📚┃suno-school

[hollow hare]

If you have any skeletal reference, it may be too spooky for ChatGPT

(well, more like a lot of gruesome descriptives are used in such scenarios and the GPT was trained to be a bit more laxed when judging those stories which misconstrued the situation. It was not designed for lyrics in mind)

[opaque pasture]

Just "rotten bodies" "zombies" and "Zeds"

[gentle rose]

No inpainting in Suno yet? What a let down. Udio is kicking ass.

[unique stone]

It's expected to be part of v4.

[hollow hare]

Does it even work in Udio? I remember lots of people complaining about the inpainting. Suno has it slated for v4 and this topic is better at #🗣️┃chill-chat

[broken lotus]

Do you come here every day to say the same thing?

[opaque pasture]

Is there like a list with words that can prompt Suno into not generating it? Or do i need to refer to a list from ChatGPT?

[broken lotus]

then have a time out troll.
Maybe the feature will be available when you are allowed back?

[hollow hare]

Hm I am pretty sure it would have the issue with the rotten corpses (I can definitely see references of it in the holocaust or with IRL wars going on).

Now, if you said "roten corps's", that may work, but you will have to check with GPT what specifically got censored and if the phonetic spellings helped stop it (as long as your lyrics abides by Suno ToS so no discrimination [against non-imaginary races], etc.)

[hollow hare]

Ah, you can say "avoid talking about corpses" as part of the instructions for example

[unique stone]

Could be nice if inpainting in v4 doesn't require a paid subscription to use.

[opaque pasture]

Zombies aren't a non-imaginary race? Should i be worried?

[hollow hare]

Well, there have been a few cases where people came back after being declared dead for a bit of time (reason for saying 'you are not dead until you are warm and dead', etc.)... then there were those cases with the bath salt people in Florida a few backs that were referenced as zombies for a particular reason...

[broken lotus]

to the AI we are all fictional.

[hollow hare]

Ah right, I wanted to end off saying that you can get a bit creative with GPT @opaque pasture

When I generate, I use a modified prompt of the following for random songs that I had the imagery for in mind: #📚┃suno-school message

[nimble smelt]

Is there a way to look up a song I did by lyrics?

[hollow hare]

Not yet sadly but you can create a playlist in the future 😦 You will have to go digging through your library

[thorny cypress]

You can also try using another AI model to write the lyrics, if you have access to one that's uncensored. Mixtral 8x22b is pretty good for song lyrics, as long as you give it info for what kinds of themes to use.

[glacial scarab]

This is the second time I see you recommending this. Are you unaware suno itself will flag the lyrics once you hit create?

[thorny cypress]

Well, I'm just saying, it's easier to get something that won't trip the filter if you generate the lyrics separately somewhere else.

I rarely actually generate lyrics with Suno anymore because it tends to do weird things with rhyme and meter, or the music styles bleed into the lyrics, and I'd rather not waste the few credits I have on stuff that turns out horrible.

Easier to use something else where you can edit the hell out of the lyrics before you give them to Suno heh

[glacial scarab]

uhm... All Paid I hope?
Or else you just requested help on breaking TOS

[broken lotus]

🤦‍♀️

[vital kelp]

please excuse me i am EXTREMELY new to making AI vm

I trained my model with RVC and I want to use it with Bark
I ran the installer (pip install git+https://github.com/suno-ai/bark.git)
Where is the GUI for it?
Did i miss something in the instructions for it?

[final hazel]

Bark is no longer used here. But one of our moderators is an OG in this: @untold briar
Maybe he can help you.

[vital kelp]

womp womp

what is used here? is it multilingual and can 'express emotions' like i n bark

[final hazel]

We "used" to have Bark bot (text to speech) and the Chirp bot (text to music).
Bark is open source now, and chirp evolved to the website https://suno.com/

Some helpful starting links: https://suno.wiki/
https://youtu.be/lj8utBd_IUM

[nimble smelt]

My playlist keeps bugging too tho multiple browsers hard to tell which ones I did to try and find my song.

[unique stone]

Ugh... the vocals on this song I'm trying to work with are great but they're not strong enough for UVR to pick them up consistently (keeps putting a lot of it in the instrumental stem) and the various things I've thrown at it just make it sound awful. EQ helped a bit. My idea of using lower bitrate MP3 for the vocal was just awful lol

[inner verge]

Hey is anyone willing to share how they might possibly get some kind of old very retro robotic voice? Like something like those old speak and spell machines?
I tried all kinds of things as far as trying to instruct what kind of singer i want, [robotic voice] [speak and spell voice], [retro low-fi robot voice] and it just mostly kept giving me results of really strong autotune vocals.
Not expecting any answers but just kind of curious if anyone has any ideas.

[wet minnow]

the most robotic voice you probably will get is monotone robotic voice or vocaloid. but its always more like a glados/ turret from portal voice instead of the super grainy speak n spell

[inner verge]

Yeah that sounds about right, It's gonna probably some really weird chance if i ever get it but thank you 👍

[wet minnow]

here is an example of what i ment 😄 #💬┃general-chat message

[inner verge]

Lool, yeah that's basically vocaloid, I really want to try and get that retro daft punk vocals but; im wondering if they are working on anything related to actually customizing the singer better?

It'd be great to actually have a separate description box for that.

[rustic star]

i have a questin...can i influence that my next song get´s the same voice as a song before?

[wet minnow]

maybe in direction of talkbox or vocoder? but i think i had no luck with these

[inner verge]

I've had 50/50 results, Usually better results when more of the song is made.

[inner verge]

Actually im just gonna go put the vocal singer descrption box in suggestions, has it been suggested before?

[rustic star]

the problem is...when i extend the old finished song by 00:01 sec he starts with the first sentence-.-

[wet minnow]

the only solution i see is a selfmade seed beginning... a piece of music taht has voice.. and then you always extend that into different songs.

[rustic star]

i wrote u in ger

[thorny cypress]

No but I discovered "male vampire vocals" always gives the same voice. So you could probably use a trick like that.

[thorny cypress]

Hmmm. I wonder if "retro tts" would get that. Maybe add monologue, there was an early text to speech thing called that but it might get confused with a normal person monologue.

[void marten]

Anyone could share the prompt for lofi kind of music?

[ebon ice]

how do you get these 3-4 mins song on suno that i see ? am i missing something

[broken lotus]

https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[hollow hare]

https://suno.com/playlist/71bbcf97-bbe1-44fc-89aa-553b723c6643

[lyric hare]

can somone please explain inpainting? Ive tried udio but i cant get consistancy between extensions. is this supposed to fix that?

[broken lotus]

This isn't a Udio channel.
This is Suno.

[lyric hare]

ok sure but @mecd asked about inpainting, ive only seen that option in udio, and i want to know what it would bring to Suno? why should i care about inpainting?

[broken lotus]

It should allow to freeze most of a clip, except (for example) 01:23 - 01:41 where a line was pronounced wrong, or spelled wrong...
it will re-generate the clip, the frozen parts will stay the same, the unfrozen part will be 'painted' new.

the equivalent of spot painting a bad hand in Stable Diffusion, or removing something, or adding...

[lyric hare]

ok cool, yea that would be a cool feature

thanks

is there any way to get the audio to not go all fuzzy after you buy a membership? if i make a song on a free account, it sounds great, even if i then extend it later on my paid account. But if i start on my paid account, its just distorted and almost painful to listen to.

are they dropping quantization for paid accounts? seems to go all fuzzy after about 60 sec

[broken lotus]

it's 1 bot. the features are the same.

and no it has not changed

[lyric hare]

ive gotten quite consistant a/b results. if you start on a free accoun the audio is much better all the way through.

[broken lotus]

it's 1 bot. the features are the same.

[lyric hare]

its the primary reason i started looking elsewhere, because its so bad its almost unusable for anything extended.

[broken lotus]

Maybe have a listen to some other people's songs?

I hear 100 songs posted every day.

[lyric hare]

i have, its pretty consistant degradation, especially for rock songs. electronica seems to be a little better. but rock just turns into fuzz

[broken lotus]

Yeah, rock is the worst probably.
I use 'dreamcore' and 'sinistercore' now instead of rock... same vibe, but it takes breaks, and that seems to clear the noise.

[hollow hare]

^ Try using a more specific subgenre as, typically, the more wide range genres tends to have more degradation IMO.

If it was in a series of extensions, going to an earlier part and extending (as usually additional parts kind of reinforces the degradation and it compounds when the AI detects and emphasizes it as it relearns the style ~1m before the continue point). This is a bit more manageable as you can usually stich the parts together in a DAW.

[lyric hare]

i see so keep the generaton shorter, so you end up with more instrumental after the lyrics?

[hollow hare]

Depends on what you are going for. I try to maximize as much time as I can because I try not to go past a Part 3 extension (well, usually I combine Parts 1 + 2 first then Extend that for 3) so it does not build the degradation much

Just remember that, when using extensions, the AI reads ~1 min before the Extend from time to relearn/mimic the style (which is why extensions are 1 kin long instead of 2, the first min used for learning).

It does not remember the initial styling so if you had a lot of degradation in your base/part 1 track, it is better to regenerate a new one if you are planning to extend

[nimble stratus]

How do I efficiently permanently remove clips from my trash ON AN I PAD I have thousands and removing them one at a time takes hours and hours. HELP!

[hollow hare]

Unfortunately I believe that is the only method available at the moment

[rain glade]

Have a problem with suno '/chirp' command. Please help, why?

[hollow hare]

Chirp bot is retired, use the website https://suno.com

Some helpful starting links: https://suno.wiki
https://youtu.be/lj8utBd_IUM

THERE ARE NO OFFICIAL MOBILE APPS. Mobile apps promoting Suno merely redirects to the main site for a fee/subscription. Do not give your critical info (e.g. passwords, financial info) through these apps.

[spring ice]

I am looking for more words that I can add in a prompt to make better nursery rhymes / children's songs. All the help is appreciated.

[grim tulip]

Hello is there any way to change the thumbnail/generated art that is produced with each new creation? Not a huge deal but would be nice to use the art from other creations and use it for my final song edit as the cover art, thanks for any answers

[paper garnet]

Couldn't generate that.
Prompt flagged for moderation suno says that, what should i do?

[solid pivot]

At least on finnish language, there are few common names associated with children and children storybooks, such as Muppe which is either dogs name or cutesy way to say "it's a dog". For some reason suno managed to generate great children song about dog named Muppe.

[supple lintel]

Do anyone know if there is a way to change parts of the song? Say I like the song but not the last ten seconds. Can I change only that part?

[graceful condor]

You can try to extend from a specific timestamp (like 10 seconds before it ends). If it's the end of the song you may get a few seconds of silence. Currently there's no inpainting

[pallid leaf]

hmm I'm experiencing a lot of issues where extending song keeps quality ok..ish for like 3 minutes, but any extension after that starts so generate a lot of overlapping noise.

