but the results i got made the song better and i dont like tha

[polar spade]

Make a persona of the output first.. Then extend the output , while adding that persona?

[ripe galleon]

im on free tier sadly

[ripe galleon]

dasvon

just this once, could you actually do this for me

i really wanna see what kind of bullcrap it spits out LOL

[graceful condor]

It seems I can't incorporate this part, any better way? [Immediate silence and deep bounding heartbeat. Then the full band returns.]

[raven turtle]

You could try the edit mode and replace the section? Or extend with a new prompt adding the melody and then paste that into where you want the melody section
As I work in the same genre (dark electropop like DM) I get this issue too but generally I add those kind of things in Logic Pro as overdubs out of Suno.

[void jolt]

Not sure if it's supposed to be asked here but is there a way to make the instrumentals not feel so... overloaded? So many times if not always you hear two seperate drum tracks ( on v4.5 ) or any other instrumentals overlap with each other and it just feels overloaded...

[tired cargo]

Hi people I tried searching but couldn't find anything on this. Has anyone encountered issues like this? Songs longer than 4 minutes, end up with hollow and hissy and sort of reverbal sound? Any method or any negative prompt to address this? I've been wasting my premium credits due to this issue and its hard to get rid of this for some reason. Noticed this on 4.5 and 4.5 plus too.

[rose ether]

hey all - is there a technique for taking a long song, like 5m, and turning it into a 3m song? I have some generations that I like, but they go on a bit too long and i want to do the equivalent of "remastering" them to a shorter, punchier version without detailed editing. Like a cover that is made to be more radio friendly.

[polar spade]

Without presently using a DAW?

Honestly, find a streaming audio recorder to record audio played online.

Separate your tracks with the Get Stems.

While in the multistem display, mute tracks you dont want, then click play and record the remaining tracks as a wav or mp3.

Upload the audio, then Cover, Remaster or do nothing.

[polar spade]

Yes. Try to just Cover it. Likely Cover will shorten it anyway. You could find a nice spot to crop and fade it as well before Covering, which may help be more exact.

[rose ether]

hmm, i've had good luck with covers and rewriting the lyrics to be shorter. i really wish you could put a time target on (all) suno generations. ah well

[polar spade]

Theres something to be said to letting the AI determine the best arranged song for your prompts, lay it all down, and whatever length you get , you get.

Other sites DO have the ability to set times, and so you will reduce any variation of song which might have sounded better if longer. But if length is a priority, we wouldnt really know the difference if we never got the longer song, right?

Its relative. When songs were 2 minutes or 4 minutes limited, we accepted them as OK, but wondered what song we'd get if they were longer.

To each their own preference is better. We'll see if that ever becomes an option with SUNO. An option with an on/off switch. 😊

[void jolt]

What is DAW ?

[rose ether]

On another note, is there a good way to say “don’t start wailing notes”? Something to put in a style tag for the chorus, probably. Like, a little is ok, but the singer goes off on these soaring notes and I want like ONE of those

[void jolt]

This is genuinenly annoying yeah... No matter the genre and style tags you use it start wailing on the chorus so you can't even do certain music genres like metal

[polar spade]

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

[void jolt]

Even if you write it information that it shouldn't stretch the notes it still does

How is (Chorus - Metal scream vocals) not enough for the AI to do that but instead it wails and stretches the notes and words and just sings it regularly... I thought v4.5 introduced more metal/heavy rock generations

[royal kiln]

what is a prompt u are using ? Genres: metel core , melodic death metal. Take a look at my prompting https://suno.com/song/dc371f75-bcdd-4193-9f8d-049e024b3331

[last bay]

hi i try to extract stems but it seems i can only download them now? in past they were in library. so can't use for persona?

[unique stone]

In your library or workspace, go into "Filters" at the top left and untick "Hide Stems" to see the stems.

[last bay]

ah thanks!

[hollow wigeon]

Guys, somente can help out here? I am a New user of the App, i have the Premium stuff and all, but my option to Record AND upload dissapeared! Someone know how to fix it??

[graceful condor]

I don't what's going on with Suno today, she's putin out nothing but stank. 😢 (1) she refused to use my "Inspo" on one render (that was a waste).

[graceful condor]

It is a mess in so many ways. I can't upload pictures for covers anymore, I can't rename my newest tracks.

[sly citrus]

I feel like a lot of the gens are like RND item drops. Sometimes you get a purple, most of them are common

[polar spade]

Im not certain, but I dont think upload and record are available on the app. Looking at the app on Android, I dont see it, either. Use a browser instead and maybe someone else can confirm.

[raven turtle]

On iPhone there’s a mic logo. If you click that it takes you to a record/upload page where you can do both 👍

[violet turtle]

Is anyone else having the same issue as I am with Suno? It is either not playing a song at all or only playing for 1 second and then stopping.

[raven turtle]

Has anyone else noticed some generations are starting with a muffled drum sound that after four beats gets clearer? Really odd. Like it’s winding up the quality and not quite there for a few bars.

[cyan jewel]

hey does anyone have any familiarity with getting specific voices from suno? im starting to notice some real same voice syndrome.

[ember summit]

Some of my songs have scrolling lyrics and some don't. Is there a way to force it to scroll? Setting I have to click?

Or have I screwed up the scroll AFTER I get a finished song the way I like it and go into the lyrics and remove all the phonetic tricks I use to get what I wanted?

[half dragon]

You mean the video? After fixing the display lyrics, you'd need to regenerate the video again.

[ember summit]

I guess I must've "accidentally" done that before. Trying to do that on purpose now and I can't get any of the "create video" or " recreate video " to work. I know I'm doing something wrong.

Well, I thought I was getting decent with this. Wonder how many other stupid things I'm doing ?

Well, I thought I found it, but all that did is created video in the form of an MP4 file. Which is nice, but I had a way before without doing anything on purpose that when I played the song on my phone using the SUNO app I would see the lyrics scroll and they're gone. I'll reread this in the morning. See if anybody's gotten an answer. I'm sure it's simple except for me.😋

[west cloak]

Suno is amazing, some of the most amazing software I’ve seen in my life. I Dabbled in it when it was early on but v4.5 is on another level. I’m trying to make remixes of my existing melodic music, or in other words I want the generated song to contain elements of my main melodies. Similar to this instrumental Clannad song, remixed into an electronic version, how would I go about doing this with Suno?

Snowfield:
https://www.youtube.com/watch?v=VNSq1qf5ZFk

Snowfield remix:
https://www.youtube.com/watch?v=0lXcsamK0t4

[sturdy dove]

Tsup all you ponks!!! It's your general PITA here to say thank you, part 1. I'll link up a project in general chit chat. I promised to post up what was finished to all those who helped me, so here's part 1.

THANK YOU to all who put up with my and helped me get through my project.

[oak fiber]

not sure if this is the right spot to ask but, if I am uploading stems or a master stem of my own audio file into suno for remixing, is it better for it to be completely barebones? Like, no EQ, no compressor, no gate nothing?

