📚┃suno-school

[oak meadow]

I was confused because it says 10 running jobs and two running jobs and there's nowhere I seen it clarifies what that means

[near flume]

You can have 10 versions being run at one time, 5 click generations.

[oak meadow]

Oh so you can just spam five clicks? But that's all. I think I'm understanding now 😅 I don't think I'm going to waste my credits like that. I typically put a lot into my prompt. Anyways thank you for clarifying.

[oak meadow]

It'd be nice if they add a feature like slot machine selection where you could generate five instances from one click

[near flume]

It's better the way it is incase you have a spelling error in the lyrics or styles. You'll only have 2 bad tracks to deal with.

[oak meadow]

That makes sense. Again thank you for the help. I have a much better understanding now.

[polar spade]

As I do, initially, as I'm finding a prompt that gives a good vibe, I'll then click off 10 gens total rather than 1 at a time. As most sound pretty good, it helps me not stop at the first good one I hear.

[hollow hare]

I feel attacked 😆 (the amount of times I had a typo in my generations...)

[polar spade]

Of course, that's on the premier plan.

[polar spade]

Oh yea.... and make sure everything is spelled correctly before clicking off 10 generations.

[tranquil dune]

how do you clear an ai's voice

https://suno.com/song/90da813c-751f-47df-8974-b16b3f554f66 listen to this

WHYY

please help i want to post some on youtube

[safe hazel]

Does anyone know what this vocal effect is called, found here at time 2:50 of this song (used on the word "I'm) https://youtu.be/Oxd6pXSYkzE?t=170

and how to do it in Suno

[restive fern]

you can use Clear Vocals in the music style prompt

[keen phoenix]

This my old Turkish song ;> https://suno.com/song/e38ad2b2-da4a-4f14-85a8-bd40bef2aaac

[tranquil dune]

this brings back so many memories

[safe hazel]

I love that song 😆

[polar spade]

Any gravelly voice is called , in the USA, " vocal fry"

[polar spade]

In the UK it's called something else. I've forgot

[safe hazel]

I always thought of vocal fry as like a higher pitched valley girl kinda voice

[polar spade]

The vocal fry is when you add a glottle stop to any voice to where the vocal cords are slower vibrating than the pitch they are intended

It will cause that sound whether you are speaking low speaking slowly or speaking, high pitched or screaming

[safe hazel]

wow glottle stop was my nickname in highschool! (this is a joke btw)

Thank you Dasvon, I will now attempt to create a Suno vocal fry!

[polar spade]

If my punctuation is missing, it's because i'm doing voice to text while driving

Probably not a good thing

[safe hazel]

probably

I just thought glottle stop was a funny sounding word

[polar spade]

If the text reads, ahhhhhhhhh, then i've hit a light pole

[safe hazel]

I will have my phone ready to call the police (and make a suno song about it)

[polar spade]

Lol

[whole comet]

So - what's the difference between "vocal fry" and full on false-fold screaming ( a la Jinjer or Arch Enemy )?

And has anyone replicated Arch Enemy style vocals in Suno? 🤣

[naive quail]

I'd say the difference is vocal cord engagement. With vocal fry, the singer is still utilizing their vocal cords - they're just relaxed and vibrating more slowly. You can still hear the natural tone of the singer's voice, just in a lower register. With false-cord screaming, the vocal cords are much less engaged and the resulting vocalization obscures their natural voice.

[whole comet]

So they are more using their vocal tract to shape and manipulate air tract resonances, than using vocal chords for tone generation? Which would explain why some female metal singers "scream" so low.

I'd love to see that in Suno 😄 If it were possible, may need to so some Arch Enemy inspired tracks 😄

[naive quail]

how low can suno go!? 😄

[safe hazel]

Vocal Fry attempts unsuccessful 😊

[modest coral]

Asking for Male vocals (or writing it in any form vocals, voice, ect) isnt working anymore. It also gives me random duets of male and females too.
I wrote it so many ways like

like verse 1 male voice or verse one Male vocals..of course I did the [...] just writing what I did in summary form am going insane!!!

100 point wasted I cant complain as their free

but I still pissy

[safe hazel]

putting male/female vocals in the lyrics box never really worked. It's more receptive in the Style box.

[modest coral]

What am saying is, I did do that.

sorry I wasnt clear. I mean I tried ALL! ways possible nothing is working anymore.

also what is up with the reddit suno..do you guys know if. THEY ARE SUCH ASSES THERE!
posts always thumb down to -0 and there is never any help.

I wanted to ask this for a while, its so sad

[whole comet]

I doubt that's an issue specific to the Suno reddit 😜 Still - sorry to hear of your experiences.

[modest coral]

yea. That was just an after thought ❤️
Am getting so sick and tired of the rudeness there!!!

I just snapped, I wanted to know if you guys are aware of that ❤️

[naive quail]

Still fun, tho. 😄 https://suno.com/song/bbad168b-4ddf-4e69-8875-abf0a595e39b

and the guitar makes up for the lack of vocal fry and false-fold 😄

[timid wasp]

Exactly the same experience. Though, sometimes there are some bright people spark, but overall it's not that kind I'd recommend to anyone.

[shut moat]

ya'll should try fake languages,

https://suno.com/song/1de39ab3-8e02-48cf-a877-60ccb1790455

[polar spade]

Uh oh. Have to think of styles that use vocal fry. Or just prompt "Tom Waites" haha. He's a popular folk singer who's voice got worse through the years.

[rich timber]

Why does my recent two songs sound familiar with a recorded song? It feels like dejavu.

https://suno.com/song/c85204cb-ba35-432f-a5f2-26f0a7e692d5

I can't place the other song anywhere.

[safe hazel]

ooooh that is a purty nifty guitar

[upbeat glade]

https://suno.com/song/c453e523-c9b9-4451-920c-a12803e4ab10

Anyone else having this issue? I downloaded the song and it's definitely a suno issue and not a playback one. The audio randomly fades out and back in during the song, and this isn't the first time it's happened and I thought the first time was a fluke.

[rich timber]

Yes, I have a recording that faded out towards the end. I don't know how I can fix it 😕 😔

[upbeat glade]

It isn't even that, after the chorus of my song plays it fades out and back in during the instrumental transition part

[rich timber]

I noticed. Cool 😎 song 🎵

[upbeat glade]

Yeah, cool if it didn't mess up lol

[rich timber]

Right

[safe hazel]

It might be time to pull out the big guns and use "glottalization" 👀

[north sleet]

I had Pi.ai talking with a valley girl vocal fry vernacular. It was wonderful.

[undone star]

only option is to take a point before the fade and extend it. if its mid-verse you'll have to cut at a point witthout or prior. kinda the only way.

[ruby garnet]

Hi! How do I unsubscribe from the newsletter «$10.00 payment to Suno was unsuccessful again»?

[hollow hare]

If you meant subscription of the Service, you need to contact [e-mail]

[lusty grove]

Hello , Will Suno provide an API in the coming months or is this not on the agenda at all?

[normal ridge]

I just tried to login to the new APP using my Google connect. It asked me to make a new username and I did. It seems to have put that new username on all my songs I’ve shared before… but when I login to the website or the app none of my songs are there, my subscription credits are gone, and it displays untitled. Is this just a migration thing I need to wait for because of the username change or did I break the system somehow?

[silent pulsar]

Hi does anybody know how can I limit the length of the song ai is creating? it always makes songs long as it wants, is there a way to limit the length?

[north sleet]

Can depend on the style, lyrics etc how long it generates. Some generations will be short and others will slam up against the 4 minute mark for v3.5.

You could also try to create verse-by-verse and use extend to build your song. Then just end it when you want it to end for your last part before you get the whole song.

[lime sundial]

Hi there,
I'm trying to extend a song after a verse, so my extension need to start with a chorus.
Suno just skips it, resulting in 2 verses without chorus in-between them.

I'm thinking of trying to put an "empty verse", but am not sure I can avoid to get some weird things doing that.
Any efficient trick ?

[strange bough]

Did you try to extend before the previous verse?

[lime sundial]

no.
But from what I tried, voice/tempo/instruments go wrong already most of the time when extending,

It looks like the first lyrics of the extensions have the same issues as the first one of the song often have, and basically rarely match styles with the rest

So I try to limit the number of needed extensions.

--
I'd just like Suno to respect what is written in the lyrics box

putting a verse with [Silence] as lyrics made the chorus reappear, but it put some instrumental here, and there's n,o way I use it without editing, which I can't do beyond cutting and appending parts, since my only knowledge of music tools is some ffmpeg basic usage

[earnest palm]

not really a technical error https://suno.com/song/ad89f76b-2bd6-4e3f-a94a-57455ce46ad2

but a question ... why is it trying to start a movie at 3:54 ?

[north sleet]

Cause v3.5 doesn't like to end.

[warm ember]

Hello community, i'm trying to update/regenerate the same song with only updated lyrics but i don't know how.
I tried "reuse prompt", "edit displayed lyrics", "regenerate video"
It's this video : https://suno.com/song/8d9bd3cd-3f9d-48e6-884d-731dfd89a651
The lyrics u can read are the correct ones, i would like the same in the music ...
Any idea plz ? 🙏

[cunning mulch]

Uh... you can't do that with the vocals. I have a song much like that just waiting for the same tools

Also, have they changed how a song's tagged when using the 'v3 Cutoff' (Extend in v3 with just 'End' in the style prompt and tags in lyrics)? It used to make the song say it was all done in v3, but not any longer

[past flame]

I can't seem to post my song in suno-showcase. the screen shakes. anyone know why?

[wind panther]

Sorry If this has been asked before, and if so could someone direct me to it. I thought I'd read somewhere that you could go back and edit the lyrics of a song that was created without changing the style or sound of that song (for say a miss spelled word or a word that doesn't quite sound right etc) But I can't seem to find that anywhere. thanks

[frozen pivot]

This is not something you can do at this time. We all want this feature referred to as “inpainting”.

We are looking forward to that in a future release.

[wind panther]

okay i must have mistaken on a youtube video that i thought they had mentioned being able to do that and thought i was missing something.

[safe hazel]

all you can do is edit the way the lyrics are written out, not edit the song audio

[wind panther]

right I knew that but was wondering if they had it where the song (sound and style) doesn't change just the edited lyrics but I guess that feature isn't out yet as @frozen pivot mentioned above.

[modest coral]

an getting so angry!!! Am glad I have not subbed yet,
as so many points would be lost. Am trying to make male vocals.
I have tired everything, everyone has told me. Its not working.
What is going on. I have male gender songs I want to make, that will not work with female voice.
anyone else getting this problem.

[white yacht]

I had a quick question.. is it possible to add a tag inside a song for it to (drop the bass)? 😹🙏 Thx

[cunning mulch]

Why not give [Bass Drop] a try?

[hot trout]

Pardon me, but can anyone tell me what the everloving brain batter "Prompt flagged for moderation" means, and what EXACTLY Suno AI thinks it is doing? Especially because it only started this nonsense after I had done more than fifty generations of the same song and was finally closing in on structural perfection for it?

[hollow hare]

It means that the lyrics that ChatGPT had generated had been censored by itself (literally)

[hot trout]

I'm not using generated lyrics.

I write all my own material.

There were exactly ZERO lyrical changes. Just structural ones. And suddenly, it's lost it's mind.

[hollow hare]

On that case, it means ChatGPT found something objective in the lyrics you provided.

Don't forget, AI is dynamic, not static, so its interpretation of it when you submitted it had RNG'd significantly than the earlier submissions

There are some stuff ChatGPT really does not like, e.g. when you talk about bones and ribs as I found out once before
(e.g. when describing a necromancer hoard), ChatGPT assumes that I was talking about a massacre (which, yes - that is likely how they got so many bodies - but no, the lyrics did not involve any explicit murders other than implied dungeon creatures being defeated)

Ugh, so many typos z.z

[hot trout]

Funny thing is that after I reverted the structural change, it was happy to work properly again... and in all honesty, given that the song I am working on is pirate metal, you'd EXPECT that there would be some things that some delusional butt-monkey somewhere MIGHT find objectionable, but not really so much.

[prisma wraith]

In dire need of assistance, the vocal type that joins in starting at 0:08, what would one have to add to the style tags to increase the odds of getting it in other various songs?
https://suno.com/song/be14c0cd-b88e-44c1-8c2f-5b6560c64c77

[hot trout]

In your style prompt, I would try something like "raspy voice" "harsh vocals", something like that. Play around with it! I've got a few thousand credits to burn; would you like me to run a test or three?

[prisma wraith]

thanks for the response, Imma give it a handful or gens myself, and will give you another reply if it was unsuccessful

It managed to give a similar one through the use of harsh voice, although it was only one gen out of 10, so I'm unsure if it'd be possible to call it a success

[hollow sapphire]

🇫🇷 Hello, I have a question, and to be honest, I'm wondering: I edited a song that was in progress, just to tag it as v 1.0, which is normal so far. But when exploring more than usual, to my great surprise, I discovered that I could upload an image and change the cover art. So I did! But is this normal for a "Free" user?

[hot trout]

Sometimes that happens. Maybe try "distorted" or "scratchy" instead? Different keywords can make all the difference. I haven't played with this one yet, but I hope you find something more dependable.

[safe hazel]

Are you using custom mode? What exactly are you using as your Style prompt?

[cunning mulch]

Hmmm... anyone happen to know what this hallucination is (and what it might be saying)?

it came up right at the start of a song that was in all English (but I wanted an Indian voice)

[safe hazel]

it's probably not saying anything real, but I would say it sounds like Japanese

[cunning mulch]

well, I'm thinking either Hindi or Punjabi, considering I prompted for the latter as a vocalist (and clearly got it)

[safe hazel]

probably then

one time I got an entire song that was just a long hallucination

None of the actual lyrics were in it

[cunning mulch]

but, if it is, it would be cool to know what it's saying

[safe hazel]

yeah, but, it's probably just nonsense words that sound like Hindi, just like we get nonsense words that sound like English. I got some Korean sounding hallucinations before, I doubt it was actually real words though

(I was making a k-pop song)

[charred merlin]

This looks like it was updated recently 🙂

Pictures of the paid version and the free version 🙏 (oh heck, I don't know if this was announced -- this was a surprise feature drop 👀)

Thanks to the team for this Quality of Life update for free users

🏆

Pro-tip for new users with the ability to upload custom images -- design your photos for 4:3 aspect ratio, then resize the canvas to 1:1 so you have the black bars beside your photo.