Here's the 3-minute song https://suno.com/song/158fa50c-ef70-4b86-ad23-f0f7e5a141d4 and here's "typical" extension for it https://suno.com/song/2ff5e454-f676-4c82-ad82-30fad930e948

hmm actually the noise starts at last seconds of the 3-minute song so maybe the extension uses that part too much

[broken lotus]

Yep, cut out that sound, and extend from a time before.

[cold sandal]

Good morning, I am a subscriber to Suno musica, and I need some advice, I would like to point out that I only use it on Italian lyrics, but now both male and female voices are always the same, even with different types of melodies, how can I change the

[fluid robin]

i wonder is there a way to make this prompt work female and male duet ?
so that it will mix betwean each singer during the song

[broken lotus]

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

[fluid robin]

thank u

[gusty wren]

is the singing that suno does on its own gibberish or actual words and phrases

[broken lotus]

it's 'hallucinations'...
something like a stream-of-thought babble, once it starts it just keeps going...

I use to get a LOT in German and Russian in v3 alpha...

[gusty wren]

I still do when i use gibberish

[broken lotus]

alpha constantly hallucinated... and then it couldn't end songs...

[gusty wren]

if they're actual words though, then they're something I don't want in my songs

[gusty wren]

oof

[broken lotus]

the alpha-stans are cray-cray.... 😅
I don't think they even used it.

[clever pendant]

anyone managed to generate the sound of metal hitting metal?

or specific sounds, that is

[broken lotus]

I have not had luck making 'factory' or 'metalwork' sounds...
Always wanted percussion on pots and pans..., I haven't figured out a way to do it other than the odd random one-off soundfx...

[pliant tartan]

How to generate Phonk using Suno

[gusty wren]

I think I barely used it when I first started here

[broken lotus]

there is a lot of discussion in the Brazil channel https://discord.com/channels/1069381916492562582/1154779408989687848

it's a pretty common question. Just search for 'phonk'

[clever pendant]

how about, having a clapping beat throughout a song. thats pretty common, right?

[broken lotus]

mmm I think
fingersnaps
works..

I would guees something like
handclaps
...?

[clever pendant]

eh nah, specific theme im going with, either claps or floor stomps

[thorny cypress]

I've heard it do hand claps during a song, it's just how to get it to do it on purpose heh

[broken lotus]

How many times have I yelled at the computer
"You KNOW how to do that! You did it before!"

[thorny cypress]

If you want it in a specific rhythm you could try

Blah blah lyrics [clap] [clap]
Blah blah lyrics [clap]

But that's just a random guess that may not work heh

Just don't do (clap) because it'd probably have a backup singer literally SAY "clap" lmao

[hollow hare]

It got censored from ChatGPT (which moderates the lyrics in Suno) not liking something. Ask OpenAI's ChatGPT itself about "what would censor the following: [content]" and it should tell you.

Usual causes are using swear words or interpreting harmful situations. Even if it worked before, it could be a small change caused its interpretation to become different.

I remember people getting censored for using bones, ribcage, and skeletons because it made ChatGPT assuming some sort of horrific tale.

[hollow hare]

You can prompt the AI on what to generate for the thumbnail art by entering the description as the final line after an Ending line like [End] (or three, preferably in square brackets as well to help try to stop the AI from saying thr description outloud)

#📚┃suno-school message (I show an example a few messages after in that link)

[gritty cradle]

why i am not geting any server and chirp for creating music

where all it gone

[broken lotus]

Chirp has moved to https://suno.com

[neon gulch]

/chirp

[final hazel]

see the message above yours 😉

[cerulean lotus]

My biggest gripe with suno right now is the background noise in the upper frquencies that builds the more you extend a song

Sometimes it can be averted by a [guitar solo] or some change to the format, but it returns pretty quickly

[hollow hare]

The issue is that the AI does not remember the original style used. What it does, for each extension, is study the style ~1m prior to the continuation point to learn/mimic the styling you had (which is why extensions are 1 min long as the other min was reserved for studying).

So if you had a loudness or degradation issue within that minute, it would start becoming emphasized as that is what the AI picked up on and all future extends builds up the issue as long as it exists within the 1m of continuing

[cerulean lotus]

Yeah, I guess that checks out, but I would also guess that there could or should be some RNG that created the overall style of the piece that it could hinge upon as well, in order to keep the base structural integrity or something. And the analyze for tempo, bridge, keychanges and such

[autumn jolt]

Me: 'Let's have a guitar solo'
Suno: 'No, I don't think so'
Me: 'Ok, a saxophone solo'
Suno: 'Hmmm, no we'll repeat the chorus'
Me: 'Ok, a keyboard outro'
Suno: 'The song will end abruptly, with a crescendo of fuzzy static noise'

[cerulean lotus]

Lyrics: (...) Style: Fuzzy Static Noise

[autumn jolt]

It's annoying when you can hear something good deep down in the mix, then there's the beat and static over the top

[unique stone]

Sometimes my style prompt gets ignored until the following extension, like it needs time to brew a bit

[autumn jolt]

I just got a clean sax solo... then saw I'd just generated a brand new clip and not extended 😆

[thorny cypress]

That's what happened with my sheep song. I should actually try to make the song I originally intended to make lol

[deep copper]

quick question there
can you generate sound effects ? ( e.g : a baby crying)

[thorny cypress]

Ask Mr. Tree lol

(it's possible but I don't know what works and what doesn't)

One of those things where you can waste 374941 credits trying to get it to do what you want.

[deep copper]

🥲

[north drum]

holy that's quite a lot

[warped remnant]

https://www.bandlab.com/mastering <- if you need a quick free algo mastering

[minor ginkgo]

/chirp

[wet minnow]

Chirp is retired. Please use https://www.suno.com to make music.
Also have a look at the community Wiki for a guide on how to use SUNO. https://www.suno.wiki/ ✌️

[tranquil garden]

How do you get it to have a male voice? I'm trying to get it to have a voice like Gunther where he's kind of just talking instead of singing.

[mental widget]

Создать

[final hazel]

🐤 🪦

[final hazel]

Chirp has moved to https://suno.com/

[wicked ridge]

So I generated a song with Suno that was meant to be instrumental, but it added random lyrics that sound like they're supposed to be Russian/Ukrainian, but my friend who speaks both says it's total gibberish nonsense. What's up with that?

[wet minnow]

these are called hallucinations. some genres love to sing. and if you dont have lyrics suno makes its own 😄 . it doesnt happen always, but some of them are pretty cool 😄 i have made songs out of some of them.

[wicked ridge]

I'm not super well versed with Suno and its technicalities beyond just generating things, that's super interesting!

[wet minnow]

i also have no idea of how it internally works. i just put in some words and a creative style and it does magic 😄

[hollow hare]

If you want to help avoid hallucination, use instrumental tags to help fill out the lyrics box instead of the instrument switch

https://www.suno.wiki/faq/metatags/instrumental-tags/

[wicked ridge]

Thanks!

I'm going to guess how/why it's doing this

It's basically borrowing from its database and frankensteining the various lyrics/vocals together based off of what it knows

Sort of like how AI image generation works

[unique stone]

Pretty much. Some genres are more prone to hallucinations due to those genres normally having words.

[wicked ridge]

Makes sense, yeah

[unique stone]

(but I don't know how it actually works any more than anyone else here!)

[wicked ridge]

Maybe a dev for Suno has some insight as to why, if they were here and wanted to be transparent about it

Tbh I'm just curious, it's almost like if an alien was trying to speak a language but didn't get words

https://suno.com/song/c30ee98b-3ae2-48cf-a3e4-3229269bbe0b

this is the song that's speaking fake Ukrainian if anyone's curious

[final hazel]

Ai just makes up things. Just like ChatGPT still does.

[unique stone]

It's "word-like sounds" basically. Sometimes you DO get actual words mixed in, including bad words on rare occasions.

Some people don't like them. I prefer them over writing lyrics xD

[final hazel]

It's simple tbh. If you try to generate an instrumental in a genre that is not often an instrumental in real life, but more a song with lyrics, it will make up lyrics and start to sing.

[cerulean flower]

hi everyone ! i'm starting to use Suno and trying some things !
Let's say I have a chorus in my song but each instance of the chorus has slightly different lyrics (i like the progression for the narrative; like "it won't work" becoming "it could work" then in the final "it will work")
Is there a way to do that ? [chorus] seems to replicate itself anywhere i put [chorus] 🤔

[barren vault]

Try [Chorus 1] [Chorus 2] and so on. For me it works well.

[cerulean flower]

ohhhhh thanks !

[neat field]

does anyone else have more examples of "angry" prompt?

only one i have is miku (which is awesome and unique) but wanting to hear any others - https://suno.com/song/6abf428d-6d62-4fa9-80c4-8b5f643154dd

[lean magnet]

can we use this channel for tech questions? im seeing an issue with Extend.

[cerulean flower]

last question for me, is there a keyword to make the song understand it must end ? like [instrumental] etc

[graceful condor]

Hello I signed into My discord account so I could get the 10 songs thing for some reason when i try to create a song its on a forever loading screen and it says an error occured. What does this mean?

[cerulean flower]

oops it seems [end] does it, i'm dumb ! thanks XD

[neat field]

sure, how do you reproduce the issue?

[lean magnet]

the extend feature isnt working. the app is telling me that its refunding 2 tokens. im on the monthly sub. @neat field

[neat field]

it will refund any generations less than a certain length, are these generations meant for an end to a song ?

[lean magnet]

yes, im trying to extend over the 2mins that it normally gives for a song

[neat field]

you could also try to extend from 01:59 or 01:58 and see if that helps

the model is probably trying to end your song with a short generation

[lean magnet]

ahh yes. its asking to extend after 1:55 and the tune is 2mins

i getcha

[neat field]

gl

[lean magnet]

thanks...will give that a go now 👍

[hollow hare]

https://www.suno.wiki/faq/making-music/how-do-i-end-the-song/

More than just [End]

[hollow hare]

I really wish they kept the old phrasing, easier to understand "Continue from"

[lean magnet]

im new to this tonight after seeing a TikTok. are there better tutorials here?

[hollow hare]

A general starting point copy-paste:

Chirp bot is retired, use the website https://suno.com

Some helpful starting links: https://suno.wiki
https://youtu.be/lj8utBd_IUM

THERE ARE NO OFFICIAL MOBILE APPS. Mobile apps promoting Suno merely redirects to the main site for a fee/subscription. Do not give your critical info (e.g. passwords, financial info) through these apps.