[wispy laurel]

In theory it shouldn't make much of a difference BUT, keep in mind that Suno isn't actually intelligent. It won't know that your instruments have EQ or effects applied already, it'll probably just assume that that's what the instrument sounds like, and work with it on that basis
Bear in mind that any EQ, FX or whatever else you start out with, will be "baked into" whatever comes out at the other end, and Suno might also decide to add its own alterations on top of yours

[oak fiber]

Cool, thank you. I was experimenting to see how close to the original audio it could stay in a remix, and was just curious how much of a difference that would make in terms of clarity.

[hoary dew]

Can someone can help me (show me), how to make a Classical or similar ,symphony of 8 minutes # Without ANY Vocals . I would like to use the Styles (1000) AND Lyrics space.. Giving into lyrics only musical intructions. It was possible in v.3.5 or in V.4 sometime. But in V4.5+ I spent many many credits , but on 10 tests I got 1 working...Vocals start to sing at a moment, breaking the music. This regression is very bad. Any idea ?

[normal lagoon]

Can someone teach me how to control the harmony direction of song generation in Suno?

[graceful condor]

Wth is wrong with Suno? Even a day later I still can't rename my Tracks or Change the cover. Always the "Unknown Error " red messagepops up. This is so dumb.

[raven turtle]

There’s plenty of mood and vibe prompt lists around that help Suno understand where you are pushing things so I’d recommend starting there. Or I have had some success getting chatGTP to analyse a commercial tracks sound and use that as a reference prompt

[normal lagoon]

To be more precise, it's the progression of chords.

[graceful condor]

What exactly do you ask GPT in such cases? Just "Analyse this track in terms of Sound, Tempo, Rhyme flow, Voice range, Instruments or how it is build structure wise?" Or?

[raven turtle]

All of it to be honest. I usually ask it to be formatted as. 1000 character suno style prompt. It isn’t precise but should get you in the ballpark and you can amend for the sound you want from there. It’s worked for me on occasions. Such as that drum loop and guitar wash sound on Depeche Mode’s “Walking in My Shoes” single mix that also features scratching riff fills. I got close to that early 90s electronic grunge sound using ChAtGTP analysis. Of course it’s a bit of a slot machine pull when generating with Suno but it can deliver results.

The other thing I’ve found useful that can help push it in the direction iI want is understanding the producer as opposed to the artist. For example for a late 80s Kylie Minogue vibe, I love the remix production work of Pete Hammond and Phil Harding, so citing producers or their methods I’ve found to help a little.

[near flume]

You need to go deep into creating the vocal style. Try this: Male Baritone vocal with warm, slightly gritty tone and expressive delivery. Emotional indie rock style with soulful phrasing and dynamic intensity. Verses delivered with restrained intimacy, gradually building to powerful, impassioned choruses. Subtle vocal rasp, clean pitch, and tasteful melismas. Include layered harmonies in the chorus—tight thirds and sixths—for richness. Production should feature light reverb, smooth vocal comping, and a natural, live-room feel. Mood is determined, heartfelt, and resilient.

[indigo gazelle]

why suno 4,5+ always do reverb and delay? i have excluded them in the style tags

[cyan jewel]

oh im going for female fox. its just like, the same vocals every time even if i try different things like contralto, or alto, or tenor, or vocal fry, or breathy, or nasally none of it seems to change anything lol

[cyan jewel]

lets say i wanted a vocalist with a nasally voice in the mezzo-soprano range, and a deep, throaty, flirty, sensual vocalist in the contra-alto range. how would you tag that?

[raven turtle]

Have you tried the prompt for “dry signal chain”?

[indigo gazelle]

No i have putted in the excluding Style Tags the prompts: reverb, autotune, delay
And also: No reverb, No autotune, No delay

[half dragon]

When excluding styles: don't use "No"

[wispy laurel]

Exclude doesn't always work too well, I find. In general it's better - if you can - to find a positive instruction with the same practical result, and put it in the main style instruction instead
Like for instance, a style instruction for "low-pitched vocals" is usually more reliable than an exclusion for "high-pitched vocals"
In that specific case, more precise and technical wording would work better than either of those options but the point stands - Suno responds much more reliably to "do this" than to "don't do that"
So the suggested "dry signal chain", or in more natural terms something like "natural, unprocessed sound" will almost certainly work better than using exclusions

[indigo gazelle]

i have tried both, but suno ignores the exluding styles, and make more delay and reverb, even with the tags: delay, reverb, hall, autotune, it does not follow the instructions.

[indigo gazelle]

i have tried: low pitched vocals for a deephouse track, and suno still make as usual the same vocals. a little bit frustrating

i will try: dry signal chain 🙂 thank you guys!

[raven turtle]

Try “baritone” vocal

[wispy laurel]

Or "deep contralto" if it's a female voice. Although that might not be as reliable

[graceful condor]

I was wondering how good is the "Japanese pronunciation" Suno does? And is there a way to improve it?

[shadow escarp]

Anyone having issues with Male or female prompt for verses? I want one verse to be all male and the second verse all female. I get split voices on a verse. Half male and half female. I write country songs. used 700 credits and still have the same issue. Anybody have a work around for this issue?

[quick sonnet]

Just wanted to try out something like this, but how?

[shadow escarp]

I use to just put [Male voice] or [Female voice} and that use to work. now i'm having issues

[graceful condor]

You can give directions [Verse 1][Male tenor voice, softly singing] etc. But it is more luck than anything if it works or not. It never does on the first try. Either it is more male more female or as you said split sections.

[shadow escarp]

If I want a duet sound I usually just use [Harmony]

[rose ether]

style tags: "four-on-the-floor beat, deep bassline, rhythmic and urgent, alluring, husky low-voiced intimate delivery, scratchy electric guitar (light overdrive), slow melodic guitar solo, glitch ambient, textural microglitch, electronic beeps and synth hits, wide stereo, modern punchy mix, no harsh highs"

does this make sense?

in particular, wondering if "husky low-voiced intimate delivery" is a good cue or if there's a better way to say that

[quick sonnet]

Should I write then [Verse][Male voice] and [Chorus][Female voice]?

[shadow escarp]

Usually chorus will bring in more than 1 voice.

[quick sonnet]

Would like the mail voice to do the Rap part, while the female one is singing the Chorus.

[quick sonnet]

So here Refrain is meaned by Chorus.

[quick sonnet]

No, something like Jay-Z and Beyonce, defenetly not exactly them, but just an example.

[graceful condor]

[Vocals: Male rap verses, female melodic chorus, emotional delivery]
[Intro – Female chorus, soft]
[Verse 1 – Male rap, rhythmic but warm]
[Chorus – Female melodic, soaring]
[Verse 2 – Male rap, more intense]

And so on. At least it work on my side.

And in style prompt this line {Vocals: male rap, female melodic chorus emotional and soaring}

[hard beacon]

i had good results with [mr.] and [mrs.] before the part starts.. but that was with 4.0 🙂

[steady surge]

Any guides or best practices on taking some of my poorly made wave tracks and using them as the basis? I just tried out the free tier and the first ten songs are completely off in style even using what Suno generated from my upload. Tried adjusting the prompts but where to begin to make this better? I'm having trouble getting the tempo to be the grunge style slog like Alice in Chains, Soundgarden but my track sounds this way.

[humble verge]

What's with this new exciting feature where remixes just completely disregard the lyrics of the song? What a fun way to burn through credits.