The bar's won't show up in the discord video embed, and as an added bonus, you can write hidden messages in the black bars to the side that only people will only see if they actually click play 😉

Sharing this song as a demonstration of the technique
https://suno.com/song/364a225f-20ce-4538-aeb7-627c07b55c53

[grave wraith]

As I know here are very techy people with the same interests 🙂 Anyone think he can make it better from the beginning or repair what is left?

USed UVR 5 but I can't get that mid to end part good...you hear extreme that something was cut out

[weary wharf]

Is this the original output from Suno?

[weary wharf]

Is this a stem?

There is a lot of data missing here, and I hear a vocal in the background, I assume this is a stem.

Red - This part has missing harmonics, which cannot be recovered, no way to correct.
Blue - Too much noise, probably because of the hihats bleeding
Yellow - Metallic sounds, mostly happens because of stem seperation, some data moves to other layers -- will sound ok with other layers, but if you clear anything, it will break other layers.

[grave wraith]

This is the stem after separating instrumental and vocals. I need a nice instrumental and I can't get it done without breaking the instruments in the "quite" parts

[weary wharf]

Can you share the original wav?

[grave wraith]

Sure

[weary wharf]

FYI... The problem with stem separation is harmonics on top of each other. Let's say you have violins and vocals. If the vocals start hitting the same frequencies, we like to hear vocals, so strings are "muted". Means, with stem separation the instruments that are muted won't be there and cannot be recovered. That's why seperation is not a good solution in most cases.

[merry nymph]

Does suno understands in a prompt BMP and B♭ major/B♭ minor?

[weary wharf]

Sometimes it does. You can give it a starting note with a piano intro maybe, that might lock what you want to do... if you know how to of course.

[merry nymph]

I noticed that quite often the tempo goes wild without any reson.

[grave wraith]

Well I understand that the artifacts I face are kinda overlapping frequencies. Just wonder if there is some better combination of AI models maybe that separate them cleaner without killing the guitar and bass in the 2nd part of the song. I think, but it's hard to hear that there is some kind of synthesizer in parts and that is clashing with the vocals, unfortunately

[weary wharf]

That's Suno... Try extending before the change.

[merry nymph]

It happens mostly during extending

[grave wraith]

And I understand that a Suno song is no clean mastered pro song 🙂 Where different channels are used and are more clear to separate

[weary wharf]

I have similar problems with symphonic songs... Thin sounds tend to break vocals.

[weary wharf]

Try moving 1-2 seconds before/after. Probably you are hitting something that breaks Suno. Happens a lot.

[grave wraith]

But appreciate that image...I haven't learned that graphic yet completely

[weary wharf]

Before stem seperation first do a fast mastering. Then bounce it and try stem seperation on that file. That might help.

[grave wraith]

🙂 Still learning 😄 I did that on that stem you heard

[weary wharf]

Spectrogram is pretty easy to learn actually. Try working on single instruments - piano, violins, guitar, vocals alone... you'll understand what is what when you see it in clear picture.

If it looks like a "midi" layout, it's most probably a piano or a synth. If it's wiggly + has a lot of harmonics it's a string instrument, etc...
If it has some lines in the low range, and some "blurred" placements around 5k/8k/12k hz, that's probably someone singing and you see "S", "T", "P" and some other consonants.

[grave wraith]

I have a feeling you use a better app for this 😄

[weary wharf]

RX 11...

[grave wraith]

It looks like this in Audition

[charred merlin]

#1256301881966198805 message

Not to the best of my knowledge. But you can use free web-based tools like https://www.bandlab.com/splitter to clip out 60s of a song that you like, set the key and bpm, then feed it back into Suno' audio upload.

speeding up for no reason
I wouldn't say "for no reason", the devs have pointed out that everything Suno does is determined by SOMETHING you've used as input -- even with zero style instructions it infers song style from the lyrics.

I've noticed that simply trying to prompt for "120 bpm" isn't recognized as a command, but likely tells Suno "bpm" as a style prompt -- while that's not a genre in itself, it might infer from "bpm" that it needs to pick up the speed, because "bpm" is largely represented in songs where changing tempos are prevalent. Just my observations 🙏

[grave wraith]

The spectral editor?

[weary wharf]

Here... https://www.izotope.com/en/products/rx.html

[weary wharf]

That looks like a low-res problem... You can update the resolution in the settings. Try 4096 and Hann...

@grave wraith this is a cleaned and mastered output. Try this one.

Try adding a "short soft reverb"... that might help with the muted vocals.

[grave wraith]

Yeah...I know RX11 but using the spectral editor gives me something about no host 😄 I guess it is not in the essential pack 😄

I have a feeling Audition just shows it bad. That's with your settings

[weary wharf]

If you zoom in to much spectrogram will not make too much sense anymore. If it's not a really high bpm song, it will be lines from start to end.

[grave wraith]

I see 🙂 Yeah, I think I got it better now 😄 Wow, I see things I never seen before 😄

RTFM they said

[fickle lake]

lets put prompt on the comment section on mp3.... agree?

[weary wharf]

Let's use AAC instead of mp3 😛

[fickle lake]

you mean flacc?

but imagine this... your song prompt is in the comment tag in mp3

[weary wharf]

flac is lossless. aac is like mp3, lossy but better.

[weary wharf]

Not using mp3s 😛

[fickle lake]

i lnow man

but mp3 supports tags like comments

[weary wharf]

But yep, that would be good anyway. I would not like to loose the style and lyrics I used.

[fickle lake]

so you agree that when you dl a song on suno, the mp3 comment tag would have the prompt in it?

[weary wharf]

Yep, that would be nice.

[charred merlin]

That sounds like a helpful idea, even if just buried in the metadata -- #💡┃suggestions-web is the channel the devs browse for ideas on how the improve the model, so it might be a good place to share your idea 😄 💪

[grave wraith]

Wow... teach me all your secrets 😄 Marry me 😄 Do something 😂

It is really a lot better

[fickle lake]

yes mr bot... please put that idea in the suggestions... i did put it 30 times there, and they didnt care...

[charred merlin]

The ideas don't get implemented immediately, and they might not reach out to you directly if they use the idea -- you're just leaving your helpful ideas in a suggestion box, like they have a mom & pop restaurants 🙏

Instead of releasing tiny updates in piecemeal, they might be doing a LOT of stuff in the background -- there's no way to know.

That said, there's the known-suggestions channel where the guides and moderators can list commonly suggested ideas to make it easier for the devs to see the good ones > https://discord.com/channels/1069381916492562582/1252298754787377225

[cold zenith]

DO you know how to maximize the chance to get a clear voice ? because sometimes the voices are clear in some generation and sometimes it is not.

[weary wharf]

Here are your stems...
https://we.tl/t-GaI7i1bd4S

The link will expire in 2-3 days. Hope this helps...

[fickle lake]

so where or who should i send this idea then?

this idea is really simple... its like 10 lines of code

[simple osprey]

where can i open a ticket for billing stuff?
is pinging team an option here?

[charred merlin]

No official support presence on Discord for first-contact issues -- you need to email [e-mail]. 🙏

[simple osprey]

ok ty!

But is it normal that the credits reset?

[charred merlin]

100%, they don't stack.

[simple osprey]

bummer

had around 1000 left and now i only have 375 extra credits...
i mean i paid for it

but ok

[charred merlin]

Just checking -- do you still have a paid subscription?

[simple osprey]

yes

[charred merlin]

Ooooh, I thought you meant that you went from 1000 subscription credits to having 0, and only seeing the extra credits. 😅 my bad

[simple osprey]

ah na i had 1000 monthly left and now instead of having 3500 i have those 2500. 😄 Sorry for the misunderstanding

[vestal rapids]

I really need to make more songs

[simple osprey]

where did u get those 10k credits!

[charred merlin]

You pay for your allotted usage during the month. Free users reset to 50 credits every day, and when paid users blow through the entire month's all at once, they get the free daily cap until their monthly allowance kicks in again.

I paid for it
100% legit -- and I used to try to spend all of my monthly credits in the last couple of days before the reset (that's when I would do a crazy amount of experimenting, and get a lot of instrumental intros to build from).

It is sort of a pain in the ass, but they don't have a lot of text on the https://www.suno.com/me page, and one of the few sentences explains how the credits don't carry over from month to month -- so less of "they pulled a fast one on us" and more of "damn, I was so excited I didn't even read to learn the fine print"

[vestal rapids]

it was compensation

[simple osprey]

ah ok yea.. i see.. its okay and understandable but still a bummer 😄

[simple osprey]

could use that too 😄

[charred merlin]

As far as I recall, it was specifically compensation for raising concerns regarding accessibility.

Just before everyone starts asking how they can get free credits, I really don't think that's a common thing, but I'm just a simple country robot.

[simple osprey]

😄

yea it's fine

i have some credits now

[vestal rapids]

yes, I basically had a month I paid for i couldn't use while I got someone to help me with the new interface, so Suno compensated me

[charred merlin]

I think I made 5 songs in August, so I'm sort of glad I don't know how many credits I forgot to spend. 😅

[simple osprey]

[charred merlin]

I never said I was a clever man.

[simple osprey]

How do you make songs?
Im going for a some text and then generated and hear how it works out.
Do you write all down or first instrumental?

i dont get that instrumental thing first and then text thing

[charred merlin]

Unless I'm making a goofy song or something just intended as a "one-shot generation", I almost always use instrumental to begin with. That way I can choose exactly when to insert the lyrics.

Also, I don't have comparative tests, but I just feel like the lyrics come out more appropriate for the song when v3.5 extends from a clip with a strong enough melodic hook, opposed to having it try to create the vocals at the same time as the rest of the song -- it's basically another instrument, so I'm just using an extension to choose when to insert that stem, and it feels to fit more natural to my ears.

[simple osprey]

Okay, how excatly are you doing this?

Someone told me to find an instrumental that i like and then extend it from 00:01 with my lyrics. But that didnt work most of the time

[charred merlin]

Extending from 0:01 leaves only a single second of the original track, then continues with yours -- accidental 'Terminator Extension', in that you only give the extension a single second of context to "listen to and extend from", so most likely those sound like ENTIRELY different songs, right?

[simple osprey]

moooost of the time yeas

[charred merlin]

If I make an instrumental song, and decide that it would sound better with vocals at 0:20, I place an extension at 0:18 or something -- that gives me;

part 1 - 0:00 > 0:20 -- cool instrumental intro
part 2 - 0:20 > 2:20 -- extensions can only be half as long (because they're spending half of their brain power listening to the second they extend from!)

But lets imagine "part 2" sings the chorus at 1:20, but keeps repeating it until 2:00 -- we can add another extension to wrap it up

part 3 - 1:30 > 2:30 -- we wait until the chorus is sung properly, then we add another extension to tell Suno what to sing CORRECTLY, unless we want another instrumental section.

[simple osprey]

Holy Moly

and im doing spray and pray until i get a version i like

[naive quail]

It's a trap!

Soon you'll be extending every 20 seconds. 😄

[charred merlin]

If you're interested in knowing more about wtf I mean by 'Terminator Extensions', I talk about it here in my pool party thread > #1256301881966198805 message there's some helpful song examples for ending songs or swapping genres INSTANTLY (well, within a second)

[simple osprey]

can i do stuff like electric callboy then?

from metal to hardstyle and back

[charred merlin]

lmao, Toker just shared the picture I was going to show -- goofy songs & one-shot generations are one thing, but I guarantee that almost every song that Toker or I feel comfortable enough to share have at LEAST that many extensions as he showed

[charred merlin]

oh hell yeah we can, watch the song example in the terminator extension section for a demonstration ❤️

At this point, I've stopped wondering what Suno can do, and now I'm more actively curious what it's NOT capable of. These are the conversations that make me drop everything.

[naive quail]

I wish it was easier to pull out the sound effects

[simple osprey]

which song excatly?

[charred merlin]

But as a caveat -- the more specific of a song you want to create, the less likely you're be to create solely through the use of complex prompting.

[charred merlin]

i linked to it in the pool party thread above

oh hell, it's called "simple sunovating" isn't it. Well ain't THAT a misnomer

[simple osprey]

uuuuh niceeee

[frozen pivot]

Sometimes, I feel as if I am one of the oldest here 😁👍🏾
It's all good

[simple osprey]

What do i write into lyrics and style in my terminator-extension?

[charred merlin]

Wrong thread?

[frozen pivot]

Only if it is not technical discussion

[charred merlin]

Sorry, wasn't quite sure what you were replying to, and figured that was meant for gen-chat or off-topic, since it didn't quite seem to apply to the conversation here 🙂

Not a worry at all, hope you're having a wonderful time 🙂

[frozen pivot]

Well, my take was how we use to create music versus how we create music now…

Just a thought

[charred merlin]

Technically you can use anything, because you're only using a single second of it 💪 The terminator extension doesn't contribute to the song style, but by extending from it and only leaving a single second of context, you prevent part 3 from "hearing" part 1, because there's "something" in between that it needs to extend from, but since it only has a single second of context to listen to, you can practically pivot the entire song's style on a dime

[simple osprey]

ahhhh yea

i see

was brain afk

[charred merlin]

I wrote those posts in the pool party what seems like a LONG time ago in Suno-days, and there's been VERY little refinement.

The ideas are there, but I wouldn't be surprised if makes zero sense to the unfamiliar reader 😅

[fickle lake]

can we please have prompts in the mp3 tags

[charred merlin]

I'm here if you have any questions though

[simple osprey]

do i still say [Verse 2] in part 2?