I found it pretty straight forward. If you do have questions, simple ones can be easily answered at #💬┃general-chat or you can ask here (but response time will likely take a bit longer)

[lean magnet]

thats great thank you for the link^ and help

[dry violet]

does anyone know how to make similiar intro as in here:
https://www.youtube.com/watch?v=eZ1IGBaFa7A

im struggling to achieve anything which is even remotely close, i've created a reddit post with more info here if someones interested. Any help is greatly appreciated!
https://www.reddit.com/r/SunoAI/comments/1cr8iua/help_needed_epic_cinematic_orchestra_instrumental/

[stoic mason]

this is not an extension, this is a base prompt. the result was 15 seconds long and there was no refunded credits: https://suno.com/song/f2ae41d9-691d-41e4-9132-0726058d1226

[shy parcel]

Hey im going to upload a song in a moment but first I would like to know how to change the graphics? Can I put anything on it or? sorry total newbie here 🙂

Anyone wanna help me out

[wet minnow]

you can add a super short end part with instructions to make a picture. suno uses the last part for the picture shown. for the last part search a silence at the end of your song and extend it with the style "end, short" ane [end] in the textbox make some free lines and add your picture description in (brackets) ... here is an example of this technique https://suno.com/song/fa4f872f-a0bf-4e4a-9705-d736c6df29ac

[autumn perch]

Sound degradation is the real problem fr long songs.... longer songs more degradation:

https://suno.com/song/268c13a6-54d2-42f2-a35e-280d14bb38f8

[marsh trail]

Is there a way to "fix" a generated song, like just keeping what's generated and the general vibe but shifting text elsewhere ?

[final hazel]

Not at the moment. The suno team is however working on V4 and the talked about adding "paint-in" to that version. And then you will be able to repair or change things. So we are all looking forward to that. But we have no idea when 4V will come out though.

[thorny cypress]

Is it even to the Soon ™️ stage yet? Lol

[sterile ice]

I'm trying to add instruments by changing the prompt after extending the song, but it's not listening to my prompts at all

Any idea?

Like starting a piano instrumental, then changing to acoustic guitar

[thorny cypress]

Yeah I had a problem with that when I tried to piggyback a song off of harpsichord. Even the masters couldn't figure out how to get it to bring in a pipe organ lol

[agile obsidian]

I have a question about the subscription. If the subscription gives X amount of credits and I don't use those credits at all throughout the month, will the credits double on the refill day or will it only refill the amount that the subscription offers? (if that makes sense)

[thorny cypress]

Any unused credits are lost. It only refills to the amount the subscription is supposed to give you.

[thorny cypress]

(which honestly sounds kinda lame... If you paid for them, you ought to get them, period. I mean if you're not using them, you're not costing the devs anything.)

[agile obsidian]

yeah, I think that a little bit of a rip off. No offense.

[final hazel]

Incorrect.. There is a legit reason.
This policy is in place because Suno.com reserves computing power based on fixed monthly credit usage. Allowing credits to accumulate can lead to unpredictable demand, making it difficult for us to manage resources effectively. This unpredictability can cause service disruptions and increased operational costs. By resetting credits monthly, Suno ensures a smooth and consistent service for all users.

[unique stone]

Other services work this way as well.
(One being my phone provider which gives some amount of GB of data per month, which I lose if I don't use it)

[thorny cypress]

Hmmm that's true. And back when phone plans had limits, most companies didn't have rollover minutes. Still doesn't mean it's fair though.

But in the end, the policy is spelled out on the website and users agree to it when they subscribe. So they're still getting what they paid for.

[unique stone]

The wording used to be a bit less clear, so they did change it to be more obvious at least (a few months ago).

[thorny cypress]

Still seems a bit sucky that companies can do that though. I use a text AI service that includes credits with subscriptions, which are used for certain premium features. The users would be outraged if suddenly the credits they paid for could expire.

But I don't know what their model is as far as how they reserve compute. I just know they've had times when everything went down because they ran out, or some system didn't kick in like it was supposed to.

(although honestly when you run an AI service these days, there's so many moving parts that could go down and shut it down. Hell one time AWS went down because it was Black Friday. Ain't nothing you can do about that lol)

[unique stone]

I just try to make sure I use all my credits up within the month (sadly I don't seem to have a problem doing that lol)

[hot cargo]

[End]

[unique stone]

I've been attempting to remove some of the fuzziness from some really nice vocals. It's tricky because of how engrained they are with the rest of the instruments. I'll revisit it with fresh ears tomorrow and probably hate it but so far I've applied a small amount of EQ, then split off the high end (where it has lots of hiss) and put it through an effect that does a kind of mp3 compression on it, then put that through a very short reverb before mixing it back in with the low frequencies. It seems... okay. The aim was to smooth/dull the high frequencies without lowering their volume.

[spiral kelp]

Im also wondering if it's possible to create loop music, i've tried it several times using the tag "loop music" but it never worked lol.

[gray sky]

hey, I'm not able to write any commands to generate with chirp. It just says, "Failed to load messages"

anyone know why this is happening?

[hollow hare]

Chirp bot is retired, use the website https://suno.com

Some helpful starting links: https://suno.wiki
https://youtu.be/lj8utBd_IUM

THERE ARE NO OFFICIAL MOBILE APPS. Mobile apps promoting Suno merely redirects to the main site for a fee/subscription. Do not give your critical info (e.g. passwords, financial info) through these apps.

[rocky perch]

Hi all. I have an issue with suno consisting in the following: at the beginning, I get pretty clean vocals. But with each subsequent part, it becomes more electronic and more artifacts accumulate in it. Are there ways to deal with this?

[fluid robin]

for the (Chorus) thingy to work do i put (Chorus x2 ) or outside ?

[fluid robin]

i encounterd the same problem had to waste 3 attempts in order to get a good voice

[rocky perch]

If only that... I waste far more than 3 attempts usually

[fluid robin]

oof and i am on the free version but i noticed with to many promps when in custom mode the voice some times gets corrupted

[rocky perch]

If the song is long enough, then I spend 350 to 500 credits to get good quality. And I'm on a fre version too XD

[fluid robin]

O_o

[rocky perch]

No, you just have to multiply 50 credits by the number of days XDDD

[fluid robin]

thats atleast one weeks grind of just using free points and i dint now that u could add the points thought they only stayed at 50

[rocky perch]

As far as I remember, there is no possibility to grind in suno? Which is... you get 50 each day but you can't save more?

[fluid robin]

I dunno havent tried to save the credits they always get used up each day

[rocky perch]

Then yeah. 50 is a cap

So yeah... up to 10 days to make a good song. I feel like a real musician, lol

[fluid robin]

i noticed when using in custom lyrics OAOAOAOAOAOAOAOAOAOAOAoaoAoAoAoAoaOAoAoaoAoAoaoAoaOAAOoAoAOAAOAOAOAOAOAAAAOAAoAoaoAAOAOAOAOAOA in the whole song the program will just make its own song based on that

[smoky cipher]

Is there a way to define pronunciation for a word? For example I want to use "toa" pronounce "toe-ah" but I can't get it to come out right

[unreal trench]

google for phonetics

[trail fulcrum]

Does anyone know when the update to version V4 will be released? And what new features are there

[grave wraith]

Is there anywhere a page to see what expressions Suno understands? And what it can do actually? Like I tried to make a duet but it's always a female voice only (I guess it just doesn't understand my prompt) Looking for the limits and to get the utmost out of it

[hollow hare]

https://suno.wiki is a great place to start.

Getting duets... that is more luck than anything else. Try to use male and female sounding names in tags, that might help, but it is still hard to achieve.

A few people were able to do it but at an expense to their credits:
3Daizy's duet
MrTree's duet
Wiki entry

Again, getting it working on purpose is hard. Alternative methods:

1. It is easier to make a duet by doing something like generating instrumental part 1 then Extend it twice (one extension where you specify for male vocals in the styles field, the other for females), or something similar for each alternating part, then edit them together in a DAW.

2. Even easier (and cheaper) is to make a full track then swap the voices gender using other AI tools like kits.ai (see #🗣️┃chill-chat message and #💡┃suggestions-web message ).
   You can use a DAW to edit which voice track plays when.

[grave wraith]

Thanks...that's something to start with

[sly goblet]

Hello, who can help me? the credits in my account have not yet been reset after paying. Is there any waiting time or would it be immediate?

[broken lotus]

it should be immediate. On the website, go to the subscription page, find a button about your billing history. That will show your invoices...

If the latest payment had a problem there should be a button to retry payment.... or there should be confirmation of the payment, and you'll need to contact
[e-mail]

(also make sure you are logged in the correct acct)

[sly goblet]

thank you for your support. Okay, I’ll do that. The problem is that when you send an email to [e-mail], it takes them 1 month to respond.

[split hound]

Hi! Just recently discovered Suno! So good!! 😄

One issue I have, which is my bad, I made this awesome song (tbh it's fire), but I didn't realise it would pronounce a lyric '$' as an S

I was wondering if anyone knows how to edit a clip and just change a couple things but keep the same tune 🥹

[cerulean lotus]

Extend from the timestamp just before that line, write the proper line and rest of song and pray to the RNG gods 🙂

[split hound]

The tune is literally PERFECT, which is the issue lol

I literally just want to take out 3 $ signs and put the word dollars in one spot

[broken lotus]

next model (v4) is expected to built around 'in-painting' where we can mask part of the clip (to freeze it), and regenerate specific parts that had small issues like this.

No ETA on v4. The Devs have been working on it, but I haven't seen a beta or alpha yet.

[split hound]

damnit, you think there is anyway dev could take it and try do an edit themselves? lol

we want to use it for an announcement

😂

memecoin drop

it's fire haha

everyone loves it

just those tinyyyyy errors

If this is at all possible we would be so grateful 🥹

[granite crystal]

Hello everyone, I'm trying to use suno locally with an api endpoint, but when I call this:

  const audioResponse = await hf.textToSpeech({
    model: "suno/bark-small",
    inputs: story,
  });

I get an error:

    at /Users/mr.t/Code/my-epic-tale/node_modules/@huggingface/inference/dist/index.cjs:260:11
    at Generator.next (<anonymous>)
    at asyncGeneratorStep (/Users/mr.t/Code/my-epic-tale/node_modules/@babel/runtime/helpers/asyncToGenerator.js:3:24)
    at _next (/Users/mr.t/Code/my-epic-tale/node_modules/@babel/runtime/helpers/asyncToGenerator.js:22:9)
    at processTicksAndRejections (node:internal/process/task_queues:95:5)```

Do you know where that could come from maybe? Thanks 🙂

[split hound]

Any way to contact dev for this to get it maybe?

Want to get this edit done asap lol

[glacial scarab]

gonna wait for V4 like anyone else, and even then backwards compatibility to V3 songs is not guaranteed.
I don't think a dev will custom fix a song for a random user.