[hard beacon]

for me the cover function works well.. 🤷‍♂️

[humble verge]

Good for you. I've been getting nothing but mispronounced and misaligned lyrics today.

[hard beacon]

4.5+ ?

[humble verge]

yeah

usually it works fine

[hard beacon]

ok.. coz i've noticed that covers to 4.5 had some of mispronounced vocals

[humble verge]

but today, for some reason, it doesn't

ah the weird and wacko ways of AI

Replace section is all borked too

[hard beacon]

@humble verge i have to correct my cover argument.. if you try a complete different genre (with Style influence slider over 80%) , then there is a lot of mispronounced vocals 😉

..and audio influences with low settings

but it's really hard to create a dubstep cover from a rock ballad 😄

[humble verge]

I’m working on a blues cover of a blues song with the same lyrics, but SUNO keeps messing it up, mixing up the verses, mispronouncing words, and making odd choices about where to pause so it sounds completely unnatural.

I’m also running into problems with the editor. When I try to replace the intro (which has no lyrics), SUNO randomly pulls the first two and a half lines from section two and drops them into the section I’m replacing. I’ve done several runs, and it either does that or fails to connect seamlessly with the next part of the song, adding awkward breaks or strange sounds right before the transition. Honestly, it’s a complete mess. SUNO seems a lot buggier lately.

[hard beacon]

ok.. that's weird

but in the end of the day it's still AI 😉

[humble verge]

https://tenor.com/view/jack-nicholson-forget-it-jake-chinatown-gif-11618710

[unborn star]

Editor is well known for being not that good for replacing parts of the music. Go for another creation is better or wait for Studio.

Btw. it could be about your prompting. Not sure how much experience you have, but having these kind of problems, doesn't look natural for me.
Maybe you want to much or you do some mistakes in your prompting.

[humble verge]

Well prompting isn't an exact science and SUNO may interpret prompts differently from gen to gen. So that's really just a matter of trial and error. Hard to say if it just decided to be stupid today. I'm not doing anything differently, prompt-wise, than I normally do. Regardless, it shouldn't affect how it reads lyrics.

[hard beacon]

let's ask on another way.. Rafaelis.. you are trying to create a blues cover of a blues song... what result do you want to have???

[humble verge]

The result I'm going for is... a blues song.

[hard beacon]

from a blues song??? or what style is the original ?

that's what i am thinking coz: I’m working on a blues cover of a blues song with the same lyrics,

🙂

[humble verge]

The original is a blues song I uploaded. I am doing a cover of it, with the same lyrics. I did some decent ones yesterday. Same process. I tweaked the style a bit today, but nothing major and nothing concerning the lyrics or instructions. But for some reason, SUNO now decided it doesn't know how to sing the lyrics correctly.

It decided that the verses can be mixed any which way it likes

[hard beacon]

ok.. i have to say i don't use the upload function to create covers out of it... so.. i am out 😄

[humble verge]

and it decided that words are strange and unpronounceable

[hard beacon]

i think it's a big difference between "suno outputs -> cover" and "upload -> cover"

[humble verge]

Yes you may be right

however, like i said, worked pretty good yesterday

[gloomy shore]

Hello, i have a question. Is it possible to create a young / kid vocalist?

[humble verge]

except for the usual cover time limit issue

[wispy laurel]

Have you tried making a playlist that contains only the uploaded version of the song, then using it as inspiration for a new song (not cover)
Use the same lyrics as the original, prompt as you normally would, set weirdness low and audio and style both high
Doesn't always work but sometimes works better than cover/remix

[hard beacon]

@gloomy shore : as far as i know suno doesn't create children vocals

[humble verge]

Interesting. No, I haven't tried that approach. I'll give it a shot. Thanks 👍

[gloomy shore]

Oh, there goes my tugging heartstrings song 🥲

[hard beacon]

maybe try to use "lullaby vocals"

but i don't know if it works

🙂

have tried to create songs with children choirs.. but.. well.. no way

[gloomy shore]

Hmm ok, I'll try that, thanks

[hard beacon]

but i think it wouldn't wok

*work

[humble verge]

I've noticed there can be issues if you set audio influence too high when doing covers because discrepancies between the original and how you prompt for the desired result confuses the AI

So best to set it at 60% or lower.

And that all depends on how the AI decides to interpret your prompt. So again, trial and error.

[wispy laurel]

Got this on the second attempt:

https://suno.com/s/M21WQe0fAXybMguW

"A nursery rhyme, sung by a little girl with no instrumental accompaniment. Very young, innocent, childlike voice and style."

It failed at ' no instruments' but I was expecting that. So it can in theory create child voices. You'd need to experiment with wording to get a more realistic voice, but it at least can get close to the style and tone

[hard beacon]

i would say these are female vocals and not children vocals 😉

[humble verge]

I'd put it around 14-15 yrs of age.

[wispy laurel]

Meaning therefore, it isn't "locked" to adult voices only. I got that in literally one minute using the first prompt I tried, so I'm pretty sure it can do better than that if someone makes more of an effort than I did 😂

[hard beacon]

suno is using vocal models... and i think they don't have children models in their outputs..

that's why all song creation ai's sounds nearly the same.. they are using the same models

[humble verge]

maybe do a persona? Find a song sung by a kid, get the vocal stem, upload it to SUNO and create a persona

[hard beacon]

try it 😉

[rose ether]

style tags: "four-on-the-floor beat, deep bassline, rhythmic and urgent, alluring, husky low-voiced intimate delivery, scratchy electric guitar (light overdrive), slow melodic guitar solo, glitch ambient, textural microglitch, electronic beeps and synth hits, wide stereo, modern punchy mix, no harsh highs"

does that style tag make sense?
in particular, wondering if "husky low-voiced intimate delivery" is a good cue or if there's a better way to say that

[humble verge]

male or female vocal?

[rose ether]

this is for a female vocal

[rose ether]

mostly trying to figure out if 4 words is too many for a single style phrase

[faint moss]

[vocal_style | gravelly Southern baritone, emotional, broken]
[vocal_type | raspy baritone male]

[rose ether]

well, and how to do the thing

[rose ether]

this is for a female vocal

[faint moss]

my bad

[gloomy shore]

Oh it's great, even though not really childlike vocal but atleast its closed enough, but i need a kid male vocal though 😂
I also planning to do a raspy female vocal too. Any idea?

[rose ether]

uh, my message got deleted and i didn't even see what i said lol.

melisma bad, anyone know how to stop it?

[polar spade]

Most all sound like a 30 year old woman faking a young kids voice (i.e. as in most animation voiceovers)

Occassionally I have gotten a childlike voice but still seems random, though i have not made any recent attempts until I get back to my kids songs. The better chance has been when we get a choir or ensemble of kids singing , then it might sould like kids.

Heres an example of call and response , but the kids vocal was one out of dozens if not 100 of NOT kids vocals. For me this mumble-like speech quality of a kids voice has been rare. SO suno , it appears, knows where its appropriate only by saying childrens song or kids song, and likely cant be prompted more than that.

https://suno.com/s/GME2fMiTFbUetqS9

[humble verge]

This is straight out of Sesame Street

[unborn star]

Hmm I don't have any problems at all but my concept is much deeper than just prompting, so we can compare each others if you give me your prompts and I show you what I get out of it (when I change them to my style)

[humble verge]

I don’t think it’s a prompting issue. It’s more about finding the right balance between the covered song and the remix settings - in other words, how much the AI listens to the original versus what it’s told to do with the result. Basically, it comes down to sliders.