[simple osprey]

i will have a lot i guess

[charred merlin]

Sorry for leaving the post hanging earlier, but all we can do is bring our suggestions to #💡┃suggestions-web -- I know you've brought it up there 30 times and they haven't implemented it, but it's not that you've asked incorrectly, or haven't asked enough -- the devs are aware of the idea, and it's whether or not they decide to use the idea.

If you really want a direct response on the matter, you could email [e-mail] and let them know your idea -- but fair warning, they're under no obligation to use your idea, and you may be likely to get a form response similar to "Thanks for your idea, we're always trying to improve services and will take your suggestion under consideration"

[fickle lake]

yeahh... do you know the joke?

[charred merlin]

There's the #1252298754787377225 channel where the guides and devs highlight the most common suggestions -- it doesn't look like they've added your suggestion to the list though.

Perhaps you could ask them to make a post for your suggestion, and then you could ask others to leave upvotes on it so the devs know how important of an issue this is to everyone.

It's not that they don't value your idea, and it's not that it's not important, but there are MANY gears turning, and things don't always happen at a rate that feels satisfying. But if your idea has merit, and you stick with it, there's a chance it will be used.

[simple osprey]

That change of style didnt work well

https://suno.com/song/67552354-89ce-4fad-b2a9-7cc810fdee98

min 01:01

[fickle lake]

yes...

[fickle lake]

sent the message now

[charred merlin]

I have a little bit of time before dinner, so I'll give it a quick try and see what I'm able to come up with.

From what I can see, acoustic fingerpicked until the last 20 seconds and then blast out with punk?

[fickle lake]

you are part of the devs tho?

[simple osprey]

yea, but it was just a try! Don't waste time on that one 😄 i was just curious 😄

initially i wanted to stay with fingerpicked the whole song

[charred merlin]

Not in the slightest -- just a user like yourself 🙂

[fickle lake]

but you have powers

[charred merlin]

Out of curiosity, what powers do you think I have?
Not upset, genuinely curious so I can help alleviate confusion 🙂

[fickle lake]

you have the powers to ban people here...

[charred merlin]

No, I don't 😮

[fickle lake]

your nick name color is shit for colorblinded people like i

[charred merlin]

Alpha users have NO powers in Suno discord.

[fickle lake]

isnt alpha the best?

[fickle lake]

send your most best song

[simple osprey]

That's "Admin"

[charred merlin]

We only help test new features before they are released.

Only the moderators can ban people.

[fickle lake]

send your best sond you ever made on suno

[charred merlin]

If you want a mod song, you'll need to talk to the mods, there's a couple who post their work here frequently 🙂 My mistake, I thought that said "send your mods best song" 😅 👨‍🦯

[fickle lake]

oh... can i be one of them?

[simple osprey]

https://suno.com/playlist/0f90f0f4-50ac-459c-bd39-865c0beb4239

Here are my public songs feel free to listen to them

[fickle lake]

i just want a song of you

[fickle lake]

did you make those all songs

[simple osprey]

yes

[charred merlin]

Sorry, I'm the wrong person to ask 😅 That's something you could reach out to [e-mail] about as well!

They never told me why they picked me as an alpha user, so I don't know how to give advice for that.

[fickle lake]

fuger you! you are great

[simple osprey]

Thanks. But most of them arent that good

[fickle lake]

why did they pick you tho

[fickle lake]

send more shaisse

[simple osprey]

wat

[fickle lake]

shaisse means "good stuff"

i cant say "s***t" here

its so stupid...

[simple osprey]

hey harber.

https://suno.com/song/aeaa20e9-e719-4a59-8661-2b244ca2ff46

when u jump here to 02:52 the song is over but suno decides to make another song there.
how can i end the song excatly there?

[fickle lake]

in finnish severs we talk with swear words all the time... nobody cares

[fickle lake]

listening...

[fickle lake]

i added you as friend

[simple osprey]

thats nice but please forgive me i don't add anyone i just met 😄 Sorry

[charred merlin]

Taking a stab at this one before I step out for food

[fickle lake]

but you talk to people irl who speak on your face...

[simple osprey]

ty ty

but when u have time can u look at this one?
#📚┃suno-school message

[simple osprey]

and im talking to you

[fickle lake]

yeaghhh

[obtuse ridge]

Honest Feedback on this Amateur edit of Mine, is it acceptable enough so far?

[pale crescent]

is there any prompt that will stop suno from adding a 30s instrumental after every single chorus

ive burned like 800 credits trying to make this thing stop stuffing as many instrumental breakdowns in as it can

verse 1 - chorus [30s instrumental] verse 2 chorus [30s instrumental bridge [30s instrumental] final chorus [30s instrumental] final verse.. its like having some 14 year old who learned guitar trying to collab

[charred merlin]

Okay, watching Gardeners World with @delicate idol after dinner, but I'll reply to your DM so I remember to circle back and get that terminator extension to work -- I looked at the example you gave and it looks more or less*** exactly*** like I would have suggested, and whilethere was a notable change, it still retained too much of the original track, or didn't pivot quite as hard as either of us seemed to have wanted.

Even if it was just a fun example, I'm still happy to dig into it to see what we'd need to get it to work (heck, might even be better to start off with a song that we're just playing with, so the expectations are low early-doors, and we can try to build on those successes)

[glacial scarab]

did you already try not having empty lines between parts?

[pale crescent]

i will give that a shot

nope it utterly refuses to complete a verse without jamming a instrumental in there

v3 will do it, but it takes the beat in a different direction every time haha

[charred merlin]

Tangentially related, but interesting enough, imo -- someone pointed out that replacing the spaces between words with hyphens seems to suggest to Suno that you want it suno quickly, or with a tighter flow.

Suno will sing this at a normal rate
But-it-will-sing-this-quickly-though

[glacial scarab]

ah good to know i sometimes used this to make it do that with 2 words.. never thought about trying with a whole line

[glacial scarab]

Aw man.. hmm.. no then i have no real idea except to try different formattings and stuff until it somehow stops doing that.. it can be very stubborn sometimes

[charred merlin]

There's a LOT of different input factors that impact what Suno spits out for you -- not saying that anyone will have a 100% answer on what will work for certain, but without knowing stuff like what lyrics / metatags / styles you're using, whether you're making a fresh song or extending from something (extending from full song or extension matters as well), using v2 / v3 / v3.5 (they all have strengths and weaknesses)... hard to give a clear answer as to how to create what you want.

Without having investigated or checked if this is a good example, something as simple as leaving ellipses like "..." at the end of a verse might suggest to Suno to let the music drag out -- there's SO many degrees of experimentation that go into creating the specific songs you want.

Like, when you say you spent 800 credits, that's 80 "generate" clicks, -- what have you tried different between those, and did you notice ANY differences between them? (which should teach something by itself, if not)

[pale crescent]

i wanted to redo it from chorus 1 because i noticed if you dont label [chorus 1] [chorus 2] etc and instead do [chorus] then [chrous 2] it will miss the chorus 2 and get lost, i revised the lyrics and it kept singing the old lyrics, or it would realize its error and just get hung up singing the original verse then hanging singing the new revision a few times before continuing

[charred merlin]

I've rarely use structure tags in my songs with vocals, and it usually infers it perfectly from the song structure (provided I'm continuining from something with v3.5 -- v3 is just wild. Creative, but wild).

Not to put them down, I know they help suggest song structure just the same that any prompt can suggest something, but I haven't found it to be a requirement by any means -- but I'd be happy to test something out if anyone has strong feelings otherwise! 😄

[glacial scarab]

That sounds like you might extend from a point where suno doesn't really know where it is in the song. Do you extend after the full chorus 1 is sung or from within it?

[pale crescent]

yea i tried to extend a few times

if i look at the full lyrics its a mess lmao

i might just end up taking a bunch of these and throwing them in ableton and seeing if i can patch something up

[charred merlin]

Reading my mind -- it might be a forced analogy, but because v3.5 craves song structure so much, and can tend to drift toward "generic, on-track" songs without efficient use of prompting, or without a good existing melody to hook onto, I like to imagine it as being on a highway looking for "offramps" -- there's natural offramps, and sometimes it can predict the natural offramp -- so for whatever reason if it's convinced that the best musical decision is to try to resolve the melody, it will keep trying to steer toward that upcoming offramp.

Unless you apply a degree of musical theory as to WHEN you introduce an extension point, sometimes it can feel like shots in the dark trying to move the extension point around.

Of course, even WITH musical theory...

[pale crescent]

yea i have another track im trying to piece together, i got most of it together but the opening is off i figured i would let that one sit for a bit while i figure out what i wanna do, so i started on this one lol

[glacial scarab]

There are generally 2 ways to extend. One is after all lyrics of the part are sung. So on an instrumental or short silence.

The second one is to extend mid line.

Sometimes one work better then the other depending on song. The first one is pretty self explanatory I guess.. just extend after a part is finished singing paste in the new lyrics hit extend.

If that doesn't work i do the 2nd method. Go back a few seconds so it's mid line or even mid word and put the lyrics like this:
Line that has alread been sung
Line that is currently been sung
New Lyrics

[pale crescent]

yea there's 2 extensions, you can hear it fumble the 00:34 but the other one isnt terrible, another song i was stiching it up outside suno but there's slight variations int eh vocals that are noticable

but fair, just a thing to deal with i guess

[pale crescent]

wow, i gave it a bit and now all of a sudden its listening with no spaces thank you guys so much

[tall bone]

How can I record and use my own voice in place of the auto generated voice? I wrote lyrics and picked the style and song sounds good, but I want my own vocal model to be used instead? Is that possible?

I’ve recorded a 60 second clip in the app and just puts my recording in the beginning of the song and instrumentals the rest of the way. That is the opposite of what I would like

[safe hazel]

are you putting in the lyrics you want into the lyric box?

[tall bone]

Yeah

[fickle lake]

you know what...

use the mp3 shaisse to use prompts on comments

[empty jungle]

You could generate a song and then get the instrumental stem and sing over it maybe?

or just do an instrumental song

[tranquil dune]

okay this is what i got for you i decided to to some rhyming words the used chat gpt because idk man i picked the genre so https://suno.com/song/7a6032de-406f-4aa1-ab4f-538c40928203

wrong chat i'm sorry

[pale crescent]

you can use kapwing to combine instrumental and vocals in a bind

[fickle lake]

its good

[fickle lake]

but

[fickle lake]

its just

[flint haven]

Please do not ping someone multiple times Tarino especially when your followup contributions have no substance.

This channel is slow, you can take your time and formulate your response. (Don't worry, you are not being penalized, just advising you that someone can construct multiple subsequent pings that as harrassment).

[Pinged in General by accident]

[graceful condor]

yo

[charred merlin]

I know we're not exactly brainiacs, but you don't need to outright call us slow.

(that was a joke)

[feral pendant]

Is it just me or has the variety in vocalists gone way down in the past few months? I swear, every "male vocalist" sounds basically identical at this time. Trying to find the right tags to do a 1980s hair metal style song, but it's either some 2000s mall rock or brooding synthwave, which doesn't fit with the prompt at all.

[safe hazel]

I think there is a huge variety of voices

[glass jacinth]

You can specify for example [jamaican vocalist] and [loose pronounciation]

At least I did both of those in one song. Maybe just the Jamaican part took care of the pronounciation.

[wheat prism]

Hi everyone)
Could you please advise me what I need to do to make sure that the new song cover I uploaded, as well as all the changed to new covers in each part of the song, are reflected correctly when I send a link to my song?

I'm also not very clear on how long it takes before the old covers are no longer reflected and are correctly replaced by the new covers I uploaded?

[hollow hare]

I assume you mean by cover, you mean the cover image of the song? Or of the video?

When you link the song on Discord, it caches the link you initially used. To forcefully change it, the easiest method would be to add an / to the end of the URL so Discord treats it as a new link.

On the website, it should load the new image quickly once you refresh the page. If it does not, you can use a cache-bypass reload (shortcuts on desktop: #📚┃suno-school message) otherwise you may need to delete your cache

[remote dome]

Is it just me or is there a lot more "Screeching echo" in the generations the past 2 weeks? 😦

[spring spoke]

oh snap, you can brain learn here? tight

[sudden bough]

I dunno if anyone ever came up with that, but some old male\female voice could be good to prompt to. Or there is options for that?

Maybe chant male\female to. Because i find no way to prompt the AI to make choir\chorus\chant sadly. Like the instruments. Rare so much when the AI actually recognize the instrument prompt.

Also when you moved something to the trash, the message about the action is appear after a pretty long time. On phone it is pretty annoying.

[atomic iris]

Spent damn near 800 credits trying to get the song to have a MALE vocal, but everytime, EVERY. SINGLE. TIME. im getting a female. Does anyone know how to get it to stop feeding me female vocals repeatedly?

[safe hazel]

put "Male Vocals" in the style prompt in custom mode

[polar spade]

You want a choir sound without music, or looking for harmonies?

[sudden bough]

Yeah, not works.

[polar spade]

Anytime something repeats, it's helps to remove other prompts which might be controlling /overriding what you're trying to do .....to at least see which one is prominently controlling the outcome

[sudden bough]

Obviously with music. Sometimes accidentally it works and i got what i want, but its like total random. More works with aggressive phonk than symphonic, weird... but it is what it is. Same as duett between female and male. But 8 out of 10 times, duetts are really good and keeping them. Its soo rare and soo unique.

[sudden bough]

Spend like 5k credits on songs what just like nowhere close what i want. Really good tactic to waste credits. 😄 I like it, just can't stand it. Every once in a while it pisses me off and leave it for a day or two.

[atomic iris]

Ive done this, 800cr worth lol

[polar spade]

For instance "ballad" or "kpop" gets lots of female voice. "Country" gets slightly more male than female, but close to 50/50. So think if one of your prompts often gets a female voice or associate with feminine qualities. Such as "emotional" ......where "emo" might get male.