[split hound]

Haha you never know, and it is a really minor fix compared to udio edits I've seen

Does dev have an @

[glacial scarab]

idk try and see if you get timeouted if you do you got a dev tagged

[final hazel]

Hailing a DEV for this, will get you a time out.. It is not a critical thing

[split hound]

We are wanting to use it to announce a token drop to a 18k community soon is all

[unique stone]

The only time I've seen a developer assist with a specific song was when someone tried to make a long piece of music and it broke the "get whole song" function.

[final hazel]

You and everybody else. You don't have any privileges above the rest.

[split hound]

Damn I didn't know the attitude in the servers was like this 😅

I wouldn't have bothered asking lol

[granite crystal]

Your attitude is pretty rude tbh @split hound ...

[unique stone]

Just extend the song from the part you want to change, or put it in another program and edit it.

[glacial scarab]

Hire a producer to recreate the song without the problem if its important 🤷‍♂️
If not wait like everyone else

[split hound]

It's just because we only just discovered suno like yesterday?

[final hazel]

I only see one person with an attitude and that is you @split hound
Thinking that a DEV should give you access before the rest.

[split hound]

We were going to do something different, but then when we found this 👀

[final hazel]

no mini modding needed... @granite crystal

[split hound]

Just to explain, I literally want to remove 3 characters and add 7 characters

[glacial scarab]

Well if you still want to use suno and need it soon put the extension time before the issue and hope it fixes it and continues in a way you like. Thats how we do it. Or fix it in post if it's something tiny.

[granite crystal]

I've just stated the obvious, haven't moderated anyone as I don't have that right anyway 😆

[split hound]

Does that keep the same tune/beat?

Like overall

[glacial scarab]

Depends where the issue occures. At the start Probably not.
In the middle good chance it will continue with a very similar beat

It will not be 100% the same but with a few tries it could be similar enough

[final hazel]

We all have songs like that in our library. Every one of us. And this discord server has more than 370 k users.

[split hound]

If it helps the removal of 2 characters is just to make the word sound correct and that word IS said at one point in the song

I've never edited a song

But could grab just that word I guess and like copy paste it over the top

😅

[split hound]

Yeah I didn't realise that it reads $ as an S

[final hazel]

You don't get the point. Anyhow.. you will have to wait for V4. And V4 will have the paint in (it will not work for V3 songs).

[honest stratus]

Songs can be uploaded to Spotify if I've paid for the SUNO subscription?

thnks

[warped furnace]

I don't know what changed over night (not my lyrics) but Suno has finally accepted them without complaining. Scratching my head. 😄 ChatGTP wasn't very helpful in clearing up the problem.

[final hazel]

ChatGPT does the gatekeeping. Maybe it was busy last night because a lot of user were trying out their new model?

[warped furnace]

Maybe. 😁 I had time and credits on my hands to make the song. It's just a tad annoying when something doesn't work without knowing what's wrong so you can work on a solution.

[thick stirrup]

chatGPT was updated yesterday to a new version. Probably just different behavior

[warped furnace]

Good to know. Have never used it before. That explains the problems with Suno as they use it as well.

[warped furnace]

Well ... Here wo go again. Just had the Song made. Wanted to have another version ... And now it needs moderation? 🤪 😅

[torn parrot]

There are a lot of "ethical" roadblocks. Why does the AI want to roadblock so much? I was trying to make a song in the style of a band (backstreet boys) but it got flagged. Tried other bands and it was working fine

Then in the lyrics it mentioned NSync lol

[hollow hare]

The only issue you will have is the distributor rejecting your work, especially if the song sounds very AI like

TuneCore and CDbaby are known to reject AI songs

DistroKid and Ditto accept AI songs

LandR are primarily known to accept AI work bur recently rejected someone (sounded too AI like maybe? We will see with Feline Music test): #💬┃general-chat message

--

Also note you need atleast 1000 plays in 12 months for Spotify monetization #💬┃general-chat message

[tranquil vector]

/chimp

[wet minnow]

Chimp is retired. Please use https://www.suno.com to make music.
Also have a look at the community Wiki for a guide on how to use SUNO. https://www.suno.wiki/ ✌️

[final hazel]

https://tenor.com/view/monkey-ape-smile-digibyte-dgb-gif-7030981627894415237

[hollow hare]

A copy-paste of one of my response so do not feel targeted please:

Using existing artists or song titles will likely be ignored, if not censored. Very likely such information was stripped out from the training to avoid lawsuits (assuming if they were included in the first place).

You will need to use style keywords in custom mode to get as close to the genres you want.

E.g. you may want to try Ibiza Trance as one of the styles for ATB-like music (https://rateyourmusic.com/artist/atb)

Though I recommend looking at a specific song for more accurate styles. There are other sites as well to identify song genres like https://sonoteller.ai and Chosic's music genre finder (Spotify only, <
https://www.chosic.com/music-genre-finder/>)

[hollow hare]

Also the reason for the musician censorship is to stop people from making 1:1 copied lyrics which would be a violation of copyright AND against Suno policies

Remember this?
https://media.discordapp.net/attachments/1069381916492562585/1232951392071847988/Screenshot_2024-03-05_233857.png?ex=66445f76&is=66430df6&hm=d036dde2daf84beee4b2aab6d9f141966996d8a68ee6f7c4e7940341bb1c66e1&

[wet minnow]

they just see : #😜┃suno-memes message

[torn parrot]

Ahh

[simple osprey]

Is it possible to reuse the musicstyle but with a complete new text?

[hollow hare]

You can try to extend it, though if the lyrics starts very early, it likely won't have time to copy the melody 😦

Which is why some people generates an instrumental base generation first then Extend the lyrics after (depends on what you are after)

[simple osprey]

Ah yea, i could try that

Thanks!

[hollow hare]

You're welcome 🙂

[azure ruin]

I have no channels lik ehow to chirp and chirp 1 or smth

is that normal

[hollow hare]

Chirp bot is retired, use the website https://suno.com

Some helpful starting links: https://suno.wiki
https://youtu.be/lj8utBd_IUM

THERE ARE NO OFFICIAL MOBILE APPS. Mobile apps promoting Suno merely redirects to the main site for a fee/subscription. Do not give your critical info (e.g. passwords, financial info) through these apps.

[azure ruin]

oh ok thx

[near tangle]

is there a way to install Chirp/Suno locally?

jut noticed I might be in the wrong channel for this.

[unique stone]

It's only available via the website / there's no way to run it yourself.

[near tangle]

got it. look for an alternative it is

[hollow hare]

I do not believe there is any open source music AI available.

There are generative text to speech open sources though, like Suno's Bark

[near tangle]

"I do not believe there is any open source music AI available." currently

I'll wait

[hollow hare]

Oh, apparently there is this

https://github.com/facebookresearch/audiocraft

[karmic lion]

Gm!!! I have a question. I understand we get commercial rights. I created a song that I want to sell to a company. They are asking how I can transfer the "copyright" aspect for the song. I'm unaware how this works as you can see. I know there has to be paper work involved for the "Transfer" of rights or whatever. What do I do going forward? Thank you!

[warped furnace]

That's a question for a lawyer. I live in Germany there is no transfer of ownership rights they always stay with the creator. You can only transfer the right of use here. Don't know how it works in your country. If the company acquired rights before they should know.

[near tangle]

If you're in the USA you need a lawyer. Usually it's part of the contract you agree to when you sign the rights over to the customer. However, whenever dealing with contracts one should always consult a lawyer.

also, unless you've already signed the contract, don't sign away your rights.

a lawyer will help better explain why

[celest vapor]

Could you guys add more information when lyrics are censored that go beyond "this was moderated"? It would be helpful to know what exactly it doesn't like.

[hollow hare]

It got censored from ChatGPT (which moderates the lyrics in Suno) not liking something. Ask OpenAI's ChatGPT itself about "what would censor the following: [content]" and it should tell you.

Usual causes are using swear words or interpreting harmful situations. Even if it worked before, it could be a small change caused its interpretation to become different.

I remember people getting censored for using bones, ribcage, and skeletons because it made ChatGPT assuming some sort of horrific tale.

[celest vapor]

haha that's pretty funny. I'll try

It kinda sucks how, despite generative AI being brand new in the consumer market, it's already being turned into this overly-polticially correct, overly-cautious school marm. It'd be nice if it stayed the wild west a bit longer 😄

[hollow hare]

It is because the company is protecting itself, hence why the server is PG13 and the service ToS explicitly forbids several problematic situations:

https://media.discordapp.net/attachments/1102676217687195679/1236146831902244874/image0.jpg

[celest vapor]

Like, ChatGPT wouldn't even generate some story I was asking for, relating donald trump to some clearly nonsensical thing about peanut butter and turning into a fish because, in its eyes, someone might take it seriously and take offense

[hollow hare]

Because you know someone would make songs about holocaust denial, pro-genicide and who knows what else.

Hell, there was a guy who made a song about a serial rapist in action (bypassing censorship) and it was disturbing. That guy made an AI video too (with implications of the scene) and posted it on YT. Don't know if it got suspended yet since I nuked the link from my history

[grave wraith]

Does it understand a difference between 90s pop and 2024 pop in the prompt?

[celest vapor]

for sure

Great product, though. keep it up ❤️

[glacial scarab]

Hm I don't know how good it is with getting it completely right. But it seems to understand that it's not the recent stuff you want when you put a number.

[grave wraith]

That would be cool as pop changed a lot since then 😄

[celest vapor]

Have you tried "medieval pop"? Maybe there's enough temporal disparity with that one 🙂

[grave wraith]

I would have to lie but I think there is no 70s pop at least 😄

[glacial scarab]

Ah yea it does something at least. You can report back if you test it for pop. I only tested with punk. I put 70s/80s... It was not really 70s 80s just you know older

and thats what I wanted anyway just not modern stuff

[grave wraith]

Well..I am playing around and it's hard nowadays as they mix everything together. I mean...to be right...they all did always but since the 2000s they don't do actual pop... more a mix of everything they can find 😄

[warped furnace]

That's not quite the problem, I think. I got flagged yesterday. Asked GPT if the lyrics were okay it said yes (after alterations). Suno still rejected them. Today I made a song with them. No changes at all. I'd say it's a glitch. Some hours before I got a new song. Now I want another style (still same lyrics) and he flagged me again. 🤓
@celest vapor just try again. And again. (If your lyrics are clean. Suno accepted f^ck^ng without batting an eye.)

[celest vapor]

Yeah, it's very confused about the world sometimes, it seems 🙂

[normal niche]

I have no problem with not allowing the use of most artists names...or maybe it's really some when it comes down to it. I tried to include the phrase "vanilla ice cream" in one of my songs and couldn't, because Vanilla Ice. And I can't use my wife's name in a song, because apparently there's an artist I've never heard of called Carla.

[wet minnow]

how about eyes instead of ice? or Karla? creative workarounds...