For example, if you set audio influence to 90–100% but the prompt doesn’t align perfectly with the original, the AI can get confused. Maybe the original has an unclear vocal, and with audio influence set too high, the AI prioritizes interpreting that sound instead of reading the lyric you gave it. So if your lyric says “fell” but the original recording is muddy, the AI might interpret it as “frell” or “farl.” And since you told it to follow the audio 100%, that’s what it will use in the remix.

As for why it mixes up verses, only the god of AI knows.

[unborn star]

Hmmmm.. I usually don't have any problems that often. But ok, was just an offer.

[polar spade]

You might look ( as the magenta colour bops through the alignment show) carefully to see if the intro is really just one cursor thick. Then try to edit one cursor line thick segment.

Or alternate, add a segment between the intro and first lyrics, then perhaps combine the two segments before editing. A better means is to zoom all the way in and add a micro-segment of less than a second.

Not sure all what you're doing, but throwing those out there, while I go to lunch. Good luck.

[humble verge]

Really just want to replace the whole intro section here, but it's doing a very bad job at it. It's all instrumental, so it should arguably be a piece of cake for the AI to handle. But, well, I guess not.

[humble verge]

As for my other problem, I've been doing covers of the same blues song and got decent results yesterday, where as today, I get a lot of weird bugs - like mispronunciations and random mixing up verses - singing half of verse 3 in verse 1 etc.

[quick sonnet]

And what about the end? Would like to have after the las chorus still the chorus melody which then becomes quieter, I don't know what that's called in technical jargon.

[quick sonnet]

Is this some feature of the Pro version?

[humble verge]

Yeah the editor is for the Pro version.

[polar spade]

As i noted in need-help room, it could simply be thats Suno gets the flu when its overloaded on Mondays when everyone goes back to work and instead, plays with Suno on their computers. We're going through 12 hours of "lunch time" zones as well.

Hard to always tell whats going on.

[humble verge]

Yeah it does seem to come in waves.

[worthy otter]

Editing sections can suck tremendously, it doesn't help if you preedit the lyrics before you open up the stuff in the editor, i experienced it can help if you split the part you want to replace if it contains an instrumental break of intersection attached to it. Then rename the instrumental split just "instrumental break" or something, after that you usually could safely edit the vocal section with less hassle. In worst case scenario's, i had to "save" the edited song and open the new version into the editor again to succeed. But the editor is not really my favorite, i rather use the cover option, toy with the advanced buttons (100% audio, 100% style, 0% weird) and rewrite the lyrics there. This type of "editing" usually give faster and better results and saves you a lot of frustration credits.

[indigo gazelle]

thx

[indigo gazelle]

thank you 🙂

[humble verge]

The less i need to use the editor, the better. I usually try to remix myself out of a pickle, but sometimes you just get that perfect result with a couple annoying glitches and that's when I open the editor. It sure can cost a lot of credits though. I spent 1300 credits once, just to get the bloody AI to pronounce one word correctly. The song was just too good to let go.

[wispy laurel]

I've done that a couple of times. But have since discovered that remastering can (sometimes) fix mispronounced words
So
Song is perfect except a word is wrong
Remaster and split vocal stem from a remaster that pronounced it right
Crop that one single word out of the remastered vocal
Use DAW to cut that word out of the original vocal stem and replace it, just that word, with the remastered one
If it doesn't fit perfectly you can maybe cover it with subtle FX, or else take your chances remastering it with Suno to hide the seam when you re-upload
It's a lot more fiddly and it's a pain in the arse if you weren't already intending to do DAW stuff, but it can save a ton of credits by just doing a stem split and a couple of remasters instead of a hundred failed section replacements

[quick sonnet]

Which instruments I can use in the prompt? I mean al wester ones like guitar, keyboard, drums, etc should be no problem, but how about some exotic ones?

[golden plover]

The copyright restriction is literally the dumbest restriction on this platform. So before this restriction it lets me use copyright lyrics and generates songs. Now I want to make stems out of it? Nope. Gets stuck on Detected Stems.

Would be beautiful if Suno could disable the copyright restrictions for one day so that I can make stems out of the existing recordings

Or at least make it so that I can make stems and extract the vocals

Becuase of this dumb restriction I now have to use third party services like Moises and it gives me very bad results.

Or at least Suno could fix their bugs and actually TELL me that the lyrics are copyrighted instead of being stuck on “Detected stems”

[ember summit]

Can anyone remind me again, how to see my lyrics scroll while the song is playing on the phone app?

I see how to create a video mp4 file ( on my MacBook Air) instead of creating an MP3 file. But I used to be able to see my lyrics run concurrent with the song while using the Suno app on my phone while it and about for the songs I created ( I paste my own lyrics after uploading my audio with the voice stripped from my creations). But I do usually have to edit my words. Maybe that loses the scroll?

[humble verge]

if you just want to extract the vocals, you can use Ultimate Vocal Remover.

[ember summit]

I have used dobro, mandolin, fiddle, banjo and they've appeared

[humble verge]

Yeah I've done a fair bit of collage work in DAWs. It has to be a really good song for me to bother though. One with qualities that don't carry over when remastering or remixing. I wish I was a lot better at working with DAWs, I'm sure I could forego the editor almost completely. I usually use iZotope to 'fix' artifacts as well as i can, and then mix the stems in MixPad after. But if I can just remaster or remix my way around the problem, I prefer that to spending hours trying to solve a pop in the vocals or whatever. I haven't tried cutting out and replacing a word with another. I'd assume it's difficult to make it match, since the tiniest differences would stand out quite a lot, unless disguised by instruments.

[unreal socket]

So I changed a first occurrence of a pre-chorus and I renamed it after, as [Pre-Chorus 2], but later on I have the old [Pre-Chorus] repetition, I wanted to have a the new [Pre-Chorus 2] also in the second occurrence but Suno generates another Pre-Chorus even if I tagged [Pre-Chorus 2] in the new generation. I don't use the extend option or it will delete part of the songs I wish to keep, which is the bridge and the final, how do you guys solve this without using a DAW and only in Suno?

[wispy laurel]

Yeah it can be a pain to make it match properly. Sometimes depending on the song, you can hide it by adding some ambient background or even just noise at REALLY low level, depends on what else is playing alongside though

[half coral]

Does anyone know how to create a concert band song with out Suno chosing to make it a march or a Polka? Have an idea to make a prog rock song for a concert band I am in so after I get orchastrations I can convert to written music, I also suck at creating melody lines . Gave suno style prompts include wanting no marches or polkas and what do I get, all marches and polkas. I tried about 10 times before I gave up. I have created about 50 compositions in different mixed genres most work but prog rock in concert band, (To me it's a no brainer)and gospel reggae just don't want to work

[remote seal]

Boys what prompts work for you to get great glam metal songs

[manic saffron]

how does one make it so suno knows whos sings certian verses? I want to me a duet but even with its like [Verse 1 Male] it'll have him and the woman singing

[polar spade]

You can just about search "duets" in here or peruse the workflow channels, but then sift through and find some good chats and examples of what different users are trying. There's probably a good 8 or so different techniques to at least try.