[atomic iris]

If i remove the prompt i have, then i dont get the style im looking for :(. i only have one "genre" input

[polar spade]

Nothing is obvious here. Lol. But try, as I often suggest for harmonies, to put "barbershop" towards the end of your prompt.

[polar spade]

That could be an issue. Lol. So what's your one prompt?

[sudden bough]

😄 😄 No... not at all. Emotional Never gets female for me when i have a promt to Male Voice. So not at all. Ballad is not recognizable by the AI, not in the list. Kpop it is, but that is not the one what i looking for. Atcually right now. Also if i want simple speak without music, that is easy to achieve. Just like use less prompts is better, not confuse the AI too much.

[sudden bough]

Well, some things are. Like we all make music. At least i think this part is should be obvious. Or not? 😄

[polar spade]

Emoti9nal = maybe female Emotional + Male vocal = maybe male. Got it.

[sudden bough]

Emotional, Male Voice, works, there is no Vocal by the AI so pointless to use. Emotional, Female Voice, also works. There is no Male Vocal also. Male Chant or Female Chant is works often on the other hand.

[polar spade]

Seems prominent to have music. But, Things like gregorian chants, chapel choir, a capella, and others will get vocals without music. It's just a cornucopeia of options out there.

[sudden bough]

Example, if i say, Cold, Sad, Symphonic, Male Voice, Male Chant, Ethereal, it always ALWAYS makes songs what it exactly suggest. Never done anything else.

[sudden bough]

Yes, they do, it is right, but! They often collide with others like Pop, Rap, Hip-Hop, and mostly never figures out what or how to do so cling to one from the two.

[sudden bough]

Also, it struggle with (text) parts, but really works with metal, Dubstep or genres what have the same music dynamic just like classical or symphonic music have.

It not change the facts. These:
"I dunno if anyone ever came up with that, but some old male\female voice could be good to prompt to. Or there is options for that?
Maybe chant male\female to. Because i find no way to prompt the AI to make choir\chorus\chant sadly. Like the instruments. Rare so much when the AI actually recognize the instrument prompt.
Also when you moved something to the trash, the message about the action is appear after a pretty long time. On phone it is pretty annoying."
Could be fantastic to fix\add.

[polar spade]

What's the one prompt you're using? Suno, also takes into consideration you're style of writing and, I propose, specific phrasing which it may associate with female than male. Or, such as flipping a coin, the results might be close to 50/50 but only if you run a couple of hundred tries.

The specific metaprompt. Bracket/verse/colon/double-space/gender/close bracket [verse1: male] has worked more often for me.

Even the double spacing, from my trials, seem to make a difference. SUuno prefers proper grammar, else it ignores it or sings it , instead.

[safe hazel]

"Ballad" is definitely recognized

[full apex]

Anyone that can give an example prompt on how to reach the same voice agian? have tried singer: / voice: but even tho i write female i end up with a male 80 % of the time or a duet

[safe hazel]

the only way to get the same voice again is to extend a song that has that voice in it

[full apex]

when i did that and extended from second 1 i got a male instead of the original 😄

so i need to make the extended in the end and then clip it out in audacity afterwards

then how is the transition from the first to the extended? if it is 2 different genres ?

[polar spade]

For what it's worth: in the auto generate lyrics window, I typed "create a song with alternative male and female vocals about love" and clicked generate lyrics and suno began each stanza with [verse 1: male] , [verse 2: female] , and [chorus: both].

So I figured that's a good format if suno writes it that way.

[safe hazel]

from what people commonly write in here though, it seems like that doesn't really work often. But, I agree if that's how suno writes it, it would make sense to copy it

[polar spade]

Oops... sorry my comment wasn't really following the stream of thought y'all were on. Ignore mem.

Me

[safe hazel]

it's good knowledge to have

does suno use it's own lagnuage model to generate lyrics or does it use like chatgpt or something

[polar spade]

I don't write a lot of intended duet love songs, but what songs I have that are in the "romantic" vein, it feels like suno bends towards making them duets. Again, maybe just a placebo effect. I haven't run enough of them to be certain.

[safe hazel]

sometimes i'll get a "duet" that has the first verse in one voice and then all other verses in another. So it's, kind of a duet...

[polar spade]

One clue that suno uses "typical" vocals, is writing Latin dance style songs, no matter whether male or female it has to sing the verses, the chorus is more often, if a "choir" style, male voices. For instance, a mariachi song with female lead, still has male backup vocals. Personally, I've never heard female backups in traditional Latin dance. So , that interpolate into thinking it does that with each "genre" it encounters.

[naive quail]

Then there are those times where you're not even trying for a Duet and Suno gifts you with one. https://suno.com/song/0ff0b618-a2c7-4969-b707-b5b0fea2aa16

[topaz hamlet]

Hi! Is there anyone who experienced Suno generator made only 1 instead of two songs? I mean, it seems an error because all my songs came with two types but only one song generated 1 😦

[jaunty thistle]

Hello guys short question: Offers Suno under the paid option the possibility to download the Singer and the Instrumental seperately ?

[frozen pivot]

Yes!
Get Stems is the option. However, still not a very perfect option

[jaunty thistle]

Allright , thanks

[hoary jolt]

Does anyone know when Suno will be upgrading their Vocals? Compared to Udio's vocals I'd love if Suno could get ahead in that category.

Also, do we think they will have the ability to take the songs that we made, that we fell in love with, and then UPSCALE them similar to upscaling photos with higher quality audio while maintaining the actual song we made and love?

[safe hazel]

This happens to me sometimes as well, it does seem like an error. Maybe it comes about from overloading the server by clicking "generate" to quickly, but I dont know

[languid edge]

can we only download mp3's as paid users now or (more likely) a skill issue on my end or ui change?

@half cedar Did we lose the ability to download anything better than an mp3, sorry to '@' you if so, either way, multiple browsers, I can't get the wav files I paid to generate and need to resolve this

[latent sand]

Hi, does anyone know how I can make the Extend function extend a song by a few minutes instead of short duration. For example, if I have an instrumental that I like and it has 3:30 and I ask it to extend from 3:20, it usually adds 10-30 seconds but not more, I would like to actually extend it by another 3-4 minutes and ideally generate a longer instrumental. Any tips are appreciated!

[safe hazel]

if you try to extend it from a point where there is silence or the song is fading out to end, then the extension will follow that same trajectory. If you want it longer, move the extension point earlier, to where the music is still robust

[latent sand]

Thanks. I'll try!

[full field]

so u know how when u play a song the bar shows up at the bottom with play, rewind, etc anybody know how to close it? it always bugs me cuz it gets in the way

[cold zenith]

well i'm gonna wait for v4, i don't have the choice. There's a music that lenght 7min and the quality decrease too badly. 😭 I hope, V4 will be out soon.

[round bobcat]

Has Suno changed the way it works with uploaded files and then extending them? I have had songs that were "perfect" for my lyrics, but had mispronounced words in it. What I did before, was download the song, edit out a 6 to 60 second clip for export, upload to Suno, then extend and enter lyrics from that point in the song clip.

It used to work fine, but now Suno is outputting a different style from my upload. This is a huge problem for me and I have a potential paying client who loves the version of their song i sent them, but there are mistakes in pronunciation.

Does anyone know how to do this?

[full field]

do what?

[round bobcat]

To correct mistakes in a version of a song where Suno uses the exact same style of my uploaded file

[full field]

reuse prompt?

[round bobcat]

Thank you, I think I've tried this and it didn't work, and creates a whole new style, but lemme try again

[full field]

just put the style in

[round bobcat]

that didn't work. I did put the same style in and got a whole new style for the song. I'm pretty sure I HAVE to use file upload so Suno has a model to work from

But I appreciate you trying to help me rad

[safe hazel]

Suno wont use the "exact" same style, it will create something from the uploaded file that is similar to it, but its interpretation of the uploaded file will likely always be different

[full field]

thats one issue i hate about suno

[round bobcat]

I know I've done it before, sometimes multiple times for one song. It created extended songs with the exact same style for me

I just don't know what I'm missing and not remembering if there's a step I'm failing to do to make this happen. I KNOW it was possible because I almost always have to do it to correct mis pronounced words

[full field]

send 2 examples of the different styles pls so i can see what u mean

[round bobcat]

Here is the song itself https://suno.com/song/9a056a91-9693-49df-9eeb-b532bc90fbcd

[full field]

ok

[round bobcat]

Here's what happens when I use the process I wrote about in my first post https://suno.com/song/0ac37b7b-0e44-43d2-8641-6f7032764453

[final hazel]

You cannot fix mistakes in a song atm

[round bobcat]

The voice is close, but it's way different in the vibe

Okay, thank you 3Daizy, are you a Suno team member?

[final hazel]

I am one of the moderators 🙂

[full field]

found your issue

lol

[round bobcat]

Awesome! Human, I presume? Thank you for weighing in here. I just found out Suno has a Discord group today.....I will be a very active member of this group and hopefully make good contributions

I've done it both ways, with the exact same kind of style paramaters and without

same results

but lemme try again just in case

[full field]

@final hazelso u know how when u play a song the bar shows up at the bottom with play, rewind, etc know how to close it? it always bugs me cuz it gets in the way

[safe hazel]

it bugs me too, but refreshing the page will get rid of it

[final hazel]

Feel free to let the devs know. It's not meant to close afaik, but it is a great #💡┃suggestions-web

[full field]

ok thx

[round bobcat]

I think I'm just going to have to hire a vocalist off Fiverr or Upwork after I remove the vocals in Logic Pro by stem splitting

[full field]

did u put in the emo rock and all that?

[round bobcat]

I did use emo, rock, exact same parameters

same results

[full field]

send the one u made with the same ones

there might be something different u arent seeing

happens to me alot lol

[round bobcat]

LOL,, I can relate! https://suno.com/song/811cb667-1cd7-45ad-8baa-f085170e3974

https://suno.com/song/811cb667-1cd7-45ad-8baa-f085170e3974

[full field]

id report it for not following style

[round bobcat]

Not even close to the original https://suno.com/song/9a056a91-9693-49df-9eeb-b532bc90fbcd

[full field]

and try again

[round bobcat]

Will do it. Thank you so much for your time and efforts

[full field]

np

[round bobcat]

I may just burn up the rest of my 1560 credits for the month using the same style parameters. Who know, I may get an even better version, right?

[full field]

it def does not sound like no emo pop rock

sounds more like hard rock

[round bobcat]

One other quick question, I watched a great YouTube video on how to use custom tags, and he said using "record label quality" increase the output quality. Is this true?

[full field]

no clue

i wouldnt think so

u can try clear vocals

that helps sometimes

[round bobcat]

As a musical style parameter? Thank you, never tried it before. You can hear how clear the vocal is in the original song? I love it and that's what I'm after.

[full field]

ya

np

oh one more thing

who are the guys in the pic?

just ai pic?

oh um i dont think thats allowed to use them

in the pic for your song

ya u need to ask about that

oh u do

howd u get permission?

so how do u know them?

nice

nothing

disabled and cant work

nah

?

ohhh

oh ya. have been since i was little

im a free user so i dont have perms to the songs i've made

which sux cuz i've got alot of good songs

[charred merlin]

Hey man, this might not be the place to self-promote in this way.

[flint haven]

Please do not talk about personal information here.

[round bobcat]

Okay, sorry guys, I'm not intending to promote anything. But I get your point. Mods, feel free to delete anything that falls in the self-promtion category

[full field]

ive made some really funny songs

[round bobcat]

I deleted all of the non-technical posts

[sudden bough]

Here the official list of the genres and styles, singing types the AI in SUNO recognizes. To those who needs some help, and to those who keep trying pointless ones what no works.
https://www.suno.wiki/faq/style-and-lyrics/styles-and-genres/

There is a pretty big Wiki site where dozens and dozens tips and tricks usefull to try if you can't get the One you seek.

[round bobcat]

SWEET!

[full field]

i did a jaws one and at the end theres a guy yelling theres a shark theres a shark then like gets all muffled

https://suno.com/song/0e18d8f3-cad0-4d48-bbb8-b70913ec39b3

[sudden bough]

Just because i read up there somebody keep pushing pointless dreams about one style what SUNO AI not recognizes, but that individual keep saying it is. No it's not. This is an AI, you have to be specific what you want, and not give too much but not too less either. Scripting doesn't care what you believe or want until you be specific about that. Don't waste your credits on pointless tries. Here the help.

[full field]

???

[hollow hare]

1. It is an unofficial wiki made by members of this server.

2. This is not a complete list. Iirc, most of these were identified during 2.0/3.0 (the PDF of the music tags on the wiki is very useful though)

3. 3.5 is quite overtrained to certain genres as you may have noticed. If you are having issues getting a niche genre, try generating in 3.0 as it is better at understanding different genres.

I recommend, in this scenario, starting off as an instrumental in 3.0 then extend into 3.5 for the better song structure

4. There are a lot of subgenres out there, try checking other sites (e.g. wikipedia) for the style most similar to the one you want.

I love using https://rateyourmusic.com, both their genre list, and the description of the songs that they have on that site.

Edit: Oops, sorry for the ping

[sudden bough]

The AI followed the lyrics what maken it to actually coming up something what fits to that duh duh duh. But when you want to make Lyrics, you need prompts. 🙂 But interesting outcome.

[full field]

huh?

i have no idea what u said there lol

[round bobcat]

Love it! very cool

[full field]

love the ending lol

[sudden bough]

I tried the official list of genres and styles but the AI was absolute chaotic and lost in it, so it only have one purpose to keep pushing pointless tries. Wasting credits, thanks for the idea but i Like to get what I want and not waste credits aimlessly. But since i only using the ones up in the list i shared, not have much problem to make the AI understand the style i want. So, yeah, that list what Only works and mostly. 9 times out of 10, the AI doesn't recognize random things and stick to the one it knows.

[sudden bough]

Maybe then read back up what you did there and you going to. Sorry, no time for games.