[broken lotus]

I had 'sputnik' get blocked 😡

vanilla ice-cream
vanilla-ice-cream
-vanilla ice cream

something like this should work

[normal niche]

I ended up going the “vuh-nilla” route.

[normal niche]

And yeah, Karla. But that didn’t make my wife happy. 😂

[thorny cypress]

Honestly, AI needs to stop taking things so literally. Like, theoretically you could do something "in the style of Bruce Springsteen" without it actually BEING Bruce Springsteen. But AI goes "oh, you want Bruce Springsteen voice? Here you go!"

[broken lotus]

make it look deliberate
Carlaaaaaa
😁

[meager quiver]

Is it wrong to have like nearly 100 versions of my deez nuts song already? xD

[sharp frost]

Hi how can i do a "duo" in my music ?*

[wet minnow]

yes but its very difficult. https://www.suno.wiki/faq/style-and-lyrics/duets/ https://suno.com/song/1b4f8607-1a8c-464e-b57d-d0f6089b28e8

[sharp frost]

ohhh thx !!

[thin harness]

I've made awesome song for my project right before a quick update you did 30 min ago! Now it's all gone. Is there a way to restore it? Messaged to billing @ sino aswell

[final hazel]

[e-mail]

[thin harness]

Thanks! Will message now!

[balmy crown]

I am trying to figure out if I'm experiencing a bug or sleep deprivation and either are possible

has anyone else not gotten credits today?

sorry for bringing the extra uncertainty of potential human error into t his

i blame my cat, who regrets nothing

BUT. I don't think I've used Suno yet today (I remember rushing to finish a version of a song before the time rolled over yesterday evening, and I haven't used it since as far as I'm aware)

but I have 0 credits

[hollow hare]

Daily credit refreshes at 0300 UTC+0 (<t:1713236400:t> in your local timezone)

[warped furnace]

Your cat ate them? 🙈 😄 I got mine today. Had used them all up yesterday. I use the free version if it's of importance. I only had problems with prompts and being flagged.

[balmy crown]

I use the free version, and it would've been before 11 in my timezone yesterday evening

lol the cat didn't eat them the cat interrupted my sleep so many times that I'm a shell of a human with only a tenuous grasp on reality lol

okay but I double checked my reality and

the last song I generated was 10:30ish yesterday evening

(ignore the fact that the song description for style is OTT)

ah well, it's only a day. if it continues to not refill by 11:00 tonight I'll be more concerned, I guess

this is just the time of day that I usually start to play around with it so I was like "wait... where.... my credits..."

[hollow hare]

I believe you can contact [e-mail] to investigate. Provide as much details as you can (when you last generated a song with the UTC offset, when you noticed the issue. Maybe wait to see if it refills tonight and let them know what happens in the same email? Etc.)

If you sent them an email saying you got no credits, you may be simply ignored. However, with having much more detail available, it is more likely you will get credits for the missing day(s).

You will likely have to wait a few business days for a response back, especially with how huge the user base is and I suspect how small the team, esp. Support team, may be

[balmy crown]

Yeah, I was just wondering if it was a common issue; if it's a one off, I don't want to bother them, since I'm a free user and have no business complaining about it haha

[hollow hare]

Understandable but it would be something that the team would definitely be interested in knowing about and checking the system for any bugs, etc. so it does not affect a wide user base in the future

[balmy crown]

oh, I guess

I hadn't thought of it that way; i'm usually the person who runs into glitches with software that's been out of beta for ages, so I just figured "ah it's one of those weird things that I have generated with my cursed tech aura" (very scientific, I know)

Okay, I will do a write up just in case

[hollow hare]

IIRC, someone had a similar issue but they were affected for several days

[balmy crown]

oh dear, I'm definitely hoping that won't happen to me, but if it does, it does; it'll give me time to compose better lyrics

[hollow hare]

One moment

Just adjust your timezone to say it is EDT (UTC-04:00)

[balmy crown]

ah, thank you, that is a better format

[hollow hare]

You're welcome. Most people forget that daylights saving has their own timezone offset.

Granted, the team should know this, but still... hope that helps. Everything else looks good in the message you had

[balmy crown]

thank you so much Orz

(Also TIL about timezone names; I always just went off what I had to select when a computer asked "what time zone are you in" and used the pull-down menu to select the one that sounded right. I had to google what EDT was and why it was my timezone but a different acronym than I was expecting.)

[hollow hare]

Heh, it is one of those things that the typical person does not need to know about. I learned it because I have to account for timezones in my profession.

Also the advocating of ISO 8601 time standard (YYYY-MM-DD as the most basic form) which helps a lot for archiving and sorting on computers (e.g. "240514_xxx_v8.doc" instead of having "xxx_v8_final_final(3)_May 14 2024.doc", etc.)

For those curious:
EST = Eastern Standard Time (UTC-0500)
EDT = Eastern Daylight Time (UTC-0400)

[balmy crown]

I feel really called out about that file name

[narrow hedge]

Hello!
Im trying to add [intro] in all my songs, but I think none of them starts with intro, it goes directly to the lyrics.

WHat should I do?

[balmy crown]

intro also could be read as "these are the intro lyrics"

the same as [outro] goes before final lyrics

you might try something like [guitar solo]

or [musical interlude] or something? idk

uh get someone else's advice before following mine though

I once accidentally made an entire song accoustic somehow

without pressing the accoustic option

(Instrumental, sorry not accoustic)

[hollow hare]

That seems about right, I usually do [Catchy instrumental, Short] in some of the songs

Other times, I just generate a bunch of instrumental only music for various styles (with various instrumental tags) and extend those. (Usually when I want to burn my daily credits without having anything specific in mind)
https://www.suno.wiki/faq/metatags/instrumental-tags/

[balmy crown]

oh, that's clever

[final hazel]

@balmy crown https://youtu.be/lj8utBd_IUM might be able to help you too

[balmy crown]

#📚┃suno-school message
just in case you missed it, there's a really good set of suggestions 😄

(sorry if it's redundant, I just wanted to make sure to @ the first person who brought the topic up)

also, ty Daizy!

[narrow hedge]

thanks a lot man!!

[tacit mauve]

I really start to hate this fucking piano in all my instrumentals, there is no way to change the instrument?

someone knows any prompt to change the instruments?

[hollow hare]

You could try defining what instruments you want in the styles.

You can also use instrumental tags to help specify what instruments you want to play and how: https://www.suno.wiki/faq/metatags/instrumental-tags/

[shy spear]

PLEASE can someone tell me a way to better insure male only vocals? The prompt I used to get them was working great, now Im starting to feel like the ai is somehow purpsoely wasting my credits lol

Also, any way to better make sure I DONT get multiple voices? telling it solo or solo singer doesnt seem to do it like ever

[hollow hare]

You can add more male vocal terms in the style section to give it more weight. Using male names or male vocals in metatag might help as well (or could be a placebo...)

Alternatively, make a full track then swap the voices gender using other AI tools like kits.ai (see #🗣️┃chill-chat message and #💡┃suggestions-web message ).

[shy spear]

Oh I kinda did that with my first song. Tried Spencer Charnus, sounded like Chris Motionless lol

The part irking me is I want a solo singer and it always seems to FORCE an orchestral chorus

I need help re-writing some lyrics if anyone is willing please. The ai will use or at least generate curse words, allow works like blood, gore, and killer, but not ones like Cannibal, flesh, and the like

Trying to do a horror core track about the antagonist in my indie horror and its proving harder than I thought

[hollow hare]

~~Try to avoid using Chorus IMO if you can. ~~
Actually, I haven't tried specifying specific types of choruses like [male solo chorus], but maybe that would help?

[shy spear]

Let me rephrase. It forces multiple singers randomly through an otherwise great track

[hollow hare]

Try giving the AI additional instruction in the prompts like saying "avoid cursing".

For the latter part, unfortunately ChatGPT is the one that moderates lyrics and can be dumb (and thinks of vivid gory images that it censors when you use stuff like skulls and bones).

One solution is to use phonetic spellings for the troublesome words, so something like Canibal, Canybal, Canniball, etc. for cannbial

[shy spear]

Which is weird af because it helped me turn the concept into the starting lyrics and included those words lol

I know your name is a different character than what came ot my mind, but I find it funny because Im currently attempting a song about Stolas lol

[hollow hare]

It's like an NPC cop. Sells you drugs, turns around, acts all surprised for possessing what it just sold you because it does not remember anything

[glacial urchin]

Would I be able to put a song I generated using my own lyrics in suno on Spotify?

[hollow hare]

If you have a subscription plan, you will have the opportunity to do so as the copyright for the base generations made during the subscription plan would belong to you. You cannot monetize and must disclose you used Suno for the free plan base generations and those youbextend from.

To upload to Spotify, you need a distributor and it is VERY important you read the terms and conditions for these services as it will directly affect the ownership and royalties. See Top Music Attorney coverage about TuneCore (sus ToS) and other distributors (#🗣️┃chill-chat message ).

Note that Spotify now (since Apr 2024) requires 1000 plays within the past 12 months to be part of the monetization system.

The distributors can deny distributing your song, especially if they sound AI like. See the following for more info #📚┃suno-school message

[glacial urchin]

Would I be able to have a distributor distribute it to Spotify if I generated the song on the free plan and then purchased a subscription

[hollow hare]

Any base generated made during the free plans are illegible for copyright, even if you extended during a subscription (covered in the FAQ).

I believe this is because Suno does imprint an inaudible audio watermark (acknowledged in #📢┃announcements ) and I would not be surprised if songs made during the free plan had one noting it down, which is why they add the line about the former situation (and to curbtail service abuse)

Some distributors may require proof of ownership. Be sure to do your research into available options and select the appropriate service for your copyright music. Spotify does provide a list of distributors that deals with them

[glacial urchin]

So what you are saying is that I cannot use distro kid to distribute the song to Spotify because it was generated on the free plan

[hollow hare]

Legally, you can be held liable for any legal action Suno may challenge you upon discovering that you distributed a song which had the base generation made during the free plan as you would not own copyright commercial license nor ownership to the music (as it would belong to Suno but it would also depend on your local laws about how valid that is)

[glacial urchin]

Thanks

[hollow hare]

Currently, the laws about AI music and ownership is murky but there are some companies that would be willing to enter a court battle to help make examples/presidence for future legal battles and/or laws (many companies would try to avoid it, of course, but if they had sufficient evidence and the potential law in favor for them...)

Well, other than about impersonation. Those are not murky, very clear, and that is aspect that the Tennessee ELVIS law was for

[glacial urchin]

Yea

[glacial urchin]

I wonder if there is a way to purchase the song from suno

Like if someone wants copyright permissions so they can have it distributed even if they don’t have a subscription

[hollow hare]

I believe there was one case that had happened but details are hushed about it. I suspect it would be more effort and cost than generating a new variant during the subscription plan

[frail smelt]

So we can't distribute our ai generated songs, right?