I, for one, like to create a vocal per an extension and add them one by one.

[manic saffron]

I didnt even know you could work on one vocal and the add it later is there any tutorial on how to do this?

[polar spade]

In an elementary way, you can add an extension each verse and prompt accordingly (based on your preference from your research) so if any issues, you tweak the prompts and try again, only needing to focus on 1 stanza at a time.

The long version is only in paid subscriptions where you can find a vocal you like, create a persona. Then as you extend each stanza (verse/chorus/bridge) ytou also attach that persona, helping to "force" the male/female vocal into thats specific stanza.

[near flume]

leave the lyric changes in both pre-chorus the same, rename the first instance [Pre-Chorus 1] and leave the second as [Pre-Chorus 2]. This maintains a structural flow.

[manic saffron]

ok thank you!

[bleak plank]

hi everyone, if i want to add an adlib (just a man speaking some intro words, not singing, no instruments), what prompt should i add to the lyrics thanks!

[humble verge]

[Female alto vocals softly hum]
SUNO: Woman screams for five minutes straight

[fringe elm]

[no music] [introduction] [male spoken] and when you finished, [introduction end] [music starts]

[uncut tulip]

is suno broken?

[copper cloud]

It's not workinggggg

Helppppp

Okay so basically my wifi is just betraying me again 😐

[fathom crypt]

I have an issue when I write male Vocal it sings in a female voice and when I write a female Vocal it sings in male voice.
How to fix it?

[honest heath]

are you selecting male or female voice in the advanced settings?

[rapid summit]

Has anybody achieved a brasilian phonk song that doesnt sound HELLA cursed 😭💀?

[winged peak]

Yes, this usually helps. And adding the information in the style-prompt as well as in the lyrics-section like [vocal: male singer, gritty voice, british accent] usually helps as well to get it done.

[near flume]

In the styles box prompt in this format: Gritty Male Vocals with British Accent, (probably won't get the accent - Suno sucks at accents). Capitalized words tend to have more impact when setting styles.

[fringe elm]

Yes a little bit but then again, 4.5+ is still in work. It will get better as it rolls out more in time

it has a thing for singing very long notes or screaming them

lmao

[mossy pivot]

Here a song how broken it is 💀

https://suno.com/song/5a0d5608-e59c-4af0-9d45-bb051251489a

[polar spade]

The obvious silly answer is to prompt for "male" when you want female and "female" when you want male. But we will ultimately see that we're really just getting a 50/50 ratio and some sway one way or another by the weights given to the genre and lyrics we use.

Everyone will give their version of how to prompt for gender, but you may also just search for "duet" here and scroll through hundreds (958 from just "duet" , I see) of chats about it. Pick one or a few and try, if you wish , to get a feel of what they all are basically suggesting. See what they do and whats in common about them all.

[fathom crypt]

Thanx man for advice

[mossy pivot]

But on pc you can select male or female in the advanced options btw

Or you can put [ Male Voice ] in the tags within your lyrics

[polar spade]

Its kind of a meta-comment, I guess. Lol. Yea as @honest heath and @mossy pivot suggest, try out the male/female switch under Advance settings to add to your experiments.

[fathom crypt]

No I was actually trying to make male vocals for one line and then the second line in female vocals. And whenever I write male vocals for that line it comes in female vocals and then next line I write female vocals it comes in male vocals.

[fathom crypt]

I solved my issue and published the song. Thank you for your advice and support 🙏.

[mossy pivot]

Oh oof hmm try

( Female voice ) I spin like a carousel
( Male voice ) it goes fast like Duracells 🤣🤣🤣🤣

Idk if that works but one time I tried ( whispered ) bablabla
and it actually whispered the part

[fathom crypt]

Check it out. I solved what i wanted.
https://suno.com/s/ZfAWMI1kDpLxgzje

[mossy pivot]

Ok just dm this to me I need to finish lunch then I’ll listen to it

[fathom crypt]

Send me dm and request. I will try to send you there

[quick sonnet]

Can I also specify musical instructions in the tags so that the sung chorus is underlined by a melody, while the rapped verses have a corresponding hip-hop feel?

[mossy pivot]

Follow this

This is how 4.5 + handles your styles and tags

[quick sonnet]

I am a free user, so the prompt is limited to 199 characters.

[mossy pivot]

😔

Then just use like this (120 bpm , Rock , Male Voice , Sad ) that’s the way 3.5 works for prompt style

Then in your lyrics just use tags like this
[ intro : Electric guitars , synth pads ramping up, Female voice]
……………………..(…..)
……………………..(…..)

That’s how 3.5 used to work

The ( ) are the ad libs it’s optional but makes your song more professional

[quick sonnet]

Well, the thing with [Female chorus] and [Verse 1 – Male rap] doesn't seem to work somehow. In the end, I came up with something in the rap part that sounds like Niki Minaj for poor.

[winged peak]

Suno always takes certain "liberties", so the only thing that helps is to keep refining and improving the prompts and trying again and again. and again. I managed to create a duet relatively quickly, but there was also a bit of luck involved: https://suno.com/s/EzO7rdtPOuQCirgE

[quick sonnet]

But duet is not what I want, every voice has it's part. The male voice doing the rap part and the female one singing the chorus. It's structured like a track by Jay-Z and Beyonce, for example.

[quick sonnet]

I am using the free version, which is limited to v3.5. Is this posibly the reason for the problem?

[graceful condor]

How can you incoporate a crazy laughter within the song. Like within a Carnival Crazy Waltz? I tried like this in verse and so (Maniacal Laughter) but it rarely works.

[quick sonnet]

Well, ultimately, that wasn't the worst result. The others were full of dissonance, but some rap lyrics just don't fit a female voice. Even though I know I wouldn't have a chance in this remix contest anyway, I tried writing something for fun. Unfortunately, the beginning of my lyrics wasn't copied into the prompt. This is what came out: https://suno.com/s/yGIf63ZvZWR9eo4Q

[unreal socket]

Anyone knows how to fix this? I changed a first occurrence of a pre-chorus and I renamed it after as [Pre-Chorus 2], but later on the same music I have the old version of [Pre-Chorus] again, I wanted to have a the new [Pre-Chorus 2] also in the second occurrence but Suno generates another Pre-Chorus even if I tagged [Pre-Chorus 2] in the new generation. I don't use the extend option or it will delete part of the songs I wish to keep, which is the bridge and the final, how do you guys solve this without using a DAW and only in Suno?

[mossy pivot]

https://tenor.com/view/midnight-the-cat-stopped-working-midnight-the-cat-cat-cat-meme-funny-cat-gif-81113296414293415

[mossy pivot]

U can use the editor

If you a paid member

[quick sonnet]

So here the closes to what I wanted, with male rap and female singing: https://suno.com/s/oaPbsPDC5xRzEiJp

[lost holly]

Have you tried copying this way?
As far as i know, the style instructions in the lyrics (Male/female vocals) plus the duet style in the style field are the key factors here to get parts sung by a female parts by a male.
In my experience it wont work out always but often enough, so you might have to give it a couple tries.

