📚┃suno-school

[atomic plinth]

If you have premium just go into studio and literally take the drums out

[trim adder]

nope

[sullen remnant]

guess you gotta burn those credits then, I do a lot of ambient and half ot it has drums

[trim adder]

this is bs

[sullen remnant]

lol

[trim adder]

Udio did proper ambient better, but it still lacks in the covers department

[humble verge]

weird how the legacy editor has all sorts of bugs all of a sudden, right when they release Studio.

[indigo basalt]

Just wondering but is there a way to get the limited time deal for Suno premium on mobile, say on iOS, or is it only via the website to purchase the limited time deal for the Suno Studio? I’m only able to purchase it on mobile, due to personal reasons, so if possible a mobile payment would be great! Any and all help/information is greatly appreciated! 🙂

[ebon sand]

I used it earlier and it was all good, what bug did you find?

[devout sinew]

Hello, I have a quick question. When creating songs before, I remember we used to have a “Save Prompt” button that allowed us to save our own prompts. I am not seeing that option anymore. Did something change with the latest updates, or is there a new way to save prompts now? Thanks in advance for clarifying!

[sullen remnant]

i just realized theres that option and idk how to delete the prompts i accidentaly saved, anyways is in between the undo and the delete prompt options, for the prompts you already saved, theres an icon with 4 lines, 3 straights and the 4th one diagonal, at the bottom left corner of the panel

[devout sinew]

Thank you, but that just for what you saved before, I want to be able to save a new prompt. Thanks

[sullen remnant]

thats the one in between the undo and delete option

[spice brook]

if i click on simple then enter my prompt will it no longer generatr with ReMi??

[summer flare]

Hey, anyone know on the website how to get around the autoplay bug to turn it off and on? Is there another way to the playlist queue? The button that I usually use no longer seems to work on the website.. on my phone it works just fine.

[ashen relic]

Sup guys, I'm a pro member, pay monthly, but it won't allow me to use any other version apart form v3.5.
Tells me to upgrade. I go to upgrade and I'm already pro plan, and it just says error. The whole app is not working g right I don't think, I've uninstall and reinstall. Logged out and in sent emails etc

[dusty sparrow]

Ayo fam! I have a query that's got me a bit confused/frustrated.

I was considering upgrading to premier and there's a limited time offer/discount for that at $12/month for the yearly but when I get to the "cart" it shows the price based on $24/month for the yearly?

[agile jay]

Did you get that resolved? I can see 144 when i get to the "purchase" pop

[dusty sparrow]

no I haven't I'm using operaGX so I'm going to try with Chrome and see if that works, I was just thinking about how opera does some weird things, but apparently some people aren't even getting the limited time offer at all

[agile jay]

Oh, yeah i've used opera in the past and dropped it cause weird things like that happened lol i'm using chrome. i wonder if the offer for 144 only if you're already a paid member? WEird that it wouldnt' show for everyone if not

[dusty sparrow]

Yeah I'm getting the same error on chrome :/ I'm not sure what to do

I'm a pro user so I'm not sure

[agile jay]

try signing out and back in?

[dusty sparrow]

yeah same thing
"From Pro Plan $10 / month
To Premier Plan $288 / yea"

[agile jay]

that's crazy. i guess send an email to their billing support. include the screenshot you've got

[dusty sparrow]

I did, it took me a bit to find their email but that's exactly what I did, explained what this error/glitch was and attached the screenshot.
I'm really trying to cut back on subscriptions so even if I can get this sorted out I plan on cancelling 3-5 other services I use like netflix and stuff, not cause I'm hurting but I think having suno studio and more credits will motivate me to spark more of my ideas I've had filling up my notebook haha more time to focus on production again

[agile jay]

not to sound like an ai, but that's a great idea lol i try to keep only the stuff i want monthly like hulu and spotify (dont get me started on spotify). i'm also paying for bandlab but i might pause that and use suno instead. not sure yet. have you had a chance to play with studio? i dont see a way to test it before upgrading, which is what i'd like to do ideally

[dusty sparrow]

HAAAH very meta of you! but yeah that's kind of the vibe of life, we have SO many of these bank drainers that we barely utilize or that actively "zombify" us.

I haven't had a chance to try Studio but I signed up for waitlist what feels like months ago and was hoping I'd get to try to pre-release but alas, never got an email, I have watched hours of people using it, bigger streamers, seen lots of complaints, people were acting like v.0.2.0 was supposed to be a full v.1.0.0 release, then I saw videos with huge improvements in v.0.4.0 then again with v.0.8.0 and now that it's launched for premier users I assume it's at v.1.0.0 but I haven't seen any pictures or videos since the launch yesterday (also haven't looked will after I go get food)

as an Ableton user, Spotify + Suno Studio is going to my entire winter lmao

[agile jay]

lol yeah it's gonna take time for it to be reall reallly good. i think anyone who expects perfection immediately is somewhat unrealistic lol i only just started to watch a video (yes, 1 literally now lol) I've been playing with v5 though this evening. maybe tomorrow i'll look into studio more in-depth

im jumping off in a bit, have a good meal and hope you hear back from support soon

[dusty sparrow]

Bless up! enjoy the experimenting and learning! also feel free to share any good videos you find in my DMs, I love learning

[limpid badger]

How can I best prompt for a cymbal instrument track (in studio)? Simply typing Cymbal doesn't work. I tried [Cymbal crash] and [hi-hat taps] - but it produces something different.

When I prompt in an instrument track, it adds all other kinds of instruments too.

[dusty sparrow]

I'll do some research and try to get back to you

[limpid badger]

I think I managed at least getting the drum track I wanted - but that's all the drums together 🙂

[dusty sparrow]

that's a start! idk if you're in the subreddit but I've seen similar questions asked there but I never really look into much because it's not a problem I've ever thought about. do you also use the "exclude styles" prompt box?

[fading star]

Guys, how to fix the bug/error every time I generate a song

Somehow the AI will sing a word or two wrong, every single time

[manic estuary]

does anyone know if the changed how the style tags work in the lyrics? i used to be able to define if something would be sung or rapped and now it just does whatever it wants. same with other tags.

[limpid badger]

aye, that I did, and it helped some. But mostly, reducing the Audio influence to 20% helped also

[dusty sparrow]

want to share the prompt and your song so I can observe and follow along?

[dusty sparrow]

this is what I was answering

[fading star]

Can we discuss in DM?

[dusty sparrow]

yeah sure

[limpid badger]

I made a Jazz version of Edvard Grieg's Holberg Suite 🙂

[fading star]

Give me a min

[dusty sparrow]

I've never personally experimented with this outside of specifying in the styles box if it's rap or genre

[dusty sparrow]

I discovered a suggestion
Add Instrument Tags in Lyrics (Custom Mode): If you are using Suno’s Custom mode (where you can edit lyrics and structure), insert the instrument as a metatag at the start of sections. For example, you can put [Acoustic Guitar] or [Violin Solo] on its own line before a verse or chorus. This acts like a stage direction telling the AI what instrument should be playing. Some users report that placing the instrument tag at each section helped Suno stick to that instrument For added emphasis, you can even isolate that tag with random characters or spacing. One power-user trick is to put the instrument tag in brackets followed by some gibberish symbols, then start your lyrics – this “noise” can make the AI pay more attention to the tag itself

For example:
[Trumpet Solo]
@#$%^&* (some gibberish to isolate the tag)
[Verse] Lyrics start here...

[limpid badger]

Yeah I am aware of that trick. But I don't get Custom mode when "Add track" in studio

I am only able to influence the style

ah never mind I found Custom now

[dusty sparrow]

Awsome! glad you found it, I hope you figure something out

[echo prawn]

Can Suno Studio write lyrics according to the original pause structure when covering a song?

[dusty sparrow]

probably not? as a cover doesn't mean the same thing as how we humans "cover" songs. if you can upload a song without breaking copyright rules you could experiment with prompts like "keep the original structure but in ___ genre"

[spice birch]

I'm on Android is anyone else having issues trying to remaster a song and it won't allow to choose a Variations strength other than just normal?

[dusty sparrow]

the apps are always garbage just use the browser model on your phone

[spice birch]

That's what I'm doing I don't never use the apps

[dusty sparrow]

I have no idea then, I use the app to soundtest generated songs in my car so I don't use the mobile browser version, I just use suno on my computer, I'm sorry

[limpid badger]

I play with ancient songs where copyright has expired 🙂

[dusty sparrow]

yeah maybe try my suggestion but other than that I really couldn't say, try varations of my suggestion though sometimes just rewording things helps

[echo prawn]

Actually, I recommend giving Pro members a chance to try out the studio. Because I don’t know if it’s suitable for my situation.

[dusty sparrow]

why don't you know if it's suitable for your situation

[echo prawn]

Because I wanted to cover existing songs (most of which were Japanese songs whose copyrights had expired), but if I just wrote the lyrics directly, the AI ​​wouldn't know where to pause. So the generated songs were very messy.

[dusty sparrow]

interesting, maybe try downloading just the vocal track for said songs, upload them to suno and use that because it should carry the melody in the song so it will allow more freedom in your covers rather than uploading a whole song and causing the AI to be confused

[frigid pivot]

Using v5 all of my songs are right at 2 mins and under, some songs are even 46 seconds, what am I doing wrong? I never had this issue with v 4.5

[wispy laurel]

You also can use commas, hyphens, .....'s and line breaks to control the timing and where it pauses

[echo prawn]

For example, I want to adapt the Japanese song "Umi Yukaba" (Chorus of Gunkan kōshinkyoku）into Polish. However, the AI ​​will sing all the lyrics in advance, As a result, the part that should have been the lyrics in the original song became the accompaniment because the AI ​​had no words to sing.

But I wonder if this song is too difficult for current AI.

[limpid badger]

https://suno.com/s/4P4tmMs8rdwftKla

made this in studio

based on this original: https://youtu.be/NO_sC9qA73E

[cursive kiln]

I don't know how to add sections before or after an existing work in the studio

[half dragon]

Add an empty track then draw a box on it where you want it, then generate that section
Same thing with "before" section

[quiet solar]

Suno's always had an element of doing whatever it wants - the style tags are more like suggestions than instructions. However, they work the same as always. If it's a genre like Nu Metal or Rapcore that often includes switches between sung/rapped sections then style tags for that type of switch in v5 seem to be getting picked up a bit more reliably than before, at least in my experience so far.

[limpid badger]

Ok, that was too messy in the syncopations, so I made a new more funky version:

https://suno.com/song/a556d3d7-12ee-4816-9bdc-902756ffb264

[toxic cairn]

Anyone knows where the get stems are now on the mobile browser?

[limpid badger]

Would be nice if we could upload Midi too

[patent leaf]

all we need is VST support and the future will be inside us

[untold briar]

Oh yeah, especially if you training a model to convert MIDI into audio, not just rendering it normally. That feels like something Suno could achieve and it would be different than any regular music software.

[lusty juniper]

wahahahaha i clicked the detect all stems or something. i saw alot and i selected vocals then it all disappeared. i definitely will need a tutorial. you guys have any links to one?

[ashen ingot]

Hey mate, have you watched the beginner’s guide made by Suno themselves ? If not here's the official youtube link : https://youtu.be/qR4BefPvSiI

😉

[lusty juniper]

thank you thank you. was talking to a publisher and i was explaining that with Suno Studio, we can pretty save a lot of time with making the songs

awesome update ❤️

[tribal marlin]

When you generate stems directly from that predefined instrument selection, how many credits does that use? It's shown at 6:10 in that demo video above.

[cursive kiln]

and where can I add this empty track, because when I add it with a plus sign, I can't add anything in a place where there is nothing on another track at the same time, which blocks me from adding new sections

[fallen glade]

okay seems I cant use Persona on any covers anymore

[north lynx]

Suno stop-word filter is working incorrectly. I tried to use the word “necro” as a description of style, mood, and genre of music, but the system blocked it with the message “copyright violation.” This is absurd.

“Necro” is not only the name of an artist — it is a word that has been used in culture for a long time and has its own meaning. In fact, it is widely used in genre descriptions (for example, necro metal, necro sound, etc.). The same applies to “Madonna” — a word that has existed for thousands of years, originally referring to the Virgin Mary in religious art. Why does your service suddenly treat any use of such words as if they automatically infringe on a specific performer’s rights?

It turns out that ordinary words, which existed in language long before certain artists, are now being treated as “forbidden.” This severely affects the functionality of your service, restricts users, and makes creative work impossible.

I strongly urge you to review the stop-word list. Not every word that coincides with an artist’s name should be treated as copyright infringement. Context must be considered. Right now your filter creates false positives and disrupts normal usage.

Leaving such issues unaddressed damages the user experience and artificially limits creative freedom.

I look forward to your response and to seeing this problem corrected.

[regal hamlet]

hello everyone i have made a song but when i open the stems in studio they wont load, does it take long ?

[lusty juniper]

50

i just tried it again

[nocturne meadow]

I recently learned about your limited-time offer for the Premier Plan at $12/month with a yearly subscription. However, when I try to subscribe through the Suno app on my Android device, this offer does not appear as an available option.

Could you please let me know how I can access this promotion on the Suno Android App?

[pine wind]

Since yesterday, I've been experiencing the same thing with my lighter (younger-sounding) male voices. Suno either uses the normal male voice or the female voice, but never the voice of the persona.

[cyan jewel]

swift, and bratty are also stop words that I keep running into which is like, really frustrating lol.

[quick sonnet]

Any idea how to let suno sing exactly the way like in uploaded acapela song?

[spiral frost]

Hey guys, did someone got access to suno studio?

the beta

I have more than 40k songs created, i'm paying near to 150 dol a month for credit, i send my music page with near 2millions of streams, they didnt even granted me. So i'm asking myself if they even made their "beta" test session

[strong yew]

Newbie here, seeking advice from experienced users! Has anyone else encountered the same problem?

1.Since upgrading to V5, I have used the built-in vocal track extraction feature in Suno four times, and without exception, each time the separated vocals contain severe noise issues.
2.There is a serious problem with the lyrics input box: when modifying lyrics, scrolling the mouse wheel downward and then upward easily triggers the collapse button of the lyrics input box. The problem is that once accidentally collapsed, clicking to expand it again makes it impossible to edit.
3.On the web version’s dark mode, if you originally set it to dark and then after the new UI update you switch back to light mode, you cannot find the option to switch back to dark mode again.

I would appreciate it if someone could clarify. Thanks in advance! Wishing all good people a lifetime of peace!

[faint whale]

I’d love on the stems feature to not include stems with nothing on them. I often find that I am downloading onky to find they are flatlined…

[cyan holly]

It's SUS for me that the team has not released any demo video about STEMs

I'm still waiting with upgrading, to see the quality of STEMs

how structured and detailed they are

[stone pasture]

if anyone wants crazier bass/808s in their generations, only use v4.
(dont ask me how i know it, i just know it.)

[nocturne meadow]

I know how to get crazy bass and 808s in v5

[polar spade]

They (stems) are all located in the Library. You can preview and delete each one you wish, before any downloading. Check the FX stem especially as it may have what we might think was more a guitar , synth, cuica , percussion or another instrument.

[cyan holly]

like, do I get that level of detailed MIDIs, or do I get random, rather poorly structurized MIDI notes as below

because the last one I can do in Ableton by converting .wav into Harmony and Melody, but it isn't very accurate and lacks a lot of notes

and if my goal is to work on the track in Ableton , but MIDIs are rather poorly structured, then this whole 'Midi' business is a whack and marketing bubble

and since Suno hasn't shown any demo about it still, I'm so very skeptical

[tepid moth]

I was very negative yersterday accoriding to create individual stems, but today I generated a guitar track for a funk song, and my lord it delievered. Yesterday I tried a drum track for a techno song and it was a disaster. Included the bass and also the full song for some versions. Will try again today and see if its better. At least the individual stem creator for some instruments are way better than the separated stem from the full song. Way way better. And Im glad for that.