[frozen pivot]

Cannibal = Can he bal (😀 ) try that an see.
Just use the phonetic pronouncement of the words..., not the right spellings

[shy spear]

Will do 😄

[frozen pivot]

👍🏻

[ruby ferry]

can this ai use non instrument sounds as instruments? having no luck getting a steampunk sounding thing out of this

ran out of prompts trying to get it to replace the background drums with steam train chuffing

[frozen pivot]

Looking for sound effects?

[ruby ferry]

I suppose

kinda just wanted to hear a steam train as drums lol

or just clockwork sounds in general would be cool really... thinking of Luis Humanoide's steampunk stuff on youtube

[lyric hare]

you cant copyright AI generated works. full stop, this has already been fully fought and the ruling is clear. If there is signficant human work, eg you wrote the lyrics, then the copyright belongs to the human. Suno is providing no human works to the song, therefore they have no possibe copyright claim to the output.

[frozen pivot]

I am not sure of what you want..., Have you tried creating just an instrumental with your style?
Otherwise...
If Suno can't do it, try
https://elevenlabs.io/

[ruby ferry]

yea i was doin instrumentals from the start

[frozen pivot]

It's not for music...

[lyric hare]

elevenlabs is pretty much the expert on AI generated voices. Ive seen a demo of something that they are releasing soon for vocals in your own cloned voice. Market is getting spicy

we worked with them extensively on a disability aid project to reconstruct voices for people with speech pathology (ALS, CP, etc)

they can do amazing stuff cloning voices from just a few phonemes.

[frozen pivot]

Yeah!!!
I don't want to discuss a potential competitor here... Just a suggestion 😀

But Both Suno and Elevenlabs are changing my life

[ruby ferry]

it seems like this is pretty good for just getting some music to slap in the background of a portfolio project

[lyric hare]

yea the quality degradation problem is still there. extending a song turns it into mud.

[lyric hare]

changing a lot of lives. My non verbal daugher has a voice, her own voice, now. and soon she will be able to sing.

[frozen pivot]

I agree, if that's what you are looking for...
The Music is excellent for TV commercials as well, and more

[frozen pivot]

This is wonderful... She can do anything with this tech

[lyric hare]

https://www.youtube.com/watch?v=d8k4Pit4_ZU

game over for suno unless they have something in the chamber ready to go to counter. Elevenlabs has 1000x the compute power.

[ruby ferry]

but how configurable is the instrument part of the music?

[frozen pivot]

Well...,
Let's get ready to create and have fun...
I currently subscribe to Suno and Elevenlabs 😂

[lyric hare]

well that is the trick isnt it. this is where we need a multi modal stem model. give it a few notes to prompt and let it come up with the various instruments, then the vocals. I still see humans needing to be in the loop, because its humans listining to the final product. AI has no idea what "is good or not good" but humans do,

[lyric hare]

we are elevenlabs partners for the disability aid project.

[frozen pivot]

That is where producers come in...
The human hear will be necessary for the final product

[lyric hare]

exactly, it actually opens up creativity, new people with new ideas can now produce things that were unreachable to them before. from an evolutionry standpoint we are going through a period of massive diversification.

[frozen pivot]

Great to know...
I've been producing for a long time...
I am loving this tech

[ruby ferry]

ill definitly sub to one of these once I get somthin that matches my exact ideal music

just need somthin to give my steampunk themed section of my animation portfolio some spice

[lyric hare]

its the creatatious period for human creativity.

[lyric hare]

If i can get consistancy on style that would be huge.

forget the autogenerated lyrics, the are garbage. work on riff editing and prompting. thats the money right there.

[ruby ferry]

if I could describe what i need in terms of artists, it would be if Luis Humanoide used train sounds instead of just clockwork

[lyric hare]

that i would like to hear lol

this stuff, lets push music to places no one ever thought of going,

[frozen pivot]

I've not even used the Auto Generated Lyrics..., I use custom!
I am now a SUPER STAR with 5 songs. With 20 more being edited after unsteming them

[lyric hare]

johnny cash, sung by my cat.

[lyric hare]

i used it for a funny song, but nothing serious.

ive always written lyrics an songs in my head, i just cant play anthing other than a soundboard or digital editor. i can finally get (most) of them out. the censorship thing is still an issue, but ive talked to eleven labs about this, because its also a major issue with disability aids, and they are going to be pretty open. we wont have to cowtow to the crapgpt censor for sure.

[hollow hare]

Ah right. I keep mixing up copyright and ownership. @glacial urchin see the reply and #💬┃general-chat message

[frozen pivot]

Ok.
We'll connect here again. Greetings from Maryland

[lyric hare]

https://suno.com/song/0f3086da-c2a2-4767-aaeb-8652145371de

[lyric hare]

greets from PNW

[frozen pivot]

This is very nice

[ruby ferry]

fairly catchy yea

ill just post the closest ive gotten to what i need

https://suno.com/song/1d0e603c-f77c-4485-8e91-40dfb44cc834

it kind of matched the rhythm of a train chuffing noise in the beginning

but idk its kind of the opposite of what I asked for lol

this was before I specifically asked it for steam train chuffing sounds tho

so idk if it knows what that means xd

[frozen pivot]

PNW on the Canada or America side?

[lyric hare]

I would try ripx daw to get the stems, then remix on blackmagic and change one of the instruments to steam or insert in some effects

[shy spear]

That worked great, found a way to make it say other stuff stoo 😄

Take a listen see which sounds better?

https://suno.com/song/c7cb05c9-0032-4cdc-92f4-0e447b896b08

https://suno.com/song/334a9d61-a608-4450-a3e6-8a8d43cbe39d

Kinda had mainly Spencer in mind writing this. No help on the lyrics when I wrote it too.

It's just verse 1 (and yes the one is supposed to say start)

[frozen pivot]

Great... I use both RipX and Spectralayers from Steinberg...
Actually, Cubase is my DAW from its inception

[lyric hare]

Canada

[ruby ferry]

oh if theres ways of directly replacing stuff I've got a few ideas of how I can go about this better, gonna try rewording prompts to specifically sound "like" rythmic train or clockwork sounds

thanks a bunch, out of prompts tho, ill try again tomorrow and sub if I can make somthin I like out of it

[frozen pivot]

Nice chat...
Will catch you later. Thanks for your time
https://www.suno.wiki/

[frozen pivot]

Try this... if You've not yet
https://www.suno.wiki/

[ruby ferry]

woa, theres a wiki?

noice

[frozen pivot]

Not always accurate, but helpful...
Catch you all later

[frozen pivot]

😀

[frozen pivot]

I like both...
You are the producer here. Suno is the Studio.
Try your local dialect and see what happens...

[frozen pivot]

😀

[azure crane]

So now it’s only possible to download from desktop mode on mobile?

Some genres starts in the middle of a song though I have added some different intro tags… It’s so annoying…

[split hound]

Damn I tried this...if only one of the errors wasn't in the second line lol changes the whole thing, sort of managed to get half of it fixed up, but it changed one vocal part which was epic, oh well

new model thing should be good hopefully for ppl to use

[broken lotus]

@deft geode

"Can you separate the vocals for each verse so that they are sung by different singers? Despite using several commands, the songs are still sung by either a male or a female singer alone. For example, I want the first verse to be sung by a male and the second verse to be sung by a female. How can I achieve this?"

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

[broken tangle]

wait guys sprry if im stupid im new to suno but is there a way to generate only 1 song at a time and not 2

[broken lotus]

you always get 2 clips on every 'job'. No way to change it...

[jagged loom]

for now? I mean, it would be great if there was a way

well, just saying im content with having 2 songs

[nocturne lichen]

Hello I want genrate voice casino tendering so I need to add emotions in voice How can I add?

where I can get tag list

[potent dagger]

Hi, sorry if this question was answered already, but does anyone know if I have credits and pay for a subsription, do they get added or do I lose them?

[wet minnow]

you have extra credits shown in the credits menu? these stay (even after the sub ends) and will be used last after the paid subscription credits. sub credits are added on top

[potent dagger]

ty!

[grizzled slate]

My Workflow:

( A Style of Music comprises the Genre and the elements that direct or describe the Genre.)

First, I find a Style of Music that suits the Song I've written, often with the help of ChatGPT and the like.

I then use that Style of Music to create an Instrumental and regenerate until I get an Instrumental that fits the feel and theme of my Song.

I use the Instrumental as my Intro.

I Extended the Intro, setting the time I want Verse 1 to start.

Next, I Click on ... to the right of the Verse I like and Get the whole Song.

This process is repeated for each part of the Song.

In the final section: the Outro I use a lyrical tag that I can get to work, such as [Fade out Outro] [End]

Tip: For each Extention, I start by Refreshing the page, then click Extention. This seems to help reduce those so called hallucinations.

Well, I hope this helps someone.

I'd like to hear about your workflow!

[unreal trench]

Tip: For each Extention, I start by Refreshing the page, then click Extention. This seems to help reduce those so called hallucinations. --> Really? sounds interesting, i must try that sometime...

[willow axle]

Which kind of attributes does exist to create songs?

[broken lotus]

Try here... covers many subjects
https://www.suno.wiki

[zinc stream]

hi

i am getting issue with payments can anyone guide me

like when i go for subscription it gives me error

[flat robin]

Sometimes I want something in the style of some known artist or song, but I'm not so good at summarizing the peculiarities I'm looking for, and English is not my first language. Not to mention I obviously can't name artists in the style prompt.
So I open Copilot and paste this:

I need you to describe a song. You should summarize its genre, voice style, prominent instruments, and every other peculiarity in a maximum of 120 characters. No need for a bulleted list, or even a complete sentence: just a comma-separated list of the most characterizing features of the song. Your output will be fed into a text box in a music library management software.
Some guidelines: omit the word "genre"; do mention the gender of the vocalist; do not name singers or players, briefly describe their style instead; do not name specific instrument makes or models, briefly describe the vibe they give to the song instead; if at all possible, no item in the list should be comprised of more than 3 words.
The song I want you to describe is "song title" by artist.

It won't always yield a perfect style prompt, but it's been of great help to me so far.

[glacial urchin]

So does that mean I own the song or suno does

[sly grail]

is there a way to scroll through pages in the "create" window? I have an older project I want to tackle again but the "create" window only shows the most recent 12 projects

[warped furnace]

Thanks for your tip. Hadn't thought of that. 👍

[limber bramble]

All my songs dissappeared any support?

[final hazel]

The devs are notified! @limber bramble

[limber bramble]

Thanks will wait

[dawn stone]

is there an issue with song generation / extension atm or is it just me?