[full field]

dude im not understanding what u are saying

[sudden bough]

For Example. I was precise with prompts and i get that. Exactly what i wanted. Just have to make it a few times to try because the beats or voices was not what i imagined. But after a few tries it was the One i looked for.

https://suno.com/song/0c2a059b-0d94-4691-ab82-ec80224127a2

[hollow hare]

Like I said, 3.5 is not great at recognizing more niche genres, which 3.0 is good at

Not to mention that the tokenization may play a factor as well. "Dark Trap" could be treated as "Dark (emotion)" + "Trap (genre)", whereas using a hyphen is more likely to treat it as "Dark-Trap (genre)"

The other major issue with AIs is that they are random. For example, there are so many variations of Pop (https://rateyourmusic.com lists 157 subgenres for it) so the odds of getting a specific type of Pop is quite low without narrowing the subgenre (e.g. Dance-Pop).

The method you showed in that song does work quite well when you want to have a more flexible style.

Actually, a lot of people do not know you can influence the music genre through the lyric prompts. Well, most people use it to influence the lyrics (like [Verse X: fast rap])

Though, admittedly, it is better to have it in the style prompt for consistency (if you know the specific subgenre) since the AI can ignore the metatag.

[sudden bough]

Yeah, that is what i mentioned a little up there. The Lyrics matters to, sometimes the AI when not sure or can't recognize, use the lyrics as a guideline.

Also, yes, i am aware that the official list is not really works, that is what i said. A first few times i tried those, but the AI was totally lost. But i was looked around and find that list, since that, not have much problems.

Into the music business for 10 years now, and still amazed about this AI stuff. Keep using the ChatGPT either so the both AI works together pretty well.

Also, a side note. An official scripted genres list could be fantastic, so those who not adept in the world of music could have so help how to achieve what they seek.

[safe hazel]

there is no "offifical list"

[strange mural]

I wish there were "threads" here for different subjects because I feel like this probably has been answered and I am going to be a guy asking an old question. But I need ideas/advice/insight.

I am working on a piece that has a CHANT before the first verse. I have tagged it as a ton of different things but Suno SKIPS is 9 out of 10 times. And maybe 2 out of 10 of those is acceptable. Can you FORCE Suno to include something? Can you put flaming arrows on something and say DON'T MISS THIS PART.

If we could FRONT TRIM stuff I would sandwich it between verses and cut the front, but alas, earwax we don't have that.

[safe hazel]

you can't force Suno to do anything

[strange mural]

LOL true - but not neccesarily what I meant - let me reword. How can I "Enhance the likelihood" it will not skip this entire section and go directly for a verse.

[safe hazel]

how are you writing out the CHANT part?

[full field]

well it shouldnt skip

[strange mural]

It does... I've labeled it INTRO, CHANT, TAG, VAMP, HOOK

Even labeled it VERSE and it skipped to Verse 2 🤦

So I've done it two ways... one in sentence case and once in all upper case... roughly same results

[sudden bough]

Uhhmm this is mostly easy to make it do something like this.
When you write the lyrics, make the fore sentence be like that.
"(your text)
main lyrics"

[sudden bough]

What you put between these () are mostly choir, chant does and the main line stay with the lead sing. I'll show an example. Here.

https://suno.com/song/1c88e8f4-9873-477b-a2c9-8b1a0c3ccc4a

[strange mural]

Ahh like when I do echos, oohs and background vocals.

I'll try that

[sudden bough]

As you see, the chant sing the lines between the (), and the lead singer stay with the main lines.

check it out what i send and you will see.

I used this trick with my Fantasy Series, as the different empires sing their main things, or laws, and at the chorus part if the empire like that, aka Nordic, Zulu or like this Arabian style, they sing with a lead singer and a chorus in the back.

[strange mural]

I don't usually have problems when it is in the middle, but didn't know if Suno had a predisposition to head right into VERSE... and skip stuff. It skips intros for me all the time, ut that is usually less problematic.

Yeah what you did there I have never had trouble doing

[safe hazel]

the [Intro] tag doesnt seem very reliable

[sudden bough]

it do, even if you use, [Instrumental Outro\Intro\Hook] it give a short break only for music hook then start with some really nice start at the next line. It's scripting. Not a biggie. Just have to know what you want and how to add that.

[strange mural]

If it was sandwiched between verses I am almost 100% positive it would do it BUT at the beginning (well its an extend from my intro) it skips it 90% of the time

[sudden bough]

you have to do it with line by line for the maximum effect. Just to be sure.

[full field]

dm me what exactly u put in and i will fix it for u

[round bobcat]

I agree, Suno skips intro much of the time. Since this is all run by software, it's perplexing to me why it can vary so much!

[safe hazel]

think about all the different kinds of music it trains off of. That varies a lot, so Suno varies a lot

[strange mural]

"Why is suno skipping my intro" is a question that will live in infamy alongside "how many licks does it take to get to the center of a tootsie pop" The world will never know

[unique stone]

The best way to ensure you get an intro in your song is to start with an instrumental, and extend off that where you want the lyrics to start.
Uses more credits, but probably less credits than attempting to get it to start at the point you want through prompting alone - plus you also get a general feel for the vibe of the music before committing words to it.

[strange mural]

I think I am resigning myself to just mixing this outside of Suno and using the scraps from different generations. Brew the coffee gonna be a long night. Hope the friend I am making this one for appreciates this.

[strange mural]

Yeah that's what I am doing with this. I wanted an extended Taiko drum intro before the rap and had to do it as an instrumental

Extending from 45 seconds into it, trying to go into a war chant and then into the rap verse...

The only challenge with that is because the Taiko is so dominant in the instrumental on about 50% of the extensions (maybe more like 60%) the vocals are BURIED.

So with a 50% buried vocal rate and a 90% skip the chant rate... I have burned a blooming TON of credits and time on this one.

I'm sure one of these generations the heavens will open and the muse will descend and then I will at least have a good sample to build off of.

or (likely) I will stop this madness, accept its AI and not perfect, and finish the damn song and move on.

[naive quail]

You have to be on the verge of giving up at least three times before the gem reveals itself...

[runic kayak]

oh wow, the word "Nerd" is getting flagged because there's an artist with that name. Same with "Avatar". Wanted to make a song about being the avatar of an old god, but nope.

I get it, some things should be flagged, but these words? really? the filtered word list needs some cleanup.

[safe hazel]

god, Suno is incredible

[strange mural]

LOL ok so finally decided to approach it a different way. Just doing the chant til I get what I want and then extending it - if there's nothing else it can't skip it.

[charred merlin]

Is custom mode blocking you from using those words as lyrics?

afaik, this only happens when you use "artist names" in simple mode -- there's a lot of artist names with unfortunately vague names, so the automated moderation flags a lot of things in Simple mode, but you should be able to use the write about / surprise me tool in the Custom Mode lyrics box to generate lyrics about whatever you want! (within reason, of course)

#1270928841614823546 message 👈 Here's a video showing how the feature works, and here's a known bug that you might run into if you use it 👉 https://discord.com/channels/1069381916492562582/1259902708765167709 (the workaround isn't written for some reason, but it's to just use under 1249 characters before clicking "write about")

[quasi totem]

is there any way to convert .pth and .index files into .npz files?

[dire carbon]

https://suno.com/@mesmerisingstrumming6162

Create//

[hollow hare]

Video Tutorials V3.5 Created by Suno Moderator

ENGLISH: Watch here
العربية (Arabic>): Watch here
NEDERLANDS: Watch here
FRANÇAIS: Watch here
DEUTSCH: Watch here
हिंदी (Hindi): Watch here
INDONESIA: Watch here
ITALIANO: Watch here
日本語 (Japanese): Watch here
한국어 (Korean): Watch here
中文 (Chinese): Watch here
PORTUGUÊS: Watch here
ESPAÑOL: Watch here
TÜRKÇE: Watch here

[dire carbon]

Hii

[astral flame]

are artist names blocked on al lversions?

[charred merlin]

Arguably, as long as they're handling compatible data types, I can't see why you wouldn't be able to reformat it if you knew what you're doing.

That said, Woodhouse yelled at me the last time I went too far off-topic outside of Suno-related stuff, so if a program is asking for a specifically formatted .npz file, probably safer to check that program's documentation. 🙏

Best of luck! 💪

[charred merlin]

#1270928841614823546 message 🙏

[wheat prism]

There are no problems with the display on Suno's website. There are problems when the actual cover art is not reflected when I send a link, whether on Discord or another social network.

The process, as I understand it, should consist of replacing the cover art of the song, including the cover art of the parts that make up the song, for a correct, up-to-date video, which should also be re-generated.

So here's the question: How long does it take for the song video and the song cover reflection to change to the new one?

[keen mural]

Feel like I’ve been blocked from creating. Using a free account because I want Suno to use my tracks. Q: how many songs can one create before the API / creations shut off?

Can’t upload in showcase sorry.

[south willow]

Hello, is there any possibility to insert negative prompts in suno?

[south warren]

Why isn't suno working?

[unique stone]

You get 50 credits that refill daily, which is enough to make 5 pairs of songs each day.

What do you mean by wanting Suno to use your tracks?

[cursive jetty]

its seems to flex.. i always force the regen video manually and then preview before sharing

[wheat prism]

I'm sure you can't do without manual intervention. The main thing is to understand - how many hours/days after the video and cover when sending the link will the recipient see the new video and cover?

[unique abyss]

today i tried to make some lyrics but Suo shwn this mesage. Also Aimusic shown some server errors

[cursive jetty]

it will try and auto do it if your uploading new image, but after you manually do it - just download the vid to check its updated 👌🏻

[charred merlin]

ffs woodhouse, missed you pal

[charred merlin]

Furrybeans had commented above regarding the daily free limit of 50 credits;

In regards to Suno using your tracks if they are made on a free account -- Suno retains the rights, titles, and interest in any songs that were generated while your account was on the free tier, but while they legally would be allowed to do, I'm not sure they're using the Free-tier songs as their own content-library.

They use their social media platform to celebrate all kinds of Sunovators, if you want to get involved and maybe see them use one of your songs one day 😄

www.instagram(.com)/sunomusic/

[cunning mulch]

ah well... was kinda hoping someone would know what the hallucination I got was saying/singing, but I guess not

[astral bay]

what are the best text or music to video/animation AI tools? I've seen some ppl recommended some a while ago

[turbid sleet]

I see good stuff from Neural something, let me check

Neural Frames

Here's the man behind it demo/explain 5 different music visualizers https://www.youtube.com/watch?v=nl5VITu3Ya8

[cunning mulch]

See, I was making a Bollywood-style song, and wanted to make sure I got a voice with an accent. I certainly got that, but I also got this non-English hallucination. Any idea what he's singing here? I'm interested so I can add it into the lyrics

[astral bay]

if its a hallucination its literally something that is gibberish and not real language

[astral bay]

thx this looks interesting I'll check it out yo

[turbid sleet]

Faster Whisper XXL says it's Hindi

1
00:00:00,000 --> 00:00:01,980
शुत्ते सब नवस किता

2
00:00:01,980 --> 00:00:03,800
होडे तो नवस किता

translates to

1
00:00:00,000 --> 00:00:01,980
"Everyone felt nervous."