[round holly]

omg ableton what is the discussion about

[cyan holly]

?

The discussion is about that I want to know if MIDI files that Suno gives to you, are actually usable in real life, aka if I like the song generated by Suno but I wanted to make it my own, and change sounds with VSTs in Ableton, add some notes here and there, then how accurate those MIDI files are

and as Suno has not provided any information regarding the level of details their MIDI files hold, I'm skeptical about all of this

[round holly]

wait hold on what

since when did suno get midi output???

[cyan holly]

you've been living under the rock lol

[round holly]

i have yes

[cyan holly]

since yesterday

[round holly]

last update i remember is v3

[cyan holly]

Suno Studio is out, for Premier Plan

[round holly]

i HAVE to try it out

[cyan holly]

V5 dropped

[round holly]

oh come on 😔

[cyan holly]

and Suno Studio

yeah, and that's why I am asking the MIDI question, because I'm not willing to pay 30$, instead of 10$, if I get Ableton level random BS notes

[polar spade]

That's a good position to take. See what other people think about it first and how they apply them.

Though, even if it's something that this or that user has found they can work with, its never easy to show that it's perfect for any one person's needs. So even a how-to video may not be applicable for evey genre.

If you can find a friend who is willing to give you at least one stem of your track and email you a file, you'd have something to work with as a trial.

But just keep mentioning in here periodically -- and check for new personal posts in the Workflow channel for MIDI use and Im sure in no time other users will note how MIDI files are working for them -- good or bad.

[cyan holly]

Been looking for videos on YT regarding MIDI quality, but haven't found anything useful yet

[tepid moth]

There was a guy that posted a video about it on YouTube yesterday if you search suno studio midi on YT. Not a prefect demonstration as the guy dont even know how to create instruments in his DAW, LOL. But it show at least the midi files.

[round holly]

that's actually very fair

can i show something im working on here?

although it's not ai

its just me

[cyan holly]

yeah, as most users have no previous experience with any DAW. I've been scrolling reddit and a lot of people are whining how confusing the Studio interface looks and this tells you a lot about the avg. user

Suno Studio, based on what I've seen, looks like every avg. DAW

[tepid moth]

I did download my first midi files yersterday, and what I can say is that its not perfect at all, since its early stages for it. But it was better than I expected. It does not add velocity or anything like that. Just simple midi files. On the bright side it was pretty good already separating the notes. Thats my experience after the first midi i downloaded. Ill keep you updated.

[cyan holly]

okay, well it's easy to automate velocity and add it on Ableton by using knobs on my midi device

i personally only care about the detail of the notes, like is it missing any notes here and there or..

is it like 'flat'

or 'dynamic'

[round holly]

what do you guys think?

[cyan holly]

sounds nice

[round holly]

thank you

[tepid moth]

You should ask Suno to get a few days trial for the suno studio, if not they missing out of a customer. I havent tested it enough yet too say go for it or not.

[cyan holly]

yeah, well, given the fact I was one of the first who signed up to Suno Studio, but then found out yesterday I need to be Supreme, not Pro, it would be nice thing to have, but I doubt they'll do it for me lol

I'm not an influencer

uhh, I found a video finally!

https://youtu.be/p5wsY9zPRt4?t=452

gonna check that one out

[polar spade]

Yes. If we have used a DAW with adjustable panels, its visually as easy... emmm....easier. If we've never seen a DAW (except maybe Audacity), there's going to be a learning curve for new users. Finding all the features is fun while also daunting, for awhile.

Given that Studio is created to work alongside the present layouts and interfaces, it is not set to replace a DAW, simply in that it doesnt take plug-ins or presets, but for being integrated to enhance what we already have, its a game changer.

[cyan holly]

Yeah. I can see Suno working towards DAW integration tho

or, who knows... they're positioning themselves strategically and...

hoping to make an exit, being bought by Ableton/Pro Tools/Logic or whoever

doubt it tho

I mean, I'd be willing to pay 50$ per month to get access to pro level Suno that I can open in Ableton, use VSTs and let it generate perfect midi files for me, extend my project that I've started etc etc etc

yeah so Suno MIDI's are simplified Midi's without velocity, dynamic, EQ, distortion or whatever the .wav has

[polar spade]

DAW makers are now needing to catch up to integrate AI into doing what Suno already does. For instance, replacing segments by blending the sounds, instead of splicing and patching in. But at the same time, thry are hoisted by their own petards to appease the music purists to never use AI. So there's that.

[grizzled tiger]

Is anyone seeing the Max Mode in Suno?

Max Mode
Edited today
Max Mode in v5 represents a premium generation option that leverages significantly more computational resources to enhance every aspect of your musical creation. By allocating additional processing power across the entire composition, Max Mode delivers superior audio quality, more nuanced instrumental arrangements, cleaner vocal synthesis, and better overall mixing and mastering. The advanced algorithms work harder to ensure smoother transitions, more professional-sounding production values, and greater attention to the subtle details that distinguish amateur-sounding tracks from polished, commercial-grade music.

To enable Max Mode, select v5 as your model. Then head down to the Advanced Options and turn on Max Mode.

*Max Mode costs an additional 5 credits per song

https://help.suno.com/en/articles/8105345

[cyan holly]

but it's still accurate in terms of notes

[tribal marlin]

i haven't used Suno for a while, aren't normal tracks only 5 credits to make? 50 seems like a lot if you only want say a 5 sec loop. I was hoping they would make the credits scale to the length of what you generate.

I could imagine using this a lot once they have it working with the prompts. Have actually been looking forward to something like this.

[tepid moth]

Oh this a new one. Thanks for sharing

[grizzled tiger]

but I haven't seen it in my end. that's why I ran here to ask (it's faster somewhat, compared to making a post in Reddit)

[tepid moth]

I cant find it either under advanced

[grizzled tiger]

As a premium user, I suppose if I haven't seen it yet then it hasnt been available. or what else I dunno?

[tepid moth]

Hopefully adding it during the day

[tepid moth]

They removed the article

[cyan holly]

mkay, I'm now, seeing this long *** video about Suno Studio, convinced that it has potential now, even in beta

#yolo , upgrading to Supreme

velocity and effects can be added, which is actually good, because you can make the sound truly 'yours'

"I wouldn't use STEM extract in Suno for songs that are not made in SUNO. The ones you do in Suno are much better, as it goes back to original source material"

[tepid moth]

I do find the article strange though. As a premium user people should already have access to the most powerful engine, but that is not the case then if that article was correct, It has more power according to what I did read.

[grizzled tiger]

probably because I posted it here, lol

dont know whether this is teasing or their mistake of making the entry go public too soon

[tepid moth]

Yeah. However this means that they have not turned V5 on full power yet. Not even close. Mistake or teasing; the words are still there because of you😀 💯

[cyan holly]

yeah, it's always like that

V3.5 was pretty poor in the beginning, but 4 months later is was banging

the same was with 4.5 and 4.5+

and ofc 4.0

[quick sonnet]

Also have noticed the problem, mentioned in 2.

[nocturne meadow]

suno studio is good

[grizzled tiger]

barely go on Discord, but tbh folks are wat nois-er than those on Reddit

[tardy coral]

send a screen shot?

[grizzled tiger]

well they have just literally deleted it. lol

[tardy coral]

I'm down for MAXMODE

or are you just trolling?

[grizzled tiger]

now page not foun if you click on the link

ok. lemme open reddit (soo fkig lazy) to get you that screenshot

[tardy coral]

cool. hopefully theyll enable this..

[tepid moth]

Reminds me of an episode on Black Mirror on Netflix, where a couple did buy a premium subscription of something. They thought they had the best ever subscription. Little did they know that the engine was not in full power by far. Then the company added Premium +, then there was Premium ++ then there was Ultimate, Ultimate + and so on. Then they where broke.

[grizzled tiger]

well, Reddit folks are now also real excited about this, but also they are sooo pissed rn as the link was gone. so *deloited *to see those cranky folks (literally every day)

[grizzled tiger]

well then, for the record, I have just bought a full year premium plan just to get access to Studo. so you tell me, lol

[tepid moth]

What I am upset about is that this tells that V5 has a lot more power than we are given.

[tardy coral]

or maybe what we have now is Maxmode

[tepid moth]

I am premium as well 😀

[tepid moth]

Yeah maybe they just added it in general

[north adder]

What shall I do to replace the whole English lyrics of a Suno-generated song by Korean lyrics?

[tardy coral]

chatgpt

[tepid moth]

I doubt they will announce it tough, that V5 is now in Max Mode 😂 . So if so they will add it in silence.

[grizzled tiger]

and remember to prompt "keep the same syllable count in each line to maintain the singable flow"

[grizzled tiger]

this is like edging ngl

[tardy coral]

who would opt for standard mode though.. essentially what they'd be doing it making Suno 5X the price

[grizzled tiger]

despite all the pulse you may feel in those veins, you are not allowed to come =))

[tardy coral]

"let me generate low quality instead of max quality" - said no one ever

[grizzled tiger]

seem like somebody is trying to use an F bomb, lol

[tardy coral]

S bomb

[north adder]

I have already got Korean lyrics with which I want to replace the English lyrics.
And I have already created an English song and want to make it a Korean song, using the Korean lyrics.

[grizzled tiger]

all same thing dear. lemme give you my prompt I use with chat

Help me adapt this song into [target language], keeping the flow natural, emotional, and poetic. Prioritize musical compatibility, emotional fidelity, cultural resonance, and idiomatic expression.

Don’t translate word-for-word. Focus on how it feels when sung, not just what it says. Make sure it fits the rhythm and phrasing of the original melody.

I want it to feel like it was written in the target language, not just translated. Use imagery and emotion that makes sense to native speakers, even if some metaphors or lines need to change.

Format: Give me side-by-side lyrics — original on the left, translated version on the right. Keep each line aligned by rhythm/phrasing.

[Insert original lyrics here]

[lone storm]

can anybody tell me how to extract midi separated in correct instrumentation?

[north adder]

I mean keeping the original English melody unchanged, and just replace the lyrics with Korean

[wispy laurel]

Cover the song, delete all the lyrics and replace with the Korean text
Set audio influence pretty high
Style instruction should start with [Exactly match structure and arrangement of the original song]
Sacrifice a black rooster under the full moon, pray to the AI gods
And just maybe it'll give you something close to what you're trying to do

[grizzled tiger]

Sacrifice a black rooster under the full moon, pray to the AI gods
=))))))))))))

[wispy laurel]

It used to be possible to do this much more reliably
But they broke it and now it's almost impossible to keep it that close to the original any more
Realistically you might be looking to get a Korean song whose vocal stem has the same tempo and melody as the English one, and then you could use a DAW to combine the Korean stem with the instrumental from the English version

[half dragon]

Not sure what you meant, but you can definitely add stuff at anywhere on any track.

[true carbon]

What do you guys think of Studio so far? Read that V5 goes back to V4.5 quality when in studio, is that right ?

[shrewd knoll]

Hey! Is there a way the write characteristic of adlibs in the lyrics? I want the adlibs to be said with a sarcastic tone.
Heres the line:
"(Wops) I think I my location just turned of (that’s weird)"

any suggestion is appreciated

[true carbon]

Fl studio user here and really excited about SUno Studio 😉

[proper berry]

like Suno in general, it's hit and miss, and kind of primitive. but also they will improve it rapidly. the split stems sound cleaner. i was impressed by a MIDI export that it did for me.

[true carbon]

Wait it does midi export ? Can we also play with panning in that ? The guy i was looking at in the video did not play with panning at all

[tepid moth]

Where did you read that?

[true carbon]

The video i was listening to, in the comments a user said that and the guy from the video agreed

[grizzled tiger]

for sure, he is one of those reddit folks.

[grizzled tiger]

Im kind bought into what she/he said: Suno in general, it's hit and miss, and kind of primitive. but also they will improve it rapidly.

[proper berry]

Yes it can do midi export. control click on a track and it offers Get MIDI.

[lusty juniper]

okay one thing that needs to come back is the the "pt 2." and "full song"
i no longer see it. thats not good

[true carbon]

Just for that feature it could be really nice, for the drums i am more skeptikal since it will do all on the same midi : hats, snare, kick

[proper berry]

i haven't tried exporting a drum track. i wouldn't be surprised if it worked insofar as it would produce midi notes on common pitches for different drums

[true carbon]

I'll have to upgrade my sub i guess, it's not avaiable at Pro level studiio right

[proper berry]

you can select multiple tracks with shift-command-click. this is vital for moving stuff around and splitting at points in time

[proper berry]

yeah, you need premiere or whatever it's called. some guy late last night wouldn't stop complaining. it does go to $12 a month from $6 but also you get 10,000 credits

[true carbon]

Since i am in the tech prompt talk, i like to do instrumental, is there a way just like when doing lyrics to ask specific element in the instrumental? Let's say in a part i want synth, in another one something else

[north adder]

@wispy laurel So there is no way to make a local song from an English one as far as Suno goes. That's too bad.

[proper berry]

you mean like square-bracketed instructions?

[proper berry]

Are you sure? Did you try covering the song and pasting in the new lyrics?

[true carbon]

Yes

[proper berry]

i remember seeing an instrumental piece hat had extensive instructions in square backets. as with suno prompts, it might've followed half of them, but they got a lot out of it. someone shared a piece overnight, although it had vocals, they got suno to do a lot with instructions

[proper berry]

https://suno.com/song/106b0e6b-6345-42ba-91b6-0e2e0d6c5d82?sh=CUSEAvHlAha1vkgc

[north adder]

@proper berryLooks like Suno tries not to allow local language translation from an English song but only allows us to replace part of the English song. So many obstacles in Suno!

I have needed to create AI songs, using MuseScore sheet music for my choir urgently. After I subscribed to Suno, I have found that Suno only creates their own chosen songs. Pretty disappointed.

[true carbon]

Yeah studio work great, if interested i'll show you what i did wit it

[dusk dirge]

The new features in version 5 are excellent. Now, it would be very useful to add another functionality. When I create a song, then another version of the same song, and then yet another… there are often parts I really like in different versions — for example, verse 1 from one version and verse 2 from another.

At the moment, I’m forced to go into my DAW to cut and combine the different parts I like, and then return to SUNO to refine them.

Could you consider adding, in the editor that separates the parts of a song (and where stems can be made), the ability to place two songs created in SUNO one under the other? That way, I could copy a part of one song and drag & drop or paste it into the other. For example, I could take the intro and verse 2 from song 1, which I like, and insert them into song 2, which I also like — replacing its intro and verse 1.

I remain at your disposal to discuss this new feature if I haven’t expressed my needs clearly enough.

[true carbon]

Just like UDIO does, it's the only feature i like in UDIO but it's missing from SUNO, be able to add intro and stuff like that, add before or after a section

A song a did yesterday with 5.0 upgraded it this morning with Studio, i feel that it give a lot more of creativity to our project, now more able to reproduce what i have on mind https://youtu.be/-1ls2-yedZs

[white dirge]

BoomBoom’s GPT just leveled up, now she got full studio knowledge baked in. Go hit ChatGPT search, type BoomBoomRaps, and watch the magic unlock.

[proper berry]

it does kind of do its own thing but if you upload guide audio and set the audio influence high then it will often/usually hew fairly closely. i don't understand your workflow, you want to feed the sheet music to the AI? can you produce MIDI?

