📚┃suno-school

[frozen dagger]

Those are both good ideas. I would like the saved prompt box to be better. Static postions (not most recent used shuffling) and be able to read the full prompts. The start of some look the same and with the shuffling around. I'm already very unorganized. The feature to save prompts is key for me.

[pure panther]

More organization would be very good

[cyan jewel]

does anyone have any prompts that help darken the mix, combat the sibilance, and harsh bright tones of female vocals, guitars, high hats, etc?

[cyan jewel]

@tender eagle ?

[tender eagle]

may be the res I dunno that eq. mine is a different,

[cyan jewel]

this is studio

[tender eagle]

oh never used the eq there, I dump to ableton :S

I just use studio for readding the stems after I eq everythign up and master it

also another thing in studio that may help is to stem it out to instruments, and see which insturment sample has the hiss

and focus on it

my wf is export wav -> UVR stem out -> abelton eq/master -> recombine in studio

I could recombine in ableton but once I have the samples properly mastered / eq'd etc I can recombine them and adjust the minor tweaks in studio

after just freezing each sample in my daw

lemme see if I can do dehiss it in studio never tried it maybe I can see how to go about that.

[tender eagle]

I can't get stems to work atm. so I'll have to wait until this gets sorted for me :\

[cyan jewel]

i tried messing with it in fl studio. the more i messed with it the more im realizing is wrong with these v5.5 outputs.

this is kinda yikes...

[sleek stump]

I've been struggling with 5.5... My current project I started in v5 was giving me exactly what I wanted, I switched over to 5.5 for the next track and golly gee, what a struggle its been, Yesterday I switched back to v5 and got what I wanted pretty quickly. 5.5 needs more time in the oven, and more issue reporting.

[grand berry]

**For heaven's sake, can someone please tell me how I can assign a lyric section to a specific vocal part (male or female) when I'm working on a duet? **

[half dragon]

You can't. Not reliably anyways.
If you spent enough time and credit, you can.
Duets are difficult.

[grand berry]

🫠

[heavy kestrel]

Hi people ! First time posting here... I'm going mad with covers, that's a common subject on a bunch of reddit threads but I really don't find a solution...
I only use Suno to make covers of my own songs :
upload
fix lyrics
don't touch settings
no style
First generations are great (no matter the model, eg I've got better results with v4-all sometimes than any v5)
Then, just for fun I experiment with styles (turn a pop song to and EDM french touch)
From there if I try to cover the same uploaded song again there's a kind of "memory" and all my previous experiments have influence to the new generated ones
Tried to clear my browser cache, no matter what I do Suno still insist on using my previous styles indications which I do not want

Any help will be greatly appreciated !

[neat nexus]

What happens if you make a new Workspace and try starting over in there?

[heavy kestrel]

@neat nexus That's what I wanted to avoid -> keep things clean/tight, maybe I should give it a try

• upload my song to the library I guess
• from there, create a workspace "mysong-try-2" and reference the uploaded from there ?

[neat nexus]

Yeah, that sounds about right.

Just to see if it restores the plasticity for you again.

[heavy kestrel]

I'll try and post here if things go right with this workflow...

Strange,

• I uploaded the song to the library (outside of any workspace)
• created a new workspace
• can't reference the song from the library, or may be I'm blind
• I can upload again (not a big deal), but so what is the library's use case ?

[neat nexus]

Upload in the workspace you're planning to work from. Otherwise it's not uploaded to the new workspace you're hoping to work from.

[heavy kestrel]

Ok, tried this

• click on the "best" generated cover on my first workspace to check if I modified any default setting (answer I did not)
• created a new workspace
• uploaded again
• copy/paste fixed lyrics (with simple meta tags (chorus, verse, bridge)
• model : v4-all
• generate

I got nothing close to the first version, I mean very very far away... Ok, we don't have a "seed" setting on Suno but how this hell is possible ?
I could accept a "slightly" different style, but going from "close to the original" to "techno metal style" is beyong me

For me Suno is too unpredictable to be useful... That's a shame because some generated are very GOOD but never reproducible

[neat quail]

thats how i feel right now too; its difficult just making isolated instrument tracks, there's always some other instrument punching itself in and it doesnt really listen to my instructions very well. difficult to make a song because it keeps trying to make a whole new song for an individual track

[heavy kestrel]

Isolated ? I tried several times, but it's a no go
But in my case I don't even bother with stems (which are useless)
Just having a new version, close to the original generated
Downgraded to "Pro", next step will be stopping at all

Ok, re did the whole process... new workspace, got another style, I think that I should spend more time on doing music by myself than trying to tame the beast

[remote sapphire]

https://suno.com/s/IZ0uU2TOyxUJehEU

https://suno.com/s/iGcQeTAOh2fODamT

Please help me select the best one
I getting too confuse to select one

[vapid steppe]

hi all

any experts here

[frozen dagger]

i was literally about to tag you and ask you how it went

[vapid steppe]

(https://suno.com/s/nofLui8LoYD16Ln0) check this and tell me how it is

total burned credits of today 1600

i was about to try also....heheh good that i didn't.... i am in another problem

welcome sparrow ....

[frozen dagger]

I don't know what kind of feedback you're wanting. Compared to the last version shared it is more consistent through-out than it was before. It sounds good, and I like the lyrics. (after I translated)

Going to be different for everyone. Music is very personal. Question is how you feel about it? (Or how does it make you feel?) Sounds like you have some more thoughts spinning already. Let yourself be free to explore, and enjoy the process as much as the outcome (output).
Did the custom model trick help at all for what you were trying to achieve?

[vapid steppe]

custom model helped a bit...thank you....what i did was that..from the first 2 minutes which was good from the previous generation i trimmed it and made the model ...then i trimmed the original into 3 parts and then generated 3 parts then stitched it in adobe audition

but the main thing is that... suno cant generate a constant 5 minutes song

if suno wont fix this issue , it will effect Suno very badly

i am thinking of sending an email to support

[remote sapphire]

well can anyone even help

https://suno.com/s/IZ0uU2TOyxUJehEU

https://suno.com/s/iGcQeTAOh2fODamT

Please help me select the best one
I getting too confuse to select one

[unreal sleet]

Hi I'm wondering if anyone is able to help me work on an intro inspired by a song?

Like, I can verbalize whats happening but not sure how to tell suno

[Intro]
[Slowly evolving ambient synth pads]
[Distant, muffled tremolo guitar riffs, low-pass filter]
[Volume swell: crescendo]
[Epic orchestral strings join]
[Heavy drum fill transition]

this is where im at atm

[quiet solar]

I have a couple of 5 minute songs that Suno didn't struggle with at all - it's not a consistent bug that affects everything, but a very annoying one to run into for sure

[remote sapphire]

You can try these styles

1. Post-Rock, Cinematic, Atmospheric, Shoegaze, Ambient Synth, Tremolo Guitar, Orchestral, Build-up, Explosive, 75 BPM.

2. Cinematic dark ambient intro, slowly evolving synth pads, distant muffled tremolo guitar, gradual crescendo, epic orchestral strings swelling, dramatic tension build, ends with a powerful heavy drum fill transition, emotional, atmospheric, post-rock influence

3. Cinematic dark ambient intro, evolving synth pads, distant tremolo guitar, slow tension build, massive orchestral swell, intense crescendo, heavy dramatic drum fill, dark epic post-rock, powerful and emotional

4. Ethereal cinematic ambient intro, soft evolving synth textures, distant hazy guitar, slow swelling crescendo, emotional orchestral strings, atmospheric build, dramatic but beautiful transition

[unreal sleet]

but you dont know the song intro im trying to work towards 😄

[remote sapphire]

well u are right

i thought these prompt might help u somehow

[remote sapphire]

Someone help me to decide which one of them is the better version

[unreal sleet]

https://open.spotify.com/track/227jOVNY1MrE5MkUcQRhsk?si=d6e5e44367e9484c

this is the intro for what its worth

[remote sapphire]

Try out this one if it helps or u like it

1. Cinematic deathcore intro, soft emotional ambient synth pads, distant melodic guitar with heavy reverb, slow atmospheric build, melancholic tone, gradual crescendo, orchestral strings swelling dramatically, epic and emotional, tension rising, ends with a massive cinematic drum fill transitioning into a heavy section

2. Melancholic cinematic deathcore intro, emotional ambient pads, distant reverb-soaked clean guitar melodies, slow immersive build, sorrowful atmosphere, gradual crescendo, large orchestral string swell, epic and haunting, builds into a powerful dramatic drum fill leading into crushing heaviness

[frozen dagger]

Using that method of building the song structure, and comparing to the Lorna Shore intro, (don't know if you have lyrics for yours or not) but if not then I'd add this trick to try and give Suno something to attach to for an idea of duration. #📚┃suno-school message
The wording looks fine. And with the right style prompts should be able to find something close (hopefully) that you can start to work from.

[unreal sleet]

I will try that since suno seems to skip everything and go right into vocals

[pseudo loom]

Is it possible to extend a song using another input audio as source? My experience is that Suno fails to reproduce well after a few minutes or maybe a tad more, which is why I would want to extend from a certain point in the song based on the same point in the input audio.

[quiet solar]

No, Extend uses the pre-existing song as the source input audio. But if you Extend from a point before where you feel things are breaking down, that's definitely a way to 'reset' Suno's inputs.

[pseudo loom]

Thanks for replying!
Right, which then means it tries to extend from the point of the input song where I intend to extend?

[quiet solar]

Yes. It'll use the prior audio from the song as an audio guide, then continue on from where you put the Extend marker using what you have in the lyrics box.

[pseudo loom]

Right. So I suppose it will try and match the same parts. I will try it. Thanks!

[quiet solar]

Yeah - for example, if you ask for a new chorus in the extend it'll reference a previous chorus for what the melody and instruments should be.

However, Suno creates everything from scratch any time it makes something, so changes to instrumentation can happen. It won't necessarily perfectly clone the previous audio for the extension. But if you use Get Full Song when you're finished, you can clean the end result using Remaster or (if it's not too long) Cover.

[pseudo loom]

Right. Because my issue is that it keeps to the intro, verse 1-2 and chorus 1-2 without much issue at all. Then there’s and instrumental solo part and verse 3, which it doesn’t get right at all. The verses are all the same melody.

[quiet solar]

Extending from a point at either the start or end of the instrumental section sounds like your best bet.

[polar spade]

If they are that close, I would flip a coin. In other words, you'll always wonder why someone's taste chooses one over the other. If you post only one and we heard only one, we wouldn't know the difference.

[unreal sleet]

suno gives me ZERO freedom if i have my own model built out of songs I fed it

[frozen dagger]

I've only made 2. One with collection of songs I made. (Wildly drastic styles) and another with the same song repeated to work on a cover.
Still new stuff, but curious about everyone's experiences with it so far. I've been describing it like Inspo+ at larger scale.

[unreal sleet]

for me, barely any word slip-ups anymore. but every song structure is the same no matter how i prompt

[frozen dagger]

how many songs did you feed it and are they all similar or different? That would (probably) be the heaviest machine weight in it's calculations, and we don't have a slider for it lol. No way to edit the model to try and remove (or add) 1 or 2 quickly to see what difference it makes. Takes a few minutes for the model to be created so slows down the learning process.

[unreal sleet]

I think the model is biased because of the songs I fed it. I was under the impression it was used to keep voices and style, not copy directly from it like it sometimes does.. completely taking the melody etc

and like i said, its only allowing me to follow 1 structure

[frozen dagger]

yeah that's what it does. that's why i compare it to inspo+. You will hear spliced in pieces. It's for honing in on, where other models has the millions and millions of songs, this kinda puts a target on songs. "Do this Suno"

[unreal sleet]

its a bummer becuase ive been fightning the last couple months with word slips and bad pronounciation and this new model handles text to speech so well

[frozen dagger]

For my use it is amazing. For my aleatoric love of indeterminate music I can put in my wildly different stlyes and create something that i don't hear anywhere else.

[unreal sleet]

Yeah, I just wish I could do a longer intro but it goes straight into vocals because the songs I used for my model doesnt have very long intros :/

[frozen dagger]

5.5 is still the model under the hood for the Custom Model i think? like it's what it is based on. the 5.5 has been like you just said, much better with pronunciation and just coherence.

[unreal sleet]

100%

[frozen dagger]

this is why i play in studio.

[covert granite]

I'm having the oppsite problem jI have a piano jam with a 12 bar intro then verse chorus and bridge.... 2 min... I'm trying to flesh it out and it refuses to do solo piano for the intro and insists on puting drums and bass at the beginining. I've tried tags in the lyrics and instructions in the style to no avail... 30 takes with help of grok..... no go

[unreal sleet]

sounds like we similar issues not getting an intro

sorry for writing like a potato i have a fever 😄

[frozen dagger]

sometimes it's better to work on the instrumental first, and studio is just easier for me to create a new track, select the area where the singing should be and create it, not prompt for the location.

[covert granite]

I did that..... 2 min of just piano with the intro the verse corus and bridge played simply

[unreal sleet]

I guess it's a principle but I would never pay more because the current is flawed, not willing to give them more money to fix my problems that should work in the first place

Like if I paid for my phone, and the data is getting drained faster because of an issue with the phone, I wouldn't upgrade my plan to one that contains more data

[frozen dagger]

create the most basic parts. each as their own gen. to capture different vibes, split stems to remove what dont want. chop them and place them in position with relative timing. they will be horrible and not match, but cover will make it better and keep the flow (mostly, with good prompts). That becomes the instrumental base I start working on to create an instrumental to add lyrics. Just a lot of iteration. (not for creating a specific melody from the start, rather finding the melodies from iterations that i like to build from.)

[covert granite]

I already have the melodies because I write my own songs on piano

[frozen dagger]

good analogy. agree. The Suno app was draining my data when not using wi-fi. I wouldn't have gotten a larger data plan just to fix that. lol

[unreal sleet]

😄

[frozen dagger]

is the problem mainly with trying to get the instruments solo? or sticking to your arrangement as is?

[covert granite]

how do I prompt it to reproduce what I playd but just clean it up a little. I play piano and then if I set it to 90pct it copies too much bu if I set it to 80pct it adds too much nd changes the music.

well both really....

[frozen dagger]

well, a big part of it the nature of what it is. transformative. Finding that sweet spot can be very difficult. I think the solution lies somewhere between a good prompt (and sweet spot sliders) and part Infinite Monkey Theorem. Where we are creating and hoping the output matches (or satisfies) what is in our mind.

[covert granite]

I wish I could say cover it faithfully or totally and have a fender rhodes piano replace my acustic

[frozen dagger]

We are not at the sci-fi movie stage of that, yet, but that is the general idea.

[covert granite]

is suno even heading that direction?

[frozen dagger]

couldn't tell ya. i don't know what they think. but i like playing my acoustic and transform it into stuff, or riffs into more ideas. i think we are at the exploring and experimenting stage, and refining as best we can as we go. it's all new territory.

[covert granite]

Well I've wasted 3000 credits on 1 song now trying to get it to do it.

I've tried a lot of prompts including negative prompts NO DRUMS NO BASS and of course it adds drum and base to my song no matter what.

[frozen dagger]

I see a lot of people feel this way. I think we all been here at some point. Most that stick with it tho, find a new approach, or things improve and gets better. Everyone has different needs tho. Wish we could analyze what you've been trying to far and help further. There are a lot of moving parts.
For me, I was struggling with pronunciation back in the day. Heteronyms, phonetic spellings, songs would be great and then absolutely butcher the word "perihelion", sometimes very comically. I try not to think of it as a waste, but as a learning experience. I still go back sometimes to listen to it scream perihelion all weird just for a laugh.

[covert granite]

I would expect if you had exclude dums and xclude bass it would do it...... and if you put it in style solo piano no drums or bass and you put [instrumental:piano only] or variations of that nad ply with every combination of sliders you would get faithful reporduction of what you played.

without a band!!!!

othe original is just piano

god how doe anyone doe mary had a little lamb with piano only

i appears to me from sound it tht it writes the band and excluded stems it out.

[frozen dagger]

i know what you mean here. logically yes, i expect if i say exclude drums, it excludes drums. @unreal sleet echoed a similar sentiment before about the male and female genders. This is what is says it does, that's what one expects it to do.
Still, anyone who works with any for of ai knows negative prompts can be "iffy". Like unwritten rules.

You're both right tho. People expect things to do what they say they do. It's frustrating.

[covert granite]

but even positive prompts..... something simple piano solo for style..... feed it another piano solo set audio influence to 90pct and style to 90pct..... isn't that what you would expect where did it get drumbs and base in that?

[frozen dagger]

indeed i would. but i don't know how this magical stuff works myself. i know it needs to be trained and worked on. And for that, I have faith they are working on exactly that. (Just me hoping and speculating tho lol)
Sounds like could be the marketing push for V6.
Check under our hood now.

[covert granite]

where is under our hood? here in discord?

[frozen dagger]

I just mean the next model v6 working with creating with stems. real stems from the start with single instruments. might be v8, i hope not.

[covert granite]

I hear you that woudl be cool..... I was just wondering where you read that where do i read "under our hood"

[frozen dagger]

sorry. "under the hood" pun for checking the engine. the suno model engine v6 engine.

[covert granite]

where do I read about 6?

[frozen dagger]

i'm just speculating (fantasizing) about the future of it. I don't know that there is any literature for it. Usually they keep stuff hush hush.

[covert granite]

ok.... sorry too literal.... smirks

laughs at self

[frozen dagger]

not you, its me.

my friendly advice. take a small break form the song. burn some credits playig with suno. give it some old stuff u had, anything. try exploring. come back to the song later.
in any case, i wish you luck with it. best wishes.🕉️☮️

[patent matrix]

bought pro to try voices but then i go check youtube’s tutorials on how to make a voice on suno out of my own voice and it’s not working pls help

same thing happens on my computer

[sweet ivy]

Have a section highlighted in Studio. I would rather slide to the right, than delete, but I'm touching the left side of the highlighted area, running along the edge, but I can't get the double arrow to appear. Do I have to hold down another key. I tried holding Shift, tried Ctrl ???

[frozen dagger]

If you go from the Create Tab should look like this clicking +Voice then see the Create new voice window.

[patent matrix]

i don’t see it on my computer

is it available in only certain countries?

[frozen dagger]

From library same thing but looks like this and when you click crete goes straight to this screen

[patent matrix]

just tried, same thing as before

[frozen dagger]

Due to copyright laws that is probably true. Would need to check local stuff

[patent matrix]

but it’s me though

oh u mean

nvm

[patent matrix]

I’m in italy, is it available there?

[frozen dagger]

Quick search doesn't show anything saying it is not, but you would have to check. I'm in US

[patent matrix]

it is available in italy i saw

but still

still like this

i can make voices out of normal songs but not extensions uploads or stems, normal songs doesn’t have my voice in it

sorry for the pressure btw

[frozen dagger]

@patent matrix Heard. Thanks for sharing that. This shows it is skipping a screen. I've seen another user in different part of the world same thing. You should post those screens and detail the issue with https://discord.com/channels/1069381916492562582/1479483412250366054 probably. Couldn't hurt to share in the https://discord.com/channels/1069381916492562582/1082775596767117312 so they can see it if they have not seen it already. Which I'm gonna guess they have. Either way. That's what I would do.

Also, try clearing browser storage (try another web browser), this didn't work for other user, but it's worth the try.

[patent matrix]

ill try with opera thanks

[sudden kernel]

anyone have any issues/tips with recording voice? ill sing a full 2 minutes, hear it fine through the monitor playback on my headphones, but then the recording playback is just dead noise.

[patent matrix]

i just tried with opera + american vpn still the same thing

[frozen dagger]

making a mental note of it, thank you for sharing that. I would include that with https://discord.com/channels/1069381916492562582/1479483412250366054
and keep an eye out for hopefully this being fixed soon. I hope you're able to create with your voice soon. 🙂

[frozen dagger]

I waiting to see if anyone else has a thought for you, but in the meantime. I know that I have before made some mistake with my settings because of OBS and other things that use the mic. Even discord itself. So to troubleshoot I would look for a conflict with settings with multiple apps using the mic. Try closing some things out; see if that helps.
If you're recording live, try using something else to record yourself and then upload the audio file instead of live recording. This would test if its doing it elsewhere too.
If using windows, can use Sound Recorder.