2
00:00:01,980 --> 00:00:03,800
"They were even more nervous."
``` @cunning mulch

[cunning mulch]

ooh... Faster Whisper?

what's this?

I figured it was Hindi, but considering I prompted Punjabi at least once, I wasn't 100% certain

sounds like it was used for sound rather than meaning then

Based. I'll add it in

[charred merlin]

The hallucinations might have little snippets of real language, but often they're gibberish -- Sunoese perhaps?

Like "Prisencolinensinainciusol" an Italian song written to sound like what English sounds like to the Italian speaker, Sunoese SOUNDS intelligible, but absolute gibberish beyond a couple passable words or turns of phrase.
https://www.youtube.com/watch?v=-VsmF9m_Nt8

[cunning mulch]

I think it might've been some leaked training data

'cause I had a different one go on for much longer. In fact, was at least two different ones that went on for about 30 seconds

and the one sounded like a youtuber, as I distinctly heard 'Subscribe' in it

anyway, where can I use this Faster Whisper thing for myself?

[safe hazel]

I've had entire songs of english sounding gibberish. You can pick out some real sounding words but thats probably just the human brain searching for patterns than coming from actual words

like in one of them i can hear the word "broken" surrounded by a sea of nonsense

[naive quail]

Suno gave me an entire unaccompanied vocal track in Sunoese https://suno.com/song/24008337-f671-4602-b0a9-e7e2a32b4992

[safe hazel]

Giving me Adam Lambert vibes

[cobalt sinew]

Hi. I'm sorry if this is not correct topic, but. Why the hell did suno not paying attention at "male vocalist" option? I got 4 songs, where i wrote "male/boy vocalist/voice" and there is still woman's voice! Whyyyyyy.😭😭😭

[naive quail]

it works well as an upload for vocal training suno

[cunning mulch]

Judge for yourself:

that's sounding like more Hindi, with "Subscribe" in there used like a loanword

[safe hazel]

it's your brain seeking patterns

[cunning mulch]

got a link to Faster Whisper?

[cobalt sinew]

https://suno.com/song/326afdb4-16af-4072-9f38-6ee7663390df
The song is actually good, but why is woman's voice...

[safe hazel]

I used mine gibberish song as a basis for an awesome sounding boy band song. While it was gibberish it had amazing harmonies so I extended it into a real song that turned out great

[cunning mulch]

so... are obvious producer tags also 'just pattern misfires'?

[safe hazel]

I would reccomend not using "boy" and "male" in the same prompt. Just use "male" and if that doesnt work try putting it in more than once

[cobalt sinew]

Well. (Привіт, земляк.) I tried. One moment.

[safe hazel]

@cunning mulch producer tags are something that suno has yet to give an answer for

[cunning mulch]

and there we go

I think that maybe, just maybe I got a bit of that, just on some other data

[safe hazel]

I dont think so. It really doesnt sound like the word "subscribe" just the general intonation of a "suh" sound followed by an "eye" sound

[cunning mulch]

well, I can lay it to rest. Where can I use the tool @turbid sleet used on the other one?

[turbid sleet]

yeah, here's the link

[safe hazel]

i wouldn't say that tool would lay it to rest. it's just a tool, searching for patterns, like our brains

[turbid sleet]

https://github.com/Purfview/whisper-standalone-win/releases/tag/Faster-Whisper-XXL

oh shit, haha

[cobalt sinew]

https://suno.com/song/51a4c742-096f-4004-8c30-4bf0d4f8eef2
https://suno.com/song/322caf06-261c-4465-a4fa-beeea72a4b08
https://suno.com/song/326afdb4-16af-4072-9f38-6ee7663390df
https://suno.com/song/a694e45e-c46e-48e8-8a9b-53f860be5053

All 4 songs have woman's voices. And at first i was writing only "male voice", ant then I thought Suno didn't understand me, so I tried to make it more clear. And yet. Here we are.😡😭

[turbid sleet]

wrong one

ah no, link is still correct

I hope you have plenty of vram

or at least 6GB

[safe hazel]

All I can say is I use "Clear Male Vocals" and then "Strong Male Vocals" (not always "strong", just some other descriptor word) and i hardly ever get the wrong vocals. But sometimes I still get the wrong vocals, it just happens sometimes

yikes

[polar spade]

I've got a great song with the "babble-scat" as I call it. So, as far as I can tell, it's Korean babble with a guest solo singing in Hindi.

[safe hazel]

try using it as a seed for a song that comes out in actual words

[polar spade]

Yes. After that though, if it alters in the babble version, it might not carry over to the real exactly the same.

I'm trying to see what happens with uploading each different section. And then Extend each separately with real lyrics. In the end, download 6 different sections of the song and splice it all in DAW.

[cobalt sinew]

WOAH it works! Thank you!

[polar spade]

I have all of them on a Playlist. I've heard enough that I'm starting to sing along in babble-speak.

[unique stone]

@wet minnow cleanup might be needed here

[safe hazel]

but what about my free gift? I'm hoping for a toaster

[safe hazel]

i feel like there is something we can learn about suno through these babble speak generations though. Im just not sure what

[naive quail]

I only looked at the first one that you linked... did two gens with your prompt and they did indeed come out as female. I change the first part of the style section from "male voice" to "male vocals" and did 2 more gens - all 4 tracks came out with male singers.

[wet minnow]

NO! @hexed bough

[safe hazel]

there goes my toaster 😦

[naive quail]

the timing on that was amazing, @wet minnow! I added an X emoji and the post immediately disappeared. I FELT THE POWER!!

[cobalt sinew]

Thank you, I'll remember Suno don't understand the "voice" word whahah.

[safe hazel]

seems weird that it responds to "vocals" and not "voice"

[polar spade]

Well, it's related, I think to how Max Martin write songs. He believes it's more important to use words that fit well in the line, rhyme and rhythm. So what might seem like a fine line in a verse, could be changed to simply "sound" more danceable or memorable. Suno, just uses common sound patterns of a language, it appears to get a good rhyme and rhythm. So, frequently, we get real words peppered in as a result.

[safe hazel]

im trying to make a max martinesque song

gimme that vocal fry britney!

[cunning mulch]

so, stand-alone... not web-based

[polar spade]

I write kpop songs recently and don't understand anything I write. With the benefit of trying to change some lyrics if the word patter doesn't feel right. So, suno does that for me now when it comes up with the babble scat stuff.

[polar spade]

Was it you wanting to use vocal fry the other day? And not having any luck. There's got to be a few prompts when combine may nudge a vocal fry vocal out.

[safe hazel]

it was me! I cant get suno to vocal fry on command, but if you listen it naturally puts some vocal fry in sometimes

[polar spade]

My song made it into the ssc5 list this week. Woop woop. But man, after hearing the other songs, I might be lucky to get 1 vote.

It's cool to hear a distinct variety though , as well. Not all one style.

For a two hour write, though, I still like my results of a song. Encouraging.

[safe hazel]

what can we learn from Suno recognizing "vocals" but not "voice" 🤔

[polar spade]

Maybe we can try bacon vocal, vocal sizzle, creaky vocal.... just go over the edge with odd prompts, to nudge suno a millimeter..... or it completely ignores them. You never know.

[safe hazel]

sixstring likes to put silly things like that in, so maybe it will do something after all.

[polar spade]

Hey it worked first run on one gen, you can hear the fry a little, but not forced... the other one was clean vocal though.https://suno.com/song/e91d2810-1124-4d8c-8a68-7a3b43359f89

New solo drinking game. Click auto-generate lyrics and anytime it writes "city lights" , "hearts entwine", the word "bright" or "street" ,I take a drink. Lol

[safe hazel]

Yeah there's some fry there.

[safe hazel]

dont forget "stories untold" and "silent screams"

AI loves it some entwining though

it's a great word, but now I dont ever want to hear it in a song again

[turbid sleet]

neonlights

soar

[polar spade]

Yes! We need a list of common suno words. I can't believe in the history of music, "city lights" is apparently the most common used phrase of us humans. I have no idea though.

[naive quail]

Bacon is not silly!

Spiders are silly tho... (looks at @charred merlin)

[polar spade]

That first feeling when suno writes your (noob) first lyrics "Hey mom. Listen to this great song I just created ....."

City lights entwine with shadows
Neon beams on streets below
Hope's the alley , thin and shallow
Where we often fear to go

[turbid sleet]

whispers, another one for your list

or 'raise above'

[polar spade]

.... and I'm going to give names to the vocalists as well per genre. Haha.

Im already thinking of a movie , like Inside-Out. Where suno has these vocalist just sitting in rooms waiting to sing these songs as they come in. And theres that one guy, Sasha, waiting for his chance to sing a Ukrainian Death-metal Polka that hasn't come up yet 🎵

[cunning mulch]

hmmm... tying to run that Whisper thing, but not so sure this little garbo lappy can actually do it

it's just sitting there after having downloaded a model

[turbid sleet]

what GPU do you have

how long is it

what was your command line?

[cunning mulch]

I'm not sure this even has a real GPU in it

probably some crappy on-board video

[turbid sleet]

yeah, good luck then

[cunning mulch]

it's running on CPU

[turbid sleet]

just as with STable Diffusion, ComfyUI, LLM stuff.....you need a GPU with plenty of vram

[polar spade]

Maybe a GNU instead, they can run
..... and butt.

Sorry, I'm in my Dad mode today.

[cunning mulch]

well why did you point me to a local version when I was looking for a browser-based one?

does one just not even exist?

[turbid sleet]

yes

[cunning mulch]

fff....

[turbid sleet]

and you said with your wav See, I was making a Bollywood-style song, and wanted to make sure I got a voice with an accent. I certainly got that, but I also got this non-English hallucination. Any idea what he's singing here? I'm interested so I can add it into the lyrics
I was just a good sameritan and translated it for you, and you asked which tool

nothing about online of offline

or i must have missed that line

[cunning mulch]

ah

okay, I see what happened. You must've missed after that

[turbid sleet]

is it big? try giving it to chatgpt

[cunning mulch]

someone wanted to argue that these kidns of things are 'simply randomness being false-positived by our pattern-matching brains'

it can analyze audio?

wound up posting a second one from another version that went on for about 30 seconds

[turbid sleet]

there is Buzz, that works with openai api

https://github.com/chidiwilliams/buzz/releases/tag/v1.0.1

there are a few like this one, https://huggingface.co/spaces/aadnk/faster-whisper-webui

that runs on Hugging face

[small parrot]

How can men and women sing in pairs?

[turbid sleet]

use duet, balled

and in the lyric prompt, point it out male/female

[charred merlin]

https://www.suno.wiki/faq/style-and-lyrics/duets/

[safe hazel]

Use getting really lucky

[cunning mulch]

not yet, anyway. Maybe with v4

[turbid sleet]

i have one, where i put on every line male: female:
only in the beginning it switched it

[jade mist]

I really hope V4 alpha is coming soon to pro or premier customers. Regarding this the model selection dropdown is completely inaccessible to blind people.

[cunning mulch]

well, I'll see wat it says in 53,000 seconds

[turbid sleet]

finding something in your suno library is the same gamble was generating songs 😛

[safe hazel]

praying for folders 🙏

[turbid sleet]

@small parrot example (in dutch) https://suno.com/song/f739289d-46f2-4ac9-b1f1-c5baaaa181fb

but it didn't listen, i didn't tell him to sing the whole chorus together

[small parrot]

Can you make a column?

[turbid sleet]

column?

[charred merlin]

#💡┃suggestions-web if anyone has ideas for what they'd like to see in future models
#🚧┃suggestions-layout if anyone has ideas for what they'd like to see rearranged with the website 🙏

Not to discourage conversation, but the devs don't often poke their head into tech-chat for ideas on how to improve the site, thats what the channels above are for ❤️

[turbid sleet]

i was working on my own library, that uses their api to scrape the data.
So that when i want something removed or out of the recycle bin, i know which page to look 🙂

[small parrot]

thank you

[turbid sleet]

now, if someone knows how to let it stop abrubtly in the first verse or chorus, and have after that a few spoken word lines, without any instrument.....that would be great

like if a band plays a song, they discover it's the wrong song, so stop and say something about it

[naive quail]

https://suno.com/song/433d5348-352c-4396-ae31-e3c4beb1b6d5 ...I think it got him. lol

[small parrot]

If a song is good as a whole, but there is a bad sentence at the beginning, is there any way to change it?

[turbid sleet]

I have it working here, https://suno.com/song/1bed6841-f5fc-4a67-88e0-08642b74dbd0 @ 2:13

but somehow he doesn't want to do it in the beginning

[naive quail]

Not currently. v4 may have inpainting, tho - you should be able to save it with that.

[turbid sleet]

uploading the part to Udio, you can inpaint there, and paste it later together with an audio editor

[small parrot]

ok

[cunning mulch]

hmmm. Got a different transcription that time

[polar spade]

Any splits I simply use by spacing down and prompt

[spoken] blah blah blah

Then space down again twice to start singing again. But that general. To split a verse and have it come back in on the same melodic phrasing, I don't know.

But have it stop , say something and then shift to another song. Maybe that's enough.

[cunning mulch]

gah. Probably gonna need a native speaker

[turbid sleet]

with extending it's possible yes

but i know it's also possible without 😄

[safe hazel]

😱 I like when it turns from blood curdling screech to some nerdy guy going "waaaaaaaaaah"

[turbid sleet]

yeah, but there is still some music playing with it

[polar spade]

Ahhhh... yes.... how to have the music stop..... while speaking. A common question.

[turbid sleet]

no, first music stop, then speaking

😛

but some shouting during would also fit here 😄

[polar spade]

A Capella doesn't work.... you can try "stand up" . I haven't tried yet. But just Stand up, alone, is a comedian talking without music every time.

I'll burn one off and see if that works. Hold on.

[small parrot]

When will v4 be available?

[neat nexus]

You're trying to start and stop? That's harder.

[safe hazel]

what about [Isolated Vocals]

[neat nexus]

Just getting a decent voice over can work, but trying to make it go from music to pausing to speaking and then singing again...that's hard.

The voice over from my recent music video: [Removed links]

But I didn't try to make it go into singing from there, I overlaid it on my music in the editor.

[turbid sleet]

I'll try

[neat nexus]

The music video in question: https://www.youtube.com/watch?v=PJVeDeNlSQY

[turbid sleet]

that Bo burnham kinda like song I shared earlier, also have it talk normal after a section and continue singing

[naive quail]

@frigid orbit those uploads are indeed extendable... https://suno.com/song/49fb46a8-5cc2-43dc-8287-fba20fbbf5dd

[turbid sleet]

lol, the 'stop this music' sounds like 'fk this music' in this gen

[frigid orbit]

Cool! That's my goal. If I find something I want to share, or something I wont use. I'll be able to share it with y'all. One man's trash..

[safe hazel]

did AI generate the video too?

[turbid sleet]

tried a few combinations with isolated vocals, but look....sadly no luck https://suno.com/song/15c1415b-8cc1-45ee-bb6c-afb580f19055

this one from earlier has the least music, https://suno.com/song/573c2018-ea67-4e1e-a726-f5279d97bea0

one of them, in the end...they repeat it, almost perfect...

but, can't use it with Suno

[neat nexus]

I used a combination of Noisee and Pixverse for it.

I made the clips for the main body with Noisee and then realized it needed an intro setup for it and that worked when I made individual clips for the most part but a few things Noisee just fought me on so I went and used my reference images in PixVerse to get them to work. The suspended animation tube with the bobble for instance. That was PixVerse image to video. The man sitting staring at the screen was also PixVerse image to video.

But the audio is entirely Suno with my lyrics. The voice over included.

[turbid sleet]

well, bbl

[safe hazel]

Thanks for lettimg me know! It came out really cool!

[neat nexus]

You might need to tinker with it but it does work...the music start up can be trimmed.