[proper berry]

It seems like this is a key use of Studio. In the demo video they lift a split stem from one song into another.

[true carbon]

i'll try to import one of the beat i made in Fl Studio in the Suno Studio to add some elemtns to it

[wispy laurel]

It can't translate it for you if that's what you mean
I've tried changing stuff between English and Arabic and it doesn't usually work well
Realistically the more practical thing is to create a new Korean song that sounds a lot like the English one rather than trying to do the same song

[proper berry]

another possibility might be to add the vocals on to an instrumental. even strip the vocals off the song in studio or using lalal.ai or stem splitting in logic or whatever.

[hoary marsh]

For anyone who's used Studio and can confidently answer. It looks like most of what Studio offers I already do externally in a DAW. However, the thing I'm most interested in is the ability for Suno to keep BPM and key between mixes, covers and extended sections. When I need a solo (or background instrument accents) and have Suno generate 5+ copies, it frequently changes both of theses things and most tracks are useless. Will Studio do this with accuracy?

[proper berry]

i only have a couple hours of use. i presume you've watched the video where the guy hums a part, gets suno to create it as a trumpet part and then pastes it back into his song. it's still hit and miss

one nice thing about Studio versus a DAW is just how you're still in Suno and it's real easy to perform Suno-like activities within Studio versus with the DAW you're very distinctly outside

[white dirge]

Here’s the short, useful answer hidden under a pile of workflow jargon:

Studio gives you a multitrack timeline, stem export, and proper tools like Extend and Replace Section that respect the existing song’s tempo and key. When you extend a track in Studio, it doesn’t randomly decide the song is now in B major at 126 BPM just because the AI felt cute. It aligns to the project’s established grid, so your solos, background instruments, or new sections stay musically consistent.

Known quirk: if you don’t echo the same vocal/instrument descriptors when extending, you can get timbre drift. But key and tempo stay locked unless you deliberately change them with the Adjust Speed tool.

So compared to generating 5+ standalone takes in the base app (where Suno happily re-rolls key/tempo every time), Studio is the fix. It’s closer to a DAW-lite environment: one project = one grid. If you need precise continuity across mixes, covers, or instrument overdubs, Studio is way more accurate than free-spinning in the generator.

In other words: yes, Studio does what you want. It won’t turn your A minor groove into circus jazz halfway through unless you ask it to.

**This answer was provided by BoomBoomRaps GPT, now live on ChatGPT.
Just search BoomBoomRaps and lock in.

[proper berry]

But Studio is pretty primitive compared to a DAW. So far.

[hoary marsh]

Thank you!!!

[white dirge]

Primitive compared to a DAW, yeah, but it’s not trying to compete head-on. It’s more of a DAW-lite playground for fast AI workflows, while DAWs are built for deep control.

[true carbon]

Obviously but that combined to a DAW i mean... YEAH!!!

[white dirge]

Suno Studio is basically an AI playground for the squad, with a clean bridge out to Reaper or Ableton when you wanna level up the mix.

[true carbon]

Usually i like to remove drums from the Suno track and add my own drums, just that make a huge difference

[viral pond]

suno seriously need to move the "studio" button, I keep misclicking it ALL the time

[proper berry]

I completely agree with you. Studio doesn't have to match features with pro DAWs to be terrifically useful. Also I expect we will see Studio gain features very rapidly.

[half dragon]

Hopefully midi-controller as input would be logical next step.

[white dirge]

Facts, Studio don’t need Reaper’s toolbox, it just gotta stack fast.

[proper berry]

I don't even know if it's possible but if they could generate the stems separately to begin with rather than separating them via methods that don't always leave the stems pristine

[knotty stag]

Use AI to translate, telling it to keep same number of syllables (if needed), and then create a new song, using the old song as a Persona, and keep the style at 100%, weirdness very low, and paste in the Korean lyrics into the Lyric section. it might work. I did something similar recently to do a Spanish version of an English song I created earlier.

[rigid mortar]

I have a lot of problems in studio. I understand the constraints of the platform. I'm not trying to edit my music in there I'm trying to get what I need from it and then go into Cubase. The problem I have is that I'm attempting to get one instrument create it. In the original song there is an accordion and a violin. In the stem it combines both of those. Whenever you try to just create the individual instrument in studio it either creates the entire song, it creates a couple of seconds and then gibberish and junk after that, or it's just complete noise and garbage. There's no way to get refunds on junk so I'm wasting my time. What can I do here? I have to be doing something wrong. I've gone from trying to get it to render the entire song worth of this one track to just doing the pieces that it needs to be and it still does the same thing.

[west moth]

How do you use the new studio? How do you upload your own audio to be used in suno???

[half dragon]

These instruments on their list work pretty consitently when I tested it -
Though I only tried small sections, not trying to generate the entire instrument track for the whole song.

[tepid moth]

The individual stems feature works good! So much better sounded stems than from the once extracted from a song. But when I try to create drums, synth stems it creates a full song version. Guitar and bass are clean. Anyone else experiencing the same?

[half dragon]

Your upload audios will be in your Library, then in Studio you can add to the timeline from Library panel, just like other songs or clips.

[west moth]

In library, oke i will check

[rigid mortar]

Accordion isn't in that list. I'm creating Louisiana music so kind of need that to be separate if you know anything about that genre.

[half dragon]

Yes, in Studio, Library has its own panel, just drag and drop the clip into the timeline./

[west moth]

Does it work on the app? Cause i dont see an upload button

I haven't even seen suno studio yet :p

Maybe I have to do it on pc to make it work?

[half dragon]

I don't use mobile app that much other than just for listening, not sure if you can upload from there or not.

[west moth]

Nooo my yearly subscription just ended 🤣

[half dragon]

Yes PC would be best while mobile app provides convenient and mobility but limit functionality.

[rigid mortar]

First time I tried to bring this song into Studio it did not see any of my original stems

[half dragon]

Interesting. Previously extracted stems songs are showing up in Studio for me.

[spiral frost]

Is Suno studio good ? Who tried it before it goes out

[half dragon]

Its already out but only for Premier Users.

[rigid mortar]

Eventually I got the stems back in there I guess it was a glitch or whatever. But it's still does not want to do what I'm asking. It's still combines everything into one. And you can't email support either because they don't respond.

[radiant breach]

"piano ballad, stripped, ballad, piano-driven, ballroom, emotional, slowed"
Should NOT give guitars and drums on the v5 model. This should never be acceptable at this point.

[rigid mortar]

Probably going to ask for my money back and to be returned to just Pro for now. They definitely did not test this

And that's the problem that I have you tell it just give me guitar and it will give you the entire band or it'll give you everything but what you want. I know the idea is to get you to spend your credits and then buy more but this is stupid at this point.

[tepid moth]

I think I solved it. Had the same problem from yesterday. If you do the whole song it will combine all in one (Not guitars and bass for some reason) . Just do not choose the whole song. Let me know if that works for you.

[rigid mortar]

It's still combines two instruments for me even if I do just one part. Or it'll cut it off or it will give me junk but it's never what the original piece was. Like I'm getting a completely different style of instrument but I'm also getting the other stuff mixed into it

[half dragon]

Did you use Custom, then type in the instrument name?

[rigid mortar]

Yep

[half dragon]

and you tried to generate for the whole song?

[rigid mortar]

Parts and the song as a whole

[lusty juniper]

It cost 50 to extract all the stems for the Suno Studio. Its a bunch

[trail junco]

In Studio, can you regenerate vocal takes with updates lyrics? Can you specify no reverb or echo to get a clean vocal?

[rigid mortar]

So as I sit here burning through credits it gave me one generation that was correct but it only gave me three seconds of it I need 10 seconds. And then after that it's back to the same garbage of putting drums or guitars or anything else you can find. It's like talking to a toddler

[half dragon]

I just tried a small section - seems ok

[rigid mortar]

I'm not running Discord on my computer. This take I got complete silence

A cajun accordion is a tin button instrument and it's different than that played in other styles of music. It originated in Germany and then came over to Louisiana and it was adopted with his current style.

It's the one instrument out of this entire song that I can't play even though I have one sitting next to me.

[twilit veldt]

trying to extend. just spinning and spinning and spinning and spinning.

[dusky sundial]

I just jumped in on Suno to try the same. To get some song ideas out to stems and midi to work on in Cubase and the output I got was horrible. Not even the drumtrack was in the same timing as the audio stem. Did uyou get the same? Random drums in a totally different tempo?

[rigid mortar]

Cubase project needs to be in the same tempo

Then add your tracks

[manic estuary]

does anyone know if the way tagginig is done was changed? i used to be able to break up a verse into singing and rapping parts with style prompts in the lyrics but it doesn't work with v5 anymore

[dusky sundial]

I had both set to the same and the midi was totally off it did not sound even remotely similiar to the stem audio.

And the bassline had totally different notes. I could have played it in better myself by ear.

[arctic wren]

I’d like to suggest adding a Time Signature tab in the Suno Create section, especially to include options commonly used in Indian and world music such as 3/4, 6/8, 7/8, and other irregular signatures.

[half dragon]

You should post in #💡┃suggestions-web

[dusky sundial]

Did you make many attempts to get it export ok? I just tried once now and trying to get to grip with if it is worth more tests.

[true carbon]

Do you folks think it would be possible, let's say i generate my track bring it to studio, then remix that same track but change vocalist to female and then use the vocal stems from that to create a duet

[rigid mortar]

I haven't tried with drums, I had a bass track that I tried it with.

[half dragon]

in Studio, on the new track, there's an option to follow the tempo, turn that on might help?

[dusky sundial]

Just one bass track and nothing else in the song? Did you get the same notes as in the audio stem?

[rigid mortar]

No

Well sorta

Im just going to play the real instrument myself.

[half dragon]

That's possible - I've seen a Studio tutorial once on that - haven't tried yet though.

[dusky sundial]

Yes I ended up doing the same, but that renders the export function useless.

[severe quiver]

What to do? I want intensity of the song to persist so I I extend some seconds before the intensity lowers but no matter what I do, the exetensions also drop intensity.

[dusky sundial]

Ok. I am going to give it a rest now for a few hours.

[tawdry perch]

🙁

[digital kraken]

I find that since V5, Suno no longer respects prompts and tags as well.
I often end up with intros with vocals, it often imposes instruments (omnipresent drums), and it has trouble respecting style guidelines (sung lyrics instead of rapped ones).
In short, these are problems I didn't have with V4.5+.

[errant kindle]

here's a style i've come up with

Warm, expressive electric guitar tone with smooth overdrive, sustained notes, prominent use of reverb and delay effects creating spaciousness, slow melodic bends, wide vibrato, blues-influenced phrasing, legato technique, dynamic melodic development, midrange-focused EQ for clarity and warmth, carefully layered arrangement highlighting guitar as central voice supported by ambient keyboards, subtle rhythmic backdrop emulating desert camel gait, careful stereo imaging with echoing reverb and delays enhancing spaciousness, lush and cinematic, with floating Mellotron strings, ambient Rhodes piano, and ethereal synthesizer swells creating an atmospheric, dreamlike soundscape, flowing ballad, soft emotional vocals with lyrical phrasing, gradual rising and falling melodies, warm clean electric guitar accents inspired by slow, expressive phrasing, lush ambient synth pads underneath, mood is wistful, intimate, deeply nostalgic, and slightly surreal, blending heartfelt vulnerability with dreamy ambient textures, dynamics swell gently over time, space and silence are part of the atmosphere, rhythm sways naturally like slow ocean waves.

[wispy laurel]

The really fun part is since v5 released I'm having all those problems with the older models too

[digital kraken]

I didn't dare try going back to 4.5. But I really feel like something has changed in the way we create prompts. No matter how much I insist on negative prompts like drum, drum kit, putting [no drums] in the lyrics, indicating only guitar and piano, it always adds drums at some point. As if, as soon as you indicate [rock] or [rap], it couldn't play without drums.

[lusty juniper]

On new updates usually something will be off. That's when we start pointing it out and then they fix it.

[wispy laurel]

Yep. Seems like every recent model has just lost the ability to follow a prompt
Or at least, dramatically changed the way they interpret them, so all the old tricks don't work now 😠

[lusty juniper]

Yeah it bases it on what it learned. However i think with Suno Studio you can mute the drums or just delete it

[lusty juniper]

Yup. Hence why I'm not rushing. It will take time to get to know the ins and out of v5. My persona at v4.5 was a full band. Like i mean if i cover a song and select that specific persona, its like a band playing it. I can cover 5 songs same persona and its like the band owns those songs. Now however that is not the case. Big difference. Glad im done with the album though

[digital kraken]

Yes, I have this problem too. What's frustrating is that I had managed to find a way to explain to him how to make a rock with beatboxing or claps instead of drums. Now I think we have to wait.

[compact shuttle]

So I know that a lot of people have asked, and haven't got an answer like myself.

For big, compute heavy features Suno often gates access first, then broadens once usage/margins stabilize. If they follow that pattern, a small Pro pilot in Oct–Nov 2025 wouldn’t be surprising, but it could also remain Premier only through Q4 or longer if costs stay high. That’s an informed speculation, not an actual fact btw.

(Referring to Suno studio usage for pro users)

[lusty juniper]

suno studio for pro users (2500 credits) thats tough

[carmine yacht]

I noticed you guys implemented audio to video via hooks, I was curious if you guys could do something like this? like mp3 audio to an ai music video? sites like revid.ai

[true carbon]

what do you mean

[lusty juniper]

pretty suno studio will eat up alot of credits. but maybe theres a way to use it efficiently. once you get the hang of it or keep it simple

[shrewd willow]

Is there a way to add intro to a song that hasnt created one?

[honest heath]

in the editor right click at the start of the song and hit "create section"

[shrewd willow]

Thank you @honest heath

[tranquil abyss]

Am I the only that doesn't see waveforms in Studio?

[raven turtle]

All my v5 generations have reverb tails cut off at the end of tracks! It’s maddening. I assume this isn’t just me it’s happening to. I guess I’ll manually adding a second layer in a DAW and adding something like a Valhalla black hole reverb to mask the cut off, but it’s a lot of work for a simple generation cutting off about a second too early! Grrr!!

[hearty moat]

Yep 100%, it's taking my same prompts very literally as well. I've found I've had to strip my current prompts right back to get it to understand. I've deleted 10+ songs in the last 24 hours. Only had like 2-3 I've kept.. Early days

I have however found that v5 is remastering some older versions 4.0 and 4.5+ really nicely, the production is awesome. Building a new song though, not amazing yet.

[glacial jay]

Yo Suno made me a great song but it has no intro how do I make one with edit or studio mode? Just a simple short instrumental beginning of the song

[raven turtle]

Could you try a cover adding [short intro] at the beginning of the lyrics prompt and setting audio influence to 100%?

[glacial jay]

I just found it it's the little blue carrot

That's a great idea though it'll probably work better

[jade mist]

And here we go, the bug is back where 5 is 4.5 plus. Same stuff happened with 4 being 3.5

[worldly ice]

Hi guys, hope you’re doing well.
I’d really appreciate your help. I subscribed to Asuno, version 5, and when I try to work with it, I get the attached image. I’m not sure what the issue is or how to fix it.
Thanks so much in advance for your help!?

[tender eagle]

So I did something I didn’t usually do. I tested it out with old prompts. Like really old. Comps where I used in-line style using asterisks which would just speak it pretty consistently with 4.5 but it really dug into it with 5 not at all speaking the *prompts* in lyrical insets here for example. Much better than [enclosed] prompts. https://suno.com/s/zCRX8aKx1zUhcZxG

[worldly ice]

??