[sudden kernel]

thats a good idea, and ironically..... i just noticed i had 2 suno tabs open, so that may have caused it lol

[dense trail]

Hi

[spice kraken]

New user here. i created a voice from a song i made to use in other songs, but it seems quite distorted or muddy and becomes worse as the song plays. anyone have an idea on how to prevent that from happening?

and no im not recording my own voice, i hate it😭

[sweet ivy]

Cam anyone help me with this. I have a region of a track highlighted, a white line on either side of the region. There is supposed to be a double arrow that can come up, so I can pull the left line to the right, silencing that part of the track, and not deleting it. I keep touching the sides of the left hand edge, but that double arrow doesn't appear, just the big hand ?

In studio of course

[frozen dagger]

Question for all the Suno Scientists in here. Curious about if the question, "How would you describe your singing?", after voice verification. Does this have an effect on the Voice we create? Would selecting Professional be something closer to the source audio? Or we thinking this is just to help with using voices with training, or something else altogether?

[frozen dagger]

Studio? or editor? This sounds like editor. I think what may be happening is there is some wiggle room when you are moused-over that thick white bar. If the highlighted selection is a section (just clicked a section that was split in the track to create a highlighed selection), then the "highlighted selection" and the break of the split overlap.
If you're mouse is in the center of the white line it is a shorter white line; this moves where the split in the audio is.
If you move the mouse to the top, or bottom, of the white line you'll see the line gets taller, this is the selection to move the "highlight" section; not the split.
Image won't be much help, but it is a weird thing to describe.
This what we talking about?

[sweet ivy]

@frozen dagger Studio

[frozen dagger]

this?

[sweet ivy]

@frozen dagger so highlighting within th track doesn't matter it has to be in the context box?

[frozen dagger]

well, right now i'm just trying to figure out which bars. trying to follow your steps but i am just not sure which bars you're referring too. If it was the context window, the playhead often gets in the way when trying to select it and the big hand icon shows like you were describing. Which we just drag it out of the way so we can adjust. Just not sure which part of the ui.

[sweet ivy]

I'm talking about the thin white vertical line that comes up when you put your mouse in the middle of a track, you can drag it left to highlight a region.area

[frozen dagger]

playhead. look for that little dot and move it out of your way.

[sweet ivy]

@okay, so I shouldn't be using that little vertical line that comes up inside the track, go to the top and set left and right borders of a region with that

[frozen dagger]

if this is the white line you are referring to. this pops up when we click anywhere on the track. this is setting the playhead to play from here. it looks like a grey circle here (above the blue triangle), when you mouse over it it turn white circle like i showed. it sounds to me like it is just getting in your way.

it will be in one of 2 places after clicking and draggin to make a selection, either at the start or end of the selection (depends on which way you drag).
The only way to adjust the tracks (similar way we can in editor) is with these white dashes at the start and end of split sections. Or adjusting the context window i showed earlier; that is always at the top and used to show what Studio is looking at when doing Cover and Replace.

[sweet ivy]

@frozen dagger Thanks I think I get it now. mark regions from the top, not from the middle of the track. Probably easier to use my DAW I'm used to

[frozen dagger]

Awesome. Sorry for the confusion on my part. ☮️🕉️🎶😊

[patent leaf]

is there any way to split a synth further apart? im trying to remove a super saw which is layered on a pluck

[pure panther]

If I were asking the question you're asking, I would upload the synth track and try extracting stems again. No idea if that would work, but it's what I would try.

[patent leaf]

I tried that but it didn’t work. A method that kinda works is putting it in studio and then generating a clip of same length but with the prompt as what I wanted so I wrote super saw, kinda works but the underlying instrument changes a bit

[frozen dagger]

I was about to suggest something similar. Like Tea said, I'd try to stem split again. If possible maybe use EQ to filter it out (I'm gonna guess that this will be ineffective if the stem split isn't splitting it after multiple attempts.)
You have the right idea with Studio and trying to recreate it. Prompt for sawtooth the way you want it. And try to recreate something that you can swap in that doesn't have the pluck but sounds the same.

[patent leaf]

EQ doesn’t work cuz they both sit in the same are, the single instrument gen was what worked best but I’ll keep looking for other methods

[frozen dagger]

play with the sliders in advanced settings trying to cover it, and this would be one of the few instances where if it's still got a pluck i would try the exclude, but i don't like using that.
if not already doing so, remove everything else but what you're recreating.

[earnest fulcrum]

Hey. I have a question. Im trying to make suno add my vocals to a previous song i made, but cover has remix, inspo, mashup and sample. Remix seems to leave the original voice in at places. When instead of cover i choose edit vocals however (add or replace song vocals) it removes either my voice or replaces the song i wish to edit with my voice... Like, how do i tell it to actually do what i want?

[sweet ivy]

@frozen dagger I was able to build the song the way I wanted in Studio. I know Suno saves everything, but if I close Studio, should the song edited in Studio show up in my library. I don't see it there ?

[frozen dagger]

look for Export to put the song in your library. The project auto-saves like you said.

[sweet ivy]

@frozen dagger Thanks it

Thanks, it's there now, so I feel comfortable closing Studio.

[patent leaf]

do it in studio preferably, then select area, prompt lyrics and style.

[frozen dagger]

You likely want to use Cover and +Voice with the cover.
We can't combine +Voice with the other stuff. (mashup, inspo, edit vocal)
If it doesn't sound like the song is using your voice there are a couple of things we can do.
One would be to extract stems (vocal + instrumental) and use the instrumental stem to Cover and +Voice, so we can add our voice and lyrics to the instrumental (without the old vocals bleeding in or taking over)
I would try that first.

[frozen dagger]

@earnest fulcrum This is a good suggestion as well if you have Studio. Same things I said can still apply, but working inside Studio does feel like it works better. (maybe i am bias, i prefer working in studio most the time)

[earnest fulcrum]

studio seems to require the premier plan...

[frozen dagger]

it does, i was assuming you didn't have it. if you did you can do same things there, but not necessary. can still do with pro ofc. (with the cover voice stuff i was saying)

[earnest fulcrum]

I have the basic, cheaper plan... not the expensive one.

so i have some access but not all i think

oh... its the editor that i have i think

[frozen dagger]

you have everything we do, except studio. you can still separate stems to get the instrumentals to cover if the vocals won't chang or bleed in when youre trying to cover the songs with singing in them.

[earnest fulcrum]

Yeah but the problem is that if i get the instrumental stem and try vocals on that... it changes the song very very badly and alot. It does not follow the same rythm, style or anything really. Though i did try it with everything but weirdness almost maxed out, whitch was at 0%... hmm... i do seem to get better results with looser constraints though.

[frozen dagger]

it should be pretty similar with the changes it makes to the songs when we cover whether it has singing in it or not.
usually start with default, see what it does. sometimes default is pretty good.
if we want it to stay closer to the song id try
boosting the audio influence up to 85 first and see what it does.
weirdness at 0 is supposed to be like saying "no freedom to make changes" but as others say it's better to think of those sliders as 50 = default (like 0) and moving in either direction is + or - from center.

going in either direction to the extreme can get weird. takes some experimenting to see how it reacts.

@earnest fulcrum another suggestion others often have is to try a lower model first and then Cover with 5 or 5.5
5.5 has been mixed reviews. so maybe stick with 4.5+ or 5 if not getting desired outputs.

[earnest fulcrum]

sigh, not all updates make things better for everything 🙁

and another sidenote... i cant even recognize where its my voice and where it isnt >.<

[frozen dagger]

things are always a bit bumpy. that's why we are here too. help each other figure it out, improve it, find workarounds, etc.

[earnest fulcrum]

v5 seems to do much better though... but im not even sure if its my voice still XD

[frozen dagger]

would use the "ChillPanic" hack 😄 with Custom Model to make the Voice and see if that is better.

[earnest fulcrum]

What is chillpanic hack???

[frozen dagger]

#📚┃suno-school message
look around here for where i told other user how.
basic is add your voice upload multiple times to a custom model first.

[earnest fulcrum]

okay i missed... all of that XD I think im too beaten down to focus on what he is saying. Thanks for the video though. Think i need to come back to it later though.

[frozen dagger]

Hopefully improvements continuing behind the scenes will be so that you won't need to worry about it. Good luck. 🕉️☮️🎶

[earnest fulcrum]

Thanks. Wonder if youve seen hannah montana. There was an episode where lilly tried singing and miley edited her heavily to make her sound good. This is kinda how i feel right now when i think about the clip i uploaded and what i seem to be getting. "What did you do with my voice?!?!?" "The edited one? A lot!" 😄

[frozen dagger]

Out of curiosity, when you uploaded, and did the voice verification it asks what kind of singer you are. I'm curious if that is any effect. What did you choose? Beginner? Pro?

[earnest fulcrum]

intermediate. Some singing, i sing alot but not to others generally.

[frozen dagger]

thanks. helps me understand how it works a little bit better to figure it out myself.

[polar spade]

Two things you may not know yet know:

1. [available to Pro and above] +VOICE
   We emulate our voices as a replacement by making a "+Voice" which was called a persona a couple of weeks back. Are you familiar with that as yet? None of your comments seem as if you have used it yet. Its almost necessary to use. You mentioned that you are Basic, so this is unavailable.

2. [available to Pro and above] CUSTOM MODEL
   An alternate option is to use a dozen or so of songs sung with your voice (recording of ytour voice is fine), which can be gathered and analyzed by CUSTOM MODEL to create a model, similar to 5.5 or 5.0 or 4.5all or any other, but only creates songs with your many songs vibe and , if youre voice is the only one in all the songs, can reproduce well your voice on the songs you create, including Covers.

Lastly, just having a knowledge of how the sliders work and getting used to their influencing the files and +Voices you attached before clicking Create, is good experience to have.

[earnest fulcrum]

no, not basic user... i am paid user still...

[polar spade]

Ahhh... good. I read that wrong then.

Then those two options are the best means of getting a voice to consistently be emulated ( relatively) to your own voice.

Seems youve used Suno for quite some time, but I couldnt follow in the thread what was being attempted and with which tools. Sorry about that.

[sweet ivy]

<@&1372996242325508176> Wow, you are still up helping people. Just wanted you to know, I finished the duet. Found a great YT video from Jamie Aplin about how to do a duet in Suno. Did a cover using just a female soprano. The original track was mostly a male tenor. So I cut and pasted where I wanted.

When I see Kanthirc in here, I will tell her. She was also trying to help me with Meta-tags and whatever

[earnest fulcrum]

Thank you again... it seems like i finally managed to get what i want, still used 5.5, just took many attempts and patience. However... final result, close enough and satisfying. Sounds pretty much like me too 😄 did use v5 to remaster it though. https://suno.com/s/RcO0zvKb1C2pLzgJ final result 😌

[tepid vessel]

I definitely can lol. It's sorta freaky to here the identifiable nuances I hadn't even realized were so recognizable from my voice until hearing it mimicked by what I know I didn't say directly myself. But as soon as something else speaks/sings like me it's so clear how much I never hear those elements in any other voice 🤣that I'm only becoming conscious of now I guess mainly because I know it isn't me that actually said it. It could be some weird placebo effect though in me just knowing it's supposed to be me 🤷🏾‍♂️ so maybe I'm listening more for it or making something out of something that isn't really there but I didn't have any real expectations either way on it going in so idk. I wanna mess around with it more but I don't have desktop access right now and I can't seem to get past the Upload portion on devices. Nor can I use my interface and condenser Mik to Upload a live vocal sample. I know my phone works with my interface with any other app so again idk. so I've only been able to do like a 22nd clip using the iPhone microphone which I am not a fan of personally but. Hopefully before long they'll be a better user interface for this type of feature. 🙂

[tepid vessel]

mine had just happened to decide to make one of my songs to a duet and now that I've heard it with one male and one female I mean it's just exactly the way it should be imo 😂. But I certainly didn't know it was gonna happen or try the best way though I've think it pretty much happens it's just in the register that you're singing so if it's a super high part it'll be a Chick and the lower if it crosses the threshold it seems to be a guy now that's just based on audio influence from my uploads that I would sing myself. I haven't used any tags to denote mail or female in any of my tracks yet and I'm not sure if I will. I seem to enjoy just seeing what the cards that are dealt happen to be. In my case with so much of the audio material already established even before uploading, It's nice to see whatever spin it might think to throw on 😂. Case and point, finding that that song's just perfect with the duet format 🙌🏾

[lapis loom]

what's the main issue with 5.5?

[shadow plaza]

Quick question:
Does remaster or cover better to keep the quality of the sound?

[wet minnow]

remaster. cover can too, but is usally used to change things

[shadow plaza]

Yes but i notice decaying in the sound each time

[wet minnow]

if you can live with the sounds getting a bit different but the overall quality gets better, then cover with the same styles.

[shadow plaza]

I can live with that i hope my fans too

😅

Thanks

[wet minnow]

no problem. and dont overthink it... usually only the creator is the one noticing if things went a bit of an unexpected route 😄 ... Thats great thing about creative freedom. It is like it is 😄

[shadow plaza]

Yeah they like everything i release.. 😅

Not picky fans

So i made realy nice track on metal 4.5+ want to improve the quality but when remaster become worse.
I tried 5.5 it make sound folk... 😅

So does covers are my only options

[wet minnow]

there are several options, but they are usually used to work creative with the input. like sample or mash up.

If you have Prem. you could stem the track apart and then take it in studio to replace the stuff you dont like.

[unreal sleet]

How do you solve volume increasing over time? Happens on different models, different tags and prompts,inspos and personas. Remaster or cover doesn't fix it

Can even get worse when covering or remaster

Getting slightly pissed that David keeps ignoring me about it when he first was thrilled and excited to figure it lol

[quiet solar]

Glad to hear the duet worked, and our input was helpful. 😉

[quiet solar]

Where I've seen it, it's taken the form of a bug (volume just kept increasing on a sliding scale from start to finish), and also an unfortunate side-effect of Suno not having a "whole song" view while it's creating. It's quite normal in music for there to be rise and fall in volume over different parts of the song, but sometimes it ends up ratcheting louder and louder if the song is a long one - my guess would be that Suno forgets that it's on a louder section after a while and considers it the new normal.

Remaster wouldn't fix either issue - cover might do, if it's not the sliding scale bug. But you'd have to try something like using a reasonably low audio influence and adjusting the style prompt and probably reinforcing in the lyrics tags to specifically try to push lower volume / lower energy segments.

To my knowledge, it's a fairly rare bug - I think you're the only person who has mentioned it / run into it on a regular basis.

[unreal sleet]

What do I take away from this

[gusty mango]

I've had it happen about 6 in every 85 outputs

[unreal sleet]

Luckily I mix and master songs in my daw so it's not the end of the world. But increasing something's volume usually comes with a quality loss

[broken phoenix]

Hi! Is there a place where different prompts sit in this channel? For example expressions to make acoustic guitar strings sound crisper, etc thank you ❤️

[polar spade]

Yes. If we are using a DAW, or Suno's GAW in this case, we can knock out a duet all day long, with a few key steps.

So, when most ask the question for how to "create a duet", its under the premise of their only having a Basic or Pro account and no access to splicing segments of different tracks together. Then it becomes a whole different mindset, for sure.

I enjoy making duets now and then. Most though, are written backwards for me. I prompt as much possible for a duet to be processed and listen to what Suno gives me. Then, re-write the song to fit the duet. Honestly, often the prompting does give me about 80% successful, so the rest as extensions and replacments is shorter than I expect to finish it.

To do otherwise is a very linear process (and mindset) and can be overly and unnecessarily frustrating. We're blocked from moving forward when we can't jump around and fill in all the gaps.

However, to write them "forward" isn't impossible. Its about writing all the parts as you initially expect them to be, of course, then using extensions and replacements, often running through several generations of one extension, tweaking until we get it right, before moving to the next extension.

Add up the credits on that and we might rationalize paying for a month of Premium to get such a project done in Studio, instead.

Remember, at the end, the ONLY song a listener hears is the one song you present to them. They wont know the difference and will think your duet is wonderfully done and wonder how you did it. If we're not musicians, at least we can be magicians.

[tawdry viper]

@polar spade

ngl at this rate i feel like alot of my suno membership time has been wasted

[quiet solar]

It's a really difficult bug to undo if it affects a song you like. Not really much more to take from that, sadly.

If it's still happening, report the most obvious example as a bugs under More -> Feedback in the bottom left on Suno Desktop (I don't know if there's an equivalent form on mobile) because that can be picked up by developers to help narrow down and investigate for fixing. That doesn't mean fixing it will be a priority, but making their lives easier to fix it can only help.

@gusty mango If it's turning up that frequently, I'd also suggest reporting an example as a bug if you haven't already.

[high snow]

Is there a way when making a songs to have glitchy effects for voices? Kinda how ppl who sing their own music would add those effects. Bc I wanted to use it for a fan-made song and idk how to do it.

[tawdry viper]

@quiet solar

[quiet solar]

?

Sorry, mate. I'm not part of Team Suno and don't have access to their systems, so there's not really anything I can do to help if this is a weird UI glitch. If Elliotstunes is on the case, I'd give him some time to look into it.

[high snow]

Report that in #🪲┃bugs-web

[frozen dagger]

This is my go-to prompt for my acoustic guitar stuff. Dunno about "crisper" sound, but this prompt just tries to include as many techniques I could squeeze in there and try to focus on just the acoustic style. If I was trying to get a "crisper" sound, I might try making some room in my prompt to list the instrument I want to use, or describing the sound texture the best I could. In the end I think that would come down to clicking Create a few times and seeing if get a different sound I like to re-use, or just record my own sample if possible.

An unplugged studio solo session centered on a single acoustic guitar, captured with intimate close-mic detail as if performed live in the room. Virtuosic percussive fingerstyle playing shapes the entire performance, integrating string slapping, finger taps, body percussion on the soundboard, palm muting, and rhythmic dead notes directly into the groove. Expressive finger control unfolds through hammer-ons, pull-offs, legato runs, trills, slides, glissando, controlled string bending with nuanced vibrato, and fluid dynamic articulation. Hybrid picking and finger independence support rapid arpeggios, double-stops, chord-melody voicings, and subtle counterpoint. Natural harmonics, tapped harmonics, harp harmonics, and campanella ringing strings add shimmering overtones. Occasional two-hand tapping passages emerge organically. The performance breathes with natural tempo movement, warm wooden resonance, crisp transient attack, and cinematic phrasing that feels discovered in real time

[quiet solar]

"Close-mic" and "studio" are good keywords to have in the mix. OBLMOUS' prompt above looks well-worth experimenting with.

Bear in mind the way Suno works isn't quite as simple as saying "make it crisper" and having Suno interpret what you mean. What it will do is look up songs that it can reference against the words you include in the prompt, so you need to be able to describe what you're looking for in terms it has the best chance of interpreting.

Prompts around the way the instrument is played will help with a more natural sound, especially for a percussive string instrument like a guitar. You also want to consider production and recording effects - things like "natural reverb", "clean modern production" and "clean mastering" might also be useful.

[frozen dagger]

Another user was reporting the same issue yesterday as well. I hope a resolution is found soon for you all. In the meantime would definitely make sure submit https://discord.com/channels/1069381916492562582/1082775596767117312 , and https://discord.com/channels/1069381916492562582/1479483412250366054. I don't know what the issue is, if it is a regional issue, or tech related, or what. Submit reports with as much detail as possible (pics in order of events, video) and try and be patient waiting for resolution.
I hope you all are able to create with your own voices as well soon. 😇

[polar spade]

@tawdry viper Same here. If we are going from Elliott again back to me, thats just going backwards. If it acts like a bug 🐛 then a direct chat with a team member or emailing Support, not matter the wait, is going in the right direction.

@naive wharf I had a chat with TreeofLife a couple days ago. His Voice module interface is stuck in "old" mode. It does say Voice and not Persona, so the module just before the prettier orange graduated border and nterim verification module was introduced.

• He said he initially started his account using VPN, but has since been logging in directly.