[Removed links]

[neat nexus]

Glad you liked it! 😄

[polar spade]

https://suno.com/song/5a7a1c62-cbc1-43ea-85b9-0b10e7983f64. A poor wxample of SPOKEN, but the only one that "paused" after prompting with [pause].

So after running about 10 different combinations here's what I found:
V3.0 will run a stand up comedian on its own. V3.5 seems to want to sing everything. "Stand up " with singing though, didn't work even in ,,v3.0

It finally followed the [pause] prompt, so that could be just random, but still, in speaking, it started in on the music to tag into to the next verse. But it gives you options to try.

After all that. I could get a silent break, no music speaking, and then start back immediately. It seems it should be possible, so I'd be interested in trying more later.

[polar spade]

And I'll say, the pause was metered perfectly to fit into the music, so it wasn't a random pause. So there's more to do to just get it to stop and talk awhile.

[safe hazel]

do you prefer "kpop" over "K pop" and "K-pop" ?

[polar spade]

It seems many type "k-pop" I've seen. I type as one word, kpop, jpop

[safe hazel]

i have some samples of "k-pop" and "k pop" but i havent tried "kpop" yet

that'll be next

im just taking the same song and using different style prompts to see how suno differs according to prompt

[polar spade]

Well, historically, K-pop has meant the pop music from Korea, but I'm sure the strong influence of "hip hop" has many spelling it "K pop" , even if they're not sure why.

I'm just lazy. So I type kpop

It's worked so far

[full apex]

Why is the extend only 2 minutes and it cuts the song i never finish rendering the song :/

[cursive jetty]

[END]

[safe hazel]

extensions can only be a maximum of 2 minutes long

[full apex]

But the recommend using it for reaching the same voice agian

[safe hazel]

yeah, but it can still only be 2 minutes long

keep extending it for more

[full apex]

that's kinda rubbish :/

[polar spade]

The initial generation is up to 4 minutes. Any extension are limited to 2 minutes. You'll need. Copy lyrics somewhere a bit back from the end of where the extension cuts off, extend from that point ( a break between verses?) and extend again for up to 2 minutes.

[polar spade]

It's the tool we have to work with. I'm sure the limitations are warranted until a future update.

[polar spade]

Members have 10 minutes songs by extending again and again.

Seasoned creators often just write one small 30 second section [regardless if it improvised out to 4 minutes] then Extend from :30 with different prompts, and repeat, creating each verse or "section" by extentions. So the 2 minute limit can encourage us to be a bit more creative.

[naive quail]

This. I'd love to be able to set a smaller time limit on extensions (at a reduced credit cost).

[turbid sleet]

@polar spade @neat nexus thanks lads, I'll have something to play with later

spending another couple of hundred credits with the change for perfection, or do concessions

hmm, lol.....eas

y

[hidden pier]

What happens when all you need left is 30 seconds to end the song, but it keeps generating 2 minute extensions with unwanted slop because it's hallucinating lyrics you didn't write or want?

[polar spade]

The problem is suno doesn't know when to stop singing sometimes, so it will just keep singing over a short extention no less than a long one. Haha. So trimming tools are needed.

[hidden pier]

I'd take it further and say that basic editing tools in general are needed, not just trimming.

[naive quail]

I'm thinking of instances where I'm building the song with an extension for each section of the song structure... and, for example, I get to a solo. The usable bits are usually in the first 20 to 30 seconds; the rest just gets scrapped

[polar spade]

If it ends where you like, then , at this time, I've had to extend it once more a couple of seconds BEFORE the ending ( and of course before the slop starts as well). It tries to then give some 2 to 15 second or so extensions, which is preferred, and end the song. It's random still, but frequent. Suno doesn't charge [or refunds] back anything less than 8 second, but you can keep the ending extension. Click "get whole song" from the [. . . ] menu and you'll have a completed song.

If you don't care if on suno, then you can option to download the audio file and open in an audio edited program to trim off the slop ending part.

[hidden pier]

tbh i'm not sure i fully understand. I've already gone through this process and burned through my daily credits twice now, trying to get it to clip at a specific point, but it just keeps giving me 1:30-2:00 junk, most of the time entirely different from the source.

[turbid sleet]

Did you try setting it on v3? To See what that gives?

Won't be 2 min long 😂

[hidden pier]

Yeah that's my next attempt when my daily credits refresh.

[naive quail]

I usually replace everything in the lyrics box with [End] and delete everything in the style box (or use End) and then give v3.5 a few cracks at it. If I'm not seeing any short ones, I'll switch to v3 and try that.

[polar spade]

Oh no, yea, I get it. It's hard to explain the steps sometimes as they are just hacks depending on what the previous results are.

Right now, we can't trim off the junk noise. It may end great at 1:12... then 2 seconds of silence and stat back in with junk until 2 minutes is over.

So you want to extend again, somewhere right before it ends on the good ending. So pick that time, lets say, 1:08... if it sounds like a good spot. So with that last extension, click the [. . . ] menu and choose extend again. Click CREATE, take out any lyrics before that 1:08 spot. Paste or keep in any lyrics you want sung or re-sung. If just you want to end the song, take out all lyrics and type the prompt [end], in the lyrics box. That will help. Then scroll down. Make sure the time stamp is edited to "1:08" , then click CReate again to run extention.

[charred merlin]

Mr Big Spender, using v3.5 to end songs like he's lighting cigars with crisp $20s

I'm just being a smart-alec, v3.5 can end a song in STYLE provided the right "offramp"

[polar spade]

We make it rain ending!

[hidden pier]

Does it matter if I stitch the extension and try to clip it from the new song, or would I have better luck extending from the extension itself to get find the branch I want?

[naive quail]

haha. After 10,000 generations, you start to get a feel for whether 3.5 might honor the end tag.

[polar spade]

Great observation. Many say, to run the last extention into a whole song first so suno doesn't work with buffering two or three generations. So yes! Many prefer it that way. Make a whole song, including the slop stuff at the end... then Extend it to hopefully get rid of the slop and junk (sometimes called hallucinations) and have a nice button ending.

You always have both a whole song, or the initial extension if you don't like something, you can extend either one at that point. The best organization is to start labeling your extensions so you can follow which ones extend forks off from the last one. I will often change the picture to help.

[hidden pier]

thanks for the info. seems i have a few days of just waiting ahead of me to get this song right. xD

[polar spade]

Hands on experimenting is the way to go. I've done this daily for four months and still learning.

[runic ocean]

I'm trying to generate an extension to my song to do a duet, (male/female) but the extension gets automatically flagged for moderation, the lyrics have been generated before with no problem and it's not using any copyrighted material like I've not mentioned any artist names etc in the lyrics it just won't let me generate it

tried to see if it's just trying to generate female vocals that flags it but changing it to metal male vocals or just metal vocals still gets it flagged

[leaden kayak]

Hoping they update the stems. I know its only V1 but can still here the vocals in the background on instrumentals

[runic ocean]

fixed it just had to generate from another song lmao

[languid jolt]

Hey folks. Is there any way to force Suno to follow the song structure? I use all the correct tags like [verse] and [chorus] but Suno ignores this most of the time and halfway through the song it goes and does the first verse again or does the chorus twice and just screws up the song. Is there a way to force it to follow the structure that I lay out?

[hidden pier]

Most people suggest doing the song in short chunks, using the Extend at intervals and updating each extension with your prompts, lyrics, etc.

[glad valley]

uuugh suno just won't stop putting a female singer in even when I specify male vocalist, male singer

Is Horrorcore somehow female vocalist focused?

[dapper notch]

Incredible software, blown away with the ability.

One request though.

"Celtic" comes out as "Keltic" when sung.

Can we get this tweaked please so I can do songs about Glasgow Celtic F.C. please?

Thanks 🙂

[hidden pier]

Where are you specifying? What tags, if any, are you using, and where?

[glad valley]

in style

cause male/female doesn't work in prompts in lyrics

doesn't do much for me

some people it does....For me? Doesn't matter if I copy their ENTIRE prompt and style

[hidden pier]

When you do try to use male/female in the lyric box, how are you formatting it?

[glad valley]

I said it very specific

I put it in the beginning, then I tried middle, then I tried end, then I tried after beginning and before mid

then after mid and before end

[hidden pier]

I'm asking you to replicate how you typed it out though. I'd like to see how you formatted your tag.

[glad valley]

Sad, Male Vocalist, Male Singer, Gothic Metal, Horrorcore, Darkcore, Minor Scale, Baritone B2-A4, 70 BPM, is one example

[hidden pier]

That's what you typed in the lyric section?

[glad valley]

the baritone part works for most songs

no

in the style

[hidden pier]

I'm sorry if I was unclear. I'm asking you how you formated your tags in the lyric box, specifically.

[glad valley]

in lyrics I've tried names, I've tried the same, male, man, male vocalist, male singer, male vocals

[hidden pier]

How did you format it though?

[glad valley]

uuuh...I guess I can show what comes before it, but I cant show the lyrics

cause then I get banned

[hidden pier]

Yeah just do like an example, not the exact thing from your track.

[glad valley]

...

[hidden pier]

Crochety old bot.

[glad valley]

guess not

I have a version that worked before I entered horrorcore, gothic doom or funderal doom

all three of those forces female vocalist

just like symphonic metal

and often pop

[hidden pier]

Here's how I format my tags

[Verse 1][Male]
Lyrics go here

And it seems to respond well to short, concise descriptors in brackets.

[glad valley]

I don't do that anymore after reading the guide to stop repeats

ok so I cant even do that

I cant put spaces, ok so 3x #

whyyyyyyyyy

three hash tags Chorus 1 (Joyful+Horrorcore)

works for everything but vocals

works very very well for everything else

[hidden pier]

I don't think that's the cause of repeats, tbh. I've got two versions of a song. The first version has a verse repeated once, the second version does not. I think what matters is that v3.5 prefers longer duration, and will fill in gaps. My second track has more lyrics in it, and there are no repeats. In fact, it gets cut off at the end and I've been struggling to extend it in a way I like.

[glad valley]

I can change tempo, vocal octave etc

the repeat reason is syllibles

and using three hash tags makes it ignore singing it by mistake

but still uses its instructions

since it becomes a headline or title for the AI but it will still process instructions in it

[hidden pier]

To clarify, the second version of my song has the same lyrics as the first version, formatted the same way. There's just more of them.

[glad valley]

wonder how my risecore and dropcore version turned out

[hidden pier]

At any rate, that's off topic. As far as getting the sound you want out of the lyrics, you're using tags differently than I'm familiar with so I'm not sure if I could provide any further insight.

[glad valley]

Fair enough. The other format has never worked for me either though

the style overrides all prompts

[cloud blaze]

HI I got question about bart suno model

[glad valley]

or if the AI hallucinates

[cloud blaze]

where can I fins any details about dataset to finetuning or tranning this model

about 2023 there was information that tranning code will be available

[glad valley]

bart? Is it just called bart?

and is it an LLM, a audio generative ai? voice generative?

https://huggingface.co/docs/transformers/en/model_doc/bart I mean here's one bart

[turbid sleet]

I think we are gonna get a video to song function:

• the call to the api for the library contains now a ?is_video_to_song=false parameter
• in the json with song info, are these 2 new keys:
  • "video_to_song_video_upload_id": null,
  • "video_to_song_video_upload_url": null,

[glad valley]

@cloud blaze dont just run away when you ask something

https://github.com/suno-ai/bark here's another

[cloud blaze]

https://huggingface.co/spaces/suno/bark

[glad valley]

so if you wanted the text to speech model

[cloud blaze]

this one sorry

[glad valley]

check my github link

oh so bark?

well the github is bark too

all it took was to google bart suno model, even mispelled

[cloud blaze]

I mean suno bark

[glad valley]

damn it

[cloud blaze]

your git hub link is correct

but I can not fine any information about tranning etc

[quartz rampart]

/chirp

[flint haven]

The /chirp command is no longer working

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

[glad valley]

I mean there are audio generative models too

https://github.com/coqui-ai/TTS wonder if bark works in that

[cloud blaze]

but I'm looking for similar to bark ( singing music with vocal )

[glad valley]

not about to link it here :p

[cloud blaze]

thx but back to suno bark ... I would need some tips on how I could train this model

[glad valley]

same way you train any text to speech model

more or less

just check TTS for their training tools and see if bark works with it

wrong chat

[pine island]

Guys someone know how to’ improve quality of voice? Is there some model improvements in progress?

[safe hazel]

You can try using "Clear Vocals" in the Style prompt of custom mode. They're working on a new model but we dont know anything about release dates

[fierce zinc]

A couple of minor complaints: too many instances of 'the 4-chord song', and too many E-D-C chord progressions

[pine island]

Thanks. Tried but not works effectlively

[glad valley]

try F-C-G-C-E Detune

or add Math-Rock too

[signal jackal]

Anyone else missing the "Follow" button on their profile? I see it for a second on desktop and then it disappears. Thanks https://suno.com/@jimpeak

[rich timber]

My record is repeating my verses. I tried suggested on suno-wiki putting (la lala) before the verse

I had to chop. Then my next extend repeating Chorus.

1k points later, still having repeats.

[charred merlin]

You're not supposed to be able to see it for your own profile -- if you see it for a second, it's likely because viewing your own profile causes some kind of CSS to hide the button, so there's a chance your browser might not update quick enough.

Don't worry, people can still see your follow button 🙏 🙂

[pearl valley]

Stem results are ridiculous to charge credits to receive not an instrumental or a cappella at all just a partial concept of the idea yet you’re charging to give mediocre results why?

[safe hazel]

You don't have to use it

[charred merlin]

I can see the matter from both sides;

• on one hand, the vocal separation just isn't there yet in terms of competing with products that specialize in stem separation, and for intermediate-to-advanced users, there's absolutely room for improvement.

• on the other, for new-to-casual users, being able to go from musical 'ideation-to-creation' in potentially under a minute is mindblowing as it is, but being able to make a "karaoke" version of their favourite Suno songs without having to leave the Suno-ecosystem opens up a LOT of really fun use cases that otherwise would have never been explored

THAT SAID
It's super helpful to have technically-informed feedback on what they're admitting is v1 of the stem separation -- I hope you follow up with this link and let them know specifically what you're dissatisfied with -- hopefully stems-v2 knocks our socks off 🙂

#🚀┃updates message

🙅‍♂️🧦

[terse tulip]

Are you not supposed to be able to search for your own profile? I can search for my songs and find them, but not my profile.

[charred merlin]

Tried from my account and a free account, Edge & Firefox -- maybe some of us are just unlucky 😉