[final hazel]

yes. the devs are investigating @dawn stone

[dawn stone]

oh ok alright, thanks 🙂

[hazy relic]

I paid for a subscription and now my unsued credits are gone. I have not shut down my account. WHERE ARE MY UNSUED CREDITS I PAID FOR?

[final hazel]

Quote from the website: Credits included in subscriptions do not carry over from day to day or month to month. Purchased top up credits do not expire, but require an active subscription to use. See the terms of service for limitations on commercial use. Email us at [e-mail] with any questions.

[autumn dagger]

I'm especially having an issue with endings and fragments of sound popping up in the fade out/silence

[echo atlas]

¿There seems to be a bug in your case. Or is it an update?

[final hazel]

The devs are working on Suno atm. The "extend" option is not working as intended (read: not working). A fix will be pushed ASAP

Because we don't have that permission as moderators.

[upbeat heron]

will you make an announcement when its fixed again?

[final hazel]

I will not officially. I am a moderator, but help people. So if I know I will tell here.

[upbeat heron]

thank you

[final hazel]

Should be fixed now. See -> #🪲┃bugs-web message

[dusty cipher]

Thanks! This is exactly what I was looking for. I was having some difficulty getting any kind of intro but I see I'm not making it the smaller steps I should. Awesome and thanks again for the advice!

[broken lotus]

it seems to have something to do with text in the Extend being a repeat from earlier text (like a chorus).

I had a stubborn song that I fussed with a while and couldn't get the lyrics to work... but then set the Extend From timestamp 1sec earlier and the extends started doing the lyrics as written...

So that's my advice... adjust the time stamp and it may sing correctly.

[neat nexus]

Adjusting the resume from time can be magic. Definitely play with this.

[karmic lion]

Thanx for the info. I had a chat with chatgpt and got the info there, too. 👏🏼

[silk basin]

What's the best way to get Suno to give you a voice with really insane vocal range? Does anyone know?

[snow harbor]

Do you know any service that does the same as midjourney /describe or clipinterrogator but for music? Would be great to interrogate an AI about how would it describe any uploaded music or sound

[thorny cypress]

I dunno, in my limited experience when it tries to do that, it ends up sounding very autotuned and hits octaves that shouldn't exist.

[lyric hare]

ok WTF time.. WHY is it singing my style metas?

https://suno.com/song/74e161f3-eb0b-4ca1-9b6e-744ec222e9aa

[harsh fiber]

If you rename a song, is there a certain amount of time it takes for that name change to appear in your downloaded video? I made the name change, but my current downloaded video has the "old" song name. If I give it time and download again, will the new name be there instead?

[silk basin]

Man, I'm hoping one day we can get ranges as expressive as people like Freddie Mercury or Josh Ramsay.

[craggy delta]

i made a good track, but, the intro seems to be cutted out, is there any way to fix this? or adding a intro that fix that cutted intro?

[grim talon]

Hi, I'm wondering if there is any way to prompt a singing voice for specific lines when creating a rap song?

[thorny cypress]

You can't edit the track or "extend" onto the beginning, no. Only way is to keep trying until you get one with a good intro.

[craggy delta]

ok, i'll deal with the haters in youtube saying the intro is broken, heck is a good song! 🙂

also send it to a masterizer

and it sounds good _;)

[thorny cypress]

You can either use stuff like [instrumental] or [intro] or maybe [fade in] in the place where you put the lyrics (on custom), or you can generate a part of instrumental specifically for the intro and extend at whatever time stamp you want to bring the lyrics in.

[harsh fiber]

Alternatively you could load it into Audacity and do some manual edits if you really want to keep things. 🙂

[craggy delta]

tried that david! maybe if i see the song become popular i'll do a intro in Fl studi

but i doubt it XD

with my luck, only 24 people will see my video

this is the song i made

[thorny cypress]

I don't know but I think [singing] would probably work?

[craggy delta]

this is the song i made with the cutted intro XD

how does it sounds now it's been masterized?

[harsh fiber]

Sounds good to me, how much more of an intro are you looking to get? It doesn't feel too "rushed" at the beginning to me . . . but that's just my opinion. 🙂

[shy spear]

I finally sort of figured out why I keep getting a major echo/reverb/ multiple signer effect when I dont call for it

Because I am calling for it, with the lyrics. The words echoes is part of my one song, so I spelled it phonetically instead of alphabetically. No echo

[coarse socket]

Hi there, sometimes EDM song aren't in good quality... Just wanted to know if someone know if in v4 there's going to be some improvements regarding audio quality? Thanks!

[broken lotus]

tags are too long
https://www.suno.wiki/faq/metatags/

[shy spear]

Now I am facing the new problem of it totally ignoring an entire verse I have entered and skipping to the next part

And it is using lyrics from a prior section that I did not include at all

[final hazel]

That happens to all of us now and then. Sometimes the AI forgot where it was. So there are a few things you can try:

• change the continue point by one or 2 seconds
• put back some old lyrics

There is no cookie cut answer. Sometimes you have to puzzle tbh.

[shy spear]

I am usign way too many credits that Im not getting refunded for this lol

[shy spear]

Its actually singing an entire verse from three parts back now that I did that

[final hazel]

I consider it normal to use at least 2000 credits per song.. at least

[fluid robin]

How long do u need to wait for a refil of credits when subscribed to the monthly billing ?

[final hazel]

Maybe my video helps. I show a few trouble shoot options: https://youtu.be/lj8utBd_IUM @shy spear

[final hazel]

It will always refresh on the same day of the month and time you started. So if you started on the 26th of April at 8 pm, you will get the next one on the 26th of may at 8 pm.

[fluid robin]

Thank u started to getting more drawned into the monthly version due to that i am getting the hang out of the music prompts now

but i noticed that some of the prompts might corrupt the ai voice completely

[final hazel]

Yes, it's not perfect yet. But V4 promises us better quality and a paint-in feature to edit and redo certain parts in the song.

[fluid robin]

i cant even imagine the stuff that V5 will be able to do

[final hazel]

Interesting times for sure!

[fluid robin]

I noticed that the ai sometimes tries to make its own song out of Random lyrics for this one as a example it tried to make a anime type of song i only used the words a-o in the first verse and randomized them in the text

[shy spear]

lol and now that I figured that out it's singing my things in the brackets []

Dude nobody is gonna fall for it, give up

[fluid robin]

haha that was fast

[shy spear]

I'm redoing Blood Moon Symphony. The first version seems to reverbish and electronic for good metalcore

[fluid robin]

decent prompt that works for my songs Futurstic Robot Voice Toaaster Rythmic electro ancient swing

[shy spear]

There we go, I think I like V2 better

https://suno.com/song/06aab661-f507-46cc-a5a6-fd807e2a5eff

I had to spell so many thing phonetically lyric video is so bad lol

[fluid robin]

how should i put the music prompts in order to get it to sound like a 1950 rock song ?

[final hazel]

I often make rockabilly. I have a playlist you can take a look my "style of music": https://suno.com/playlist/57f59e80-2535-4e46-9a89-b4870c347979

[fluid robin]

https://suno.com/song/2f91f15d-7b71-41be-9ade-acb3bbd6f97f Just wonder what i could have done different tried to make it more a humming song but the vocal ai decide to whistle most of the song

[opaque canopy]

is it possible to "force'' or influence a custom song tu put a guitar solo or a drum breakdown.........

[untold sonnet]

Please help! I have a dilemma, but not sure how or if it can be resolved. I'm creating an original song with my own lyrics. One of the renditions Suno created was a perfect vision for what I had in mind. However, the lyrics need to be tweaked and finished. I have no way of repeating that particular song it created. I've burned through hundreds of credits trying. I also tried using the Extend option, but even that comes out sounding different than the original version it produced. I thought maybe I could piece it all together in Audacity if it did. I wish there was a seed option or something that would allow me to consistently get the same sound and tempo as I experiment with my song until I feel it's perfect. I'm really impressed with the potential of Suno, but right now it's feels like I'm just stumbling around in the dark hoping to get lucky.

[opaque canopy]

thats how it work luck... but i want to know how to influence a certain behavier something like ... [verse 1] [guitar solo] ..... [chorus 1] [drum breakdown]

is that possible

they said the less of music style you specify the better the sound quality is ... i tried and cant really tell i still ear that 'organ background noise' all the time lol

not complaining but want to help improve it with feedback

[final hazel]

https://youtu.be/lj8utBd_IUM
https://suno.wiki

I think those will help you @opaque canopy ^^

[opaque canopy]

i read all the faq

it has give me some clues but still...

[final hazel]

your questions are answered there..

[untold sonnet]

A seed option would be great. If that would possibly work. There must be something behind each rendition that would help separate it form the others and allow you to use it again.

[opaque canopy]

yess something like ... you like this one the voice the tone but not the feel just re-roll with the same input but give somethin new

[opaque canopy]

i was hoping getting better answer here than the online FAQ sorry for asking

im watching your video now maybe ill learn something new thanks!

[final hazel]

AI software, like Suno for music generation, isn't like traditional software with fixed rules and predictable behavior. Instead, it learns and evolves over time, making it hard to create consistent documentation. The developers know what they taught the AI model, but they don’t fully understand how to interact with it yet, so there's no manual. We're all still learning, and sometimes users know more about it than the developers. Traditional software is stable and predictable, but AI software is flexible, adaptable, and constantly improving, requiring continuous feedback and updates.

[untold sonnet]

Just finished watching your video. Will try some of those tips! Already burned through 1,000 credits today. Still have 1,500 to go... hope I get what I'm looking from before reaching 0.

[opaque canopy]

if i write [verse 1] [guitar solo] does it influence it ?

i tried and i cannot really tell

[final hazel]

I usually for a good song burn at least 2000.. but I am extremly picky

You can check other people lyrics.. it shows how they do it. Just click on this link, and you can see how I add {bass walk] in my lyrics, and you actually hear them too: https://suno.com/song/4be44af9-641d-4071-b362-e6030c851ce5 @opaque canopy

So yeah, metatags work. However, not always... Sometimes it ignores it, but if you ask things that work in the music you are trying to make, you will have more influence...

For instance, double bass works in rockabilly, but it's harder if you want to add it to techno

[opaque canopy]

should i try [guitar solo] or [guitar] [solo]

[final hazel]

I often use [guitar riff} but I used [guitar solo] in the past.

[shy spear]

Im so mad, I did some work fixing the music overshadowing the vocals and it sounds fine. After export I have a terrible overly loud record scratch sound effect part way through my sound that was there before T_T

[lyric marsh]

It is normal if you try to customize everything. Once the Ai starts adding its own artistic flare you can't get it back on track.