• we cleared localStorage to help update clients files and any corrupted files. He logged off, closed browsers and started over but with no change. Still old UI.

Thats where we finished our voice call.

[naive wharf]

Sending his info off to the devs. Thanks!

[frozen dagger]

"glitchy" can u give an example? we talking post-production effects, or like stutter speech? both? the answer is yes most likely.
"I am per-per-per-perfect. Perfect in im-im-imper, fection!" as an example of how you might type out to create some glitchy speech phonetically. Post-production can be as easy as prompting "glitchy vocals" "vocoded vocals" or something that just describes what you're after.

[high snow]

Yeah, best example of like a stutter like a purposeful glitch in a voice would be ultimate an example of the song. You’re gonna have to listen through it though.

https://open.spotify.com/track/6vJpHOvUI6txkR3nSNXICY?si=ByobidyYRaCg_xiozNQqyw

(the glitch part is at 1:10secs)

[high snow]

Do you think you can send me in DMs some prompt ideas on how to do that?

[plush field]

can someone explain what happened to chat mode?! i need it back. sorry for offtopic, but no one reacts in genral chat to this question

[uneven crown]

whats the max amount of songs i can upload to the custom model?

[polar spade]

100? I think.

[polar spade]

Were you given it as a trial? It may have been just a trial and pulled, perhaps? I guess look around for current others who mention using chat. See if they still have it as well.

[polar spade]

OK the word " dome" about 1:15 was a glitch. Is that it? Kind of a Max Headroom stutter?

[high snow]

Yes.

But @frozen dagger gave me some tips

[polar spade]

I was insp-sp-sp-sp-ired https://suno.com/s/Ey0d2CUXxNa2xCgB

[sweet ivy]

@polar spade Had a general question. I finished the its draft in Studio late last night & saved in the library as you advised. Opened in Studio and made some volume adjustments. Do I need to save again in the library, or are the changes I just made, already saved in the track saved in library ?

[unreal sleet]

anything that helps remove all the hissing sounds that make your ear bleed? something in exclude maybe

except for using EQs on the high ends on a daw

[quiet solar]

Synths and cymbals are often culprits in my experience. I'm not aware of any prompting that can specifically target them to screen them out, but you could give it a go. If it's obnoxious I tend to...hah, you literally just typed about EQs so I'll stop there.

[unreal sleet]

Yes but having to EQ out things can also remove hihats which is in the same high end range, and I dont want to have to download full stems to see if its any good all the time

since hihats can be in the drum stem and u dont wanna touch too many elements

but if theres nothing i can do pre-maturely thats all i wanted to know 🙂

[quiet solar]

Agreed, I don't like doing it either. If I get a hiss-heavy version, I'll usually take a few different attempts on cover to try to get a better starting point, so I might only have to EQ a particularly bad section. Same approach to dealing with the cursed shimmer in v4.

[polar spade]

The changes are auto saved as you go in and out of Studio. Obviously, you wont have a final file unless you save it. You can check your history and restore a version if you wish. Click the menu of the Projects and choose Versions

[unreal sleet]

Yeah, happy to have used a daw for 20+ years in addition to suno, helps a lot ^^

[sweet ivy]

@polar spade Maybe I wasnlt clear. I'm in the Project I want, just made a few volums adjustment on the instrument track. I know they are automatically saved in Studio, but are they being saved in the library or do I have to do what I did last night and export, saving it with another name, same name with a 2 added , so I know which one is which

[polar spade]

I wasnt as clear. They don't save in your library until you Save. Some edits will save, but I cant think of any auto-save to save the whole file. There's no means of knowing what to auo-save with several tracks either muted or not needed.

[polar spade]

Yep. Same. I rename and save different versions of whole files. No way around that.

[sweet ivy]

No problem, I just wanted to ake sure I saved what I just did in Studio. Thanks

[plush field]

It was there, but now it is gone

[polar spade]

unknown

[plush field]

the best thing about chat mode is that there is no character limit. also you can revise the idea with chat bot before it start generating

like this

[polar spade]

Thats cool. Maybe we'll see it again later.

[frozen dagger]

Tesla Coil. interesting. Learned something new looking into that.

[polar spade]

Thats a red herring. I had used that as a sound effect and it got dragged into the Song as a first FX prompt.

[frozen dagger]

It was a good idea nonetheless. Very fun. Matched the source materials vibe very well.

[frozen dagger]

Looks like they took Chat down. I don't have it on my alt account anymore either. Maybe it will come back — better, stronger, faster.
I didn't use it much myself, but I liked the potential. Producer.ai uses this chat method for song creation as well. The songs created there are not on par with Suno (imo). I can see it coming back tho, and I can also see other new ways to create coming.
(Node based modular creation UI; like the switch from creating images in Automatic1111's UI and then switching to ComfyUI)

[plush field]

never mind, all of mine works better 😂
the only advantage that one has is that he knows how the generated songs really sounds, but mine are better for everything else 🙃

[waxen jackal]

I am trying to make cover of my song with additional lyrics but it's not taking full lyrics. 🙁 what to do?

https://suno.com/s/h66RDZzEgqZtC2R1

[plush field]

if you want to make song longer with new lyrics, use extend

[frozen dagger]

made your own I see 🙂

[frozen dagger]

if you added lyrics to change the song, Suno may have a hard time trying to recreate when the new lyrics don't match the previous bars in syllables.
could try matching the syllables, or stem split to remove the vocals entirely, use the intrumental to Cover (or remix) with your new lyrics in a Cover.

[plush field]

i would share this gem if there were a way to hide the gem prompt and reference material. i don’t want to share them, because i don’t want knock-off gems based on my work popping up everywhere. if there were a way to protect them, i’d share it immediately

claude version goes a few steps above, since it can revise lyrics and prompt (even a few times if needed)

[half dragon]

If you have Studio, load it in there, then export a new clip.. that'd strip off most things if not all, lyrics, prompts.. then probaby safe to share that version.

[plush field]

i am talking about Gemini GEM, not song 🙂

[half dragon]

My bad ..lol carry on
gem , gen... tomato, tomaato.. samething..

[plush field]

request

result

https://suno.com/s/5tDVfxPRzhksx99j

[tame verge]

Does anyone have any tips on how to get meta tags working properly when you prompt them? Like how to format them? For vocals, mood, instruments, and things like that? Sorry, I’m new to this…

I kind of know about square brackets, but do things go on the same line or different lines? And do they go before the entire verse? Or on the line that you want them to be on? Is it better to be more concise or more specific in what you want? Any help would be greatly appreciated.

[plush field]

"prompt is a suggestion for AI, not a command"

you can only sugest to ai what you want

today i got complete lyrics in prompt for the instrumental 🙈

[Verse 1]
[Feminine diva pop vocal]
Digital heartbeat, a rhythmic design

[half dragon]

When I first started, I learned from others by listening to their song and looking at their prompts and metatags.
Then try them myself... there are things that always work, there are thing sometimes work, and sometimes don't.
That's just a nature of Generative AI - you can't command it, you can suggest or describe what you want.. and it'd give you options.

[tame verge]

Wow… That’s crazy…

[plush field]

i had this in lyrics

[Instrumental]

[Main Theme]
[Korg M1 Organ Lead replaces all brass]
[Bright rhythmic stabs]
[No horns]

[Drop]
[Super-saw synth melody]
[High-energy digital textures]

refrence audio was instumental with audio influence of 20

whre lyrics in HUNGARIAN came i have no idea

[plush field]

see this crazy thing i am making right now

[Intro]
[Fast melodic death metal guitar riff][Driving eurotechno kick]

[Verse 1]
[Feminine diva pop vocal]
...

[Pre-Chorus]
[Masculine aggressive guttural growl]
...

but in order fo this to work properly i need to have melodic death metal and eurotechno in style as well, otherwise this tags in lyrics means (almost) nothing

same for vocals

https://suno.com/s/ELHay6IX3MKQBbf1

[tame verge]

Oh wow… Interesting. One thing I’m trying to work on is how to get male and female voices to sing in the correct spots, kind of like call and response.

[plush field]

call and response works, but the problem is you cant control who will sing what

you can have more or less luck, but not proper control

[tame verge]

Yeah… Unfortunately, that’s kind of what I’m figuring out… 🙂 I wonder if there’s a way I can splice bits and pieces of each version of the song I have and just put those pieces together to get the voices where I want somehow…

[frozen dagger]

https://discord.com/channels/1069381916492562582/1252267444773326848 Not a bad idea,as a new user, to skim through these as well. Same as VRKing saying, look what others do. There are multiple workflows for duets and multiple singers. Find what works for you.

there is more for Studio specific workflows https://discord.com/channels/1069381916492562582/1468670591354535957
Studio is DAW styled editing. This makes doing thing with splicing things together achievable. (Can export your song to an external DAW to edit as well. I think there may be some examples of that in here as well.)

[plush field]

https://suno.com/s/BV0Yc6XqqMVwbpjp
i just made this one for you, check the chorus, it has call and response

[tame verge]

Oh, OK! Thanks! 🙂

[plush field]

you have two approaches here, but as you see, neither is more accurate than the other
https://suno.com/s/zefVVTjUt3y2LyqO

i really like how this one came out

[quiet solar]

You might have already stumbled on it, but Workflow C in here describes exactly the process you're thinking of. It requires an external DAW/Audio editor if you don't have Suno Studio: https://discord.com/channels/1069381916492562582/1466598465969590465

There's a certain dark art to steering Suno to make duet generations that are reasonably accurate, and in structuring songs that are more likely to be friendly to switching between the correct singers, but there isn't only one way to do it.

The main thing to note is as Viking said, you're probably looking at a realistic maximum accuracy of around 80%-90% correct by prompting alone, so if you want precise control over who sings what then the only non-editing route for that control is to build up the song piece by piece through multiple extends, which can be painful if singers switch over a lot, and typically requires a bit of cleaning up with cover/remaster afterwards.

If you're already comfortable that your process is going to involve editing, then it becomes a question of what the most efficient way to do that is. There are lots of potential moving parts here (covers may not have identical tempo or key signature, they might deviate in melody at the wrong moment, you might be trying to keep instruments the same while only switching vocals, etc.) so again, no single right or wrong answer, but various potential approaches you can use depending what best suits the problem you're trying to resolve.

[pure panther]

Public service announcement

Mashing a song with itself is an excellent way to refine it while keeping the exact same vibe. You can make minor changes to lyrics too! 😄

[frozen dagger]

Same theory with Custom Model going on. Doubling down and using the same input when it asks for more than one good to "trick" it.

[pure panther]

So use the same song over and over?

Or for all required slots?

If that's the case I may need to try that. I use the same song for inspiration for all the songs in the album I'm currently working on. It might be even better if I could base a model off just that one song.

[frozen dagger]

For Custom Model, it needs at least 6 inputs. You can do all 6 the same, add it 20 more times.... dealer's choice. But you're able to kinda create your own influence in this way. Using the same song can be good for working with Cover to stick to the vibe of the song and not drift off in another direction after a couple minutes; or creating a Voice using you voice upload multiple times so it can try to sound closer to you.

[pure panther]

That's amazing.

I love it

I'm going to combine that with mashing the same song together 😂

[frozen dagger]

in theory adding 6 songs of "song A" and 4 songs of "song B" is sorta like a 60% "song A " vibe blending a 40% "song b" vibe.
mix and match. and then the prompt field give it something that is not already present in the audio, see what happens.

[pure panther]

Very interesting. I suggested insperation song strength sliders, this may be one way to do it.

[frozen dagger]

I have not gone this far... yet, but absolutely! It's not as straight-forward thinking as adjusting a slider to a %, but since Custom Model uses the audio to create output, it is very similar to the mashup and ispo funtionality. (with more inputs, and it being on a separate layer in the stack of "ways we guide" Suno.)

[frozen dagger]

Sliders for each song input? That would be awesome.🎧

[pure panther]

I know right?!

Lol, combining two identical songs being mashed up with a model created using only that song seems to have broken it. The original song is 3:31, the resulting songs are about 2:20. Lol yeah, it's glitching out 🤣 We're going to call that one a fail.

[frozen dagger]

I heard, "more testing and experimentation may be required." 😂😇🕉️

[tame verge]

I did find work flows, and looked through them a bit. I’ll have to explore more later. Unfortunately, I am also partially blind, so the website and studio is not accessible to my screen reader, so I have to use the app, and try through either prompting or extensions.

[pure panther]

Public Service Announcement #2

When your tracks vanish suddenly, just press the "space" key in the search bar and voila! They reappear! 😄

[thick plume]

https://suno.com/s/D0AkbsaFgAxF3ABg

[sweet ivy]

Was listening to the stems Version 1 & 2. hit the keyboard by mistake and lost that page. I have no idea how to get back to that page, so I can decide on what version of the stems to download? Any help would be most appreciated

I saw all the stems in the stems layout. After I lost them I swent back to extract stems and expected to see See Extracted Stems but they weren't there??

I don't want to waste 50 credits again

[frozen dagger]

from library go to the song you originally split stems from, right click and say get stems (same as before but there is a small white text at bottom that says "See extracted Stems")
/ho i see if i can find to SS

[sweet ivy]

Right click where ?

[frozen dagger]

If you try to split the stems from a song you already split this white text will be at the bottom.
Right click on the song that you already split the stems off. Get Stems. this window appears

[sweet ivy]

@ I made my way there, but there was no See Extracted Stems ?

No no splitting , nothing fancy, just the track from the library and extracted the stems. I saw them all

[frozen dagger]

I still think we are talking about the same thing. Did you "Extract All stems" or "Vocals + Instrumental", or am I completey off base and neither? We are saying stems.

[sweet ivy]

It was an nistrumental track I extracted All stems. I saw the grid come up, with all the individual tracks , over top Version 1 & Version 2

Where the hell did they go ?

If I unclick Hide Stems in the library, they are all there, unnamed, both sets

[frozen dagger]

I don't know which UI you were looking at, and I don't want to send you to another one to lose the song, or the stems. So, I'd suggest follow the same steps you did to split your stems and when the screen comes up asking which orange button you want to click, look under the orange button for the white text to see the screen you were on.
Or look in the workspace you were working in, and make sure filter of "Hide Stems" is unchecked so they should show.
There has been some issues of songs dissapearing from our library list. (Usually gens that are still being created.) So give it a moment and reload the page and try again.

@sweet ivy circled them for clarity.
The stems you already split, are in your library tho too, and can find them through library (check filter)

[sweet ivy]

@frozen dagger yes, I know, I have been there three times and that See Extracted Stems in not there, yet I have about 14 stems (unnamed in my Library, when I uncheck "Hide Stems"

I played about 5 of them and they are all messed up. I will have to go in, delete all of them and start again

[frozen dagger]

Hmm. Well, the stems in the libary is easier. Sounds like you have them, not sure what "unnamed" means. They usually auto name with type of stem and stuff.
It's possible we are clicking the wrong song, or page needs refresh.
But either way you have the stems you were looking for now, right?
You have Studio if i recall correctly. The stems we separate are also showin in the details window there on the right side.
I'm trying to look to verify, but not all UI windows look the same, but usually I see the song that had stems split from it is marked with a "Stems" icon along with the icon for Cover if it is a cover, and the marker for which model was used.

[sweet ivy]

It was some sort of glitch, played one and there was nothing there. Thanks for the help, will delete all the stems in the library and start over

[frozen dagger]

I hope it's sorted for ya. Glad to try and help. Hope find success in your endeavor.

[sweet ivy]

@frozen dagger Just thought of something the track I wanted to extract individuals stems from, was a stem itself, the instrumentation stem. Does that make any difference ?

[frozen dagger]

For the cost of credits.
Splitting stems has a one-time credit cost (can choose to retry extract and pay again; we are not doing this right now).
If you split the stem those gens that are made are yours forever.

If you split those stems, that is a new audio source. Have to pay credit for splitting stems from this new source. Keep those forever, not pay again. (Unless you optionally choose to try and do the process again and hope for better quality stems than you got the first time it tried to extract.)

And lastly ofc, each time you hit "split stems" you pay that credit cost, you get 2 versions of each stem. (Like each create click gets 2 versions of a song gen)

[sweet ivy]

@frozen dagger So I can extract stems from a stem. I just have to pay again, because the stem I created previously, splitting the full song into vocals & Instruments, that instrumental stem counts as a new source. That's what I did, but something messes up somewhere

[frozen dagger]

I'm sorry you're running into that "something messes up somewhere".

I hope it is not the current bug list. Without troubling you with details, you're creating stems of stems and this can populate your library with stuff pretty quick. If you are having an issue locating files, this bug may be the culprit.
I try to wait for the full generation to complete.
You know how we can listen before song is done completing? Stems usually load much faster, but I try to let it fully load. Refresh my page every now and then.
You're not alone into "something is wrong here". Some UI changes and stuff and new bugs needing to be squashed.

[sweet ivy]

Oh it was complete, I listened all the way through version 1 and then was half way through version 2

[frozen dagger]

As far as the technique involved in splitting stems, and further splitting those stems; the application of.
I am hoping some of the Suno Scientists find the time to create some more video tutorials or something to follow along with. It can feel like a lot.

[high snow]

I’m more of a visual learner. I know this might be a pain in the ass to ask, but do you think you could visually show me in DM’s on how to splice and add parts either through studio or song, editor? I know how to use extend. But another problem I have is if you use extend and you change one part of the song, how are you supposed to make that extend part sound like the rest of the song? Like say if you have one verse in your son, it’s only supposed to be female song, but it only sings one lyric of it in female, but then the rest of it. It changes back to male how are you supposed to edit that?

[high snow]

I suppose you’re to a challenge, that you could possibly if you have time to make a small clip on what you’re describing so I can see it visually? Because yes, explaining in depth in detail is one thing that’s great and all but I’m more of a visual learner.

If you can spare some time.

https://tenor.com/view/squidward-spare-change-spare-some-change-begging-poor-gif-1281766816980485400

[quiet solar]

I'm guessing you're new to using Suno and to audio editing as well?

If that's the case, I'm going to take a step back for a moment and say this - perfectly performed duets where you need to specify the singers are HARD. They're one of the more difficult and potentially frustrating things to do in Suno.

The twin headaches of "new songs don't sound identical to existing ones" and "I can't control who sings what" means there is no simple solution, and the best approach to take will vary from one song to another depending on what you're trying to do and what problem(s) you need to fix.

I wouldn't recommend trying it as a Suno newcomer - at least not until you're more comfortable with how to get similar sounding songs via Cover or Extend, and are familiar with making simpler edits like replacing a drum track, or replacing one version of the chorus with another. There are lots of potential problems you might (or might not) run into - trying to cover all of those in a guide would be a bit too much for both of us!

[eternal crown]

Guys, which version of Suno do you use for longer songs, v4.5 or v5.5? They are limited to short songs, what should I use?

[nocturne bobcat]

Hi everyone.
Is there any way to reset or break Suno’s melody “memory” when generating music?
I’ve noticed that once I generate a song with a specific set of lyrics, every new generation with the same lyrics tends to keep almost the exact same melody structure (same phrasing, progression, and section layout), even if I completely change the prompt, genre, or style. It mostly just changes the instrumentation.
I’m trying to get a completely different melody for the same lyrics, but it feels like the model is “locked” into one interpretation.
Has anyone found a reliable way to force Suno to generate a totally new composition (not just a variation)?

[quiet solar]

Suno shouldn't have any inbuilt "memory" of past generations. The structure and lyrics of a song will influence what is created, but not usually to the point where it feels like it's generating variations on exactly the same song.

You could try clearing the cache in your browser, in case that's somehow causing problems. Otherwise, adding/removing Voice or adding/removing Inspo should help push it towards something different.

[worthy kraken]

How do i save the song editor edit as a current song, and not as a new song?

[tender eagle]

I have no idea actually, I figured it may not be doing anything, since skip is an option, which if it is doing some modification to the vocals, I'm unsure if skip would make sense, I considered it, but tbh I imagine it may actually just be a poll. and the main reason I think that, is not cos I have any idea, but... the way it says "Thanks!" after you select the options, feels like it's "answering a question" moreso than telling it something it needs to use less quality / better quality singing abilities to adjust the generation parameters along with the voice used.