[half dragon]

That's generic server error

• try again with hope that it doesn't persist
• if generating from uploaded audio, maybe bc of copyright issue with vocal, maybe?

Does this happen on everything or just that one song? I got it too but usually a few tries later it'd work again due to server issue (loads or during updates...etc..)

[hollow hare]

Edit: Misread your comment
You will need to convert the image as HEIC is not one of the formats listed

[tender eagle]

If you’re comfortable with it, hit F12 you can right click and clear the errors and console. Or just refresh, it should be clear. Try to generate then, and it will probably have more information in the error output under the hood there. In the console. But as he said, it’s probably just a temporary hiccup from load, and it might actually be making the track sometimes. And it’s just showing an error due to a Race condition. Which there’s a number of still that show up as toast errors cos they see it as an error cos they don’t have the expected response in time

[viral pond]

hmm... when you create a persona.. what is actually included in that persona ? .. is it just the voice, or all the hiccups and personality in the personas voice as well? does music you use it in need to use exact style prompt or is the style prompt on the persona seperate and stored+

[true carbon]

for me i don't even have same voice on persona

[wispy laurel]

As I understand it a persona includes everything from the song it was created from. So the voice is what most people use it for but it'll also carry elements of melody or rhythm or instrumental style that might show up in songs made using it
When you select a persona, by default its original style instruction will overwrite whatever you have in that space. But you can (and should) change it/add to it, to reflect the differences between its original song and the other one and also to highlight which aspects of the persona you're trying to use
As long as your final style instruction is reasonably consistent with what the persona "wants" to sound like, it should still work at least to some extent
(If you want a persona that only carries a specific voice with no other elements being passed on, what you do is create the persona from an isolated vocal track, cropped to remove all silence - including between lines, therefore breaking the rhythm and melody and stopping other songs from 'inheriting' them)
All of this was correct as of last week. Recent update and release of v5 seems to have broken a lot of stuff, including in older models as well as 5, so I can't guarantee that personas will still work the same way they did before

[worldly ice]

Hi, I couldn't do it. I would really appreciate some help.

[viral pond]

yeah that did happen me many times as well... different voce even with persona,.. even male when persona was female

[viral pond]

@wispy laurel basically a second cover

[wispy laurel]

It's definitely less reliable lately. Not sure if it's working differently or just not working properly though

[viral pond]

open in mspaint save as png

[viral pond]

someone allready uploaded this file

[viral pond]

new suno interface is very buggy after the latest changes... i always accidently clicking studio, making the song i work on... prompts, styles, cover etc have to reset and redo....

usually before song keeps playing when browsing, but in your own library the song abruptly just stops playing when clicking into library

it looks like they are putting users to test the user interface, and not use theire own testers for this

[wispy laurel]

Kind of yes but mainly no?
Cover is much more focused on keeping similar (ish) melody, structure and overall arrangement
Persona theoretically can influence those aspects but generally is better for replicating a voice or a vocal (or instrumental) style
Think about an imaginary pop star called John Cool. His imaginary first single was I'm Awesome.
Cover, here, means copy the song I'm Awesome, but with a different artist doing it
Persona would mean, it's the one and only John Cool, singing a different song instead. He'll have probably some aspects of similar style but it won't just be the same song (unless his writers are really bad)

[rustic yacht]

I'm starting to go insane ... I've created a persona on a track I made (trap/metal in short) this track got a classic introduction -> instrumental + no lyrics singing build-up ... but now every other track I want to create has the same introduction pattern whatever I try to input in my new songs.

[south crystal]

When 5.0 dropped, I tried switching to more conversational style prompting. Spent about 2500 credits on that strategy.

Went back to my usual style specifying Genre, Instruments, Tags and ive got better songs than those conversational prompts in all of my last 40 credits spent.

Back to the old strategy!

[wispy laurel]

If you're using the app, close it and force-stop it
If using the browser, close it (completely, not just minimize)
Clear cache of either one
This might not help but sometimes it does

[rustic yacht]

Ok I'm gonna try, thank you. But If I need to do this for every new track it's gonna be funny

[wispy laurel]

Yeah Suno's just a pain in the @$$ sometimes 😂

[viral pond]

pure genre locking ?

[viral pond]

does a persona use less seeds, i.e. not creating a complete song in one go, does it help to just stem the persona to filter away the other noise ?

[rustic yacht]

Sadly it doesn't work 🙁

[viral pond]

ahh... clicked studio again.. this button really need to be put somewhere else where you not accidently click it all the time

[viral pond]

maybe you allready uploaded it... check library

[wispy laurel]

Usually the main thing you want from a persona is the voice itself
You can expect that the song will use some aspects of the persona but usually not everything from it
If the voice is all you want to copy, then make sure that the voice is the only thing that's there to copy - that's why I mentioned breaking the rhythm and melody of the original song - if the AI can't find a tune in the persona, it can't copy that tune so it looks for something else instead. If there are no instruments either it really narrows it down and, still doesn't guarantee, but definitely improves the chance of getting the same or very similar voice

[rustic yacht]

Uploaded ? Why would I ?

[viral pond]

yeah I want to have the personality traits of a persona which will apply the certain way this persona sings differnt words, however not sure howmuch audio sample you need in order to teach suno these traits

[viral pond]

sorry, mixed you by rayzo

[south crystal]

Thats how I prompted 4.5. I think 5.0 handles conversational better than 4.5, but using the bare bones framework is still way better

[wispy laurel]

For this, yeah I think the best thing would be to use just the vocal stem, but cut it down and remove some or all of the silences, so it's nothing but that voice singing those words without any particular melody or rhythm
I personally find, my most effective/best-working personas usually are made from about 60-90 seconds of speech or song

[winged peak]

@wispy laurel Oh, did Suno just sprinkle a pinch of “more variety” into V5? And branch it off from V4.5+? For the past few moments, I've been getting slightly more varied tracks in V5 with the same prompt. Hmm... what's your feeling?

Okay, I take it all back, it was probably just a glitch... I clicked 10 times, and now I'm getting the same 20 identical results again 😉 - I'm going to bed, nothing more will happen today. Good night, everyone.

[wispy laurel]

Personally I (currently) hate v5 with a passion
BUT, a big part of that is because most of my old tricks don't work so now I have to learn or invent new ones
I'm sure that technically it actually is just as good as they claim it is
But each model responds differently to the same prompt, or has to be prompted differently to get the same result, so whenever a new one comes along nobody knows how to work it so we have to reinvent the wheel
Probably in a few weeks we'll all be saying how amazing v5 is and that's partly because they'll fix some of the MANY bugs but also because we'll just understand how to use it better
But speaking as of right now, I don't think I've ever felt such intense personal hatred for a piece of software before 😂

[worldly ice]

https://statusgator.com/services/suno

i need help please

[polar spade]

I assume the sub page is being updated? For Basic and Pro.

[half dragon]

Mine shows the info.

[polar spade]

Dang. I forgot to pay the pit boss , I guess.

OK. Ill give it a day. See what tomorrow looks like. My premium is fine. Just the Basic and Pro accounts.

[viral pond]

I took the 3 month bait

I tried to design my own vocaloid... did I succeed ?
https://suno.com/s/twdP7wFXqpbS3Cq8

[tender eagle]

On your keyboard there should be a print screen button. If you hit that and just paste the output in here, it would be a more compliant image that could be seen within discord and not have to load it into a viewer app

[tender eagle]

Whenever a new model comes out, I like to go back through several songs that I have always remade with each new model. And then grab another one from the last model. Kind of gives me a spread of all my different prompting styles. Surprisingly, one of the best outputs was a style prompt that was literally four words

https://suno.com/s/NAO38nK6EAVgMQdE

Every version has a very similar feel, which is the feel that I would expect from that prompt, which surprised me given how bare It is. But I'd suggest trying a few very small and minimalistic prompts and let the flow of the actual song guide it a little bit more than the prompt. It worked surprisingly well.

I expanded The prompt to see how it sounded with a more complex prompt and none of them sounded as good as the minimal prompt. It's probably not something I would have tried had I not just grabbed an old song that I had only used a few words for style.

I think this model does pretty well in figuring out a style for the lyrics and cadence of the lyrics, surprisingly well and a lot better than the past models

https://suno.com/s/0evD61Ynr91E7UHu This one did well too. It does surprisingly well for how I expect the lyrics should sound just with the minimalist style prompts. Which it didn't really do before

[viral pond]

I think v5 has a slight bug tht sometimes appear near the beginning of the musc. there is a very subtle glitch/skip

[tender eagle]

One of my tracks has a almost discord message notification sound LOL at about 30 seconds in or so

Very much out of place

[viral pond]

not everytime tho.. but i noticed in a few songs

https://suno.com/s/xv9McZfiI3BFOa0T at : 01:00 mark

i did not edit, or do anything here

[tender eagle]

https://suno.com/s/yFUx7QFlT2k2s0hK check this one at :40 - :41. It sounds so much like a discord message notification 😂

[viral pond]

i did not hear anything

[tender eagle]

Yeah I definitely hear that. It definitely double taps.

Really? It is almost exactly the same as the discord sound. I have a very full range sound system here though. Let me listen to it on my phone

Oh yeah it's very very subtle on the phone. P

[quick shore]

Has anyone else had uploads fail with v5? I’ve tried multiple browsers, phone, different locations…

[tender eagle]

Have you tried it from a desktop? And if you have tap f12 and check your errors and console and see if there's anything kicked back that might explain what's going on.

[quick shore]

Yep MacBook

Like always

[viral pond]

maybe i need headphone to hear it, i just hve a cheap jbl here

[quick shore]

I mean it could be just my files

But never had issues before

[brave ridge]

Bro please help me

[tender eagle]

Yeah when I listened to it on my phone I could barely hear it. It occurs right at the: 41 mark

[tender eagle]

What kind of file are they

[quick shore]

Tried both mp3 and wav

[tender eagle]

Well it's probably not your file then. On the off chance that it is, however, do you have any files still stored locally that you have definitely uploaded before and worked?

[quick shore]

Yeah good idea

[tender eagle]

I'm pretty unfamiliar with apple, but here it says if you go to your preferences and enable developer, you can hit option command C and it would open the jscript console so you could probably see if there's any errors kicking back that might explain what's going on

If you have Chrome installed, f12 should open it. Just the same without having to enable anything

[quick shore]

Yeah I’ve opened it but no idea what I’m looking for haha

Maybe it’s just timing out

That’s why I asked here

[tender eagle]

Your best bet is to probably refresh the page, so that it's completely clean and empty, and keep the developer console open. And when you try to upload the file, see if any errors kick back, because you're going to have a bunch of noise if you've been doing a bunch of different stuff since last refresh of the page or cleaning your console

[quick shore]

Sounds like it’s on my end

Yeah I’ve cleared cache. Rebooted

Tried other browsers and phone app

[tender eagle]

Wait, what exactly is it doing.

It should be model independent.

[quick shore]

Gonna try on incognito browser

The one thing I haven’t tried

It’s just not loading the pop up

I click and it either doesn’t do anything

Or an error box pops up

[tender eagle]

What's the error?

[quick shore]

Last time it just said error can’t upload or something. Orange box popular

Pop up

[viral pond]

its such annoying thing that if you do a single modification / edit, i creates a new song with no likes andlistening. the original song should been replaced with these, and users who lked should get a notification. I think that would make things better and avoid having to publish and spam users with same new variants

[quick shore]

What should I look for in the console?

[viral pond]

I have given this as a feedback, hopefully they can implement a feature that will keep likes and play counts

[tender eagle]

This is the in browser JavaScript console right? Not your OS console right?

[quick shore]

An got it

Sorry k didn’t click on the tab

[tender eagle]

Ah ok

[quick shore]

Can you read that ?

[tender eagle]

I figured it should be pretty obvious cuz it should just have the error message dump out. In a big red bar

[viral pond]

yes, the entire suno page is a big javascript bundle

[tender eagle]

Okay, the issue isn't what you're looking at there. It's the two below. It the actual red errors, not the warnings

[quick shore]

Yeah the I caught encoding error

[viral pond]

first thing to check a different browser

[tender eagle]

The I'm guessing the audio data decode

[viral pond]

or incognito modus

[quick shore]

Yep

My app just auto blocks

Thanks I’ll have a crack

Just odd it’s on multiple browsers and Devices

[tender eagle]

What language is your browser in English? I guess right?

[quick shore]

Yeah

[tender eagle]

There was an issue before with a language unicode something or other that mangled the names of things . But if English prolly not the issue

But that was a couple months back I think

And the ones that did work before aren't working now correct?

[quick shore]

Non incognito throwing up more

Anyway thanks

[viral pond]

enalbe the ai assistance, maybe it can solve it

[tender eagle]

Drop the error message down a little bit. It has more info in it

[quick shore]

At least I know it’s not a Suno issue

[tender eagle]

Some of that is just race condition jitter. That they haven't fixed on some of the promise stuff.

[quick shore]

You mean scroll up?

[tender eagle]

No, you see the little black error that's pointing to the right. It should fold it down

[quick shore]

Gotcha

Maybe I’ll try using a vpn

Who knows

[tender eagle]

What are the name of your files. Try naming them something very simple like a.wav

I wonder if this is similar to that other problem that I ran into a couple months ago

[quick shore]

Yeah unfortunately no good

I’m just picking random files each time too

Not the same ones

Annoying when I’m on the premium sub

And haven’t been able to use fully for my own tunes

Anyway thanks for the assist

I’m sure it will solve itself soon

I can record audio in

So I’ll workaround by playing the music into the mic for now

[tender eagle]

Don't do that. That'll sound like crap. Use a loopback. So it plays directly into the input

[quick shore]

Yeah you haven’t heard the quality of my garage band
Recordings hahaha

[tender eagle]

Lol 😂

Like if you use something like rogue amoeba. I'm pretty sure there's some open source versions, but I'm not really sure with Mac. But you can record directly from your speaker output as a loopback input. It's much much better than microphone

But I think it has a free version but I don't know if it's gimped or anything.

[quick shore]

It’s kind of amazing how Suno can basically make them sound like they were recorded properly. Only wish I could input my own singer for
Songs that don’t have vocals. Once it sticks a voice on there none of the personas etc seems to do
Much

I mean i can stem them later

But just for the input stage

[grim pawn]

Does anyone have any tricks to up the song duration in v5? even with uploaded songs it ends up cutting like half of it

[tender eagle]

Yeah. If I really want to change the vocals I'll just run it through cover a few times and stem off the vocals I want and stem the vocals out of the instrumental and shove those together in my daw lol 🤣

[tender eagle]

Is it ending at the same time every time? Like 4 minutes and 18 seconds or something like that or 5 minutes and 49 just arbitrarily. But always the same when it won't finish it?

[quick shore]

Ha yeah I do the same. I’ve had some luck with a persona for instrumentals but not with tracks with vocals already.

The issue is the vocal lines then

[grim pawn]

it seems kind of dependent on the song, it likes to cut out bridges and skip straight to the chorus etc, but it's pretty consistent. v4.5 usually covered the whole song but it feels like v5 is "impatient".

to rephraseit usually skips the same section if i upload a file

[tender eagle]

Well yeah what I mean is on a single song. Does any cover you try to do of it. Always end at the same time When it is cutting it off. Or is it different for the same song. How long it will generate for

Ohhh

It's skipping the bridge and such.