[polar spade]

If a paid sub, one option is to duplicate the chorus 2 in the edit UI and paste where needed, as @mossy pivot mentioned. Lots can be done in the Editor.

[quick sonnet]

The problem is, for free users the styles prompt is limited to 199 characters. Also to much information is not always giving the best results, for example the instrumental intro was ignored.

[polar spade]

When we had only 2, 3.0 and 3.5 that was true. So, many songs were either formed around the limitations ( rather than seeking a solution to fit our song) or we would start over often and make the best of extentions. -- slowly and meticulously completing one at a time until done. Combined eith a bit of the former (write to fit the limits) It seemed to pay off for many.

[quick sonnet]

As a free user I can only use v3.5.

[polar spade]

Thats what I addressed. Maybe I've forgotten, are extensions not available in Basic?

.

[polar spade]

Oops... apparently i'm logged in as undefined again. Love it when that happens. My secret identity , I guess.

[quick wedge]

Free users should be able to extend with v3.5 that’s what I always used before the switch over to pro unless extend isn’t there anymore 🤔

[polar spade]

OK. Thanks. Yea, I was logging into my Basic account to check and I got this. 👆

[quick wedge]

Something must be wrong or it’s a visual bug. Try reporting it in the channel and let them know of it 👍

[polar spade]

Its always been best to plan out the song first around using extensions, and prompt through each extension . . . rather than prompt for the whole song, run a few generstions and then try to go back and figure out how to fix or augment the parts thst didn't work with just a couple of tools.

Ytou know more than you did a day ago, so starting over will be much easier.

[quick sonnet]

As a Pro user I would most likely do it exactly like this. Looks like the extend feature is also avaible for me. But what it is good for? Can I just add an istrumental intro, which I forgot, to my track without changing the rest?

[polar spade]

I usually just wait it out and it goes away after awhile. Its like getting carousels and mushrooms for Suno artwork, it just means they're making system adjustments.

[polar spade]

I always want to create a song while im logged into this account. Hahah.. just to screw with it a little. 🤣 But thats not my style.

[quick wedge]

If you ever get bored on anything else you can always test and experiment on it 👍

[polar spade]

Hey I had a moment of "torschlusspanik" . I realized, if we log into one account, bring up a Playlist, open another browser page and log into another account, the first account playlist will paste in that new account. Totally useless trivia, but interesting to find out,

Can get the same result by just copying the Playlist url, open/login the second account and pasting in the url in address bar. Like I said. Useless trivia.

[quiet solar]

Without the Editor, it's quite difficult to re-do something from the start of the song, as the Extend feature effectively forgets everything that happened after the point you choose to extend from and regenerates it all over again. Even with the editor it's still a bit hit and miss changing things at the start of the song.

As a free user, you can get the song to fade right down to silence and then extend from that point, prompting for an instrumental intro going back into the start of your song. But I think you'd have to use an external editor like Audacity to stitch it together into the correct order.

As a general rule, getting the start of the song the way you want it is absolutely critical - it's much easier to then build on/fix subsequent sections using extend from the bit you are happy with. Doubly so for a free user, as you don't have access to the Replace function.

Dasvon's advice is really solid.

[quiet solar]

Also just going to add that this is incredibly difficult to get consistently right in Suno. Doesn't matter how tightly you prompt, the AI struggles with switching singers - be that a male and female, or two males with different vocal qualities. It'll ignore half your prompts, or quite happily mirror them the wrong way round...expect to go through a lot of regenerations and possibly a fair bit of compromise.

Best advice I can give is try to limit how frequently they change over, and either;

a) Give one of them a unique section (e.g. a single rapped chorus, performed by the male singer, with the rest by the female singer), or

b) Try to get Suno to associate one with being a backing/alternate singer. That way, you can use style markers in the lyrics to ask it to switch between them, e.g.

[Chorus]
The lead singer does this part "BACKING SINGER SHOUTS THIS"
The lead singer does this part (backing singer comes in afterwards)

[polar spade]

It was just a necessity when all we had was 3.0 and 2 minutes, or 3.5 and 4 minutes and no other way to edit.

But having those limits also made us focus those skills at hitting things in one shot as well.

I believe 3.5 was Pro level though, so we had the credits to work with to hone those skills.

Now im spoiled with Paid sub so I hardly use old skills much. Skip lunch one day a month pays for a month of Suno. A good trade.

[quiet solar]

Yeah, I feel the pain. To be honest, I've been taking things created out of Suno and stitching the pieces together in Audacity since v3.

The editor has made things a lot easier now its functionality has grown.

[quick sonnet]

I see. But what about outro? There is a truck whre an instrumental outro is really mising, so adding 10 or posibly even 12 seconds outro would do it good. 12 seconds is the maximum time left for that track to 4 min limit.

[quiet solar]

The time limit only applies to single generations - if you use extend and combine the two pieces together, you can go beyond the 4 minute limit. So you're not limited to only 12 seconds for the outro.

Advance warning; you may find that Suno doesn't always like doing super-short generations, so if you take a point near the end of your song and just put [Instrumental Outro] in the prompt, you might find that outro ends up being about 40 seconds long. Or the song sort-of fades out, but then comes back in again as a hallucinated set of instrumentals.

Again, if you have the Editor it's quite easy to work around that. If not, you might want to set your Extend point a bit earlier on in the song - for example, if you give the extension a Verse and Chorus to get through before asking it to add an instrumental outro, it may play along far more nicely.

[tranquil leaf]

I'm trying to confuse it even further by having the voices switch languages.

[cyan jewel]

hey um when using a persona are you only able to generate a song as long as the persona?

[wispy laurel]

No
But the persona itself takes up some of the space that Suno is working in, meaning that there's less space available for the new creation. Songs using personas very rarely generate more than 5 minutes long at most
You can still extend past that point though

My most-used persona's source song is about 1:20 I think (because I cropped it down to just the best bit), but I have no problem generating normal-length songs with it
BUT with that particular persona it refuses to generate anything longer than 5:02 in a single generation because the "space" in which the other 3 minutes could have been generated, is being used to hold the data from the persona instead

[cyan jewel]

i was meaning without using extend. whenever i prompt using a persona it always makes the song as long as the persona.

[wispy laurel]

When you make a song using a persona, Suno isn't hearing "make a piece of music that sounds similar to this".
It hears "read this really long list of variables, then generate a new list where most variables' values are equal to or within X of their old value"
And then that list of variables gets translated into sounds
So it does make sense that the length would tend to be equal or close to the persona's length, because the mathematical process involved deliberately points that way

BUT, not in every case - my 1:20 persona consistently generates normal-length songs, say 4-5 minutes, because it won't match the length if the lyrics and structure that you give it can't fit into that length

[viral pond]

how can i avoid playing all the replacement clips and editor artifacts from mobile phone app ?

[unreal socket]

Using the Beta Editor to cut or erase [Pre-Chorus 1] and replace it manually with [Pre-Chorus 2] does not solve the problem. Suno doesn’t reuse the existing section — it either generates a new one or, if I try to copy/paste, I often get glitches/distortions, even in parts of the track I didn’t touch. It feels like the system re-renders transitions between blocks, introducing artifacts and a kind of “tractor engine” distortion randomly across the song.
Even when I try to carefully adjust the beginning and end of pasted blocks to glue them smoothly with the surrounding sections by using replace on its start and finish, it's almost impossible for the results to be good and other parts of the track are unexpectedly affected, and new glitches appear elsewhere. This makes it almost impossible to get a clean result. It's giving me an headache, if someone knows any trick that works, I realy apreciate it.