[tender eagle]

What do you mean? like Save it so it remains as the same media object in the workspace?

i don't think it does that, if you mean like how google photos will let you make some edits, and save it as the same photo instead of saving it as a copy.

[slate juniper]

I have the same problem with vocal chops. Somehow Suno think I want to have this in my songs independently of the genre or prompt.

[worthy kraken]

Then they have a crucial miscalculation in how their system works. Because if you edit the edit via song editor and save it as new song, you can no longer get the full song out of it

[tender eagle]

you sure something else isn't going on (unrecognized masked area?) cos I just edited this song and it came out proper,

[worthy kraken]

You editted whole song or edit?

Extend, i mean

Try to edit any extend instead, and you will see.

[quiet solar]

Can you do the exercise the other way around? Get Full Song and then put that into the Editor?

[worthy kraken]

Yeah, but it gives bad results, idk why

[quiet solar]

Can you be more specific?

[worthy kraken]

It ignores my prompts and puts its own on top of my replaced lyrics. And it creates very distinctive transition on the place where new part of the songs starts, like a sudden loudness change

[quiet solar]

Not sure what's behind the first issue, but Replace Lyrics / Replace Instruments was always a little bit janky, in my experience.

I've always found it better to do minor tweaks like that by covering the song on low weirdness/high audio influence and making the lyric edits in the cover, assuming they're minor.

[ripe galleon]

could suno's voice feature be able to recognize this pitched down sample of me as a voice and be able to use it in a song at all?

or will it turn into noise/be ignored

[polar spade]

Both Pro and Premier have an 8 minute limit. Either really have no issues to use extensions to make longer. Its when you Cover or add +Voices , for example, the song can systemically become shorter anyway.

[sweet ivy]

@quiet solar I found a YouTube video by Jamie Aplin on doing duets. He did a cover of the track with a male vocal, typing female vocal in the style. I ended up with the same track, with a female vocal. I could cut and paste from each. I was surprised how closely they lined up in the majority of the song, so I was able to use them together in the chorus. I also had some nice harmonies from the duet version I did. Like you said a number of generations and piecing something together

@polar spade I extracted stems for the first time in 5.5. Maybe a half second before the track started there was a click? Is that a DC offset? Since Suno has to "re-synthesize" the audio to pull the instruments apart, does it create a tiny burst of noise or a "DC offset" (a technical glitch) at the very first millisecond of the file. The full track in the library sounds fine because it’s the original "baked" mix, but the stems are essentially new, "calculated" files.

I had not heard it before, on other stem extractions?

[polar spade]

Aplin notes that one needs Premier and Studio to do that.

Good instructions, but many who ask dont have Premier. Thats like making a click bait title. "How to move 50 miles per hour." Then in the video say, "First, buy a car." 😊

[polar spade]

If the original song doesnt start out wifh 1 instrument, I could guess that stems might get a click. Its still new territory, this AI music thing. Ive read a few more posts talking about clicks and pops, but each have different parameters. Its hard to really say.

In real live, we can get clicks as well, so we just mask it out. Old school editing may be the simple means of fix in edit or Studio and move on.

[sweet ivy]

@polar spade My issue is, I believe it might have pushed the track back that millisecond. I also loaded a male & female stem in with all the other stems from the instrumental track. It seemed to my ear they were singing very slightly ahead of the instrumental track. Is it possible the DC offset cause that?

[polar spade]

Sure. Its assumed common enough. Suno doesnt have an offset tool. Fade it, edit it. Drop the Volume over it. Cut it. Whatever works to then reset the best possible to zero, even if just manually by ear. Old skool rules!

[spice kraken]

how do i keep the vocals from getting muddy or distorted half way through the song? i've tried adjusting the custom voice's sample size and place from where it was taken from, remastering, editing in the editor, but it always ends up the same 9/10

[shy badge]

Hey all. A new google gemini gem. Its like a mini gpt but for gemini instead of chst gpt. Optimized for tight ptoduction and attempt for non cliche/stereotype lyric writing.

Note this also mashes genres and styles as well as converts input as well.

https://gemini.google.com/gem/1pTWOU0d2DXBEER-9Eg19P3hdnD7yaegM?usp=sharing

[sweet ivy]

@polar spade Yeah, getting rid of it is easy, I was more concerned with the possible offset affecting the matching on the two vocal stems I added. When you say Suno doesn't have an offset tool, do you mean a way of correcting offsets. Is what I'm talking about possible, there was an offset created, when Suno recalculated to create the extracted stems?

[frozen dagger]

Agreed. Thanks for response.

[plush field]

if user writes <artist name> style, it will add the artist name in the style, which is a serious issue

you need to instruct it not to do that, and to use style description isntrad

btw. do you have some refs file that this gem uses? and how did you hide them if it does?

also, if the user writes <artist name> style
it might use artist's name in lyrics as well

[shy badge]

Totally fogot about that part. Easy fix fix tho. Will update after plasma donation later.

[plush field]

can someone explain where those two "tips" came from? AI hallucination? something that worked in 3.x?

—Rain tapping on the window—

*Sound of a distant train whistle*

5.5 sings those as any other lyrics part

[plush field]

to clarify, i am asking about * and — notation

[urban quest]

Hi there! I have a question about this new "My Taste" feature. Yesterday I generated many songs with vocalise in them. Like mmm.... ohhh... But today I'm exploring other stuff and most of the songs have a "mmm...." vocalise at the beginning without me having this in the lyrics. Could be this "My Taste"? How do I turn it off, because it's not what I want?

[plush field]

click on my taste

check what it is writen in this field

[urban quest]

great! thank you 🙂

[plush field]

i disabled that, since i have 3000+ failed experiments, and 10 normal songs

i experimetn 95% of the time

with strage blends and effects

[urban quest]

I will disable too for now. I also experiment a lot. Still there's noting in it about vocalises... strange.

[plush field]

there is one more thing, do not put too many things in the same project, it will copy the style of previous songs

[quiet solar]

Hallucination, I'd say. I wouldn't expect those to work in any Suno version.

It is possible (but rare) to get Suno to perform SFX by prompting for it in [ ].

Suno is able to contextualise prompts to some extent, if you do unusual things like putting prompts in * * or - -, but as a general rule if you're trying to prompt Suno for something and don't want Suno to sing it, put it in [ ].

[urban quest]

Actually I made a new workspace, because I'm experimenting creating voices.

[plush field]

when i had acces to beta chat mode, it created a new workspace for each song, so i guess that is suno recommendation, put one sone per workspace if you do not want them to share traits

[quiet solar]

Any prompting suggestions provided by AI need to be checked rather than trusted, because every LLM I've seen is very comfortable including suggestions that probably won't work (while telling you they will), or which could actually be detrimental to what you're trying to achieve.

A good case in point being AIs happily suggesting you put "No (xyz)" in the style prompt, which is known to backfire because Suno doesn't do very well at interpreting negative or conditional prompts in the style box.

[plush field]

i have 300+ pages of documentation in obsidain vault, the issue is that * notation came from there

[tender hearth]

songs keep infilling 'whoop' 'whooo' and stuff where it isn't needed, and constantly adding on to lines 'blah blah blah - ah ah a a a' etc - how can i stop this

[quiet solar]

My Taste only takes effect when you click the magic wand in Style or Lyrics.

It doesn't dynamically affect generations. The vocalisation intros are a known quirk of v5 models, and not a result of something you've done.

[plush field]

and this is the page which is guilty for * notation

https://sunoaiwiki.com/tips/2024-05-02-how-to-enhance-song-production-using-suno-ai

so it found that in my vault and said, ok lets add it looks useful

[urban quest]

yeah, I had a general workspace, default, since years I have this account. Then yesterday I started working on an album and made a special workspace for it. Now I'm testing voices in a fresh new workspace. I still get vocalise, lol :))

[quiet solar]

No specific way to prevent it. Some people have suggested that adding "ad-libs" to the exclude field makes it a bit less likely, but it's been a feature to some extent in all recent Suno models.

[remote ivy]

i have 7 sections like this, am over the word count limit...what is the non-essential parts i can cut from this style of section meta tags the lyric magic wand gpt4.1 gave me (through some genre shift and instrument add on iterations of the prompt)....Does that function tag really do anything based on anyone's own testing?: [Section V – Strain Crest
Form: single climax, 16–24 bars, then an abrupt dynamic drop that feels like a power cut
Orchestration: full small ensemble: sax choir in warm, breathy but tense chords, cello in wide, weighty voicings that lean into the darker harmonies, rumble bass present and throbbing like a distant engine, a broad, dark trombone line carving a slowly rising countermelody against the jar’s main pitch, synth pads opening into bright, glass-like clusters near alarm; jar blown at its most resonant pitch near the edge of breakup, recorder as rare, piercing sparks cutting through like warning calls
Function: realize the “glass begins to strain” moment; reach the loudest point (strong mf, never harsh), then cut to solo jar, soft rumble bass, and one high recorder or cello harmonic, hanging in the air]

[plush field]

results of AI agent research on the web about effects, from few seconds ago

Based on the search results, here's what I found about Suno AI's techniques for combining sound effects, style prompts, and lyrics:

Key Technique: Separation of Sound Effects from Lyrics

Some creators experiment with Suno sound effects cues, but these often create unpredictable lyric behavior. If your goal is clean songwriting and clear Suno tags, keep Suno sound effects out of the lyrics field. If you want impacts, risers, or texture, put that intent in the style prompt or add it later in production.

Structure of Suno Prompts

A Suno prompt has two parts: the style prompt (genre, mood, instruments, vocals, production) and the lyrics (with section tags like [Verse] and [Chorus]). The style prompt goes in the "Style of Music" field and controls the overall sound. The lyrics go in the "Lyrics" field and control what gets sung. Together, they tell Suno's AI exactly what kind of song to generate.

Style Prompt Best Practices

Suno style prompts define the sound: genre, tempo-feel, instrumentation, energy, vocal intent, and mix direction. A well-formed Suno AI music generation prompt can imply motion, contrast, and dynamics without cluttering lyrics.

Combining Elements

You can freely combine various musical elements to create your own unique Suno style and lyrics with Suno metatags. The search results indicate that effective prompting involves treating these fields strategically rather than mixing everything together.

[urban quest]

Oh, I see. Well, since it's a new thing this 5.5 model, I assumed is that thing's fault. Well... maybe I should try to control it from the style prompt.

[quiet solar]

Pretty sure it's AI-made, mixing in good advice with terrible advice and hallucinations.

Just eyeballing a few pages, based on my experience I'd say a lot in there is correct, some of it is outright false, and some of it is perpetuating myths that I've never seen reliably demonstrated in practice (or been able to replicate myself).

[quiet solar]

You could remove pretty much all the metaphors and similes (e.g. "present and throbbing like a distant engine", "glass-like clusters") and I expect it would make very little difference.

[jagged lantern]

If you search "how to prevent humming or ad-libs in suno's vocal track" in the chrome search bar, chrome provides various answers. I tried creating a vocal track myself, and even when using the methods suggested by chrome, there are instances where humming or adlibs still occur. In such cases, you can either cut them out with an editor or continue creating the song until the humming or ad-libs stop. I hope you succeed.

[remote ivy]

you can have the lyric magic wand (chatGPT4.1) analyze effectiveness of tags for suno.ai in the lyric box, by prompting it to comment on it...it infact doesn't recommend to put no XYZ...commenting as it being unlikely to be effective

[urban quest]

Other thing folks, have you managed to create mature or old person's voices? If yes, how?

[quiet solar]

Suno's inbuilt magic wand (and the in-development Suno Chat) are a bit different, since they may be referencing material the Suno devs have which external LLMs don't have access to. We'll probably know more when Suno Chat comes online.

I don't know if I'd go so far as saying you can 100% trust anything it suggests, but it's definitely more reliable than, say, asking standard ChatGPT.

[tender hearth]

thank you

[remote ivy]

the lyric box magic wand is just generic GPT4.1, you can prompt it for it's system info, the style magic wand seems to the be proprietary one

[polar spade]

You can delete the click, if its additional, and then turn time warp/quantizing on to see if it off sets the rest of it to where you need. The tools are new, so your testing is as good as mine. We all have to learn what works in new situations.

[misty glade]

Do you guys know if its possible to get two voices in two diferent languages overlapping?

[plush field]

try those sugnestions made by AI
• “slight tremble in sustained notes”
• “short breath phrases, like lungs are tired”
• “occasional spoken-word breaks”

sounds right (at least in theory), but you can experiment with more of those and combining, and putting some of those in the style as well

[urban quest]

Thanks. But I've experimented a lot with AI suggestions, I can affect the voice timbre, but the base voice is still a young person. I'm trying to get into really old voices but so far no luck. I was curious if anyone done that already 🙂

[jagged lantern]

My pleasure.

[quiet solar]

Having two languages and/or two voices in the same song is certainly possible.

But overlapping as in harmonising? Probably not without editing, no.

[high snow]

Not really I’m not asking you to somehow. I just suck at using studio. I actually joined this server in November. But I started actually using it in September of last year.

[high snow]

Well, I know how to use all cover and extend. I was just kind of curious cause my duet is like quarter away through my song and I don’t know how I’m supposed to crop that part out when I replace it with an extent portion and then have it sound like the rest of the song at the same time.

[tender hearth]

[intro] seems to always read as [laugh into the mic a few times]

[quiet solar]

Ah, sorry - I misunderstood what you'd posted, and didn't realise you had Studio.

If you wanted to do it in Studio, the route I'd take would be covering the song but only prompting for the singer who is supposed to sing that verse. If you do the cover on low weirdness (15% v5 or before, 30% v5.5) and high-ish audio influence (60-70% v5 or before, 50-55% v5.5) then you should get a cover which is close enough for editing purposes, since you're only interested in the vocal stem.

If you keep getting the wrong singer cropping up in the cover despite your prompt changes, turn down the audio influence and/or turn up weirdness slightly until you get to the point where Suno makes the switch.

Once you have a cover with the right singer, open Studio and import your original song. Split out the vocal/instrument stems and add them in. Then import your cover, split out vocal/instrument stems but only add in the vocal stem.

Find the verse where you've got the wrong singer, delete that part from the vocal stem on the original and then delete everything but that part on the vocal stem from the cover. Then you should be able to export the completed song with the singers corrected and minimal other changes.

[shy badge]

Updated. Can no longer use artist names.

[high snow]

I’m glad that you explained that detailed, but would you be able to spare some time and maybe except a challenge to do a example screen recording of that? I know that sounds like a pain in the butt. I’m so sorry. But I’m glad that you did put that in depth in detail with writing in words, but I’m more visual. I know that that’s really tedious to ask. But if not, I’ll try to figure it out.

[quiet solar]

You might as well just share the Studio project and I'll do it for you at that point. xD

[high snow]

Honestly, that would be great if you could. How do I share this to you? Because I never shared anything before. And I’m sorry to ask such a tedious task. I honestly hope you don’t mind.

Honestly, I feel so bad asking now. I have a feeling you probably have your own projects to work on. I’m so sorry. 😭😅

[frozen dagger]

@high snow If he does the edit in studio, can share the project then you can look at the version control and see every step taken, that's pretty close too Screenshot storyboard.

[quiet solar]

Click the title of the project in Studio. It opens a dropdown from which you can choose 'Share Project'. Drop me a DM with the link, and I'll see what I can do. xD

[high snow]

Ah, ty. Okay

[high snow]

Ty. I’ll also give you all of my lyrics and stuff so you can see it from there and you’ll see the part where it’s supposed to have just a female part. I’ll actually take a screenshot of the lyrics and I’ll circle it in yellow for you.

[high snow]

I sent you a friend request and I’ll send the shared link in a second. I’m just regenerating a song because I messed up at one part and I’m so mad. I just have to work on this song really quick and then I’ll send that over.

[high snow]

I’m having a hard time trying to share the link because every time I try to press the part where to share for it to generate a link it keeps glitching so I have to wait to see if they can fix that for me.

[frozen dagger]

curious about this myself. I suspect that if we transposed a few semi-tones, no big deal. that low could cause issue. (if that doesn't work with just down-pitched, i'd try stitching non-transposed as well, so it has something to match.
but if you tried to create the voice with it. curious how it went for you? gonna try some myself. i made some Suno ones, would like to do it with my own voice as well.

[ripe galleon]

im onf reetier

freetier*

i wish i could do it but i can only just theorize

[frozen dagger]

heard.
"To the musemobile!" (gonna see what it do.)

[ripe galleon]

alright

[abstract flower]

Can anyone please help me? Suno in the desktop browser. Yesterday I opened Suno, and found all my pinned songs rearranged as if randomly. Certainly not sorted by plays, by likes or by release time. I might have not opened Suno for 3 days, no more than that, and the order was as expected (last pinned on the left). I unpinned and pinned again 3 songs, and, when I pinned them, they returned to their seemingly random places — though this be madness, yet there is method in't 😁 Anyone else sees the same thing? Is it a known issue?

Another question: we used to have a channel called technical-chat or similar, where I usually asked and answered these kinds of questions, but i cannot find it any more. Is it gone? 😭

[frozen dagger]

Renamed to this channel. The descript is the same. "talk about prompting", it's for questions and answers; learning.

[abstract flower]

Got it. Whew! 😤 That's a relief! And I'd appreciate help to figure out the issue with pinned songs. Do you perchance use a desktop browser and pin any songs to the profile? If so, has they not reordered in the last 3 days for you?

[frozen dagger]

I had similar issues. Seen others saying same. Particularly last 2 or 3 days.
When I go to my profile now, they load out of order (most of the placement of things are not where I put them), but after a moment it loads more and shuffles them to where I placed everything.

[abstract flower]

Huh! What I see is the opposite: while the page loads, the pinned songs are shown for an instant in the right order, but then jump to these strange shuffled places. Did you happen to hear if anyone opened a ticket? Will do if no one reported opening one.

[frozen dagger]

the more ppl report the better.

@abstract flower prolly post in the bug report. u said desktop? so https://discord.com/channels/1069381916492562582/1082775596767117312

[summer pond]

so I made my custom model but i'm a bit confused on how I would use it. Would I use simple or advanced?

[frozen dagger]

I'm gonna give this one the stamp of approval, with a caveat.
I was able to transpose my voice -13 semitones, and use it to create a voice. Voice verification didn't work when I read the lines with my normal voice however. (about 5 attempts with normal voice)
When that didn't work I read the voice verification in a low register to imitate my transposed vocals... first time was a success.

For comparison of what you showed and what I used:
https://suno.com/s/4kwa5dq7B5vMuwgx

(yours sounds GOOD. Sounds like Little Nicky dreaming.)

[ripe galleon]

damn

[plush field]

has anyone else noticed that v5.5 has a bias toward live? it keeps adding a live audience to my track even when i didn’t ask for it

[ripe galleon]

so it works

this is something you genuinely generated with suno

?

[frozen dagger]

looks like you can use Simple or Advanced mode with your Custom Model. Select your model from the list and create as normal in either mode.

[frozen dagger]

the audio i shared was my recording i took into Suno Studio and transposed -13 semitones, exported as new audio. then used that to create the voice.
It does work. You will need to imitate your down-pitched vocals probably tho. So there is room for it not to work, but it was not hard.
(I had other clips ready with my normal voice and the transposed ready to test in case, but i never needed to use them.)

[ripe galleon]

oh i see

hmm

is there a chance you could just make a song with that voice

and show me how suno really handles it

[abstract flower]

There used to be a bug report form with a simple single-word URL, like suno.com/something, but I forgot it, and cannot find it any more

[frozen dagger]

i tested it i can send in a DM. it used it but it went to a normal range back and forth throughout as well. would turn audio influence up... and i have some other tricks that would prolly keep it low thoughout if i wanted.

[ripe galleon]

send me in dms i wanna see

[polar spade]

Suno.com/feedback ? Or click [ . . .More] in the left menu column and Feedback ?

[abstract flower]

This is it, thanks a lot!