[quick shore]

Still better than the days when I had to try and stitch together 2min partials that were different sounds and tempos

[tender eagle]

Lol

[quick shore]

Every segment I fed in sounded different

[tender eagle]

Try knocking the audio influence down about halfway. Take it from the default 25 to maybe like 15

[quick shore]

There was a process I had haha

Back in the v3 days

[tender eagle]

Oh God

[tender eagle]

The wait are you doing a cover or is this just like straight generation and it's just skipping it

[grim pawn]

The durations as a whole are also just shorter, averaging around 1-2 minutes \whereas 4.5 gave me the duration of the uploaded song.

[tender eagle]

Yeah reduce the audio influence in the sliders

[grim pawn]

Covers, but generated songs are also just shorter

but i mainly do covers

[quick shore]

One file just worked for upload after trying like 20!

Gonna suss what’s different

[grim pawn]

doing this works sometimes, but i find it loses track of the melody often if it's too low

[tender eagle]

First, before spending too much time trying to figure that out. Try uploading one that did not work before

[grim pawn]

especially complex stuff

[quick shore]

Yep I did

Tried another 5

[tender eagle]

Yeah. So what I think is that this model actually plays a lot of its choices in what to do based on the lyrics as a part of the influence of the way the music sounds. Almost as if the style of your writing kind of affects the style itself

[quick shore]

Haha

Lucky one

[grim pawn]

yeah, i've felt like that's the case, i'm glad you feel the same

this problem is really bad with instrumental music though, which is what i'm primarily covering

[tender eagle]

And I'm wondering if the audio of the covered song is not congruent with what it expects the lyrics to sound like cos I had it doing some wonky stuff that felt like this. Hah

Yeah what I mean is that if the instrumental does not feel like, hey the bridge here. Skip it because it can't skip the musical influence as much as it can. Just the words

[tender eagle]

Well it seems to do really well doing exactly what I think. The lyrics that I've written should sound like. Even with only three and four word prompts for style

[tender eagle]

So only that one did work? And the other's not?

[grim pawn]

yeah its way better with putting the lyrics in intuitive places

[tender eagle]

Version 5 seems to really really care about cadence and placement

[grim pawn]

ive seen other people suggest writing lyrics differently too, but since im working with stuff liek Video Game OSTs, there usually aren't any

and it ends up skipping some of the most important parts of the tracks, when 4.5 didn't do that really

[tender eagle]

If there's no lyrics then why are you using bridge?

Just use a different name for it like properly name it.

As a prompt or something

[quick shore]

Another one just worked. Interesting thing is these are files that output from Suno lol

[grim pawn]

i guess i mean the theoretical bridge, not written

it will just skip "chunks" of the song

[tender eagle]

Well yeah what I'm saying is don't call it bridge call it something else more logical for the song you're using

[grim pawn]

so even if its instrumental should i still write lyrics?

[tender eagle]

Maybe it's skipping it because bridge suggests something to it that it isn't hearing

[tender eagle]

Strange

[tender eagle]

No no ni

[tender eagle]

[verse]
[Chorus]
[Bridge]

Section titles. How are you labeling it in the instrumental not do you not? Do you just set to instrumental ?

[grim pawn]

Yeah, i just set it to instrumental

and it ends up skipping sections more often than not

[tender eagle]

Oh okay. Usually I structure it within the lyrics even if it's instrumental. With no lyrics to be spoken.

[grim pawn]

i see, i could try that

[tender eagle]

Using instrumental in the style. But giving it sections with titles

[grim pawn]

thanks

im glad to have an idea to try lol, ive been pretty frustrated

[tender eagle]

That are logical for what I'm doing. For
[Build]
[Synth Riser]
[808 Fill]

And such

[grim pawn]

gotcha

that sounds like it could work

i wasnt aware that instrumental songs benefitted from that

[tender eagle]

Also surprisingly. And I found this earlier. Try using *Asterisks* for effects lol. This used to work and then it stopped working and it would just sing the enclosed things. But now it seems to work really well too. So you can try using that for different elements of the script per se.

Oh they definitely do. I'll use it to bring in instruments that I don't list in the style for example
[Trumpets Build over Melody]

And such

[grim pawn]

nice :) i used to love adding record scratches, but after a certain update, the singer would just sing "record scratch" lmao

[tender eagle]

Here check this ))

[grim pawn]

nice to know i might can do that again

[tender eagle]

Yes you absolutely can

[grim pawn]

LOL thats the first thing you did

amazing

[tender eagle]

Lol

[grim pawn]

well, thank you :) im gonna go do some testing

[tender eagle]

That's wrong on

https://suno.com/s/zCRX8aKx1zUhcZxG

[grim pawn]

thats really cool

[tender eagle]

This one actually did what it was supposed to do almost at every single point that it had to effect listed

[grim pawn]

i missed my record scratches 🥲

[tender eagle]

Yeah me too

This is a really old one that I had done. And I just didn't modify it. If I did this in 4.5 it would sing about record scratches over and over

I had to enclose it in the brackets for 4.5

[magic nest]

trying to get that gritty raspy vocal style with country but only came out once by accident with v5 lol sooo tricky. i used every phrase so far🤦‍♂️

[civic spire]

why does not entering lyrics for instrumental now always give my music a female choir voice

[lost vault]

Studio was awesome in the morning, right now is completely unusable, don't even try, you're just gonna be wasting credits

[grim pawn]

so, adding [very long Instrumental] actually gives me a very long instrumental usually lol

[tender eagle]

Haha

Now it is?

Okay so I guess that's addressing the problem you were having right?

[grim pawn]

seems like it, i was able to get my first 4:30 generation

4:38 actually lol, with a 3 minute input

adding long, full-length, and ost to the prompt also probably helped some, but after doing some testing i feel that [very long instrumental] seems to affect it more

[tender eagle]

Nice

[strange mural]

So are Suno 5 songs limited in time? Seems like I haven't had any long tracks and sometime it will ditch stuff to cram it in. Was hoping for more than 5000 characters, but still rockin.

[tender eagle]

I don't think they're inherently limited. I think the model likes to duck out early. You got to kind of twist. Its arm a little bit sometimes to keep it as per @grim pawn

[viral pond]

how you save as a new song when you edit something in st udio : )

[tender eagle]

You might urge it along within the lyrics themselves. With some inline prompting. Hubert says that works pretty well for instrumentals at least. But I don't think it would not work for lyrical tracks. Just the same. It's kind of hackish but I mean if it works

[tender eagle]

I actually have no idea lol

[viral pond]

haha 😄

[tender eagle]

Does it have a get full song option in the little three dots of in the corner of it.

[viral pond]

yeah, I mean i can se a export option, but i mean. i want to save as shareable song

because now can finally add sound effects to the songs

giving it atmosphere

[tender eagle]

Does export export it from the studio to a file or to your storage online? I actually don't know. I've never had a chance to really play with it much before my account did. Its little little hiatus for a moment

[viral pond]

oh

https://suno.com/s/i0DyJzQJROXi2mdr

export did save it to library as a new song

i just added some sound effects to test

[tender eagle]

Yeah I thought it might, exporting it out of the studio. But still on system

[viral pond]

at around 25 seconds

[tender eagle]

I just never actually did it. And I can't at the moment test it so. But thanks for letting me know

[viral pond]

developpers and theire naming schemes 😉

oh haha something strange

why is it playing way too slow

[tender eagle]

Did you accidentally adjust the tempo LOL

[viral pond]

hmm not that i know, unless by accident

[strange mural]

Also anyone know good ways to prompt a shift in style mid song. I am doing an East Coast vs West Coast vs Chicago thing and in 4.5+ was kinda getting it there.

[viral pond]

playing norway in studio

[tender eagle]

This song feels very opioid at this speed.

Okay, that glass shatter is pretty cool though.

[viral pond]

yeah

[tender eagle]

Even at half speed LOL

[viral pond]

so now when I create lyrics with LLM I can instruct it to insert [SFX: some sound effect] but need to make sure it is ignored by the AI engine so I can place sound effects at that position

[tender eagle]

I know it has some options in there that have to do with like the tempo, whether it's at Base, original tempo or a modified tempo. Not sure if that did something.

[viral pond]

hehe yeah . not sure if the engine do some enhancements

the tempo is set to follow beat

[tender eagle]

Hmm. Is the beat set to original or?

I think follow beat means that it maintains quantization. I'm not sure I didn't get to play with it much yet. I do know that there's several different things related to beat, and tempo,

[viral pond]

not sure wht the original beat was

[tender eagle]

I know, but there's an option switch on the right hand side earlier when I was playing with it that had original beat. Versus a input beat.

[viral pond]

this is the original song https://suno.com/s/i0DyJzQJROXi2mdr

supposed to be 68 bpm in prompt

[tender eagle]

That's not the original. That's the same one you linked earlier

[viral pond]

hmm

clipboard bugging again

https://suno.com/s/gcp1xkuiC6usvSWp now then

[tender eagle]

I mean that is about 62 or 63 beats per minute I used a little metronome thing but I might be double timing. It. Just because it's so slow it's hard to really follow

[viral pond]

what does it cost in credits to save i wonder

[vivid wasp]

So are we getting closer to having better chances at sfx in v5

[viral pond]

i set it consistently to 120 and now it plays normal https://suno.com/s/NUzLk2irxMPO5zVm

@vivid waspwith studio you can upload sound effects like the test i did on this song

[tender eagle]

Yeah that's actually about 124 beats per minute according to this here

[viral pond]

at around 25 seconds

i guess prompted BPM are not accurate, atleast on the earlier versions

[tender eagle]

No they never really are. I think it just defaults to genre standards.

When the drums kick in they are definitely right at 124 BPM. On the original that you sent

[vivid wasp]

But if I need a demonic chuckle won't that flag as a voice

[viral pond]

probably,, make persona was disabled

but those can be done with meta tags i hink

atleast i done that before

uploading any files used in it probably triggers it as an uploaded asset

[honest tendon]

Saaayyy. I think this goes here....did anyone notice/experience an algorithm lull/ dampening between 150 and 200 followers?

Noticed a pretty hard flatline and was just curious.

[weak eagle]

i just tried this with a v5 cover of a song originally made with 4.5+, style influence W50 S60 A70 and results are blowing my mind, is even taking into account things that 4.5+ didnt even bother

[dusky sundial]

What is the format for adding tags in vocals?

Is it with [] in the middle of the text?

[Smoky]lyircs sung smokey.[Glassy] Do they change to glassy for this part?

[fallow sand]

So, ive got this song that I want the stems extracting for. However, it seems to get corrupted and sound glitchy. Ive cleared my cache and tried other methods I.e different device but still happens. This is on the v5 version. Any ideas?

[astral moss]

why does suno take my credits for stems I already paid for and I am only able to download them individual. Not all

[fathom crypt]

Post in the help section #archived-ticket-help

[polar spade]

If we go back into the same song we originally got stems from , within, a few minutes, I believe you'll still have access to click for past stems. Id say within 4 or 5 minutes or until we move on the making other edits. Some say it can be longer than that.

Also, there should be a "download all" which saves as a zip file.

Update: 1:15am and still have a " See extracted Stems" button.

[ornate pond]

whats good group? some one can share me some info or a link to learn how to interpolate samples with suno? example if i find a song i want to sample , gave to suno to make something with that feel?

[untold briar]

Upload your sample and Extend is the most straightforward, as well as Covering it. And Suno Studio seems to be building towards this specific use case (turning a short sample into a longer sample or whole song) but even Extend and Cover can do cool things

[ornate pond]

i try it put gave me copyright

[untold briar]

ugh, could be a false positive. Or if the Extend being blocked it's something in the lyrics you typed instead.

[ornate pond]

Uploaded audio matches existing work of art.
Please upload a different audio file.
i shows me

[jade mist]

Is the max time for v5 8 minutes?

[untold briar]

If it's a sample from a real commercial song than that's what is supposed to happen. If it's a sample you made or something, that's unfortunately just a bug, a false positive. There's no specific place to report that kind of bug but you can make a post the bugs channel.

[ornate pond]

theres any way i can do what i want to achive? , example locally or other ia?

[untold briar]

I haven't kept up with the latest open source local music AI though I'm probably gonna check it out soon, sometimes the black box of Suno just isn't as much fun as a worse model that you can dig into and muck around with.

But if you just want something "with that feel" of the sample and not something literally using that sample, you can probably get Suno to give you that with the right prompting.

[mint yew]

yes cause i believe thats what plans are capped at

[ornate pond]

any source? i would love to suno interpolate but keeping the same vibe as certain records

[vivid wasp]

Why is it, I put the word "maniacal" in my lyrics box . But all generations sound like they're saying "monocle" lol.

[untold briar]

Maniacs are rule breakers

[vivid wasp]

Yes, yes they are I watched Animaniacs and that showcased much rule breaking

[quiet solar]

There are a few words Suno struggles with pronunciation for, or will get the context wrong and prononuce it out of context (e.g. "live" as in "I want to live" versus "live" as in "I'm playing live music").

You may get some mileage changing the lyrics to something phonetically similar to the word; e.g. man eye-akull

[jagged narwhal]

how you guys try to keep original voices?

[ebon sand]

try is the correct word... extensions or inspo, both is not very satisfactory/bugged but might get the job done for you. don't bother with persona, it doesn't work.

[reef tree]

Would love to see an automix feature. Either, with my own tunes, and/or all of suno tunes. Perhaps also in the radio section too.

[polar spade]

@grand tiger re: barking sound in vocal stems

If asking other's experience, thats really best for the tech-prompt-talk. (Here) https://discord.com/channels/1069381916492562582/1102676217687195679 channel to discuss if anyone has come across the same before posting it as a bug 🐛 .

Otherwise, the tech channel (here) is better to see if it was a one-time glitch as well. I ran two multi-stem tests and listened to versions 1 and 2 of both with no issues.

You could run a stem separation on another track to see if may be a one-time glitch or even download the vocal stem as an mp3 and see if playback shows the same sound.

If a second alternate output test sounds normal, the culprit likely is a glitch. Its also possible there's just a reaction to Suno regenerating a vocal stem from your specific output that singularly is causing an anomaly that won't happen with any other output.

But certainly, see if anyone else is experiencing the same here.

[viral pond]

V5 is very buggy, it skips beats

i figured that this becomes a problem especially if you remix

is the editor broken also? tried to replace area, and it replaced the much more of the area and when i decided to slide the edit it just restarted the song mid ofit

[polar spade]

@mint mauve You must contact [e-mail] directly. [Im just posting here as need-help channel has a 2-hour delay on weekends.]

Accidental or inadvertent payments can happen. Please review the billing /refund page regarding conditions for refund https://help.suno.com/en/articles/2550209 and contact billing with details for a review. https://discord.com/channels/1069381916492562582/1373016771103166575

[viral pond]

is there a way to fade out single tracks in studio ?

when i insert a auido clip

oh yes, happen to me before as well. accidently had a card not working any more, and could not update the payment. had to downgrade first, the upgrade agan after

[polar spade]

You should post this in the Suggestion Channel. Until then, use one of many available "DJ" style apps that blend or mix and download your favourite mp3s and mix away.

[tender eagle]

For sure )). It pays a lot of attention to detail. And as I was saying earlier too, it seems that it takes the lyrics and recognizes how it "should" sound along side the prompting too. Takes a lot into account really.

[viral pond]

nice, saving song from studio masks the original style prompt. so you can hide some of your secret sauce 😎

[tender eagle]

There is nothing rigging them to sound bad. I mean I'm not sure really how it could do that.