[modern spindle]

I have a friend who wants me to create some songs for him. He is wanting me to put male and female vocals strategically throughout the songs. How do you notate where male vocals are in a lyric, and female vocals. He would like to have certain parts where the female vocals will be singing for the female lines, and the male vocals for the male lines.

[honest heath]

I dont think there is a reliable way to tell suno if a voice should be male or female for a specfic section.

[fathom crypt]

Everyone is still talking about it.

[fathom crypt]

There are many things you can tell suno. Even you can tell suno for a single line should be in male voice or female voice.

[unreal socket]

I tried that but with no success, if you do copy-paste in the beta editor since I'm a paying customer and I can it won't be perfectly aligned with the rest of the song because of variations on the endings, then if you try to replace the touching parts to try to align it, you'll get like "0,1% probability" of success and the rest of the song gets glitches and distortions since I think Suno reender all the song in a replace and detects something is wrong with your copy paste since it wont be perfect... this is realy hard.

[polar spade]

If i may respectfully put a positive spin on your comment which you needed to make twice:

All one can do it suggest things to a user that have worked well for us, or worked at least 40% of the time. Thats where I'd put the benchmark of success. So far, Im making more money than Im spending on a 40% success rate.

So, yea, we can disclaim our suggestions as ....well...just suggestions and honestly note that Suno is not a DAW but uses algorhythms to blend things as best possible lending it to have issues quite often, but still worth a go, in the end.....or.... we can say, "theres nothing can be done mate...." and move on.

The balance is to try to be encouraging, have them try on their own and hope they hit a lucky roll within that 40% bit, right? Thats how I usually go at it.

Using edit UI works very well for me for my needs and duplicating choruses work spot on, at minimum, 75%, if not 100% of the time. The other 25%, I've hopefully the skills that I do well to manually splice it so well and include both remastering tools or clever composition that its indistiguishable and sounds fine, or at minimum, makes any imperfections that might remain just increase its authenticity. All without taking it out of the Suno world. Its literally no different than what one would do when splicing in a DAW.

So, fair enough , your comments. Sorry it doesnt work out for you. Perhaps suggest or add something then that you prefer?

[cyan jewel]

im having some real difficulties with Soprano x Contralto duets. Anyone able to impart upon me the Knowledge™

In general [Contralto] seems not to work on Suno. At least, not consistently. Even if I use specific vocal ranges.

[ebon sand]

You are correct, when you go into the editor, even when you change nothing and just export the song as it is again, it will be re-rendered. Do not use the new editor, is my call

[mossy pivot]

Some tags work once in a bleue moon lol

It’s either Suno couldn’t find this pleasant enough for the overall vibe or the tag arrangement must be in a certain way ( wich I’m still asking myself that is it this [][][] or
[]
[]
[]

[cyan jewel]

hmmm

i wonder if i just have to put [alto]

[trim glade]

What do I have to do to make the songs longer than just 4:44? Every song I generate is longer but ends abruptly at 4:44.

[opal orbit]

Covers don't really go past 4:40 - 4:50ish. Is been a long time issue

[trim glade]

ah understand thx v.v

[opal orbit]

you can still extend the song after getting a version you like though

[trim glade]

yeah i know but Extended is Difficult , its change the Style near Completly

[opal orbit]

Are you doing it from the end, or from a section before it?

if you extend from a spot that is a little further in, it will usually take more of your songs style into consideration

[trim glade]

ah thx for the information , i try it , i make it most from end

[quick sonnet]

Wondering if extend feature can help me to fix a track, where the first verse an corresponding chorus are mising?

[quick sonnet]

But you can't be sure that it will work that way. The probability that it will work the first time is very low.

[spice timber]

Anyone know if, in the new song editor, there's a way to actually confirm a lyric change? For me there's just a 'Cancel' button below the New Lyrics text box, and if I hit Regenerate, it uses the lyrics in the Lyrics box (which haven't been replaced by the New Lyrics) instead of the new lyrics. It's my first time using the new editor so maybe I just messed things up (initially I didn't know about the 'Replace Lyrics' button and typed straight into the lyrics box and it messed some things up I think)

[quiet solar]

Yes, if you have access to the Pro tools. You could use extend to recreate a 'missing' verse and chorus, combine them into a single song, then use the editor to move that verse and chorus to the opening part of the song.

If you're only using the free version, it'll be a lot more tricky. You'll need to use Extend to effectively start the song again using your original generation more like it was a template. Then recreate the rest of the song afterwards either in that same extension, or by getting one bit right and then using extend again to do the next part and so on.

The problem that leaves you with if you merge it into a single song is that your song will 'start again' after the original version finishes, and there's no easy way to remove the original version without exporting the whole thing as an mp3 into a separate audio editor.

Genuinely can't stress enough that getting the first part of a song right is really important in Suno, and will save you a lot of headaches. Re-doing the beginning of a song is one of the most difficult fixes to try.

[spice timber]

Thanks for the tips! I'm on Pro, and thankfully not trying to change the start of the song (I remember how hard that can be!). Trying to fix one mispronounced word at the end of the choruses, and using the new song editor, I'm finding it very difficult to change the lyrics carefully. Just need more experience with it probably, I'll keep fiddling.

[fathom crypt]

Wow what a broadcasting everything.🤣. Revealing everything. 🤣

[fathom crypt]

It works for me and good luck to you.

[mossy pivot]

U speak in riddles i do not understand what you speak of

[fathom crypt]

Nothing leave it.

[quiet solar]

I find Suno can be pretty hit and miss with vocal range tags. The pitch of the singer seems more dictated by the key/pitch of the overall song, rather than something that you can easily prompt independently.

It's not something I've tried extensively, but best advice I can give is to stress it early on in the style description and tags in the lyrics right at the start of the song. Alternatively, you could try to upload audio that's from the correct vocal range and build off that from the audio directly, or from a persona created from that audio.

Expect this to be difficult and take a lot of attempts, though. Doubly so if you're also trying to get two distinct vocals to duet or sing parts independently, which is a difficult exercise in it's own right without also specifying their vocal ranges.

[mossy pivot]

Weird I don’t get an issue when I extend from the end

For me it always keeps the style persistent

It must be your prompt the problem

If the style prompt is clear it won’t deviate

[quiet solar]

If Replace Lyrics isn't playing nicely, try extending the song and running a few different variations of the chorus with different phonetic spellings of the mispronounced word.

Fingers crossed you'll get a chorus that sounds very similar to your previous choruses, and you can use the Editor to switch that chorus in for your earlier ones (or just the offending section if you're happy working with a bit more precision) and then get rid of the spare choruses and anything else you don't need for the final version.

[mossy pivot]

The vibe will shift abit but it will blend perfectly within the song if u generate few attempts

[quiet solar]

You can also get it to generate multiple choruses in the same extension, so it doesn't burn through your credit allocation so quickly.

[mossy pivot]

Oh that’s something I did not know

How do you do that ?