[severe narwhal]

looks like the credits balance/counter (in the top left of the site) has disappeared? 🤔

[prisma scaffold]

Hi can anyone help me here please 😎, trying to get some stems from a song that was created using Suno. It only lets me extract 6 stems rather then the 12 stems as advertised am I missing something here? 😆😁

[half dragon]

You don't see the option to extract All Stems? not just vocal + instrumental?

[prisma scaffold]

Yes I see that , buuut It only extracts like 6 out of the twelve stems constantly which is kinder annoying or is that up to 12 and not 12 every time?

[half dragon]

I don't think you can control that. So it extracts whatever it detects

[prisma scaffold]

Oh I see 😆

So I would need to go somewhere else to extract the other stuff I want then I guess? 😀

[frozen dagger]

mine is still there. check the top left of the UI for an arrow to expand the sideboard. Collapsed it shows the icons only, when expanded and has icon and text 🏠home for example.
but some said their UI changed, so if the collapsing sidebar UI isn't the issue, than you may have a different UI than others. Some have different, rolled out changes.

[red crown]

ok that wasnt a promotional link

woodhouse you stink

i made an ap which will rewrite words and sentences either phoneticaly, or dictionary or IPA pronounciations, for those tricky words that just wont pronounce correctly, and in the rewrite you can click any word to get multiple other variations if anyone is interested

[torpid hollow]

Is there a method SUNO would accept for cleaning up a reference file? It doesn't seem to accept negative prompts. In the following song, I don’t know why he added so many whooper/hooler effects (vocalizations), even though I’ve accepted the country-trap style from the start across multiple versions—but this noise ends up being a form of noise pollution (often added randomly)... Thanks in advance for your help

https://suno.com/s/BJfDy4EF0mXqBIHQ

[lone nova]

Why does suno say hook failed moderation? can I see the reasion som where?

https://suno.com/hook/a1208763-1437-44d7-97b0-52875f5ad29a

[pure panther]

#🚀┃updates message

...This is a terrible change

[dry hound]

so beyond Shady buisness practice...
Not to mention that not saying anywhere visibly how many credits the creat button costs being against EU Laws.. but ohhhh welll I guess...

[pure panther]

It's kind of baffling.

[dry hound]

Luckily soon the EU will also ban weird Money->Credit conversion rates... so you always know what you're actually paying at any given time.

[lone nova]

Do you guys know any documentation about hook moderation?

[frozen dagger]

I really want to play like devil's advocate here, as I try to see it from all sides as much as I can, but

oops tapped enter haha... but.... while I can understand the message of "we don't want you worried about running out of credits".... i'm gonna go with making things harder to find not being the solution.
feels again kinda like test driving the car, "what is that knocking sound?"
Car salesman says, "ya know how to fix that" and turns up the radio.
Out of sight, out of mind wouldn't be the route i'd take here. Maybe more ways to "earn" credits, or something that counter-balances the "dread of token burn" without just buring my head in the sand.
IF that's how we are looking at it. I don't see it that way. Not really eye-balling my credits, but I get plenty with Premier for the time I have, and how much I use Suno. Kinda run low at the end of the month and then burn on image animations and instrumentals.
Not my monkey, not my circus. Doesn't really bother me. But feels like we are solving problems that don't exists a little bit?

[pure panther]

An easy fix is to offer a toggle.

[frozen dagger]

yeah it should be a choice. if ya agree with not wanting to see em then poof. or toggle off by default, but let us who want to see it at a glance have it on. done.

[dry hound]

"we don't want you worried about running out of credits" This entire message was not said by anyone ever once... That is marketing psychology nothing else

building a plugin to get the credits displayed again right now...

[proud crow]

My fix for v5.5 sounding "metallic" or "thin" and genres sounding artificial and "off." 🛠️

If v5.5 sounds like garbage, it’s likely the new "My Taste" feature over-tuning to your history. I've found that the "Taste" feature seems to be siloed per Workspace (Not documented though. Only says "Account-Wide").

The Fix for me: Started a brand-new Workspace for a clean slate. It seems to kill or severely weaken the "sonic bias" from your old likes/uploads from other workspaces and the audio quality clears right up.

Pro Tip: Don't upload or "Like" anything in that new workspace unless you want it to start influencing the sound again!

[pale trench]

how can i stop the stupid intro vocal hums and adlibs

[gloomy jetty]

whats the prompt you are using ?

[pale trench]

Im not even prompting for an intro, its just always giving me one. I've tried to put in [Intro no humming or vocals in intro] but thats not working either

[gloomy jetty]

is the weirdness high ?

[pale trench]

nah, always around 20

[gloomy jetty]

if ytou try to generate blues you might have tape hiss at the start

[pale trench]

im doing pop mostly, so i know thats going to be tough to break lol

[gloomy jetty]

try adapting this prompt toward something you would like : brill building, r&b/soul - woman psychedelia, studio recorded, good quality, professional, harmonic, mastered, masterpiece, polished, afro house, slow piano and soft strings, grand piano
upbeat

[pale trench]

I've been using this and its' been pretty good, i just cant get those damn vocal chops in the intro : [Audio Engine: Ultra-HD 96kHz], [Buffer: Lossless], [Dithering: None], [Frequency Response: 20Hz-22kHz], Dark alt-pop at midtempo with detuned synth pads, muted electric guitar arpeggios, and a punchy kick-snare groove. Thick sub-bass, reverse piano accents, and wide chorus-drenched lead synths build an eerie, glossy texture. Tight verses, explosive layered hooks, and tape saturation add a moody, darker edge.

[gloomy jetty]

in the lyrics portion try [chorus] or [wide chorus-drenched] i got soul * or whatever

if this doest not work or it doesnt fit with your other lyrics,

Only generate one song with the chorus nothing else
Then remix the song with the lyrics

[gloomy jetty]

try creating your own model using some of your past song it's 100 credits but i see that i dont have the humming issue when i use the model that i generated

[spring fable]

Is there any way to turn off the my taste thing? I tried making a new workspace and it's still adding the weirdest bs to the prompt

[spring fable]

thank you so much, I'm not sure how I missed that

[polar spade]

Where would prompts be found in a new workspace, itself? I suppose you mean then , in the style prompt box of the Create module? Do prompts just suddenly appear, or are you uploading a file?

[warm jolt]

🪇 Question regarding search :

Is there a way to include AND if I search for songs? & or + doesn't seem to be working. Neither does "word word". It's always an OR search.

The search for songs is always some random mess.

[ripe galleon]

is there a way for someone on freetier to get suno to use only 2 specific instruments?

[spring fable]

B̤̿it̺̕B͓̚ur͍̒neȑ earlier said that creating a new workspace can help clear up if things are getting messy, I wasn't sure if that was a normal fix. For me, I have a bad habit of choosing 3-4 things off the prompt list that I don't recognize just to see how it sounds so "my taste" is all over the place. When a song isn't going where I want it, I tend to use the magic wand to get ideas and it was driving me crazy because it was adding really strange stuff. Even without the magic wand, it was adding choir and reverb to everything from my "music to sleep to" phase. Sorry if none of that is what you were asking, I have a very disorganized way of dealing with these things

[viscid ravine]

I can't log in my account - https://suno.com/@noudvankuppeveld, What can i do

[quick wedge]

Hi Noud. I’m sorry it’s not letting you log in. Try these troubleshooting tips first https://discord.com/channels/1069381916492562582/1417459233544339537 if nothing works after those try logging in with a different browser see if it helps. Make sure you’re using the correct login info as well.

[ripe galleon]

can i make two instruments like an upright bass and a bass guitar do a call and response-esque song?

everytime ive tried, suno just couldnt do it

[daring bramble]

It's possible but it requires some structural input.

For example, in your lyrics field, structure the song for your call and response like this:

[Instrument 1] [solo call]

[Instrument 2] [solo response]

Remember to have spaces in-between them to make the call and response between them effective.

It can be hit and miss with Suno, so try generating it a few times and it should eventually land.

[ripe galleon]

used up all my credits for today

so ill do that tomorrow

thank you lizi

[daring bramble]

No problem 😊
If you have any issues when you do try to generate it tomorrow, let me know

[ripe galleon]

alrightt ill do that

👍

[hasty cove]

can someone help me with a basic feature? I want to just entend to a new part. Say i have chorus and verses done and i just want to add a bridge section to take out the song... when i select EXTEND, it basically doubles the song and repeats the 1st half of the song. i selected the entire existing song as "keep" and just wanted to add a new section. It also changed the existing song musically which i didnt want. It would be so much better if SUNO had song sections you could select "add bridge" "repeat chorus" for faster song struture etc. can anyone help me out?

[ripe galleon]

why does suno think this has anything to do with jazz

[polar spade]

Thanks for the response. That is helpful to know.

Ahhh..... then, yes. If the magic Wand Button is Blue 💧, you can pull ideas from your My Taste register that has been analyzing your past songs as you create them.

If the Magic Wand button is Orange 🍊 , and we have not typed any prompts in , then it will give random suggestions. Sure. It is starting with a blank style box and doesnt know where to go.

But if we bring up the Create module and Suno automatically throws in prompts as you described , without typing anything in or clicking the magic wand, , then that is very odd.

We can get random prompts when we upload a file as it tries to get a vibe of the song, but thats kind of rationalized. But just tossing in styles out of thin air is strange.

If we go to My Taste and toggle off the My Styles , that turns the Blue Magic 🎩 Wand Button in the Create module (styles prompt box) back to Orange 🍊. But either way, nothing should appear on its own, without our typing something in the prompt box or clicking the Wand button.

If it randomly is popping prompts in, try to record the acreen while its happening , or screenshoot the prompts and give feedback in the bug-web channel https://discord.com/channels/1069381916492562582/1082775596767117312.

[digital osprey]

Hey everyone, I’m currently working on some tracks with Suno and I’m trying to push the final quality a bit further.

I was wondering how you usually handle the “mastering” or final polish stage after generating a track.

Do you use any specific tools or workflows to improve things like volume, clarity, or balance, especially for phone listening?

I’m mainly looking for simple and effective approaches, not overly complex studio setups.

If you have any tips, tools, or habits that made a real difference for your tracks, I’d really appreciate it 🙏

[polar spade]

Pointing to the obvious in terms of "simple", you may not have used the Remaster tool as of yet? Found in your song menu under Create/ Remaster.
Its there for that purpose. Depending on what you feed it, it can give some interesting results. You have a choice of Subtle /Normal/ High. Each can work differently depending again on what you're working on .

[nocturne tulip]

Hey guys, I'm on mobile and I was wondering, how do you make it where to people sing in the song

[digital osprey]

Yep, I’ve tried this feature, but it honestly feels like it does nothing on my songs. Even after generating 10+ versions in normal or high, I can’t hear any difference. I tested with SunoMaster and Mixea (via DistroKid, the first master is free, then it’s €89… lol), and there I can actually notice a real difference.

I’m not sure what’s going on, but my tracks sound bad on phones, everything gets too bright. Even after mastering with those tools, it still sounds overly sharp.
I guess there’s no shortcut and I’ll have to learn mastering myself…

[quiet solar]

It really depends what effect(s) you're trying to achieve. Unfortunately, the whole mastering process doesn't lend itself neatly to "just always do this one trick".

If you're not working with stems, your options for balance and tweaking are somewhat limited. Suno natively includes an element of production effects on the song, so it's not straightforward to make tweaks and changes without breaking stems apart. Distrokid's Mixea or Bandlab's free equivalent might improve things (they're effectively AI-driven retuning of EQ levels), or they might make things worse - the only way to find out is to try.

Things being overly bright is a common problem with no one-click fix. Something you can try is covering the final version (low weirdness, high audio influence) and tweaking your prompt a little to steer more explicitly away from brighter sounds, which can be helpful if you've got an obnoxiously shimmery section thanks to the cymbals or synths.

[nocturne tulip]

Hey guys, I'm on mobile and I was wondering, how do you make it where to people sing in the song

[quiet solar]

Just to add from your updated post, depending on the genre you're working with, a final pass on Cover also gives you the chance to focus more heavily on production, mix and mastering effects in the prompt without having to worry so much about the specific details of the instruments and vocals (you're using the audio as a prompt for all that). If you like long, detailed prompts that can be helpful in making space.

[quiet solar]

I'm not sure what you're asking. If you've put lyrics in the lyrics box, then Suno should be singing them?

[quiet solar]

Make sure you're only including the new sections in the song's lyrics box (ie, don't include anything from the section you've marked as 'Keep').

Slight changes are to be expected with Extend - you can help limit them if you reduce the weirdness slider and increase the audio influence slider, but often you'll be able to hear a difference when the extended part kicks in. Easily fixable, though - use Get Final Song to combine the extension with the original into a new song, then either Remaster it or Cover with lower weirdness and higher audio influence to get a version that should sound more consistent throughout.

It's a good idea during this process not to get too attached to any specific version of the song Suno generates, until you're making those final changes. Otherwise you open up the question of "How do I keep this part exactly as it is while making changes around it?" which is very difficult to do without your process involving Suno Studio or using a DAW of your own.

[untold briar]

For duets, try the simplest thing first and sometimes you get lucky and it works. Just ask for it.

MAN: I am singing.
WOMAN: Hello there. Now I am singing.

[ashen cliff]

Good Morning Suno Team,
I experience from time to time, issue, when I create a song in normal mode (not in Studio),
THen I move this created song to Studio, and then add extra track, e.g. Guitar,
However it produces in with shorter timing, as it is shown on printscreen
What do I do wrong? Could you please help out?
Above there is a song
Underneath there is created guitar track, which is shorter

[digital osprey]

I guess by using Remix with the same lyrics and specifying the mastering work to be done in the style section? As well as setting Weirdness to 0%, Style Influence to 100%, and Audio Influence to 100%.

[quiet solar]

I wouldn't put the sliders to 0/100 - especially not in v5.5. That'll more likely cause you new problems than fix your current ones.

There's a certain dark art to tweaking slider settings to "keep everything the same, but change it please" since it's inherently contradictory. If you're looking for a close cover with only subtle changes, something like 15% weirdness, 80% audio is good for v5 and before. I'd start with 30% weirdness, 50% audio on v5.5, which is much more sensitive to slider settings.

If you're making deliberate changes in the style section, maybe nudge style influence up to 70% or so for v5 and before, and I'd personally just leave it on the default 50% for v5.5.

[unreal sleet]

from nowhere, my gens are now repeating the lyrics twice making the songs like 8 mins long, something happening at the headquarters?

[earnest agate]

one thing is for sure, 5.5 can tell/follow a story way better than before
https://suno.com/song/8308e8ac-b889-42fb-889b-139d5b34f7a6

[slender fable]

HELP! How do I make drumless music??? 😭

I just want soundscape synths without stupid drums added…

I’m using the 4.5 model, by the way.

No matter what prompts I use, it always adds drums halfway through the song.

[unreal sleet]

is there a genre that doesnt use drums that you want to replicate?

[slender fable]

Synthwave/soundscape/sleep music.

[unreal sleet]

okay then u cant mention synthwave first of all

[slender fable]

Alright. I tried that. Even asked ChatGPT to make some prompts for me and it still didn’t work.

[unreal sleet]

the older model is known to add surprising stuff to the gens tho

[slender fable]

Should I change the weirdness and style influence sliders?

[unreal sleet]

depends on what you put in the style mate

and dont mention bpm if ur using 4.5, it will think of rhythmic and come up with drums. even excluding drums/percussions adds a thought of adding it to the ai. its very bad at understanding include or exclude. a common thing back in 4.5 was excluding "humming vocals" and that just made it include it even more

[slender fable]

Does “drumless” work as a prompt or does the word “drum” just trigger it to add them?

[unreal sleet]

percussion works better

and yes adding it as an exclusion can actually make the ai think of drums and add it anyway

[slender fable]

Is it like googling something where if you put “no [something]” it shows that [something]?

[unreal sleet]

yeah

add what u want

dont add what u dont want

kind of

[slender fable]

I see.

[unreal sleet]

i think v3 is better for minimalistic ambient music tho

[dim vigil]

Hi. Can please somebody help me with my issue. Suno creates songs that do not match the specified style parameters. I set the tempo to 120 bpm, but the song comes out everytime at 90-100 bpm, which is already a lot of wasted points. I don't know what to do

[unreal sleet]

but i would still try to get it to work on 4.5 or higher cause of the quality

[unreal sleet]

are you doing a genre that is normally slower than 120?

[unreal sleet]

Like google says; Avoid just typing "no drums," as the AI can sometimes ignore negative prompts. Instead, use positive keywords that define a drum-free environment:

[dim vigil]

not actually, i just want same genre but slightly change to difernt style

[unreal sleet]

hard to help without seeing anything though

[slender fable]

I don’t see a v3 option. Is that website-only? I’m using the iOS app.

Also, I’m a free user.

[unreal sleet]

ahh if you're on iOS sorry then no

[slender fable]

How about the website version?

[unreal sleet]

u can even use the "vintage" v2 there 🙂

[slender fable]

Haha, I’ll give that a try. Thank you ^^

[unreal sleet]

Good luck!

[slender fable]

Thank you!

[quiet solar]

It's not easy, since percussion is so prevalent across so many songs in so many genres.

It can help to lean into prompts for music that specifically doesn't have it, even if that's not perfectly aligned with your target genre. "Acoustic" is a good catch-all to include - but ultimately, the more your target genre does typically have percussion, the harder this will be.

It also helps to specify the instruments that you do want, and not mention percussion - for example, a cello/piano acoustic track will fairly often not have any drums in it. Also worth noting that your goal probably isn't to eliminate drums from everything you make, but to increase the chances of getting a track without them from "incredibly unlikely" to "happens enough that I can pick a workable song from what gets made".

And echoing Bored Ape's point - no matter how tempting, never put "no drums" or variations on that in the prompts. That'll actively work against you. Adding drums or percussion to the exclude styles field may help, but the impact is fairly minor in most cases.

This could be quite a frustrating exercise with limited credits on free tier, so just be aware in advance. 🙂

[quiet solar]

All style prompts are better considered as suggestions rather than instructions. 120 BPM is at the upper end of average for some genres, so can't really give to much more detailed advice without knowing more about what you're trying to make.

However, if Suno is really struggling, it might help to approach the problem from a different angle. Try recording (or borrowing) a 120bpm click track or very basic 120bpm drum loop as an audio feed when you create the song - that should get you to something that you can make further iterations from using Cover.

[slender fable]

I see. Thank you. Unfortunately, it said I don’t have permission to use exclude styles when I tried to use it in the past, and now the option is gone entirely.

[polar spade]

Free user only has access to 4.5all

[slender fable]

Should I try upping the style influence slider and lower the weirdness slider? Or something else?

[slender fable]

Ah, I see. Thank you for letting me know, especially since I couldn’t find it on the website.

[polar spade]

It may help to see most things will reference Pro, if a Pro/Premium tool.

[slender fable]

I guess I should probably specify that I use the free tier when asking for tips, then.

Thank you all for your help.

Honestly I wish it were easier to get the ai to actually listen to your prompts.

I have difficulty making drumless music and songs with acoustic guitar without it automatically turning it into an electric guitar.

It only seems to work with classical instruments like piano and violin/strings, not synthesizers. (Regarding generating music without drums)

I’m probably going to make a heavy metal song to express my frustration, hahahaha

[polar spade]

oh.. ahhh... No I meant within Suno's menus we will see the word Pro to the right of many things to show what we cant use as Basic tier. I didn't mean comments in here are to be assume Pro answers. 😁 Help here is for every tier and every user. 'Ask away

But yea, it's great when any Post-er will say what tier they are using.

[slender fable]

No worries, I was clarifying just in case. Thank you for your time 😊

[slender fable]

Ooh, I like how they turned out, hahaha

#🐣┃suno-showcase message

[pure panther]

Have we always been able to get .wav files for extracted stems or is that new? Have I missed that this whole time? 😂

[frozen dagger]

think it's been around for a good minute. dunno about always. 🙂

[pure panther]

Better to know now than never I suppose! 😄

[pure panther]

What's the best way to add drums to a drumless track? Without changing the existing track. (other than adding drums)

[balmy pelican]

How do I get Suno to stop Humming in the Into?? All I want is Instrumentals.. No humming, ad libs, melisma, chanting, choir,?