[viral pond]

not sure if it keeps any remix origins still, or gives notifications to who you remix tho

thinks tend to become very spicy and weird if you remix, cut edit songs / extend... it's always a dice roll, but prompts just nudges the ai engine to go certain direction, you could try to reinforce it using meta tags in the lyrics section

[tender eagle]

The reality is is that it's AI. Stochasticism is real. Yes casino in a sense, so far in that there is some randomization which does as random does. But you have more control than you realize. Your complaint doesn't really include any examples that could help you address whatever it is that is causing issues for ya though

[viral pond]

we all sturggle with this from time to time 🙂

[tender eagle]

Another option if you want the v5 appeal but want to maintain the control in the way you're used to. Gen it in 4.5+ or 4.5 and cover it

[graceful condor]

I think they actually should do a little more effort and explain all the changes to us customers with each of those major changes. Like what has changed and how to approach different aspecs now. But so far what are the biggest changes you noticed and how to approach them differently? I would really like to know. What I noticed is the voices may might be better but they rush all lines in song especially if they are longer ones. It is so annoying because they sing songs lines like if there is no dot or comma. Like a long plain straight line which has no ups and downs. I wonder how you can fix that. Did you noticed that to?

[tender eagle]

Well again. It's AI changes are in the training. It's like any AI. It does as it does and even those who trained it can't really speak for it's full abilities or what does what specifically. I'll agree in some regards to how the training may limit or remove limitations in prompt methods as well as inline prompting but even with that. You may find it does not act as you expect

[undone musk]

its not really a suggestion. But depending on how it might be beneficial in a suggestion channel. I would explore that. Why in a suggestion channel?

[tender eagle]

You have to consider a few things about how it all works too.

If it's rushing verses split the verse

From
[Verse]
[Chorus]

To

[Verse 1]
[Verse 2]
[Chorus]

Because if you think about how it is trained. Or at least how we assume it is trained. It'll have a verse listed. Then chorus. If 99% of the training data that it is trained on, the verse falls within a specific set of bars before switching to chorus. A long verse may be compressed in delivery to match the standard of that style type in verse chorus timing

[viral pond]

You can always send them an email, even if you don't have time to wait, they will still read your feedback.

[polar spade]

i had responded to @reef tree . . . . So you may have thought I had posted to you? Yes, Your post was not a suggestion.

But since you mentioned: The suggestion channel is where the devs can read and consider adding a toggle or button to adjust the blend of songs in the radio or elsewhere. This Tech-prompt room is mostly us users just having a chat.

[viral pond]

we are actually not support, but we help whenever we can if we know any answer or have suggestions

[graceful condor]

Makes honestly not a difference. It is just V5 is weird. And I do not know how to make it work. They f-d up something really hard because it is not normal to nearly talk singing or rushing lines in a straight line so fast. The example I mentioned here was a song actually with Verse1 and Verse2 after another and in Verse2 it rushed the whole Verse like "Lets get it over with"

I'm with a friend in a world rebuilt,
He says I'm just a hollow shell,
A story that I cannot tell.
The city's healed,the scars erased,
But all the damage is my own,
A seed of time that has been sown

This is not a Verse which is that long or anything. And it worked without any issues in 4.5+

[viral pond]

V5 is a beta tho... it sounds absolutely gorgerous, but it does bug out when I cover songs, and crank the weirdness up a bit

[slow merlin]

I notcied this last night, noticed that it was not giving time to the [intro] and treating it like another verse, rushing right from [intro] to [verse], also noticed the rushing of vocals as well with longer lines. Have not figured out a workaround as of yet.

[tender eagle]

Link me to the song that you're having an issue with and let me see if I can rerun it and get it sorted. There may be something that isn't obvious leading to the rush.

[tender eagle]

Yes it does do this. And again, I'm wondering if this is because the more they train it, the more it moves away from understanding things which are not readily present in the training data. Eg. Maybe they don't use [intro] as the header for that type of initial section before it moves on.

And the reason that I wonder this is because when I use that header. I find that it often does the first verse as an intro before shifting to the body of the song at the second verse, which then continues to use the verse and similar styles as it does the second but not the first for the remainder of the song

[graceful condor]

What do you mean link you to the song? Also what I noticed too with V5 it makes way more misspell errors. Not only in foreign ones like Japanese but English too. it is singing stuff that is not even there. Not written in the lyrics at all which makes no sense. Or singing absolute gibberish.

[undone musk]

obviously, you didn't read it. it is highly possible. Not that it is but it is a real possibility. I don't know all know is recorded it for a while. Without resolve my theory stands and that's how I will present it.

[tender eagle]

Share the link to the song you're talking about

Click share and copy the link or copy it from your browser URL

[slow merlin]

I have no way of backing this up, only by from what I hear and have experienced, started with v4 and it seems that with every upgrade to a new version the first week is a mess in terms of songs created and gets better over the coming weeks, AI training its self? I thought that when you create a song, when you hit the thumbs down or thumbs up your basically training the AI each time you hit them thumbs up or down, again no way to verify if this is true

[graceful condor]

Send you a detailed message 😉

[graceful condor]

Really thumbs up or down to a song trains them?

[slow merlin]

I used this in v4.5 to give the music a cue to not rush into the next part but also to give it more length to the song. Have not tried this in v5 yet
[intro]

,...,...

[verse]

[slow merlin]

so i thought i seen that some where but again not sure if this is true or not

[tepid moth]

I’ve been testing the MIDI feature the past few days, and I have to say I’m really impressed—especially considering this is just the beginning of Suno Studio. For piano and electric bass it already works great, though I haven’t tried it yet with other instruments. The individual stems feature feels a bit hit or miss. For songs outside of modern genres (like techno, trance, etc.), it does a really good job—the stems are much cleaner than what I’ve been able to separate from songs before. But for modern genres, it still struggles. For example, I tested it on a techno track and couldn’t get a single clean drum stem that was only drums. Anyone else have any experience too share about this features?

[slow merlin]

i treat it as if it were true and actually shape my workflow around it, when i start a song project i start with music first, creating the music style box first and hearing if the music is close to what i want, i hit the thumbs up or down depending on what i hear and like, it seems that it helps to get the music to where i want it but again no way if this is actually working or just getting lucky

[graceful condor]

In my case it is not really a rush from one section to another like from Intro to Verse. But more like rushing the Verse as it is. In 4.5+ it was like a dynamic wave if you wanna imagine it like that - It came forward, got backwards, held still, gained dramatic in energy, got highs and lows and was like a Roller Coaster but now it feels like if some is eager to get it over with and just wants to tell you the most important new gossip as fast as it can without any pause or comma.

[undone musk]

the only examples i have control of are weirdness which it carries its own leeway but it's excessiveness is not normal. gender select probably 70 % + have ignored selection. and rude obnoxious weirdness noise probably the same. I spent many (enough) hours here. esp. this past month almost 24/7. . i recorded it for awhile. i don't know what you expect me to bring in here as some sort of evidence

[undone musk]

ok, thank you for the response. i'll check it out

[viral pond]

are you extending the song, or just creating from scratch, is it a cover of previous song ?, you could maybe try add a persona, and reinforce female if that's what you want, both in style prompt, and as a meta tag on top of the melody + each verse/choruses etc

also I have noticed that the actual prompt contained something that just pushed suno to forcefully use a male voice, example didgeridoo growls for some reason

[tender eagle]

Give me a second. I'll be right back

[undone musk]

i didnt say the sound,

[tender eagle]

Hmm? How is audio bad if not how it sounds

[graceful condor]

What really annoys me is that the AI can not handly complex detailed prompts the more you go in to detail the more likely it will cut off the song and make it shorter. Which makes no sense. So in the end you literally can't be as creative unless you wanna extend at least 1-2 times within a song.

[keen zephyr]

Hi all. Has anyone else noticed increased data usage on the app since 5.0 released?

Ive got 74gb of data used in the last couple days

And my usage of the app has not changed. I listen to my songs while driving in the car

[empty anvil]

is there any problem with the web app? not loading

[polar spade]

Assuming theres a higher fidelity presented overall, it makes sense more data would be used.

[jagged narwhal]

I left emtpy lyrics but it adds lyrics??

[dreamy sparrow]

Use these lyrics instead:

[SILENCE]
...

[jagged narwhal]

lets try

[dreamy sparrow]

or even giving the entire song structure within section titles []

[half dragon]

You might want to switch Instrumental on.
Or use tag for song structure in lyric box without any lyric such as [Intro]... [End]

[jagged narwhal]

thanks! I guess worked now @dreamy sparrow

[polar spade]

This is Australia? I see there are a couple of different offers depending on location, it seems.

[empty anvil]

what is wrong with this today? all blank

[dreamy sparrow]

You can search Google on how to write the perfect lyrics for Suno AI. There is documentation about Suno.

[jagged narwhal]

ah no again added 😄 weird

[dreamy sparrow]

Try clearing your cache and refreshing the page SHIFT+F5.

[empty anvil]

even from my mobile app its not loading

[polar spade]

@jagged narwhal Then use [instrumental] and refresh so you're starting with no buffer.

.

[polar spade]

There's no support staff in tech-prompt-talk. This is just a community channel of users.

My Library loaded in about 10 seconds. Im Premium so a bit faster. If you're Basic or even Pro and there is a loaded Server at the moment, there may be temporary slow moments.

[vivid wasp]

I thought about that, I'm going to go into the editor and see if using "manyakal" will work later today.

[empty anvil]

now loading . it was not loading my account so i searched the suno subscription link and opened it . then it showed my account which was already subscribed , then it opened 😄

[restive heath]

Hi, I’d like to know if it’s possible to make an instrumental like this with Suno?

[true carbon]

I would probably upload that in Suno, then do a cover with it and it will show you the ''style'' that Suno analyzed in it, maybe i am wrong but i would do that !

Or just remix it or cover it when imported? Should give you something similar but different ?

[restive heath]

He says it’s copyrighted, unfortunately.

[true carbon]

Who say it's copyrighted? Suno wont liket you upload that? What did you put in the prompt to have copyrighted ? Did you add lyrics or artist name ?

[true carbon]

ah got it, i don't know then i mean it's so generic, where is it coming from?

[restive heath]

it's songs

[true carbon]

ah that's why! Then try to analyse the genre of that song and hope, it will be hard to have Suno give you that

[restive heath]

I already tried with ChatGPT etc., but it creates an instrumental that has nothing to do with it.

:/

[gritty narwhal]

Having a ton of issues with Studio. I try to remix things and it gives me an entirely different song If I remix in the sidebar, or completely changes the melody if I remix on the track. Help? Im trying to remaster songs that ive made in the past (yes the hard way, but the remaster button has always made my songs worse rather than better.) Is anyone else having trouble with studio?

[weary glacier]

guys,anyone know why the image i have uploaded cannot display at the library. ..well some can but some cannot...

[winged peak]

Suno has applied some filters which are a bit too 'strict' and misinterpret some things, often - for example - it doesn't allow me to upload pictures of persons in tight clothes, male or female... so, best create a few alternative versions and test 'em 'till one finally is been accepted.

[weary glacier]

right.thanks..

[torn spoke]

Noticed stem extraction is pretty bad at recognicing guitars

It has to be electric guitar for it to be slightly coonsistent, and even then, half of it ends up in "FX"

[weary wharf]

Sharing a prompt that I've been using with 4, 4.5, + and 5 now. It works pretty well.
Just put it in a custom GPT, or use as the prompt itself. Ask GPT about what you want like "I want a prompt like Aerosmith's Dream On". And try the 5 different prompts it gives you.

Have fun...

[slow merlin]

OMG i am so sick of v5 giving me these ooooooo's and ahhhhhhh 's in EVERY start of any style i pick, so annoying

[torn spoke]

Odd I don't have that problem

[jagged narwhal]

when we will able to use same voice 🥹

[proper berry]

have you tried remove styles? also explicit instructions in square brackets but not of the "don't do this" variety

[slow merlin]

9 out of 10 songs will give me ooooo's and ahhhhhh's or some type of blend of that no matter how i prompt

[torn spoke]

and what are your lyrics?

[slow merlin]

its should go right in this.... (Music starts with a dark, distorted synth pad and a distant, rumbling 808. Sound of a car's ignition and a low engine growl.)

(Intro)
(Ad-libs in a heavily autotuned, spaced-out voice)
Yeah...
(It's lit!)
Night vision...
(Straight up!)
Alright...
Yeah...
but instand gives me something before it that is just oooooo'h and ahhhhh's

[torn spoke]

Put into in brackets

intro

And, specify the adlibs

[slow merlin]

Style box: style | Trap, hip-hop, psychedelic rap; atmospheric production blending heavy beats with immersive, futuristic textures.
harmony | Minimal harmonic movement; static tonal centers enriched by synth pads, samples, and ambient layering.
melody | Autotuned vocal lines, chant-like hooks; hypnotic refrains intertwine with layered ad-libs and echoes.
rhythm | Heavy 808-driven beats; syncopated hi-hats, rolling triplets, steady kick patterns define rhythmic identity.
dynamics | Builds from subdued verses to explosive choruses; textural layering amplifies contrast and momentum.
expression | Hypnotic, moody, surreal; conveys swagger, vulnerability, and escapism through atmospheric vocal delivery.
technique | Autotune, layered vocal effects, sampling, electronic synths, heavy basslines, meticulous trap production.
form | Verse–chorus frameworks; extended outros, atmospheric interludes, and beat switches create evolving structures.

[torn spoke]

Have you tried making it a cohesive thing?
Also if you want to use sectioning like that, LLM's are used to title formatting. So try:

#Harmony
description

#Melody
description

You could also just remove the "oooohs" in studio

[slow merlin]

#style | Trap, hip-hop, psychedelic rap; atmospheric production blending heavy beats with immersive, futuristic textures.

?

[torn spoke]

#Style
Trap, hip-hop, psychedelic rap; atmospheric production blending heavy beats with immersive, futuristic textures.

[slow merlin]

too poor for upgrades lol

[slow merlin]

Ugh so annoying, gen'd 4 times so 8 songs [Music starts with a dark, distorted synth pad and a distant, rumbling 808. Sound of a car's ignition and a low engine growl]

[Intro]
(Ad-libs in a heavily autotuned, spaced-out voice)
Yeah...
(It's lit!)
Night vision...
(Straight up!)
Alright...
Yeah... And they all like this https://suno.com/s/svEnm0B7I2B9zdtm aaaaaaaaahhhhhhh in the very start grrrrrr

[torn spoke]

It's not good at doing non-music sound efx my guy

[slow merlin]

wow just realized im in the tech prompts chat lol

[torn spoke]

if you want to do that stuff you'll have to either find or generate those sound efx yourself and include it in the context

[slow merlin]

didnt really think i was going to get.... Sound of a car's ignition and a low engine growl........ the car sounds, just wantted to see what it would do

I do the same prompts in v4.5 not changing a thing and its perfect

[wraith seal]

I've been trying to generate the spoken word with an instrumental backing, rather than a classic song. Nothing I try works, I've tried adding Story and Reading to the style, making it very obvious, adding Character Monoglue, Spoken Word and NO SINGING to the lyric tags. Any ideas on something consistent?

[torn spoke]

It's going going to be consistent for like, 2 models into the future

It's not good at adding spoken stuff, it's super hit or miss

sadly

[proper berry]

you might try recording the part spoken

[torn spoke]

True

that would be interesting

to see if it can essentially do a voice changer

[wraith seal]

Yes, that's what I'm looking at now, recording it separately

[proper berry]

and try cranking the audio influence up real high, 90%... 100%

more to the point, experiment with audio influence and be willing to crank it up to all the way

[clear robin]

Does suno now make AI music videos when making songs? I need to know because i got the membership again today and saw people posting videos and stuff but i do not know. Would appreciate the help.