theharber, theharber., @theharber, @theharber. all gave the same results. 🤷‍♂️

[terse tulip]

So it's not just me. I searched for you and couldn't find you as well.

[charred merlin]

I have 3 songs made public, and they seem to be searchable, at least.

[terse tulip]

All of the songs I have public are searchable as well

[terse onyx]

Does anybody else get songs that are great but the last minute or several seconds are all a mess?

[sand lintel]

Hi guys and gals - And anyone else.
I am new to SUNO and ready to subscribe but...
I need some help before I do so.
My monitor is set at a resolution of 3840 x 2160.
Unless I zoom out the page in Firefox while using SUMO I am unable to view the part of the page were I can play back my song. I can't even scroll down the page to make it visible.
However zooming out makes most of the text so tiny that it is unreadable.
Any suggestions?

[pearl valley]

Hey 👋 thanks for your response and a nice new prospective on what to do with these results at this time. Karaoke 🎤 would be great 👍 can you download a karaoke video now ?

[charred merlin]

The instrumental stem include the timed lyrics, but you can edit them back in afterward- - trying to find a good example from my (potentially goofy) backlog of songs for a demonstration

-adding them now 🙂

[signal jackal]

Is there a way to make your own cover art? Some artists seem to be to have consistent art on all their songs and it makes me wonder if we have control over that ever.

[signal jackal]

Thank you so much @safe hazel - I had not seen that on my phone app because the three dots don't seem to work... nothing pops up. But on my desktop, it works! Thanks again

[charred merlin]

#1256301881966198805 message 🙏 Definitely a ramble, but also meant to showcase use cases to other intermediate-to-advanced users like us who might just brush the entire thing off as "just a toy feature", without embracing how that "toy" might be the catalyst that leads to someone falling in love with Suno as an instrument 🙂

[leaden lark]

Mobile app doesn’t allow edits or extends

[spring ingot]

Howdy, folks. Is there any advice how to prevent suno from speeding up the BPM when extending an existing uploaded track?

[untold briar]

I don't have any specific advice. One idea might be to experiment with different lengths of your uploaded track. Try the range, 10s, 30s, 60s.

Is the BPM speed up a slow drift, or a very dramatic speedup of like 200%?

[spring ingot]

Thanks, will try that. my seeds are 25s long at the moment and i've tried 10 to 25 so far. The speed up is not consistent, I get it in ±30% of generations. It sometimes does 200% making the track inaudiable, but an average it's like taking it from 120 to 140-150 bpm.

[untold briar]

If it's 200%, I associate that type of speedup with repetition in Suno text prompts. Something it thinks it's doing on purpose. But separate from a slow drift of BPM, which feels more like just a Suno weakness.

[spring ingot]

Do you think if I will trim the seed to be just a single loop it will make it better?
Probably worth experimenting regardless

[untold briar]

No idea really. But differences in duration in general can have big effects. 10s Vs 60s. 10s will be more of a loose inspiration, 60s will be more of a continuation. (This effect is also true in regular Extends, not just uploads). The specific time the Extend starts can also have a big effect. So that idea is at least plausible.

Also worth trying a completely blank prompt, space in lyrics, blank style prompt. If that Extends with perfect BPM, then possibly something in your text prompt is effecting the BPM

If 70% percent of your generation don't have a problem, honeslty I'd probably not worry about it. That could just be Suno imperfection.

[spring ingot]

I'm actually doing instrumentals. I can will try going with a blank style prompt though. Thanks for the suggestion

[untold briar]

Right, I understand. If you Upload an instrumental, you can Extend with blank lyrics/style (well you need a space for the UI to let you click on it). Often that makes kind of boring, repetetive Extensions, but not always, and is useful for debugging as a 'control' prompt.

My use case for blank prompts is particularly when I upload non-music. For example, ambience of a tennis match. If I want Suno to Extend the tennis sounds, can be difficult with any style words at all. But a blank prompt... can work.

[spring ingot]

I'm just generating and for a repetitive instrumental it like goes into speed-up once it get's bored lol. It generated a 2 minute extension and the first minute was fine, but then it's like "nah, this is boring, how about some real dancing" lol

[untold briar]

The vibe I get is Suno thinks, "Oh I'm repeating something over and over? That can't be right. I probably should be supercharging the speed - that's what real songs do when they repeat the same notes/drumbeat/whatever over and over..."

[spring ingot]

oh! repetition penalty! nice! maybe that IS the case. although diffusion models that (i assume) generate the spectrograms under the hood don't have such concept, unlike language models. some concept of repetition penalty could have been introduced to the part where they do the continuation synthesis, but iiuc this usually works similar to inpainting (or rather outpainting) in image generation. so that would be a feature that's not there out of the box.

anyway, thanks for all your advice. really appreciate it. will experiment more

[woeful burrow]

Does anyone else have he problem, that the second verse always gets repeated twice lately?

[nova vapor]

Hi everyone, I am recent on Suno, still trying to fully understand some functionalities, and I have a question wrt EXTEND function: let's say I have my own song lyrics and created a 3 minutes song for which I am not happy with the last minute. I am then using Extend function at 2:00, then request the "Full song" when I am happy with the proposed extension. My question is: how can I manage to have no text repetition in the final Full song version? Thanks!

[azure lion]

hello everyone!
when is it planned to increase the bitrate of the output music to 320kbs?
now the music sounds like 64kbs and when listening for a long time, my head starts to hurt.

[unique stone]

The MP3 bitrate is 192kbps. The model output is around the same quality as that.
WAV download is available on paid subscriptions but is only slightly higher quality (not enough to notice.)

As far as I know, overall quality would only be possible with a new model. v4 has been in development for several months so maybe that will help when it's available.

[mighty solstice]

Hi. I need some help. How i can back to free plan after my paid plan is over?

[frozen pivot]

If you don't have a subscription plan, you should be on a Free Plan. If you currently have a subscription plan, you should cancel it before the next renewal kicks in.

[verbal hull]

Why is my pro plan account still active? even though I no longer subscribe

[unique stone]

Did you cancel the subscription on the website, or just block the payments somehow? (I think if the sub remains active but payment can't be taken it'll show as an active subscription but won't be usable.)

[verbal hull]

I have canceled the subscription and payment but nothing happened, and the pro plan is still active

[unique stone]

Ah - did your subscription month end / what date does it show it'll renew? I missed what might be an obvious question 😅

[verbal hull]

August 24 ended, but the pro plan is still active

[charred merlin]

You'll likely need to contact [e-mail] -- we can try to give advice and guidance, but there isn't a system set up in Discord quite yet for official first-contact support -- and the official Support team are the only ones that can assist with account related inquiries.

[charred merlin]

(*I'd personally still be curious, but if your bank says you haven't given them any money, and Suno is still letting you use pro-level features (.wav download, entering a prompt to generate your own AI-images in the "Edit Details > Upload Image" section), maybe it's a happy accident? 😛 *)

[azure crane]

How to manage the Google login? It says Access blocked: Suno’s request doesn’t apply with Googles policies

[charred merlin]

That's not an error message I've run into yet... could you share a screenshot and list what steps you took to get it?

[charred merlin]

After your subscription ends, without any additional action on your part, your account reverts to the basic free tier (50 credits per day, Suno retains ownership).

[charred merlin]

Looking into the matter, it sounds like it might be an issue logging into your google account, unless the error message says something about Suno.
"Your sign-in settings don't meet your organization's 2-Step Verification policy"?

If that's the case, you might need to contact the admin for your google workspace, https://support.google.com/a/answer/10710447?hl=en

[graceful condor]

Got the app. Same user name. However the app shows me on basic free (50 per day) while the site shows my subscription level. Thoughts? Suggestions?

[charred merlin]

Sounds like you're not using the same account -- you can sign out from the iOS app and sign in with the account you use on desktop 🙏

[mighty solstice]

for me it's just not work. I can't create smth, say - i need 10 credits, but i have 2500+ credits.

[graceful condor]

Okiedokey I’ll try that, thanks

[charred merlin]

Oooh! The single "known-bug" I was able to add when I was a guide 🙂
https://discord.com/channels/1069381916492562582/1263857907011813517

Last couple times I've pointed someone to this link it seems to have resolved the issue, but it doesn't look like they followed up to give the link an upvote so others knew that it worked for them 😅

[lament fern]

Hi, is it possible to add VAT number and company name when subscribing?

[hollow hare]

Hello, this is something you should ask [e-mail]

[charred merlin]

☝️

[polar spade]

Sunonhas often repeated full verses, fulle lines, or combined words of various lines, in both original and extended lines.

You appear to say, though, that you like the original, and the extension, but when you "get full song" it throws in addition text/words/phrases. That it should not do, by definition, as it's just stitching the two original (to the timestamp of the new extension) and any extended pieces.

That said, there's a longer process that has had some anecdotal good results for some. It's complicated, but in short, members are putting in a short instrumental extension then follow that by extending from the 00:01 second mark of the previous extension. Giving a "pivot" point to borrow the term from @charred merlin We are trying to keep Suno From looking at any previous lyrics in the original and just focus on the extension.

I , also , before clicking CREATE on my extension, have gone to my original, click edit lyrics, and delete out the lyrics and save blank. [I've also try combinations of leaving partial lyrics. Occasionally this helped it not throw in more text that I offered in the new extension. But often, it will just give a 4 to 12 second ending -- Suno will sulk 😄sometimes when it doesn't have lyrics to read -- but then I use that "pivot" and extend off the , let's say, 4 second, extension. Again, nothing perfect, but it has helped more often than doing nothing more.

For you, first try to put in your extensions, prompts in the beginning before your lyrics, such as [pause], [end], [interlude], [Female vocals] and so on that won't affect the song as much as distract Suno from starting with random text that you didn't include in the extension.

From the beginners to the experienced, we're all still testing and experimenting.

[charred merlin]

☝️💯

Even the government scientists still haven't figured out how to stabilize the time vortex to acquire v4 in it's physical form as of yet. devs are learning just the same as we are!

(I'm seriously gonna take this keyboard back, it keeps making weird typos.)

[broken raft]

Audio Processing times out when trying to upload Audio?

[jade sapphire]

question for those in the know: i am on the free / basic plan (I know, I know) and I got stuck yesterday with 5 credits left bc one time it created only one song, instead of the usual two versions? is that a thing? I have not seen this before and was mostly annoyed bc it takes minimum 10 credits for Suno to work at all. So I couldnt do anything any more but ride it out until today. Bug or feature?

[unique stone]

If you got a song shorter than 8 seconds, you get 5 credits refunded. It's designed to be that way, though it is annoying to just have 5 credits left (as you can't do anything with them.)

[jade sapphire]

nawh it was a song about 2:30 min, roughly... i dont know what has happenend, just a bit frustrated bc it was the last song 😄

[unique stone]

Strange. Somewhere in your library you should have something labelled "credits refunded"

[jade sapphire]

cant see anything anywhere in my library but maybe bc i got my new 50 credits today?

[unique stone]

Nothing in the trash?

[jade sapphire]

not that i can see it... idk if it happens again I will email support and ask them about it but thanks for the replies, i really appreciate that you were taking your time to help some newbie 🙂

[verbal mortar]

Hey all, not sure if this is the right place to ask or if this hasn't been asked already. How can we get SUNO to do things like massive drops or breakbeat slicing?? I can probably do it in my DAW but I'm experimenting with this app and want to see if something like this is even doable

[safe hazel]

Yes this happens sometimes. It's annoying, I doubt it's supposed to happen. Maybe the devs dont know about it (maybe they do).

[alpine venture]

yea, its suno

I mean, its both lol, suno has composition, udio has refining

ill put this in offtopic but I consider it technical

a curse upon your house woody (im jk, ur cool bro)

OK

Original song:
#🐣┃suno-showcase message

Upscaled/Mastered/whatever:
#🗣️┃chill-chat message

If anyone is interested, I can go into workflows, I have like 30 minutes or so and could do a quick stream

[turbid sleet]

yeah, i'm interested...

[grand zephyr]

/info

[glad valley]

mmm sub-harmonics

gain levelling, frequency carving, frequency dropping/lifting, EQing, limiting, compressing...so much time, good knowledge to know though

[alpine venture]

I'll have to do it another time, its too late at night now , might wake my kids up

[pine island]

What about a Suno integration in a DAW as a plugin? I think the future Could be generate tracks and mix them

[safe hazel]

Why was my message deleted?

[glad valley]

the other way around would be better...There's other places which have built in DAWs with their AI

their AI isn't as good at making the music though

but their limited DAW functions would be indispensible on Suno

[rose jasper]

Think I see daily questions regarding releasing the API, and so far there are no plans

[turbid sleet]

No problem

[neat nexus]

Warning: "unofficial" APIs are something doing something that no one knows what it is but you run a risk of losing access by using those and you could find yourself in for way more than you ever expected. Because unofficial APIs are in violation of the Terms of Service. So if they're willing to violate that, what else are they likely to be willing to violate?

[violet tundra]

"So if they're willing to violate that, what else are they likely to be willing to violate?" You're right - I'm locking up my daughter.

[neat nexus]

You're right...BRB

[violet tundra]

I suppose you can request a refund.

[violet tundra]

8 second 'song' = refund. So a song using version 3.5 that = 3% of the maximum possible seconds of 240 seconds = a refund. That's an interestingly specific threshold.

[frail vine]

Hello, I think, I've got someone else's song, that was generated simultaneously with mine.
It completely doesn't follow the lyrics or language of initial input.

[hollow hare]

Did you try refreshing the page/library? Often these glitches can be fixed by that.
Regardless, please be sure to report it to #🪲┃bugs-web with as much detail as you can 🙂

[astral flame]

If i upload a 60sexond song and have suno add vocals at 20 seconds will it consider what the other 40 seconds pf my song sound like after it starts adding?

I am looking for assistance first so that i can save credits.

[untold briar]

If you mean you Extend your upload at time 00:20, then no. If you want the missing 40 seconds to have an influence on the rest of song, you could use an audio editor to put it in front of the 20s vocal and Extend at 00:60 instead though. Even though this doesn't really make sense (the song will have a hard clip) it can surprisingly work out in Suno.

[astral flame]

It seems the whole first verse might end up missing. Is this what is likely or are you saying it still works out to a full song?