[frozen dagger]

This is the best suggestion for that imo, #📚┃suno-school message

[quiet solar]

Not always, but since v4.5. 😉

[acoustic egret]

yo guys in suno is there a way to upload an acapella and tell suno to modify it and add harmonies etc...

[rough depot]

seriously? 14 syllablse and it STILL insists on SINGINGEACH ONE AGONIZINGLY SLOW???

[sacred swift]

https://suno.com/s/KmUofxGWkmwV5o0L

I tried it, and it looks like I was able to sort it out just with the prompt. If you set Audio Influence to around 80%, it should probably work.

[acoustic egret]

so what you just upload your acapelkla and say for it to write =vocal harmonies and set audio influence to 80%?

[sacred swift]

https://suno.com/s/SPSGkQD8gOkq4RZu

Sorry, I got it wrong. If I cover it with the backing vocals I brought into Studio, I think I can make it work. The original had a male vocal, and I layered a female vocal on top.

[acoustic egret]

im confused af

[sacred swift]

What I meant was: if you use Studio to cover it based on the original vocal, it can turn into a harmony if it works well. The first prompt I tried wasn’t creating harmonies, it was just replacing the main vocal line. Sorry about that.

[split flower]

D M S open

[quiet solar]

Could you confirm whether you're on Basic, Pro or Premier? There are a couple of different options you could try, depending on what tools you have access to.

[acoustic egret]

Suno Pro

[quiet solar]

In which case, you were thinking along pretty much the right lines. Upload the song, cover it with something like 30% weirdness and 50% audio influence as a starting point if you're using v5.5 (it's a lot more sensitive to sliders), and use the style prompt to describe what the song should be. So I wouldn't only put harmonies and a capella, I'd describe what the voices should sound like (e.g. smooth male baritone, breathy female soprano) since Suno will be replacing your audio with an AI singer.

If Suno generates instruments in the background, then you can split the vocal and instrument stems so you've just got the vocal parts. If it's important to have the final version on Suno rather than as an MP3, you can always download the vocal stem and reupload it as audio.

[acoustic egret]

Ok thanks so basically do a cover of an uploaded acapella

Yo btw how come it doesn't show how much credits I have left anymore? It use to display it on the top left on suno but doesn't anymore

[young geode]

Short answer: you don't.

Longer answer: you need to find something that will substitute the drums. Anything that can add that symbol sound/kick drum sound to your song can effectively be tooled to get rid of the drums.

Additionally, targeting genres/methods, that have no drums will also aid in getting almost no percussive sound. However, if you're trying to target anything that uses drums, you'll find that you're likely better off just stemming then out: most songs want to at least add the 'Million Dollar Beat', which is the most basic, and common pattern, across all genres.

[cursive frost]

there is no planet on which these vocals are bad. If they annoy you, that's an opinion, but they are objectively good. The fact that they're good doesn't negate the validity of your opinion finding them annoying, but if you think they're not good, that just isn't the case. https://suno.com/song/b6812f9c-de8f-43e2-98ea-1749e2d157cd

[junior moon]

is it an AI written song

[devout sinew]

I uploaded my vocal, but every time Suno generates a song, it turns into a duet.

Instead of using only my voice, it adds another unknown voice. How can I make Suno use only my uploaded voice for the entire song? I am not referring to background vocals. The other voice is singing solo in the verses, bridge, and choruses.

[stark osprey]

Place tags before every verse, chorus, bridge or where ever there are vocals. In the lyrics box like: [verse 1]
[Male vocal]
Lyrics here

[devout sinew]

Thanks for the suggestion.

Even with vocal tags, Suno still adds another voice that is not mine. It is not a gender issue. The problem is that it switches to a completely different singer for full sections of the song.

I want only my uploaded voice to be used throughout the entire track. Is there a setting or method to prevent Suno from adding another voice?

[slender fable]

Like, replace the drumbeat with a pulsing synth sound instead?

[wide stone]

hey guys

anyone do brazillian phonk

[hollow slate]

it's under account now

[hollow slate]

not yet

[stark osprey]

The only other option i would see is the audio influence slider

[wide stone]

huh

[young geode]

Correct.

And if you name drum machines/synths, like Roland TR-808s, it'll help focus it further.

It's not a guarantee, but SUNO wants a beat, do you need to find a way to get it's attention onto something beat-like

[slender fable]

I see. Could you give me an example, please?

[young geode]

If I dug through my failed gens, I could probably find you some.

I tend to have the other issue: I want specific drums in my songs. (I'm currently testing which beat patterns regularly register.)

[left hornet]

Hello, I’ve experienced this too. In my case,it actually makes sense. It happens when my voice can’t reach higher notes. So when the song goes into higher notes, it switches to a different voice. I think it’s because the reference voice doesn’t have enough range for those parts. Even when the audio slider is at 100%

[stark osprey]

Then specify the vocal range of your own voice if its like Bariton or Low tenor and make sure to exclude falsetto

[slender fable]

Thank you!

[sturdy sentinel]

I'm a little late to the party so if this has already been resolved, I apologise. I found the only way to do it is by doing it in parts. Like a verse at a time and using the extend function.

[slender fable]

By the way, in case anyone is wondering how to get a piano-only song (such as an emotional classical style) without any other instruments or any percussion, here are the prompts I used that worked successfully:

piano, grand piano, piano solo, solo piano, heartfelt piano, melancholic piano, emotional piano, classical piano, piano only, dreamy piano, romantic piano

You can probably play around with the adjectives/mood, like replace “melancholic” with “lively” for example, and still get a piano-only song. I would recommend keeping “piano” with it, just to enforce it.

(Note: I did base the song I made on an audio clip of piano music.)

[spring ice]

So for using our own voice, what kind of voice sound input is needed? Singing one? Or talking one?

[unreal sleet]

Sing in the style and language you want

[torpid hollow]

I'm noticing some issues... I'm loading a vocal file... I know why SUNO applies nonsensical metadata (it decides on the instrument type, etc., even though the file contains none of those details). All of this just to generate a clean, unsaturated audio track... Well, this metadata persists and pollutes all subsequent generations: by systematically adding instruments (the acapella and No_Instruments instructions are ignored). The weight of the metadata interpreted by SUNO takes precedence over everything... Is there a solution to prevent it from overproducing, to clean up this strange metadata? Thanks in advance for your help

[polar spade]

Speaking can work as well, with the caveat that the song can then get a few "ad lib" or non-melodic style spoken lines here and there, instead of singing the melody. It may though have the benefit of recognizing your voice "characteristics" better.

From my tests...If you sing a rock song, and you prompt a rock song then Suno tailors the voice pretty well to the rock song, but might not sound as close to "you" next time, in a folk song. Suno actually suggests to sing for a +Voice in the style that you wish to produce in.

Of course, the solution is to make a +Voice each for a few genres, to also include one spoken version. Use the one that fits the work.

Also, Suno hasn't worked out my "range" and just boosts my ego by singing three octaves above what I could possibly sing. Then, though its my dream voice, its no longer reproducible for me in real life - i.e. using it for voice training.

Im working on the guess that then using Custom Model would work better by uploading lots of songs within my range. Testing that too.

[unreal sleet]

I added stuff like [Chugging Accents] after the first verse and almost all gens are repeating the lyrics twice making the songs 8 mins long.. anyone know if this is the reason or have any idea?

[polar spade]

I think just a glitch. I keep getting 6 and 8 minute songs that repeat since yesterday.

[unreal sleet]

Feel like more people would have said something then 🤔

or are we special mate 😄

[polar spade]

I wasa on quite all day yesterday. I recall about 4 or 5 noticing this.

[unreal sleet]

Ahhh ok!

Would be nice to get a pop up notification saying when theres glitches around so people dont waste their credits

[polar spade]

Just after maybe noted the quality seemed to be better as well. So maybe an update that helped one thing and opened a can of worms elsewhere. Its a common occurence that a gate gets opened like that. They just need to catch that when Easter is over is my guess.

[unreal sleet]

2 out of 8 songs are doubled ish

Yeah

[polar spade]

Meh. Good idea. Not sure what to do with it though. It starts with user reporting it.

[unreal sleet]

I would have liked to known there was a glitch since I think Im doing something wrong

[hard beacon]

well.. it's sunday.. i don't think they are working 24/7 on 7 days 😉

[unreal sleet]

I mean does the ai stop working cause they are off? xD

[polar spade]

You're pointing it out is a start. And you said, you havent seen others say much. So the more people report, the faster it gets noticed and reverted or fixed.

[hard beacon]

nope.. but does the ai can repair itself (if there is a bug?) 🙂

[unreal sleet]

ah i was thinking another thing haha

I also seem to almost be the only one whos gens go from 50% volume in the beginning to 100% towards the end, been talking about this to david for months but he's been ignoring me the last 2 months

[polar spade]

Glitches and bugs are baked into anything like this. Theres a loss when we have to wait for a correction. But there's an expense to provide 24/7 maintenance as well. Im OK with what I'm paying now and waiting a couple of days for each correction than to pay more for the immediate fix. 😊

[unreal sleet]

Yes problem is I didnt know something was off, I always begin to think I broke something

[polar spade]

Yea. Im with you 100% on that. I read that feedback every now and then for quite some time. Must be a tough one to crack on how to make it consistent on all prompts and genres. I certainly dont know. Occassionally I've had to split up segments and adjust volumes -- tones too.

Occassionally Remastering has helped, but not always.

[rich eagle]

I have a song which is supposed to be a male/female duet. Is there a way to get Suno to have the lines sung by the singer you want to sing a particular line? I've not found a way for Suno to get it right yet.

[unreal sleet]

Yeah weird thing is that remaster at any strength and covering at any slider doesnt help, as if the song is marked with "remember this volume increase"

its been months trust me i have tried everything -_-

OK not everything, I haven't made a new account, but I have a new PC at least lmao

[unreal sleet]

this works 100% of the time after adding some songs to a model. dont need anything in the style box about it
[Chorus: Female clean vocal, melodic resonance]
Call it life

[Chorus Response: Male deep guttural growl only]
BUT IT'S ALL A LIE

[polar spade]

I compare that to asking if we can get a tenor sax and a bass sax to play on alternating lines. So far, Suno hasn't devised a way to say what's "female" and whats "male". Its good at getting a range, and seems to depend on the genre and lyrics as well -- And for that, Suno has created a specific tool to click Male/Female so its forces something on the backend to at least start in that range and try to stay within it, unless otherwise prompted to veer in another direction. A vague way to say, its not a perfect science yet to just prompts for it.

For fun I even got a third "frequency" of human voices by prompting for "androgenous" , while including broadway and ensemble in my prompts, but that may have swayed it a bit.

Two favourable techniques that are about 15% ....hmmm... maybe closer to 5% productive is to simply use [male vocal] and [ female vocal] each time before each segment or line.

The alternate then is to use parentheses where you want the alternate voice to sing as an accompaniment or response. such as:

[male vocal]
I want you here
(I want you too)
I need you near
(Im here with you)

[female vocal] " Its where we need to me"
[male] "just us , just you and me"

[polar spade]

@unreal sleet makes an excellent point that seems to help. Thats to give additional description to the voice itself. Clean, gutteral, falsetto , soft, whisper, even child and kid can work a bit more in 5.5.

[rich eagle]

This was what I'd tried but it ignored it for some parts

[unreal sleet]

yes and my male is always in CAPS

[unreal sleet]

I would spam gens until you get a few ones that work and then use them as inspo or use them in a model and suno wants to fullow that structure

[polar spade]

Great. Then you are updated to a level with the rest of us. 😊

Seriously, I have done most of my "forced" duets in Pro level by using extensions and extending before every vocal gender change. Its a 50/50 change of it getting it right when, like you say, it doesn't ignore the prompts.

Once, it works I move onto the next stanza or line. So credit wise, its spending more time and credits probably when we want to do duets. Since we just dont do duet every day, its not bad to just set aside time and use extension to slowly write the duet.

I do more often, do a backwards version. Prompt as much possible, let Suno do the "ignore some parts" bits while generating at least 10 outputs. Find the best one that has the most things correct, then rewrite the lyrics to fit the rest. Cover and sometimes we're done.

[hard beacon]

and if you think an older version give's you better results for your idea: use it and cover it with v5.5 to get better sounds 🙂

[timber violet]

is there someone I can message? I need to ask about some stuff/support. I'm having issue logging in.

[half dragon]

Have you sent an email to [e-mail]? and visit #ticket-help

[polar spade]

@torpid hollow Often we can just add a +Voice we have previously created to help replace the voice there is.

Have you experimented with making and adding +Voices

[shrewd willow]

Guys, how do I create persona now? I know its called voices, but how to create a new one?

cant find the option

I only find create inspiration, but not voices

[polar spade]

From a songs menu you can go to Create / +Voice.

If its greyed out or missing, then the song is considered ineligible for making a +Voice from it.

[shrewd willow]

oooooh @polar spade Thank you!

[vague scaffold]

is the v5.5 extend bug fixed

its been days

[shell fog]

I'm not sure if this is off-topic or not, but how can I add just the bass guitar and drums to a song I recorded?

[polar spade]

There are some text based guides here in FAQs as well.https://help.suno.com/en/categories/2327233-v-5-5-voices-custom-models-my-taste

[shrewd willow]

I was trying to create from a 4.5 song I had, even remastering it didnt do the trick

[polar spade]

I think we can make a +Voice from most any model old or new.

[shrewd willow]

mine is not there, this is a 4.5 song

[polar spade]

Yes. Then its ineligible for some reason previous. It could be that we made a remix of a song that had a persona in the chain of iterations/generations. Or one of our own had an upload in the beginning.

Find one that works then, to practice on adding a +Voice. Then keep looking for a voice and output that works.

[torpid hollow]

It used to handle "power-prompting" well, but now it gets completely tangled up in the weights… What’s funny about this piece: SUNO adds persistent metadata (description and style); even though I modified the script, these old patterns reappear like ghosts rising from their graves (I iterated through three tracks, the first emphasizing the concert patterns) which I faded out (along with the scratch effects); yet, in version 3, where there is no sound of this type, they sometimes come back randomly...

Clearly, there's a problem with the weighting settings... In places where I managed to emphasize the pronunciation (making the lip movements noticeable: the tongue clicking), SUNO overinterprets this as the microphone being too close and therefore sometimes causes clipping (when parts of the track are consistent).

I don’t know all the script tags to target the right one and fix this microphone saturation—which is intentional and interpreted by SUNO, not by me (I was able to reduce the reverb this way [reverb:-0.8]; now I just need to find the saturation, compression, and microphone as generative parameters—the sound is so distorted that I can’t do anything in CakeWalk).

[polar spade]

Im not doing any magic, but would just asking AI the same question. Grok and others now run daily scrubs of social sites to pull what users are suggesting to try. Thats a start.

Heres some suggestions: For the exclude box- concert patterns, scratch effects, old metadata artifacts, close microphone intimacy, heavy compression, clipping, distortion, saturation, excessive reverb

Prompts to help: smooth stable vocals, natural clear pronunciation without exaggeration, stable even tone” and add section cues in lyrics like [Verse: even tone, natural delivery].

Intentional tongue clicks and lip movements may be tinuated by staying mid-range more from the mic. Pair with “controlled natural dynamics” in the prompts, or tags, for consistent levels across the track.

Add “medium distance microphone, dry vocals, light gentle compression only, clean balanced studio mix” in Styles to counteract the close-mic over-reading that causes baked-in clipping and saturation.

Remember if you have Studio, you might find some use in the FX Remove tool for drying out the vocals as well.

Also, using Remaster might find its way in the chain of tools, but probably only if we can make such a clean vocal from it that it benefits us to make a +Voice out of it and swap vocals later.

[torpid hollow]

ok

[polar spade]

Just askin' but you have a spit screen/ pop filter for your mic?

[frozen dagger]

Just in case... if that is a song you made using a Persona/Voice the option to Create > Voice will not appear. Unless, you open the song in editor and Save as New Song. Which will create an exact copy of the song, but without the Persona/Voice attached to it; allowing you to create a new one.

[polar spade]

Thats fine to help in here, but no promoting your wares, thanks.

[crude hazel]

Been trying for days can’t get Suno to do vocal fry. I was able to do it in v5. But not on v5.5

[small bridge]

Does anyone know Orion from GLXE? I would love to get into contact with them and ask them a few questions if anyone can ask them and see if they would be open to a convo! I saw their interview on Dazed!

[stable steppe]

someone help me. I'm able to get Suno to make a powermetal song I want, the lyrics don't need to be changed. However, I'm using Google Gemini to help me with the "Styles" prompt for Suno, I've had a lot of successes and fails with Gemi, but Gemi has failed me like 40 credits here

I can give you the songs, and label what exactly it is that I want you to incorporate, and fix

[tame venture]

On my personal site I have a breakdown of taxonomic seeding, we identify animals with a structure kingdom, phylum, genus, etc... try the same with your seed, start with the genre, then sub genre, then tempo, actually I forget it's on https://jray.me/what-is-hybrid it has served me pretty well

[cyan jewel]

the fx remove tool ends up nuking surrounding frequencies whenever I try using it. Is that's intented behavior?

[acoustic egret]

Where does it show how much credits you have left? I've got suno pro but can't seem to view it

[quiet solar]

#🚀┃updates message

[acoustic egret]

Legend thanks

[lunar cloud]

I've just started using 5.5 and am trying to make instrumental tracks. No matter what I prompt, it just keeps generating songs with weird semi-legible singing?! The lyrics field is blank.
Is this a known-thing and are there any solutions? It's completely unusable as it is...

[rough depot]

What is going on here causing the lyrics to be incomprehensible?? https://suno.com/s/6k2V8KkPTGiSqPKh

[calm tusk]

Hi I have a question. So with the recent update the process of persona's/voices is different. Now when I wanna make a persona from a song I have made I to verify the voice with my own. But my question is, the ai voice isn't my own so how would I verify it? Would it still work?

[distant gale]

Click on the 3 dots on the song, then choose Create — then Voice. It’s not called Persona anymore.

[calm tusk]

Oh yeah another thing. I can’t click it. The option is grey.

I do have a pro

Oh nevermind😅

[calm tusk]

Thank you!

[minor lodge]

I almost exclusively produce instrumental tracks so I'm pretty familiar with the bleed through that can happen when using the lyrics section to try to structure songs.

The two pieces of advice I would give are this:

1. Whenever you can, trust that your STYLES and other settings (there's actually a lot of them) will give you the sound you're looking for. Some of the best songs I've created (imho) have been when I let Suno take more control and just let my style description ride on it's own.
2. If you absolutely must ensure certain hooks, song elements, or other aspects of the song come through - make sure that you encase all elements of the prompt you input into the lyrics section in BRACKETS []. Every single line/block of text MUST be contained in brackets [] or it's going to run the risk of vocal "bleed-through".

If you want more help, please link some examples and I'll take a closer look to try to figure out what's causing the vocals to show up.

[unborn star]

Is it possible to address the audio file which you have uploaded in Suno? So maybe you can tell him, what to do with it like [audiofile: Command]?

[polar spade]

You mentioned before you dont have Studio. How have you come to this conclusion?How are you using Studio? corrected. User has Studio.

Its designed for voice. But it can help instruments, however if it removes any overtones and dissonance which , such as using a church bell as example, that would remove things that are wanted.

So its like remaster or any tool. Its unique to each sound you apply it to.

Reminder: its for single , clear, stemmed out sounds mostly, not a full combo of instruments and vocals. Though it could work in occassions. I just ran it on a full song and it basically removed the vocals and some instrumental riffs. 😅

So we cant make a general conclusion based on a specific action.

By the way, you can mention this also in Studio-feedback and see if someone is familiar to address it further. That's a good channel to use for Studio questions.

[quiet solar]

There’s no way to reference an audio upload in Suno via prompts. Maybe one day…

[lunar cloud]

Thanks so much for taking the time to respond, but I'm not putting anything into the lyrics section at all! Suno 5.5 is just making up it's own completely random unwanted vocals.

[minor lodge]

what do you have in styles?

Also have you tried putting [Instrumental] in the lyrics section?