[south crystal]

I make my videos using Gemini Flow Veo platform. I have a few up on youtube already, it's fairly easy. I think Suno just makes a short looping video.

if you want I can DM you the link to one of them, so u can see what Gemini Flow does

[proper berry]

Randy do you edit afterwards? I find I have to

[wraith seal]

I'd be interested. I've used Flow a bit, but interested to see how you can use it using existing audio

[clear robin]

Sure

[torn spoke]

Anyone else experience suno adding random artifacts to audio, specifically vocals, when seperating them out? It's fkn infuriating

Random burst of "woosh"

like 200ms of someone blowing into the mic

[polar spade]

No. Suno does not create or edit videos. All videos you see are uploaded.

[south crystal]

Sent u a few

[true carbon]

Can's seems to find the answer, if i have a premium plan and i don't use all my credit in a given month, do they carry?

[unique stone]

They don't.

[south crystal]

yes i'm attempting to do what someone suggested, and pull another vocal stem.

I might also try generating a new vocal stem but I love the original vocals so I dono yet.

I remastered this song in Studio and added a way better bass line, some keyboard riffs, but the original vocal stem got some artifacts chilling lol

[torn spoke]

I literally tried 3 times to extract the vocal stems from the song, it created the artifacts every single time

last hail mary, gonna download the original as wav

reupload it

and try to extract again

[south crystal]

i wish you luck. I'm currently attempting to generate a new vocal line with 90% inspiration

lol when i recreated the voicals the first time it just recreated a new version of the entire song on the vocal line

attempting again

[torn spoke]

it STILL does it. at the EXACT same spot

you kidding me

what is the point of extracting the stems if it breaks them

I guess I'll try to extract the stems with a different program, and then I can upload them to suno studio in post

I guess I'll technically save credits on that

lol

[red crown]

https://www.youtube.com/watch?v=_riKXE86Nlk&list=RD_riKXE86Nlk&start_radio=1 can anyone help me out with getting a punk band with hevy accent on vocals like this

[torn spoke]

yeah, add the year the song was released in the prompt

like, add the period into the prompt

If you don't know it, give the song title to gpt and have it do sum research and come back to you with the period and style

[south crystal]

I agree with that sentiment

please fix vocal stems @cedar bramble ❤️

Also when I am prompting new Vocal lines in Studio, it keeps generating entire songs on the new vocal line, or like the last one is just string instruments and no singing

[torn spoke]

make sure to select the section of the track to generate on

[south crystal]

I did, I selected about 20 bars and generated 4 new vocal segments

all of them were jacked up, only 1 had singing, and was wrong lyrics

[torn spoke]

o SHI- one of the versions of my stem extractions inside suno didn't include the artifacts

such pog

[south crystal]

Next attempt, instead of generating individual stems, I'm generating Vocal + Instrumental stems. Let's see if the main and backup vocals combined will fix the artifacts

[torn spoke]

should be noted that none of the other services I tried to use to extract the vocal stems created any artifacts

[south crystal]

lol the original artifacts went away, but new artifacts appeared

in good news I can cut and splice the stems together without the artifacts

hahahha

hilarious

[torn spoke]

silver lining

[red crown]

thanks guy gpt nailed it when i asked it to include the years the band was popular

[south crystal]

ironically somehow the new combined vocal stem has some random faint synths at like bar 74

[torn spoke]

you're welcome <3

[torn spoke]

that strat works for most things btw, if you want a particular sound, find a popular song/band that made that song in that period, and include that period in the prompt.

Oh, and area

Include the area as well

[red crown]

hmm i made a persona witht the vocal stem hoping that would put that ccent in other songs but it seems to be ignorin it

[torn spoke]

I guess increase the audio influence, and also include the period/area in the prompt

[red crown]

when i do psychobilly it auto thinks i need a southern usa style voice,

[torn spoke]

so specify the area where the accent you want is

[torn spoke]

u watched aot?

Holy sh I just had the most extreme case of voice stem destruction

We need a name for this type of artifact

[spring wedge]

Ok can anyone tell me why the hell my perfect made tracks from yesterday sound iggy today
Is this a new way for suno to get people generate/use more money so they mess up people's perfectly made songs ????

[south crystal]

Trying to generate new Synth stems, it keeps generating the entire song on the synth line.

This is my prompt lol:

Synth Stem Style: Roland Alpha Juno-style synth stab; heavy gated reverb, 80s techno.

[vague palm]

hey everyone I dont know if this is the appropriate place to post this but I'm having trouble trying to get songs that uses the voice I input in my persona (my own voice), is this even possible with this?

[azure crane]

I use the extend and cover functions and with V5 it goes completely crazy after say 1.5 minutes... like 10 musicians playing solos at the same time. It happens almost every time I extend for one of the created tracks. This is a bug that needs to be fixed...

[south crystal]

200 credits later, fresh off the Studio with a new Bass and Piano line... see what u think

https://suno.com/s/6458jYh9dwaWhxk6

I was unsuccessful in trying to change the synth line, after a bunch of credits in the trash the studio just kept generating synth lines with the entire song baked into them, not synths lol

[past forge]

@shy yew

💀

!help

[hollow hare]

Please post your help question to #archived-ticket-help
Note there is a 2 hr cool down during weekends (if it is not a prompt question for here)

[rustic shale]

Not sure of the right place for this feedback, I feel like v5 is step back in a couple of ways. I’m pretty new to Suno, but this is my impression:

• it seems to ignore anything I specify as an [Intro] and plays very similar guitar riffs and vocalizations for every song
• male voices seem to ignore my prompts and they all sound very similar, I’m having trouble making distinct male voices
• I’ve had much better luck with these things in 4.5
• some of this may be new user learning curve, but as I said 4.5 seems to work better

[hollow hare]

Feedback about v5 would ideally be through https://suno.com/feedback according to #📢┃announcements message

[rustic shale]

Thank you

[hollow hare]

1. Yeah, [introduction] can be a pain to get at times. Note that that the AI is quite flexible with commands (through square brackets []), though ideally you want to use musical terms.

2. Try adding emotions to help distinguish the voice, or a city/country/location to get regional accents, or try to at least. E.g. [Smooth British Male]

Tip: If you want backup vocals also or echoes, use parenthesis (), e.g. Echo! (Ecchhhoooooo)

[pine island]

Hi! Is there a way to use my voice in Suno studio direclty?

[rustic shale]

I’ve been using things like [Intro, drum only] or [Intro, drum solo] but getting guitar riffs

[proper berry]

Sure, just record with a microphone or upload a file

[rustic shale]

I’ve also tried various versions of [British male vocals] and gotten something that sounds like Nickelback every time 🤷‍♂️
Still experimenting though

[hollow hare]

Wow, talk about Suno rubbing salt in your wound 😆

[hard beacon]

"Live Concert prompt" with v5 is lit 🙂

[hollow hare]

But yeah, seems like the old vocal tricks is not working, hm

[pine island]

But I have to record all the song or i can sing just a piece?

[rustic shale]

I’m also looking for a prompt/tag for hard rock singing without getting hoarse metal screams

[proper berry]

you can just sing a fragment

[pine island]

So I go in studio and sing a segment and then?

[proper berry]

the lack of documentation is a problem but this video shows what to do. i'm sure they'll make it easier.

https://www.youtube.com/watch?v=qR4BefPvSiI

[pine island]

I see this video is not corse to me how use my voice to sing a generated song…

[proper berry]

do you want to add your voice to a song or do you want to create a song based on something you sing in

[vague palm]

so far I have not managed to get my voice properly into suno, I even extracted the voice stem on a full song I sang, not even close

[proper berry]

did you sing directly into suno or did you record something locally and upload it?

[pine island]

I want to add my voice as lead voice of a song

[vague palm]

in my case, I have the cubase file, muted all instruments and left only the voice, then upload it into suno

[proper berry]

what do you get when you upload the whole song from cubase including your singing, as an audio upload, and then you create covers from it

[vague palm]

checking

using cover, it creates the whole back of instruments, notes are sang very accurately but the voice texture, tone and interpretation are very different from the original, so its still far from my original voice

[raven turtle]

I get the same thing from uploading WAVs of my Logic Pro work and hit cover.

[proper berry]

ok well if you don't want it to embellish your singing then i would recommend moving your song from Suno to Cubase and adding the vocals as post production.

you might try setting audio influence to 90% or 100%, but it is still going to recreate your vocals, although they will probably be a lot closer

[raven turtle]

I’ve found stem splitting the vox out and using a voice cloning Audimee or Kits.AI to not be clean enough either. I’ve found best is as said, stem out the instrumental back into DAW and lay the original vox from there back over

[proper berry]

there ya go

[raven turtle]

It does mean you have to have a decent original vox stem in your DAW, as opposed to trying to grab the vox stem from Suno as with the best will in the world, the vocal audio stem just has too many artifacts for a clean stem if the backing instrumentation is too full

[proper berry]

perhaps some day Suno will be able to generate the stems separately rather than trying to extract them from a mix

[raven turtle]

Agreed! Or maybe it’ll allow some voice cloning of yourself at some point.
Or will feature Kit.ai style singer choice for consistency. Particularly if you could blend singers like kits allows to enable you get a vocalist that’s a close approximation to your own voice.
I managed to do this and it works pretty well in Kits. Although I prefer Audimee generally

[vague palm]

the most close that I´ve been is to use Extend, but of course it feels like a part of the original sound

[halcyon sparrow]

yes i do techno hard techno,acidcore, acid techno, hard acid techno and all of my intro are really similar looks the same each songs even with trying to change ambiance theme and intro in the prompt

[true carbon]

First track done at 100% in studio https://suno.com/s/4rq9xNVO7JMKcJZq do you think it would be possible to add vocals on that in Studio? If so did you try to add vocals on an instrumtal beat and what was the result ?

[raven turtle]

In this case I’ve used extend and started with the prompt [stop]. Then extended from there. Then chop off the beginning. You get a similar vocal sound with new instrumental bed. It’s been quite effective as it keeps the timbre generally

[proper berry]

Try covering your song and paste your lyrics into the lyrics box

it will change your song but i have been pretty blown away by how good a job it does. if it mutates your beats too much then try pushing up audio influence, dialing down weirdness, and keeping the audio prompt close to what you used

[true carbon]

thanks

[jolly seal]

Hello

[warm jolt]

Hey there, I am not an audio engineer, but I noticed that the song always sounds worse if I download it via stems (doesn't matter if 2 or 8 stems). When I add the stems into my DAW, I can clearly hear where I regenerated parts of the song or small hiccups in the vocal stem. If I play the song without stem separation in the browser it is not that much noticeable. Is this my fault and I should just try to master the full song without separation in my DAW?

Thanks for any hints on this. Most guides recommend to use stems in a DAW but if I do so, it seems that I am starting far behind of what Suno gives me "out if the box" in the browser.

[proper berry]

Hi. I think this is just a function of that the stem splitter can't 100% accurately distinguish what's what among overlapping parts. The simpler the song, the better job it will do. Think Led Zeppelin versus some big complicated Frank Zappa piece.

[warm jolt]

Hm

[proper berry]

Stem splitters include Suno Studio, lalal.ai, and later versions of Logic Pro have stem splitters. They all have their strengths and weaknesses.

[wispy laurel]

Suno stems are imperfect. Almost always there are some artifacts and glitches that aren't audible in the full song. You can try regenerating the stems which might help. Or you could try downloading the whole song and using different software to split the stems, and see if it comes out better than Suno

[warm jolt]

Okay that sounds not like a real plan for the solution. I might be better up using just the final song without stems

[proper berry]

that's right

[wispy laurel]

That might also be the case. A few times I found it easier to just do what I could with the whole song rather than using faulty stems. Varies case by case, unfortunately there isn't really anything we can do to improve the quality of the stems so you do have to make the best of what you have sometimes

[warm jolt]

Okay thanks @wispy laurel @proper berry for your thoughts on this matter!

[raven turtle]

I’ve tried both methods and found remastering in suno to get it as close as possible and then drop it into a DAW for polishing up. I tend to use Shadow Hills and some Ozone while using a reference curve in Tonal Balance control

[queen wyvern]

Can anyone answer? On website https://suno.com/pricing
Pro is $6/mo
Premier Plan LIMITED TIME OFFER $12/mo
IS it the case that if I refund the Pro plan which I paid $96 on 8.30.25???…and get the Premier Plan which would include studio.??.. And pay an annual fee? If so what would that be?

[proper berry]

Hi Shane, it will prorate you what's left of your existing Pro plan and then charge you the difference if you buy a year on the limited time offer premier plan. it will make it all clear before you have to hit "buy"

[silent mortar]

Can you specify syllables in the prompt?

[hexed oriole]

You'll only pay the difference if you upgrade. I did that myself.

[proper berry]

in the prompt or in the lyrics?

[silent mortar]

Like often times I find the AI merges 2 lines and ruins the structure

[proper berry]

Yah sure, i know what you mean, or it mispronounces a word. you can spell phonetically. insert dashes, play all sorts of games. like if you're trying to say "gametes" but it says "gay meats" you can spell it "gam eats". PS don't cancel me bro

[silent mortar]

Like instead of it's going to be me. It's going to be may

[proper berry]

i'd try "mee"

[waxen minnow]

Hey all, Hope I'm posting on the right thread I'm not super hip to being social.. My biggest question with the studio rollout is the MIDI file. I use superior drummer in my DAW, none of the drum tracks I've downloaded natively work. Is there something I'm missing here?

[silent mortar]

Yeah. I think it's broken

[jovial yew]

I can see there's been some discussion in the past, but has anyone had more recent success with kids voices? Pre-teen preferably (it's a thank you song for my kid's kindergarten teacher who had to retire due to illness). Seems easy enough for choral voices, but hit and miss for individual boy and girl

[hidden yarrow]

I heard Suno v5 does it better now. I guess you can just try out your prompts again...

[jovial yew]

I'll report back 🫡

[sour wyvern]

suno studio Error recreating clip No focused clip????

[manic estuary]

is there any way to get rid of the vocal intros it always adds now?

[fading pendant]

I'm making a piano piece, and the noise like noise is too much, is there a solution?

[proper berry]

hard to say, do you want to share your stuff? but you can try adding to the prompt something like "state of the art recording technology" or "highest possible recording fidelity", smetimes that sort of thing helps

[fleet jacinth]

Anyone getting really poor quality vocals for the first few words or first line then it's fine. Also getting periodic issues with volume just being quiet and getting extremely loud out of nowhere

[calm moon]

Hello Everyone!
I'm diving deep into Suno's capabilities for a complex theatrical score and need some technical advice on implementing dynamic style changes within a single song.

1. Genre Transitions Method:

What is the current most effective method for reliably transitioning the primary genre/style mid-song? For example, moving from a [Jazz Trio] accompaniment straight into a [Full Cinematic Orchestral Score].

Should the new genre be inserted directly into the lyrics prompt using brackets (e.g., [Transition to Salsa rhythm]), or is the Style Prompt the only place to set the initial style, and we just rely on instrument tags for changes? Is there an optimal syntax that reliably forces a style shift between sections (e.g., between [Verse] and a custom [Part X])?

2. Advanced Prompting Resources:
I'm looking to master both the Style Prompt and the use of inline instructions within the lyrics (like [tempo change], [acoustic guitar breakdown], etc.).

Does the community have a go-to resource, wiki, or advanced guide/documentation that details:

• A comprehensive list of tested style tags and synonyms?
• Rules for the maximum complexity or length of inline instructions?
• Best practices for directing voice/singing styles?