[quiet solar]

Just repeat the chorus multiple times in the lyrics prompt.

When you combine that extension with the rest of the song and open it in the editor, you'll have multiple new attempts at the chorus at the end to pick and choose from for fixing the issues in the existing parts. Then you can crop out or delete the extra choruses from the end when you're done.

[mossy pivot]

Ok that’s some serious surgery 💀

[quiet solar]

Hah. I take most songs I make into an external editor and stitch them together to fix any issues that come up, or use the best parts from multiple generations - sometimes from the separate stems. This is fairly mild surgery in comparison. xD

I don't necessarily recommend doing that for most people - it just depends how much of a perfectionist you are.

[mossy pivot]

Hmm I have questions 💀

Côme to dm 🤣

[quiet solar]

Sure

[polar spade]

I hadn't thought of such an easy prep-work idea. Ill have to add that in to some of my next projects. Seems a good time saver. Thanks.

[quiet solar]

Don't set a vocal gender for the song, then tag in the lyrics where Suno should be using a male or female singer, e.g:

[Verse 1, Male voice]
*Random lyrics here
And here
[Female voice]
Female singer does this bit

However, be aware that Suno struggles quite a bit with switching between male and female parts - v4.5+ is the best it's been, but expect it to repeatedly get it wrong, get the parts backwards, ignore the prompt to change singer, etc.

If you have the patience, you might want to get it right section by section using Extend to gradually build up the song rather than trying to roll the dice and get it right in one go.

Alternatively, have a polite chat with your friend about the merits of having just one or two isolated sections sung by a male/female voice, rather than switching them regularly throughout the song.

[trail canyon]

Okay, I will try the editing way too. So far I have had no success, and I have been ignoring this imperfection.

[teal saddle]

If I may, I have been trying in earnest to download a video with my image instead of Suno's image. I have regenerated video several times. How do I do it, please? It's important for my project for the videos to have my images. Thanks in advance.

[cyan jewel]

use something like kling

https://suno.com/song/695658b5-1ac4-4371-bbf1-db9f35b105aa thats how i got this video file on mine

[quiet solar]

If you want your own image to appear on Suno's generated lyric video, then:

1. Under Song Details, replace the image Suno generated with one of your own.
2. Under Download, choose Generate Video. Wait for it to finish, then back to Download and Video.

Your image will also appear on the video if you link to the song somewhere that supports embedding links.

[modern spindle]

Thank you very much for this, this is very helpful.

[quiet solar]

Good luck 🙂

[teal saddle]

Yes, I've done the replace image and generated video, but it still has Suno's image. I've tried both with phone and computer. Does it take a bit after generate for it to come in?

[quiet solar]

It's not instant, but usually only a couple of minutes. If you go back to look at Generate/Download video, you should see the loading icon while it's generating/regenerating.

Is Suno's image still showing when you look at the song in your Library, or on the song's detail page?

[modern spindle]

When changing the vocals throughout the song, is there a notation that I should be using to do that? So that Suno knows when to change the vocals, and that it’s not a part of the lyric?

[quick sonnet]

As I said, it was a stupid copy-paste error. When I copied the lyrics from the text file, it only copied from the second chorus onwards. Well, the verses are rap, so they have to fit the original beat somehow; having a different voice would actually be intentional. The chorus is always the same on the track, so I would only need to duplicate one of the existing ones and then cut everything together in Audacity. Apart from that, I can of course only agree with what you wrote at the end. But I didn't expect the first attempt to produce the best musical result for the track—no off-key notes, no dissonance, and perfectly executed transitions. I certainly won't be able to achieve the same result with Audacity again.

[quiet solar]

Suno understands that anything in [] are style prompts, and 99% of the time won't sing them as lyrics.

If you want one singer for a whole section, add it to the section label, e.g. [Verse 1, Male Voice]. You can also do that to prompt it to start that section with that particular singer.

If you want to switch singers mid-section, use [Male voice] or [Female voice] on a separate line just before the lyrics for the new singer.

Expect a very hit and miss success rate though.

[quiet solar]

I swear there's a certain branch of Murphy's Law that ensures Suno will always produce a perfect execution of the song on the version where you make a glaring mistake somewhere in the lyrics which is difficult to fix. 😂

[quick sonnet]

Isolated sections sung by a male/female voice also don't always work. I tried three times, using a total of six tracks, but only the last one met this requirement. Yes, I'm a free user, but I've read that pro users also have this problem.

[quiet solar]

Being a free vs pro user doesn't make a difference in this case, except that pro users can be more relaxed about trying multiple times and failing.

[modern spindle]

OK, thanks! This is given me food for thought.

[quick sonnet]

I can only use v3.5, which isn't quite as perfect, while pro users can use v4.5, which is supposed to be more mature.

[quiet solar]

The last time I attempted a song with male and female singers switching and occasionally duetting, I was at 44 separate regenerations before I had a compromise that was close enough.

And yes, fair point. Having v4.5 does make a difference - it's success rate for switching male/female vocals is higher, though it's still very unpredictable.

[mossy pivot]

https://tenor.com/view/screaming-cat-cat-screaming-black-cat-meme-memes-2023-gif-5774825797503791230

[quick sonnet]

I'm new here, but is it possible that v4.5 will be released to free users at some point in the future? If I understand correctly, v3.5 was also released after v4.5 was released.

[quiet solar]

So far, Suno have rolled out older versions to free users as they've gone along. So yes, you probably will have access to v4.5 as a free user in time. But then you may be looking jealously at v5.5 xD

[mossy pivot]

Definitely

[modern spindle]

Good to know! Thank you very much for all of your feedback.

[mossy pivot]

5.5 when ? 🤣

With perfect words pronouncing

🤣

[quick sonnet]

I know, but I am doing this thing for fun, so would just like to know then how much better v4.5 then v3.5 is.

[mossy pivot]

V4.5 is miles ahead basically better vocals less mistakes, more complex prompting, better tag understanding, better voice switches

[modern spindle]

I forwarded the first part of the conversation where you were discussing the syntax of what to put into the lyrics to the note, mail and female vocals to my friend. He is not ASUNO user, I offered to create some songs for him. Actually, he asked me to do that. I shared with him my creations, and he very much enjoyed them. I am currently using the Suno pro version.

[quiet solar]

Basically as Requiem says - the new prompting style for 4.5 is the biggest change, and rewards approaching style prompts very differently to how we did it before.

There are still times when (IMO) v4 does a slightly better job than v4.5 in certain genres, though since 4.5+ fixed a lot of those issues I haven't really gone back to v4 very much. I don't think I went back to v3.5 ever since v4 came out - that was a MAJOR improvement that you'll appreciate when you have access to it.

v4.5 also fixes the 'digital shimmer' bug from v3.5 and v4, which was hugely appreciated.

[quick sonnet]

Well, sometimes a slightly incorrect pronunciation gives the song that certain something. I just produced a track where the rap part could have come from me, in my prime, with just as bad English.

[polar spade]

Sunos Inverse Law of Composition Dynamics: Suno's outcome of an increase in quality, fidelity and adherence to prompts correlates with a heighten probability and awareness in the chance the lyrics are screwed up.

This is a first for me. Text in the Suno artwork?