[shy dagger]

hey! i’m wondering if i’m the only one getting this: when i trash a previously generated song, the songs currently generating would suddenly disappear :0 they might pop back up later but might not… with page refreshes

[polar spade]

Theres has been some recent latency on the outputs showing up. Some popping up and then disappearing and re-emerging once we refresh.

For sending something to trash to affect the next outputs is odd, for sure. But could be related. The unknown is the unknown. Just have to see what comes of it as the days go by, probably.

Keep reporting through the right channels. Bug channel, feedback rooms and the Feedback form found in . . .More in your left panel menu.

[onyx finch]

when uploading audio, for example a voice... if i fill in the optional tags with a unique string can i use that to reference audio in my prompt? trying for backing vocals, sample drops, duets etc

[cyan jewel]

I have studio. You may have mistaken me for someone else.

[polar spade]

Thanks. My comment updated. I did mistake the thought of your not having Studio with a user convo at 2am this morning. 😳 sorry about that.

[cyan jewel]

OK. If the intent is for it to be used on stems then it nuking surrounding frequencies is probably intended behavior because what I'm doing is applying it to the initial suno output in an attempt to get a v5.5 output that doesn't have the harsh treble overtones and scratchy sloppy inarticulate transients past 2:10. That's my mistake.

I hadn't gone separate steps > remove fx because just stemming out the output in the first place is already destructive

[dense nexus]

Is there a way to take a v4 song and remix it in v5.5 with my cloned voice keeping the same composition, rhythm and style? I've tried recreating the song with the same style prompts and using my voice but it will not match the original composition results.

[cyan jewel]

Or, "It's probably intended" isn't the right verbiage. More like, the expected outcome?

[polar spade]

Really, who knows. We're doing this in reverse. Normally we add effects on top of a dry vocals, so its may be easier to remove to make it dry again.

An AI generated vocal has everything baked in and may be , if even slightly, more difficult to do the same -- there is no forward, so there is no reverse. Its like getting gold from a stone (ai generative) as opposed to getting gold from a radio (handmade integrated parts i.e. DAW). The latter is probably slightly easier.

Its like remaster.... its a one by one test each time we try it. Likely with never a guarantee it will work. But nice when it does.

[polar spade]

Yes. Expected results, but not entitled results. 😁

[polar spade]

Right. We start with any issue that stemming may already throw us. So if the stem is close to being artifact free, the fx removal may deal with both enhancing the vocal to be dry AND remove more artifacts.

Its counter-intuitive, but I think if the stem has more artifacts, then the FX Remove could be more aggressive and leave a strange sounding vocal.

Yea. Same mental state applies. Its a tool, like Remaster, that is available to see if it helps. If things afterwards seem too altered to work with, then we abandon that tool for THIS one application and find another route. Trying it again another time.

[valid flax]

How can I get rid of hard s sounds in a generation? Just cover it with "deesse" or "de-essed" in the style? Will that help? Do I need to re-run the originL prompt and deesse the best word for this in styles or is there something better

[lucid bone]

Hey everyone! 👋 I run a YouTube channel creating 2-hour uninterrupted dark ambient / cyberpunk lo-fi mixes for coding and deep focus. I mainly generate hypnagogic, completely percussion-less drones.

I'm trying to get really rich, multi-layered cinematic textures using the V4 model (to keep that tight 4-minute structure), but sometimes the output feels a bit too "simple" or flat. If I use V4.5, it stretches to 8 minutes and sometimes loses the dense atmosphere.

How are you guys prompting for deep, complex atmospheric drones without any drums/percussion? Any specific tags, prompt structures, or workflows you'd recommend to get professional, high-quality ambient layers? Would love to hear your ideas! 🎧

[final rose]

I think this is the best channel... I have a simple goal.

How do I take this sample and make it a lofi song where it's repeated/cut up/whatever creative stuff the AI does? I'm getting this...

Uploaded audio is too short (currently 2.5 seconds). Minimum duration is 6 seconds.

Should I really add 4 seconds of silence? This doesn't seem like the best way, but I also don't think the sample option I see in Suno is what I need, either.

Ideally I'll then be able to do this with multiple samples in the future to make for example, meme songs.

[frozen dagger]

i think you best bet is too duplicate what you have to meet the minimum criteria. You are feeding it context. (Especially with such a short sample) It will stitch the audio within and iterate from, so duplicating feels like a better option than stitching silence to meet the minimum.

[cyan jewel]

v5.5 outputs tend to be bright and harsh be default. there are things you can to influence a chance of a track NOT being as harsh but it is truly genunely a mix of genre and rng. bass-heavy tracks without a lot going on and without a lot of high frequencies are less likely to result in a harsh overly bright, treble-spiky track with terrible compression, sloppy transients, and model drift/attention loss.

if you have an output you want to save: some options:

1. download it, bring it into a daw of your choice, try to work with the existing high end and tame it with things like a more aggressive de-esser, a lowpass filter cut to around 12~16khz, and a narrow Q cut around 9khz. These HELP but it WILL result in a less "exciting" track as you're actively fighting against the exciter in suno's fx stack

2. you can consider remastering/covering it with tags that infuence a darker/warmer mix.
3. you can stem/de-fx the tracks but that's very crapshoot
4. pretend you don't hear it like 95% of sunoers

[digital osprey]

Yo 👀

Quick question about Suno:

I’m trying to make an intro that starts with voice only, then instruments come in progressively (like layer by layer, not all at once).

What meta tags or structure do you usually use to make that work cleanly?

Ideally something like:

• voice only at the beginning
• then gradual build (kick → hats → bass → etc)

If you’ve got tips or examples that actually work, I’m interested 🔥

[lament nymph]

I’ve created a song using v5.5 and want to extend it but not by 4 minutes. Is there a way to shorten the extension length? I’m new to this latest Suno but I know there was a way to do that in earlier versions. At this point I’m wasting credits trying to do this. Thanks

[cyan jewel]

extend section by section is one way to do it. there's also going into the editor and stipulating on extend length

currently v5.5 extend's like, barely working. not worth wasitng your creds rn.

but thats generally how you can influence a shorteer extend length

[lament nymph]

Ah ok thank you. So I could go to an earlier version and use that instead, even though it’s a v5.5 song ?

[cyan jewel]

i've done extending an older model to a newer model but i havent personally verified extending from a newer model to an older.

[lament nymph]

It’s good to know about the 5.5 limits atm. Thank you

[minor moth]

where do i get my vocals it doesnt show my any option for this im a pro plan

[frozen dagger]

From Create tab look for +Voice and then it should pop up a screen that has Create Voice. Can access from Library > Voices > Create Voice too.
(The 2nd way shows the voices in you've made in case that is what you are looking for as well.)

** Using your own Voice is only available for accounts that are 18+ yrs of age.

[minor moth]

how do ik the date

[frozen dagger]

the date? as in your birthday? I am not 100% on how Suno checks, but I would imagine it looks at the info submitted when creating the account, associated with Google account or phone number. 🤷‍♂️

[keen sundial]

Does anyone know why I can't upload any audio right now?

I mean any sounds~

[polar spade]

Personal experience: You'd have to start with what you're literal intent is and decide if its do-able.

So layering, not soloing.. such as voice, then voice and flute, then voice flute and guitar simultaneously playing... then drums... and so on. It would be easier in a DAW than in Pro.

You could get a somewhat feel of it building by extending every decent break point of 32 beats and "add" another instrument to a new extend. Eventually, it starts to use the previous bits as stronger weighted vibe and the next wanted instruments may not even show up.

So, ultimately, we make some separate covers of just the instruments wanted. Load them all into Studio and edit as needed.

You can't really go in a forward linear direction in Pro with extensions to get a 6 minute song of a layering instrument song. If you want it down to three minutes, it might be possible to get that feel of solo instruments coming in, but again, the continual layering every 30 seconds or so, I dont think can work easily without a means of splicing or sequencing.

Suno even made a video using recording your voice for all the instruments and then putting it all together. So its not so difficult with Studio.

[near siren]

All my generations always start with the same vocal chop and it’s really annoying 😭😭 anyone know a good way to prevent it? Nothing in my prompts asks for this.

[brisk dagger]

Hi everyone. I’m having an issue with the Extend feature when trying to generate a short instrumental outro.
Previously, using the [Outro] tag in the lyrics would generate a short closing section without any issues. Now, every time I try to extend from the end of a finished song to add a brief instrumental fade out, Suno generates a full 3-4 minute track instead.
I also tried using an [End] tag but it wasn’t recognized.
Is anyone else experiencing this? Any workaround to get Suno to generate very short instrumental sections?

[digital osprey]

Yeah so basically I do my own thing, extract the stems in Studio, cut the instruments during the intro, and add them back manually.

It’s just that right now I made some electro swing, and I can’t get the track to not start full-on with trumpet, piano, and everything right away. I’d like to have just a calm voice at the beginning, and then the sound comes in gradually.

Electro swing always feels like it’s at 200% energy from second 0.1.

[severe narwhal]

yep the same here, and haven't figured out unfortunately 😕

[high snow]

Go to #🪲┃bugs-web and report it there.

[brisk dagger]

Thanks! I did that, but someone told me that I have to leave the comment right here 😅

[high snow]

Huh. Well, I’d wait until they fix it.

[azure jungle]

SUNO did not respose to my issue "I was paid member of Pro and they just took away my membership" Mailed them so many times but they do not reply. this is really bad on their part. They cheated me. realy bad of you SUNO

my ID is kartikey_boddh

please help. I am not able to use the paid membership.

[quiet solar]

@azure jungle Hey mate. This isn’t a tech support forum, and we’re fellow users rather than Suno staff.

Your best bet is emailing [e-mail] with a description of the issue.

[azure jungle]

i have mailed all the the mail "Suno [e-mail],
[e-mail],
[e-mail]" but i did not receive a single mail.

[quiet solar]

Suno used to have this behaviour regularly in older versions. A good workaround was to expect that Suno by default would want to create at least a minimum length of audio and put your extension point 15-20s further back into the song to accommodate that.

[quiet solar]

If you’ve emailed them, your issue will be logged in the queue and they’ll contact you when they get to it.

Nothing we can do here to speed that process up I’m afraid.

[azure jungle]

thank you for your advice.

[rough depot]

god you would think it would sing faster when a line is longer but nooooo

[rough depot]

Melisma = Instant trash.

The broad way version sounds nice https://suno.com/s/cVzrYmTkTGiCQpUt

[high snow]

Are you trying to speed up the whole song or just one part?

[rough depot]

I want consistency.

none of this random mollasses crap

[high snow]

Well, if you know how to speed up the audio through there, you can do that. Just make sure the sliders onto keep the pitch the same so you can get chipmunk voices. Or you can do that through audacity

[rough depot]

sliders? wot

[high snow]

If you click the three dots on the side of the song on the right hand side, and this part where it says speed, and if you click on that you can change how fast or how slow you want the song to go, but the thing is it changes the entire song, and then very underneath that the slider of how fast or slow you want it to be there’s a slider that says keep pitch, make sure that’s turned on or highlighted instead of grade out so when you change the pitch to the song, you don’t get chipmunk voices

[rough depot]

what?

not even remotely what i meant

I just want it to stop taking 14 syllable lines and stretching each syllable out to like 8 bars

[polar spade]

We previously went over a few ideas of speeding the pronunciation through lyrics, from writing lines without punctuation, rewriting phonics so word sounds are combined , and using dashes. We discussed how line spacing and words per line can make changes in how suno can squeeze words together whe singing. We also talked about, depending on where rhyming words are found, whether Suno will pause in certain spots.

You tried a few and things changed for the better in some instances.

Id say, look at what you got and see where you can try these changes again until something clicks.

[high snow]

Oh. I wish you said that earlier, my bad.

[fierce ibex]

hello I signed up for the Pro plan. All the songs I make right now I’m working on instrumentals...

Can I use these songs to monetize my YouTube channels?

I mean, is there any way this could affect my copyright monetization? I imagine this is a question some people have, and it might have already been answered, but I’m new here.

[half dragon]

It's quite simple, all the songs you created under Pro subscription will allow you to use AND monetize.
Or are you asking about the songs you create as a free user before you subscribed? If so, no those songs owned by Suno by default. You can still use them, but can't monetize.

[fierce ibex]

Perfect, I have a Pro plan and I create instrumentals for my videos.

[frozen dagger]

the prompt is for Opera styles, which is commonly associated with the powerful vocals and arias.
try prompting for "recitative opera"?
i'm not a fan of negative prompts, but could add arias to that as well.
i think that is the struggle you've been dealing with tho. trying to get that power ballad, operatic sound but the prompting is associating with the arias, melisma, and sustained carry on the vocals.
i hope you find a way for it to stick to your leitmotif. Lots of good suggestions given.

[keen crescent]

👏🏻 Good Morning Fammily - Can anyone Assit and tell me why my Style Prompt triggers a LIVE Audience with Cheering? Would be greatly Appreciated 🎯

Style Prompt:

[Studio Recording] Country Pop, 125 BPM, upbeat bouncy groove, dual acoustic guitars (one tight strumming driver, one off-beat accents), dual bass layers (deep root bass + melodic moving bass), full organic acoustic drums (deep thumping kick, tight snare crack, complex rhythmic hi-hats, strong cymbal presence, dynamic fills, forward and full), dual electric guitars (prominent twangy lead hooks + rhythmic accent guitar), short fast violin chops only in chorus peaks, short low harmonica bursts (rhythmic, not melodic), bright uplifting tone, catchy melodic phrasing, wide stereo image, dynamic arrangement, clean dry studio mix.

[pastel plaza]

I get 7:59 minutes songs.
when the song finsh It just start singing the song all over again.
How do i fix that ?

[slender fable]

My song cuts off abruptly in the middle of a tune, but when I try to extend it, it makes a really long extension, but I just want a brief ending. How do I get it to just conclude the song? The song I made is already really long anyway and doesn’t need to be much longer.

[valid flax]

Thanks, I ran my vocal sample through a de-esser before uploading it and that helped

[rough depot]

On the symphonic metal version?

[rough depot]

It just seems ass-backwards - it sings SHORTER lines FASTER instead of the other way around, and randomly shifts its tempo even when the syllable count is exactly the same.

[rough depot]

usually i find mashup works best for the leitmotif part.

Hmm what i am going for is something like "March of the Swordmaster".

Where it's this kind of epic storytelling followed by a melodic chorus

Or "Mystic Prophecy of the Demonknight" in parts

[slender fable]

Wait, hang on, there was a glitch I didn’t notice at first where the song I had made (which is actually an extension of an uploaded audio) didn’t actually get made into a full song and was still just an extension. I couldn’t find it in my library and now I know why. So when I tried to make an extension of that extension I mistakenly thought was a full song, it just made another extension of the source audio instead. However, the extension I thought was a full song wasn’t listed under the extensions to the original source audio, and it only showed up when I searched for it in my library with the search bar feature. It wasn’t even giving me the option to make it into a full song on the app. But God, I finally was able to make it into a full song on the website version.

It was confusing AF.

[rough depot]

I have also noticed that glitch. No idea the cause I'm afraid

[slender fable]

I guess I forgot it was an extension instead of a cover and that’s why I got confused when it wasn’t listed in the library? But again, I created it from the prompt page by selecting the uploaded audio source and selecting the extend option, so I don’t know…

Maybe the way I extended it from this screen is why it didn’t show up on the source audio’s list of extensions.

Better screenshot example.

[rough depot]

Whats this called: https://suno.com/s/y22b93BsrkNt02in?time=256
Where instead of JUST singing the line, it has that choir in the background sing the first word before the singer sings it thereby making it repetitive?

Like in technical terms

IS it "chorals"? I'm getting a lot of it in this one, (i LOVE the instrumentation here) https://suno.com/s/lsp81O0ZW5eMP60o

It also NAILS the "Red Lightning, STRIKE!" part

gah! but then it messes up one of the verses.

[rough depot]

brute force it is

[quiet quartz]

😶‍🌫️What scares you more: failure or regret?

[rough depot]

I just want to prevent a random bunch of guys from joining in on singing on the final line of the chorus. Or if they;re gonna join the main singer, they should just do it for all four lines. But i cant make any directives if I dont know what the heck technical music thing I'm hearing.https://suno.com/s/ZajJqf9jnGGNkRub?time=135 "Mal Casts Illusions" getting a whole-ass boy band, see

[polar spade]

That may be considered " Call and Response"

Its helpful if you can mark the point in a 5 minute song where you're indicating this occurs. ( at the 4:00 mark in the final chorus).

[modest tusk]

hello, so is there a way for me to make my Personas... i mean Voices sound the way they used to even with the v5.5? because now my Personas dont even sound like artists anymore and all of her songs are just basically the same repeat of her first one

[rough depot]

i copied link at current time didnt I?

[wet panther]

Using PRO, does anyone know how to redoo a song keeping 100% of arrangments?

[polar spade]

Since its new, there some experimenting, but try out making a Custom Model.

You need uploaded songs for some reason, so download your artists songs to a single folder to keep organized. Then bulk upload all the songs back in the Custom Mode ummmm.... module where indicated. You can go as far as separating out music and only having vocal only files.

Then using that Model as your go to source for emulating your artists vocals.

[polar spade]

Absolutely you did. I didnt look closely at the link to figure out the 256 second mark. My fault for not noticing that.

The link itself didnt go directly there and started the song at the beginning.

Ill report that as a glitch in the app. (Im using the app at the moment. )

[modest tusk]

so i gotta really download 70... well i have all day anyways, thank you much, ill try it see if that works

[polar spade]

Minimum is 6 before it works. A good number is 25-30. And its a good idea to have hard copies away from Suno anyway, so it will be a good thing, in the end , to do , when you have time. But for now try the 24. 🙂

[modest tusk]

will do

thank you again

[rough depot]

This is close i think https://suno.com/s/3MF3BFtnu3zwu39M

Gah

[mellow hill]

What's the trick to keeping a Persona voice consistent? I saved a voice Persona that I really like but when I make new songs the voice varies too much from the original. I use the same exact Style promtpt as the original song for the Persona but it still changes too much with new songs. I've tried a lot of different settings but can't figure out what the issue is.

[rough depot]

It absolutely botched the outro

[quiet solar]

There's no guaranteed way to do it. Covering the song with the same style prompt, low weirdness (15% v5 or before, 30% v5.5) and high audio influence (70-80% v5 or before, 50-60% v5.5) can get you close, but presumably there's something you're looking to change in making this repeat version of the song, so you might need to make adjustments on the sliders depending on how much of a change you're trying to apply to the original song.

[paper dragon]

hi guys im a pro user, generated a song, opened in editor and replaced a section with new lyrics, then made a cover from the edited version and now the new covered version has both the edit and the old lyrics in it, how do i fix this? it makes the cover version longer than than the edited version and screws the general flow in that section. how do i fix this? using 5.5v

i wish the suno crew would make it more userfriendly

[quasi jolt]

i just remix my song after i edited it to make it flow better

it wont be perfect 100 % but will make edited sections flow into older sections better

using a cover it tries to regenerate a song as similare to the last but can change the lenght pitch and so on

[paper dragon]

i see thank you i will try that

i love the current beat

will the remix significantly change it?

[paper dragon]

ACTUALLY BEST advice

[burnt musk]

guys does anyone know to make suno sound depressing as all hell?

Like, "i just ruined the love of my life and i hate myself" kinda thing? like REALLY depressing?
i tried playing around with styles and they all sound too "studio" / "high production"
i have this specific "singing whiling bleeding out in an alleyway" vibe i'm going for and i cant seem to get suno to make something like that
any suggestions?

[rough depot]

Trynna do more cover experiments. Like keep vocals, swap instruments.
Or reversed

[frozen dagger]

I have a pretty good imagination, but still, I can't say that whatever I imagine that sound to be is likely to match your idea... and Suno has no imagination. (literally, not figuratively) Finding a reference point that Suno would recognize and building from that would probably be the key. Think of something even remotely close or similar, same ballpark at least to start with, and then construct from there asking what is it missing or what needs to change to be closer to your idea.
That would be my prompting suggestion. But we can use our own Voice now, and that is probably what I would do to get (at least that ballpark area) what I wanted it to sound like. (In my mind basically sob-singing; tm pending if not already; dibs)

[burnt musk]

wait we can use our voice now?
is that a paid feature per chance?