Any links or shared knowledge would be hugely appreciated! Thanks in advance for the technical guidance. 🧠🎶

[tawdry scarab]

Something that comes to mind is mentioning that the more change you intend to make the more you may need to be willing to take 1 to 3 second extensions even if you get long ones, if you don't get any sign of what you want to change to. V5 seems to make this a little harder than it used to be, but making several very small extensions, repeating the lyrics for the next part each time, it still seems to get the job done when it won't otherwise listen to what you want

Sanity the workflow channel has community resources in the form of different people's perspectives on how to do things, there are some consistencies, but no rules

[untold briar]

It's harder but also easier because there's new tools like styles influence slider, Inspo, The ability to combine different features, ensuno studio to set up context

[bronze helm]

@steady solstice I figured that this channel would be a better fit for your question

[steady solstice]

ok, Suffian. Thanks for your help. Yet no one has replied.

I got Suno to talk like a little girl, but not the "high pitch girl tone".

[untold briar]

This is a long time gap in Suno's official help. I think it's partly because AI models are strange and Suno users often discover better ways to things than the developers intended, but I also think Suno could help get people pointed in the right direction even with just a lot of good examples. Sometimes I find an incredibly helpful example like this one from 4.5 https://suno.com/song/3e96d1ee-d6ed-41c7-bc07-89ca7d65e0e7?sh=BL8kRYWp5NCx3TBT from Suno team

[calm moon]

As a power user, I have to point out a major hurdle for truly maximizing Suno: the huge gap in official documentation for proper prompting.

It’s incredibly frustrating. We have all these powerful model versions (v5, Expert Mode, Advanced Version, Intelligent Propmpt) but zero clear-cut, official rules on what controls what within the prompt system and how.

Right now, the entire AI Engine relies entirely on what individual users happen to discover through pure trial-and-error (like specific instrument tags, tempo shifts, or forced stops). Then, the community shares these findings, and Suno essentially validates them after the fact.

The development team needs to release a proper Technical Wiki detailing the prompt syntax, parameters, and limitations. Relying solely on user discovery, while cool for the community, is an inefficient way to truly unlock the creative potential. We need to know the intended rules.

Are there any other advanced users out there who feel the same way? What’s the biggest "undocumented rule" you wish the Suno team would clarify? Let's talk about it. 🧠

[warm jolt]

I wonder how valuable Suno studio then really is after finding out that creating stems is causing the song to have artifacts or creating volume issues.

[hearty moat]

Funny you should say this because for the first time today I started building my prompts in chatgpt, more so around creating the lyrics first based on a GENRE of music and adding in artists I wanted to replicate the sound of - I've been getting great results for sure NOT using Chatgpt since I began using SUNO about 3-4 months ago and a lot of my best songs came from using the option in SUNO that would re-write my STYLE input.

But today I feel like I got the most out of SUNO than I ever have using Chatgpt first.

• Prompting CHATGPT to become an expert in the genre of music I am looking to produce - From the history, festivals, specific artists etc Almost like becoming a historian of the style of music you were going for. This really helps because then chatgpt will begin to suggests songs from artists or mention a moment from a particular album, and you didn't even ask chatgpt to do this!
• Then I got chatgpt to write the lyrics for my song idea - Based off what it now knows about the genre of music
• I then got chatgpt to write production notes for the song (1000 character) for the style section literally describing the song from start to finish and how all the instruments should be played, how they should sound, vocals, everything!

I then cranked up the style metre to 100% and kept the weirdness around 50%.

The biggest most noticable difference was the quality of the song, the way it was written, the vocal and the lyrics were much more in line with the genre of music, the lyrics felt written from the heart, not slapped together by a.i

And yeah you're right - It's all been trial an error from the start, burning through credits haha!

[calm moon]

That is absolutely brilliant! And, funnily enough, that’s almost exactly the strategy I started using to overcome the documentation void!

I began tasking every available LLM—ChatGPT, Gemini, Grok, Perplexity—to scour the internet for user comments, feedback, forum posts, and know-how specifically on Suno prompting tips. Initially, they would give me generic nonsense, but after correcting and refining them dozens of times, I managed to get solid results and formulate my own set of internal "rules."

It sounds like we've both landed on the same conclusion: The trick is outsourcing the historical and structural expertise of the genre to a highly trained model, and then feeding that context to Suno. It makes total sense that this results in better lyrics, writing, and vocals that feel "written from the heart." The quality jump you described is exactly what I've been seeing!

(And yes, full transparency: while I speak English fluently, I use AI to save time and to help me articulate my most technical or passionate responses, because I love being specific, here we say: "Ir directo al hueso"... Taking it directly to the bone..

I would be more than happy to jump on a voice chat with you and anyone else to learn and share methods and opinions on this. I love music with all my being—it’s what saved my life, honestly—and a tool like Suno arrived at a point where I truly needed it. I’m incredibly grateful for what it lets us create.

Thanks for sharing your process! It validates my own path of trial and error (and credit burning, haha!). 😅

[pallid inlet]

hey, is it possible to take a acapella & use sunp studio to create music that fits the style of the acapella?

[hexed oriole]

I haven't done it myself, but I've seen some videos on YouTube of people singing into their phones or a mic, then uploading it and Suno improves their vocals and wraps the song they generate around it.

[naive ibex]

Regarding pricing. I would like to pay upfront for a new year, but there is no option. And system is kind of "weird" that there is no option to create an account with another credentials (ie user/password). So, the question is, if I cancel my current plan, or better say will not renew and create a new one (with another email) for the next year, will I keep all my created music on not paid (not renewed) account?

And what that "limited time offer" timeframe?

Should I cancel current plan, and it will be still active untile renew date, and will able to purchase a new year plan in that case? Should be an option "pay upfront", why it's more complicated as supposed to be...

[half dragon]

Can't you just upgrade your current plan?
What do you see when you click Upgrade? I see the option via web interface.

[polar spade]

This may help part of your question.

From: #archived-ticket-help message

Hey everyone!

We’re so excited for you to check out Suno Studio! We're receiving a high volume of inquiries about pricing, etc, but let us assure you that if you've reached us via email before the sale end date of October 5, we will honor the pricing for you.

If you haven't emailed support about subscribing/upgrading to the Premier plan, please reach [e-mail] and include:

• “Studio Discount” in the subject bar of the email
• The email/phone number associated with your account, and your suno handle (—eg @example_handle) or a song link from the account you need help with
• A screenshot of the discount pricing found on suno.com/account

If you have sent in a support ticket:

• Please avoid sending multiple new emails, as this may delay our response by requiring us to reroute and merge duplicate tickets.
• If you did not include the details mentioned above, feel free to gather them so you're ready when you hear back from the support team!

Thank you for your patience as we get everyone up and running; we can't wait to hear what you make with Studio!

[quiet solar]

Yeah - the broad tag "children's music" works fairly well.

In a test using v4.5 to help someone else out I was able to get it to pretty consistently use child singers with the more detailed style prompt: "Upbeat, innocent and playful children’s music, filled with bright melodies and simple cheerful rhythms, The song should be mid-tempo, based around simple acoustic guitar strumming with straightforward percussion and incidental use of Instruments such as clapping, tambourine and xylophone, The vocals should be youthful and cheery, sung by a child vocalist with group harmonies"

[quiet solar]

There are no "official" prompts. Suno treats anything included within [ ] as style prompts/suggestions, but it's not a pre-coded list with preset syntax and parameters and rules. You can put whatever you like in there, and the AI will try to find reference sources to interpret what it is you want.

As a result, prompts that are more 'music-reference friendly' like [Chorus] or [Rapped by a smooth male alto] are typically more successful than [Singer yowls like a raccoon that just burned its paws on a hot barbeque]. It's also influenced by the genre as a whole - trying to introduce musical elements that are very atypical of the genre tend to involve more trial and error, whereas elements that are common are more likely to influence the output (e.g. including a [rapped] section within a Nu Metal track typically goes smoothly, but doing the same in a country song can sometimes cause some weird genre shifts).

Suno also does the same for guidance within the text of the lyrics, where part or all of a line being in all capitals may be shouted or given greater emphasis; words or a line in quotes may be handled differently (e.g. another singer diving in with a different cadence), and anything in () being handed to the backing singer.

Again, to the best of my knowledge that's a reflection of the model learning how lyric notation is used to describe what's happening in the song rather than being hardcoded. So it won't always work like an instruction in a programming language, but the AI is able to pick up that you're asking it to do something different and will often try to reflect that in the output.

[dawn kernel]

I have lyrics for a male-female duet, but the problem is that almost every time one part, like a verse meant for the male, ends up being sung by the female and vice versa, and I’ve marked it in brackets like [chorus her part]. Is there maybe a better way I could try?

[pine island]

Any exoerience to create personas with own voice to create cover with own voice? Possibile?

[cobalt pollen]

You nailed it, there's not ever going to be a 100% effective prompt...to me, it seems like SUNO prioritizes trying to grasp the concept you want, themes, settings, styles, genres, content, verbiage, looks at the project from all angles simultaneously.

I got this format from ChatGPT back in the day...like 8 months ago lol

In Lyric prompt for vocals:

to prompts the whole chorus, verse, or any section:

(Vocals: Male Vocals explode aggressively rapping each line in rapid-fire cadence, with clean emotive deliver)

You just swap male for Female in that same format.

Prompts for specific lines - I use the same exact format, but put the prompt on the same line as the lyrics, pasted at the very beginning...i don't separate the prompt from the line...

(Vocals: Female Vocals screams the corresponding line...or whatever)

The most effective vocal prompt format I have been using for months now and it's pretty accurate and effective most of the time...sometimes SUNO just says F your Street Rapper, I'm making this into an all female acoustic folk ballad...

[cobalt pollen]

It wont make a persona if there ANYTHING that wasn't created by SUNO 100%...as far as I've seen

[dusky sundial]

What is the prompt to make suno write a solo concert for a single instrument?

[sharp merlin]

PSA: if you wanna look up style tags based on artist this website is amazing: https://rateyourmusic.com/
i suspect the old heads will know this one already

[polar spade]

If i remember tonight, I may try uploading a kids song i did with just a young boy singing. But he was about 12.

I have a few kids choir songs and persona, but only a couple of songs with just an interjection by a singgle kid vvoice i think. That may be the best way though. Generate many kids songs with call and response and prompts for kids lines. Until you get some single kid voices.

Edit and merge only the single kids lines and put in a Playlist to try to inspire more of the sound or just make a persona from and see if that helps "vibe" them into the output you need.

[polar spade]

I just dont rely solely on prompts much now. And all depends on what your final product is you want.

• I'd prompt for the instrument, add "solo" , quartets, baroque, whatever works. to get one to four instruments.
• Stem out the tracks.
• Find your solo instrument track and try extending it to have a longer track.
• If preferred, make a persona from that.

If its a concert, then why not try a Simple create. Ask Suno to "write a concerto for solo cello"

Cover it again to exclude the instruments, percussion, etc you dont want.

Both of these are 15% effective, but I dont know what a "solo instrument concert" means to you as well as what sub level you have.

[ornate nimbus]

Hi everyone, a technical question. I'm working on a very long track and I'm having trouble with the Extend option. The generated tracks are completely bugged; there's no text tracking, they're randomly inserted at every point during the extension, and the text is gibberish, as if in a different language. Sometimes I had to create tracks from three parts and didn't have any such issues, but now every generated extend is broken. Does anyone have a similar problem or know why this is happening? Version v5, of course. I'm concerned about these problems because I've used up a lot of credits, and I care so much about this track that I don't know what to do next. I've stalled my progress, so I decided to post here because maybe someone will know how to help me, or the technical department can investigate why I and perhaps others are experiencing these problems. Please excuse any spelling errors, but I'm using a translator.

[abstract sable]

I actually have questions on how to properly use the extend tool myself. In some cases, it works and gets something I want, but in a lot of cases it doesn't. But in my case, as someone who really likes Prog Rock/Metal and more integrate movements in movements, I would really love to be able to extend a specific section instead of just the end, or be able to add extra sections between sections in Suno Studio but I've not yet figured out how to properly do this and get something usable. It seems like the model just loses the flow when you do it

[torn spoke]

Anyone have any tips on how to avoid suno adding artifacts to vocal stems?

During extraction

[wet dune]

I've heard a number of theories on quality may be effected, ie- Suno generates higher quality tracks when there is low demand on the server/hardware. I could see some of these being true, or totally being the result of bias, like so many other things around prompting.

I am always shocked at how quickly I can go from hitting "generate" to "play" on a track (usually a few seconds), and then the song starts playing, and clearly finishes rendering while I listen. I was curious if anyone thought (or knew) if it might be better to not hit "play" right away, and let the song finish generating first, potentially giving it more time to edit the beginning or the song, or use more computational power.

[torn spoke]

The whole reason it is able to let you listen as it generates is because it's doesn't edit what it's already generated

[raven turtle]

I think there's a lot of truth in this. Particulalry in light of the implication of the mooted 'Max Mode'. As server resources are pushed, the point would be that more credits bumps more resources to the generation. So the implication is yes, server load can affect the quality of what you get.

[polar spade]

If we use personas with our extends, we can have better influence on making instrumental, conceptual and tempo changes , especially with prog rock concepts.

Openly, I haven't check if inspire works with extends. If so, thats another avenue.

[ornate nimbus]

Up

[polar spade]

At some point, theres a tip of the scale where the cost (assisting with $20 worth of an answer, working out the issues, figuring out why your present configuration is bugging out, trying various possible steps, redirecting with a different idea) exceeds the effort to resolve an issue. If your song is hovering in the double digits, and your still working on it, it may be time to move up a level for an extra $20 for one month and just paste tracks into the Studio sequencer, clean up all your other outputs as well, make some new killer songs, all in a couple of weeks, downgrade back to Pro mid-month before the next renewal and call it a day.

[torn spoke]

What exactly is the point of having the option to generate specific instruments when it's unable to only make that one instrument? I was so hyped for studio. so hyped! I watched the youtube video, and it looks so good.

[ornate nimbus]

Hmmm, I'm still using the Premier version and have access to Suno Studio, but I'm not entirely sure what its capabilities are because I haven't tested it yet and I'm not sure I'll be able to handle it. I wanted to test the studio once I had the entire song ready, but I can't extend it, and the lyrics are long, so the entire song will probably be over 10 minutes long.

[torn spoke]

And now I'm realising how broken it is. The generation features don't work consistently, the stem separation introduces artefacts and essentially cripples the song.

I could forgive the generation flaws if it didn't consistently introduce random wooshing noises to the vocal stems, cause at least then I could use studio as a way to copy paste things

Such a let down man

Recreating a violin section shouldn't introduce additional guitar.

Adding a drum segment shouldn't introduce synth.

Man it's a rollercoaster of emotion for me lol

Though, I do suppose I could combine this with adobe audition, and use audition for the actual audio editing, but it's such a shame because suno's beat syncronisation is actually pretty good

and saves a lot of time

[abstract sable]

I haven't fully figured personas out yet. I've attempted to use them with 4.5 but would generate songs that sound way too similar. Haven't experimented with it yet in v5

[torn spoke]

I mean, turn down audio influence

[polar spade]

Thata a tough call, as extends keep using the extend that came before, so that doesnt seem an issue but the song itself is over 8 minutes already amd we keep regenerating it to add another extension. Im stupid at that point to know more.

At least when extending, assuming youre using one verse at a time (?), instead try duplicating the verse a few times, pick the best ones (while in Edit UI) and remove the rest? That may same some time at least, until you learn more of Studio and use it instead to splice up your extends on different tracks to then extend off of separate tracks to hopefully help keep it from getting buggy.