[frozen dagger]

Pro+ plan, and 18+yrs of age, but yes. a thing.

[burnt musk]

oh dang.
the thing is that i pretty much have the entire song planned out tone, voice and theme wise
if i could just convert my voice to a proper singing voice using AI magic i'd rather do that over hoping the AI would maybe perchance somewhat understand my entire concept
but alas

[frozen dagger]

Not being able to read your mind I can't know 100%, but I think using your own Voice in Suno may get you what you are after, if not then maybe another Voice AI application out there can do the vocal and can stitch the vocal work and instrumental work together in whichever way worked best for you.

[valid flax]

anyone else living on the edge?

even with max characters my gens are around 5:30

either than the bug it must be hard to make it produce an 8 minute track

[burnt musk]

This is basically the idea
But yeah I get what you mean

Cant really get pro rn so I'd have to look elsewhere
But I'm sure there are ai tools that can do concept to final song conversions at this point right?

[valid flax]

Do you guys think that the new voice clone really changes the voice that much? Have you experimented with generating the same song/style with the voice turned on and off? With max infl? I feel like either my voice is generic and sounds like the suno voice or it it doesn't really impact it that much

maybe Im crazy lol

anyone with a really unique voice try?

[frozen dagger]

i had a 10k character i squished into 4999/5000 and was still making edits. no room for any tags in lyrics. not doing extends. it has all my lyrics in the song and it's around 5.5mins.
mine is spitting all the words out, but have no room for putting in tags to create music in between the stanzas/verses to make the song a little longer than 5.5 and stretch the vocal delivery out a bit.

[frozen dagger]

i heard some really unique voices in https://discord.com/channels/1069381916492562582/1489701055435378929

[valid flax]

maybe suno trained their musical voice on my voice XD

[frozen dagger]

i listened, and you may be able to prompt some of that, "staccato phrasing, voice cracking, melisma" and with mixing different singing style prompts maybe associating with the genre of music you are aiming for. Right mix of prompting and audio influence can get you there, or at least close to there. Would probably need experimentation (as usual). Suno (imo) pretty good at taking a rough idea and iterating on it. It will sometimes give every variation BUT the one we want it feels like sometimes, but eventually it finds the needle in the haystack of infinite possibilities... usually.... sometimes.... maybe... experiences vary.

[frozen dagger]

if it didn't before it will now probably. 🙂

[valid flax]

I assume when I pick "professional" while uploading my voice they get flagged to experiment with it, they're welcome 2

I am going to try something cool 🙂 I am going to make different covers of the same song a man and a woman, and maybe two voices of each. Then I am going to stem them and use them to add an ensemble singing in the background for certain parts

[frozen dagger]

I haven't seen any info on that, was curious myself, we can only assume.

[valid flax]

I read the terms and conditions, by uploading your voice you give them permission to use it to train the public model

[frozen dagger]

That's a great way to do it. Show others how ya do it and share it in the https://discord.com/channels/1069381916492562582/1252267444773326848 with the others. 😇

[valid flax]

It's my first time experimenting with it so Im not sure if it will work out

but I will take screenshots for a tutorial

[frozen dagger]

https://discord.com/channels/1069381916492562582/1069381916492562585 or https://discord.com/channels/1069381916492562582/1083129341870878780 better place to share

[quasi jolt]

oop

[polar spade]

Its like putting ice cubes in an 8oz drink. Subtracting the use of the ice cubes , you really only have 5.5 ounces of drink.

The more lyrics and prompts , the more tokens get used to make it all work. To compromise, the songs might be shorter.

[burnt musk]

Yep, it do be feeling like finding a needle in a haystack..

Thanks for the help man, cheers!

[rain hollow]

Chat does anyone know how to make suno regenerate my acoustic rhythm guitar layer? The rest of the song is done I just want a clean recording instead of my rough one

[frozen dagger]

Cover your acoustic audio sample with high audio influence slider and a prompt that details what you are looking for.
example acoustic prompt techniques:
An unplugged studio solo session built around a single acoustic guitar, captured with intimate close-mic realism as if performed live in the room. Virtuosic percussive fingerstyle drives the entire performance, integrating string slaps, finger taps, soundboard body percussion, palm muting, dead notes, and rhythmic accents directly into the groove. Expressive control unfolds through hammer-ons, pull-offs, legato runs, trills, slides, glissandi, microtonal bends, nuanced vibrato, and fluid dynamic articulation. Hybrid picking and finger independence enable rapid arpeggios, double-stops, chord-melody voicings, counterpoint, and campanella ringing strings. Natural, tapped, and harp harmonics shimmer between phrases, with organic two-hand tapping passages emerging naturally. The performance breathes with rubato and aleatoric phrasing, warm wooden resonance, crisp transient attack, cinematic flow, and virtuosic freestyle energy, fusing speed metal, djent, deathcore, black metal, math metal

[drowsy quest]

Hello! New here, having a lot of fun making music. I am having difficulty guiding the vocals when I want more than one in a song. For example, I want a specific set of lyrics to have one voice, and another set to have another, alternating between them per my direction using bracketed tags. It's not working quite so well! any tips?

[valid flax]

Its best to make the song as one voice first, get the version you like, cover it with another voice and/or gender, then combine the parts from each you want to sing. On top of this, you can stem one vocal layer to play on top of the first track for harmonizing. Thats what I'm doing now

[drowsy quest]

Oh that makes sense. Like in Audacity, cut up the parts you like in each and then put them together? I haven't tried that yet, I've just been trying to let Suno do all the work!

[quiet solar]

If you're open to using an external editor like Audacity to fix where Suno doesn't allocate the voices correctly, you've got a number of options for getting that kind of control over the singers (which isn't really possible just through prompting). You might find this guide helpful for laying out some potential routes: https://discord.com/channels/1069381916492562582/1466598465969590465

[high snow]

Is it possible to have the AI sing a solo voices or is it gonna try to add multiple voices?

[drowsy quest]

Oh great there's a guide! Thank you for linking that

[unreal sleet]

Im still getting songs that repeat the lyrics, are the devs aware of the glitch?

[high snow]

#🪲┃bugs-web

[quiet solar]

If you mean just performing a song with a single solo voice and no instruments or other voices, that can be a little tricky - Suno leans towards making songs rather than solos. One way you can try to move towards it is expect Suno will do this, so prompt for very light instruments as well as the singer, then extract the vocal stem.

If you have Studio, you can also cover or regenerate the vocal stem in isolation and then export it on it's own - similar effect, but may help avoid any issues with loss of audio quality from the standard stem splitting process.

[high snow]

Well, I mean, I’m still gonna have instruments and all that I just wanted to know if I could just do a solo voice.

[unreal sleet]

people are still reporting it today, yes?

[high snow]

Yeah. I had that problem also but mine got fixed.

[unreal sleet]

By itself or did you have to do something?

[high snow]

It was by itself.

[unreal sleet]

Ok good for you at least 😛

[high snow]

I heard that supposedly they reply faster if you flag the bug report on the website itself. I don’t know if you did that as well, but I did both ways. I both flagged it in here and through the website and it was fixed in like a day or so.

[high snow]

And rip for u 💀.

[unreal sleet]

Yup i flagged it on website few days back

but they dont really communicate back or give any updates when stuff like this happens

[brisk nova]

All my songs are in the Workspace>My Workspace, after I publish them I move them to a play list by month. Over 500 creations and NOW I'm thinking I need to reorganize or start a new system. What's yours?

[frozen dagger]

Over 20k gens at least, start a new workspace when i have a new idea, never delete anything, and get lost in the chaos.

[quick wedge]

<@&1082792303976001547>

[paper dragon]

hi guys, a pro user here. is there a way for me to get rid of a short less than a second weird moan in the editor?
i try to replace the moan but it just replaces it with a slightly different moan.. i just want it gone completely and have normal melodic flow instead there

[languid junco]

Let them moan 😛

[paper dragon]

not even a hot moan, something autistic or down syndrom weird stuff

[languid junco]

Probably be easier in studio.

[frozen dagger]

pro user can stem split. could try to split stem and remove the sound (if it get it on separate stem) but still need a DAW to put back together. editor alone wont do that.
Could try to do a remaster sometimes that can fix stuff, but it changes other things as well. =\

[languid junco]

Can be loaded into studio and put back together

[frozen dagger]

alternatively? trasform the weird sound into a "french horn flourish" lol

[frozen dagger]

yes, but Studio is premier. They said pro user. Trying to solve within the limitations.

[languid junco]

Oh. Forgot

[frozen dagger]

@paper dragon Stem split and Studio (or another DAW) is probably the go to move for these case tho.

[paper dragon]

sometimes i listen to different creations from people published in the home page and it makes me think im doing something entirely wrong with my stuff

[frozen dagger]

"What can not be destroyed, must be contained." If can't remove it, try to cover it up, or transform it.

[paper dragon]

i figured as much, was hoping it's a last resort for me to split it and rearrange all stems individually

[languid junco]

No wrong way

[paper dragon]

what do you mean by cover it up or transform? you mean strictly using a DAW or a studio version?

[gilded falcon]

can't you make a new voice out of a track that uses one? I'm trying to create an original voice and the one that this generated was from back when personas were still a thing, because of all the prompting and covering, the voice sounds different and I really like it and want to use that as the base. How do I do it?

[frozen dagger]

Trying to generate something new to replace that section, either replace in editor, or extend before the mistake, or Cover the entire song with high audio and hope for better result (cover a remaster maybe).
Hiding it would need a DAW I guess. Not sure what I was thinking there. Just thinking in general layering overtop, EQing the sound out (or not recognizable).

[languid junco]

Prompting can be funny.
New models seem to work with tags differently.
Having a tag like [No Sound] ot [No Voice] can work, but is taken by the AI and kind of thinks it has to put in a sound there.

[frozen dagger]

Open the song in editor and Save as New Song. Then you can create a Voice from it again.
Can't create a Voice from a track that has the "Voice" icon tag when we generated it using Voice feature.
Saving as new song removes icon, keeps exact song same.

[languid junco]

You could clip out the noise. Then cover the song, maybe it'll follow through without the moans.

[paper dragon]

i will try that also

am I the only one who feels there should be more instruments in the editor? like it feels dull and non user friendly

i understand the monetary gain approach by implementing studio, not asking for that kind of level but maybe if they put a little more love into the editor system

[gilded falcon]

okay I'm almost there, now I need to make her voice pitch lower is that possible in the editor?

[frozen dagger]

.... not that i can think of. Studio has transpose.... you may need to right click the song and look for Adjust speed.... slow the whole thing down. leave "Keep Pitch" checked off probably...

[gilded falcon]

yeah What I ended up slowing down by 0.94x with pitch then 1.06x keeping pitch.

[frozen dagger]

outside the box thinking. that's what i like to see.

[gilded falcon]

hmmm well it was a nice try, the voice is there but now I can't save the voice. Does suno not let you save voices that have changed pitches?

[frozen dagger]

can definitely make voices out of down-pitched vocals. done with Suno genned voices and my own.

[gilded falcon]

fixed it. put it in editor and created it again. Now the option is back!

[frozen dagger]

same trick worked twice! i was gonna say it. lol

[full hound]

Is there a trick for switching between spoken word and singing? I start with spoken word for the intro, verse 1, 2, and 3 are all expected to be sung - Break/Pre-verse has spoken, and then ends with singing Chorus and a final spoken word

I cannot for the life of me get the first verse to be sung - it is always spoken word while verse 2 and 3 are sung

[gilded falcon]

@full hound probs the fastest way is to record it yourself and then cover yourself.

You could also extend it rather than generate in 1 big bulk

[frozen dagger]

@full hound extending from the point or regenerating in editor are options.
Seen some say using a ~ tilde on a line break can to suggest a change in structure. (along with the prompting in brackets.)
[verse 1: spoken word] Spoken lyrics here
~
[verse 2: singing stlye] Singing lyrics here

as one example of how might write it.
Sometimes just takes a couple tries. Sometimes just a line break space between can help signify the change. It's being guided, not commanded. May take additional hand-holding with structure in the way we write it.

[full hound]

I'm giving the tilde a try

[full hound]

I abandoned the tags altogether and rolled the dice with just parenthesis

[frozen dagger]

and?... did we find a measure of success? i hope.

[full hound]

It was much better - I added the finishing lines and am pretty happy with the results

[sweet siren]

Anyone else having issues with a song starting over again? I tried minimal tags already, also a prompt that has hardly directions in it because I though I might have given too many instructions for Suno to handle. Seems last 2 days it is worse. Is there anything I can do?

[unreal sleet]

its been buggy like this for a few days, wish the team could atleast ackonowledge they are aware so we dont waste credits over and over

its been reported

[sweet siren]

Thank you for your reply.

[inner moth]

Is there something wrong with V5.5? I keep trying to extend a song and it keeps giving me 4:00 minute extends that replay the first verse… is this a known bug and if so, is there any way to get around it?

[sour dune]

Hey when you use Suno Studio to replace a words. and it generate 2 new tracks. when you listen to them, the flow of the lyrics is good. but you can hear the differenc of the production before and after. Is there any meaning to do that? Or i have to generate 100 tracks and hope to have a good result. Is there any kind of re-mixing when we replace?

or i have to keep the difference an remaster everything?

[full hound]

I noticed that if you put in stuff that "confuses" the model, such as bad tags or too much for a line (like spoken word), it will create something over a breakpoint (probably around the 4:38 range) it will demand more latent space and then hallucinate.

[stark osprey]

Anyone else also having issues with EQ changes during the coarse of a song? Song starts of correctly and the further you get to the end eq has changed (more highs, less mids, more lows). Same thing also goes for increased distortion

[stark osprey]

Same thing here.. totally different mixes every time. Very frustrating. Havent found a solution yet

[half dragon]

Like everything else... you can keep generating until you find what you like or whatever works with the song
If the difference is minor and you really like it -- export to new song with the new part then Remaster see if that improves or smoothes out the differences.
If not, back to Studio or accept it.

[sour dune]

😄

[stark osprey]

Burning credits... guess thats the business model 🤣

[jagged lantern]

When creating tracks, try setting the Weirdness to 0%. Regarding EQ issues, it would be ideal if they could be resolved within suno studio, but if not, I recommend using a different editor to improve the EQ. If you search for audio editors on chrome, various options appear. there is AudioCity, which is free but performs well enough to be used by actual producers, and Adobe, which is also widely used but is paid. There are many other editors available as well. The choice is you.

If you encounter anything you don't know while doing suno, you can search for it on chrome, and chrome will find a lot of information and provide you with useful information.

I hope the problem is resolved successfully.

[full hound]

Reaper is nice and cheap - I recommend it. Very good VST support and made by the people that made WinAmp

[hasty cove]

can someone help me, im just trying to extend my song, to just add an outro. I choose "extend" choose up to the end of the song, under lyrics i ONLY have [OUTRO] thats it. It remixes the entire song into a remix version of the song i just want to extend. how do i avoid this?? i do not want a remix, only want to add an outro to an already done song. why can't there be an option of "add section" or something. if someone can please help me to do this simple task i would appreciate it!

[frozen dagger]

Extending from the very end of the song is not recommended usually. That is part of it. Other is that extend (and normal gens) are bugging again as well. (Seen in users' posts above)
I might try to open the song in editor, add a blank section at the end and try to create something that way.
Otherwise you're probably gonna wanna try moving your extend point back a little bit (which will overwrite of course.)

[hasty cove]

thanks ill try 4 bars earlier, ill give it a shot. but why is it remixing it into an entirely different song?

[frozen dagger]

Likely because it is creating something new and adding it to what was there (by looking before and after the point of extension). It's not really picking up where it left off. Like a new seed that is trying to get close to the original. (But also looks at the current prompt; any changes to prompt can change it obviously. Same prompt still can sound different.) Usually a Cover or more work is needed to blend them.

[hasty cove]

ok thanks. No idea why there isnt a simple "add section" option with generation. and can choose verses/chorus/bridge etc , would solve this problem entirely. you should be able to add sections to songs you are working on easily.

[frozen dagger]

Sometimes I find it better to Duplicate a section of the song already there and then Replace that with a new generation.
This way I know what audio it's looking (clip not whole song) and I'm able to set the duration by how long i make the duplicate go. (Just creating audio space for context, its going to be replaced.) This can make it sound much closer to our original source. Add some some duplicate noise at the end, then transform it to be an ending.
GL

[hasty cove]

gracias, thank you

[frozen dagger]

de nada

[stark osprey]

Weirdness to 0 has no influence at all. Using an editor is not a solution because the complete sound spectrum changes during a song. It can't be automated either. The problem lays within the generation itself. I own Wavelab, Cubase Pro and Spectralayers but a bad generation can't be fixed afterward with any tool without compromising quality

[quiet solar]

Suno does sometimes go more into the highs as the song progresses. And yes, it can be really irritating and not straightforward to fix. I wouldn't expect reducing weirdness to fix it either - I strongly suspect it's not an audio factor Suno is deliberately introducing to shake things up from genre norms, it's more of of an unintended glitch in how some songs are generated.

The only approach I know of to fixing it is to accept you have a deeply flawed starting point and try to un-flaw it step by step, using mid-to-high audio influence covers (and likely covers of covers) to rebuild the song. If there are things you can do in external tools to shift the EQ balance without driving yourself up the wall, that'll help - you can also regenerate any stems that are particularly badly affected in Studio and see how that goes, if that's an option for you.

Ultimately, it doesn't matter if any individual step you take tanks quality along the way, as you can mess about with that in what you prompt for in subsequent covers. The goal is to eliminate the EQ issue, or at least rein it in to the point you can deal with it / compromise over it, then fix up the song from there. Main downside is that it takes a bit of time, and it's unlikely the song will sound exactly as it did in your original generation due to inherent variations from using Cover.

[full hound]

There's a guy whose channel is ThereIRuinedIt where he does real singing and then runs it through a processor with a voice trained on a singer's model.

Oh - on YouTube

[jagged lantern]

Yes. that must be frustrating. In that case, have you tried changing the style prompts and meta tags? If you haven't yet, I recommend giving it a try. I hope the problem is resolved!

[jagged lantern]

I have never seen a single song where suno perfectly sang a vocal track from beginning to end. he either changes the lyrics or sings without singing at all. so, I create many identical tracks, listen to them all, and then use the editor to combine the sections I like to compose the song. I also create tracks without vocals by editing the sections I like using the editor. If you haven't tried this method yet, I recommend giving it a try.

[paper dragon]

hi guys im a novice pro user and got a question. a song i generated has both female and male vocals in different sections, i want to turn the female section into male section while keeping the chorus female. is that an option without hurting the beat or flow of the song?

[unreal sleet]

its too random which gender sings what if you assign two, just rng

[valid flax]

bros I am editing the crap out of a suno track, extracting vocal stems from voice covers and mashing them together

[shell loom]

edit tracks:
https://www.youtube.com/watch?v=1u4YgdUXGHg&t=142s

[quiet solar]

The only way to guarantee the end outcome is editing, but you can tilt the RNG in your favour with good prompt guidance - which at the very least might reduce the amount of editing you need to do: https://discord.com/channels/1069381916492562582/1466598465969590465

[unreal sleet]

is the repeated lyrics/8 min songs bug fixed? still no communication 🤔

[wide stone]

Anyone try this prompt: Beats 4/4 and 4/1, Brazillian phonk-funk, pounding drums, melodious, spirited, vox, minor keys piano, agressive ambiental, Brazilian-funkstyle, mega bass, super song structure, reverb effects on the vocals and the piano, Brazilian Funk Carioca at 130 BPM in G minor, The track features a heavy, distorted 808 kick drum playing a syncopated syncopated beat characteristic of the genre, A sharp, metallic snare hits on the off-beats, A repetitive, high-pitched synth whistle melody loops throughout the arrangement, Male vocals are processed with heavy compression and slight distortion, delivered in a rhythmic, chanting style

[unreal sleet]

That could be written way clearer for the ai to understand mate

what result is it you want us to see?

[wide stone]

how