[copper kernel]

2000 is crazy

I use 50 and I get music I genially listen to

[lyric marsh]

Nice tunes. 🙂

... (@angryknights)

[shy spear]

Wow the censor is ridiculous lol I can't use the normal spelling of the word pig haha

Or porcine wtf XD

[somber crane]

where can i make suggestions to the devs

[opaque canopy]

there is a lot i would like also to ask the devs and suggest lol 🙂

am I the only one who try to make metal on suno?

[eternal badge]

Thank you for the tips. They were very useful.

[inner jay]

I got charged after cancelling and I don't have the credits for the month I just got charged, premier plan. Need help

[shy spear]

https://suno.com/song/fb24d621-19fa-4194-bb29-bc2737cf9861

Cleaned up, easier to understand, removed annoying record scratch, background sounds in spots

[hollow hare]

Contact [e-mail]

[grim folio]

Hi everyone! How can I change my song longer than 2 minutes? If I extend my song, it makes a new song.

[hollow hare]

You can combine part 2+ with the previous parts to make the entire song via Get Whole Song in the ... menu of the part 2+ song.

See https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[grim folio]

Thanks, but can I change my song lyrics?

[steel night]

⚠️ Please extend the lyrics limit of 1,250 characters ‼️

While crating a song and switching the Menu, eg. from Create to Library and back to Create, all the editing is gone. Lyrics, Style and Title are empty. I start from the scratch 😩

[thick kettle]

Hello guys, i just see Suno website to make IA music, and i want to know if it's possible to select the language of the song, like french, or japanese, chinese voice ... ?

[silent flame]

hello, I lost credits when my subscription ended, why ?

[somber canopy]

Anybody knows about "Prompt contained inappropriate material" I couldnt pass this 🙂 it doesnt have show or autocut options.

[radiant hornet]

https://suno.com/song/3c2ee254-1cf4-4201-9a0f-62f678c0a9cf

[foggy salmon]

Hi! Today I was somewhat taken aback when I indicated in the prompt in direct text "male singer" and in 4 attempts out of 5 I got female vocals...

[thick kettle]

yes i just test it with only japanese text from chatgpt, i just want to ask you guys, what kind of plan you have ? and if the 10$ one is good 🙂

[frozen pivot]

All depends on how many credits you might need. The songs are all the same

[thick kettle]

Okay ty 😄

[unique stone]

I am on the $10 one, paying monthly. I wasn't sure how much I'd use it so that's what I started with and it seems to be enough for me

[glacial scarab]

https://tenor.com/view/too-many-counting-gif-15929632

how?

[unique stone]

I don't actually make that much stuff or I overwhelm myself with ideas. Only time I run out of credits is if I'm repeatedly generating different extensions off the same thing to get variations to play with.

If I'm doing other stuff I might not make anything for a couple of days

[glacial scarab]

Hm.. maybe I should stop putting random stuff into suno when bored 🤔

[vivid charm]

Any news on the API? Eleven labs are launching their API soon

[glacial scarab]

I don't think the API has a high priority.

[vivid charm]

Alright, do you know why that is?

[glacial scarab]

As I am not a dev no.. But If I had to guess other issues like the next version and imroving quality is more important.

[unique stone]

What do you want to do with an API? (Just curious!)

[vivid charm]

Build cool stuff? I think most devs prefer an API to a consumer interface

Just look at OpenAI - they focus alot on the dev community, following the Apple playbook

[broken lotus]

I would like wildcards... it's a simple thing, but I make a lot of variations on my current lyrics.

I'm not even sure I need them randomized, but I would like to have some of my standards in format where I can select them quickly (like a prompt wrapper, maybe – a style shortcut for my own prompts)

(I also want a playlist downloader for my own clips)

[unique stone]

I wonder what additional terms might apply when an API is available. I mean, there could be some creative uses for it, but I suspect a lot of developers might just want to scalp the service (offer same functionality as the site, nothing more, but charge extra.)

I expect it to be paid subscription only at least.

[gusty leaf]

Hi everyone, I'm fairly new to Suno, been playing with it for a week or two. Today I subscribed because I was getting pretty good results. But since today, whenever I extend a song, it completely looses the melody of the verses and chorus. I'm not sure what I'm doing wrong since it worked fine before. I did look at the faq and saw tutorials.

[final hazel]

The song you're struggling with probably has a slow tempo. The AI only has a memory span of one minute. If you're trying to recall a melody from a verse or chorus that was played more than a minute ago, it won't remember. To improve this, create full songs in between and then extend them. This helps the AI retain the melody better.

[gusty leaf]

ah that makes sense, I was working on a song I started a few days ago. I'l try it your way, thanks 🙂

[azure crane]

Does [Drop Beat] works only for a specific genre? I try to silent the drums in a track.

Maybe there is a better tag to use?

[thorny cypress]

Heh, if I had unlimited use I'd burn through credits like water. Sometimes it can take a day's worth of free credits just to get it to pronounce one word right. And if you do nonstandard stuff like pipe organ in a heavy metal song, credits don't last long. It'd be awesome to be able to experiment the way I want to, but I don't see that happening any time soon.

[gusty leaf]

with a finished full version, is there a way to correct the lyrics or text that displays when you play the song?

[final hazel]

Not atm.

[near tangle]

is there a way to force a short instrumental intro before the lyrics of a song without creating an instrumental generation first?

I've tried using [Intro] but so far it's getting skipped and it goes straight to the lyrics

[broken lotus]

[intro] isn't enough. you need to prompt it with more instruction.

[synthpop buildup]

[instrumental hook]

etc...

but metatags can't 'force' anything.
the only 'force' is to use an instrumental as the 1st clip.

[near tangle]

:/

[sacred socket]

So, I just made this in free - if I now sign up for a paid plan, can I use the same prompt and get commercial use?

[near tangle]

if I understand correctly, only of the new generation

free gens are that, free. non-commercial use only.

[sacred socket]

but if I use the same lyrics and prompt in a paid plan, that output would I would own, yes?

[near tangle]

by my understanding, yes

[sacred socket]

👍

[soft arch]

[intro] [interlude] might work, or try repeating [intro] a few times.

[upbeat shore]

afaf

[soft arch]

It seems the sound quality degrades with every extend. I've wasted dozens of credits trying to finish a song but the sound is just extremely poor (dirty, muddy, staticky, buzzy) compared to the start. Doesn't matter if I extend an extend or extend after combining each extention to a whole song and going from there. Anyone else have this issue? Any solutions?

[near tangle]

that's normal. How many extends are you doing?

I usually have decent-ish results if I keep my parts under 5

[azure crane]

Hmm, drop drums is just impossible to get...

[soft arch]

The song is just under two minutes right now, not even that long. This only recently became an issue for me. I have longer songs without that issue.

[near tangle]

try to use the full 2 minutes for your first generation and then each sequential full 1 minute gen instead of breaking it into several second pieces.

the more you utilize of one generation the less degradation you'll experience

[soft arch]

It's 1:57. I can't remember for sure but I don't think I've shortened it. I think I'll just start from scratch even though I really liked the start. Sigh.

[near tangle]

you can go back to the first gen and reuse that prompt

[broken lotus]

maybe
[percussion break]
is what you want?

or are you trying to have a break without percussion?

[next kettle]

Every single song I've been generating lately is just beginning with vocals - Common?

[broken lotus]

Maybe? if you're not changing your prompt.

[hoary fiber]

I am extending a song, but the extension starts after few lines of my lyrics... how to make sure it stars from the beginning of the lyrics?

I mean the next segment starts from the +-middle of the lyrics I added for that segment...

[next kettle]

Seems like 80% of them, after trying a few random different prompts. If I use custom lyrics it starts very suddenly, ignoring all brackets.

[broken lotus]

I can't read your metatags or hear your music from here.
😏
the advanced technique is to use an instrumental for the 1st clip and then use Extend From Time to insert the lyrics exactly where we want.

[broken lotus]

try Extend From Time, and set the timestamp before those lyrics start.

[mystic flume]

Think of it like a conveyer belt. Your initial generation builds a rhythm and meter, and it's very difficult to interrupt that, or to force the model to change its mind (for lack of a better phrase). So if your lyrics are being cut off, you are jumping in at a bad place, even if you're just extending from the end of a clip. Find a better spot with trial and error.

[hoary fiber]

no, no... I am stepping into right place at the end of original track I want to extend... Lyrics from the first segment are ok, and music in the second segment is smoothly continuing the first one. But the lyrics which are written for the second segment does not start from the beginning in the second segment... I have 4 lines of lyrics and I hear only last 1.5 lines...

[hoary fiber]

I am unable to find any guidelines or manual for this...

[broken lotus]

Mybe this helps:
https://www.suno.wiki/faq/making-music/extend-from-time/

[mystic flume]

Right, but if the model is skipping lyrics, there's something wrong with the placement in my experience. One thing I will try is to add [Instrumental] or [Interlude] before the new lyrics to see if it'll convince the model to do what I want.

[sacred socket]

Hey all, on the Pro subscription, if you need more than two minute songs, what are the options?

[wet minnow]

free and paid work the same. you always have to extend the song https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[sacred socket]

thanks for the speedy reply, much appreciated.

@wet minnow oh, the link is great, but the Extend option is a whole heap of complicated , eg "Each generation will be labeled next to the clip’s name, but 2 clips are created with each job. You will see two Part 2, and two Part 3… If you re-roll a generation you may see more.

Selecting a continued clip will highlight the earlier parts, visible on the Create page.

[hollow hare]

It is one of those things that is easy to understand once you do it, though I do agree it is a bit overwhelming to understand initially

[sacred socket]

@wet minnow with Extend, do I have to crop the lyric box to only what was missing, or does the extend just go over the whole thing up to three minutes?

[wet minnow]

i remove all text from the textbox that is already sung and leave in there what needs to be. and i add some more text if it fits the 1 minute of a extension.

[sacred socket]

that's a good, clear answer. Thanks. 👍

[jolly tundra]

Every time I use it,[Chorus]After the command is given, the sound effects will become very confusing. Do you know this?

[dull coyote]

Hello, I need your help, I just created a very good song this morning, I am a Paid user but the song just disappeared and is nowhere to be found, not in Library, not in the trash, what happened?

[wet minnow]

have you disbaled all filters in the library? and have a look at all pages.

[dull coyote]

I already checked everywhere, is just not there, not sure why, I didn't remove it, nor dislike it, The trash is empty and the Library just has all the disliked songs

[wet minnow]

there are filters for the library

[dull coyote]

I already checked them all and it is not there, not sure what is going on here.

[dull coyote]

Is there a way to search the song by the name?

[wet minnow]

not at the moment. sorry

[dull coyote]

I already found it, it had a different name in the library not sure when I renamed it, oh boy I was scared 😄

[grim folio]

can i make the song longer only from the end or is it possible to change the beginning as well? for example add an intro

[opaque canopy]

you have to get it right from the start you cant cut and paste what you want where you want

[alpine wren]

So bark is the v2 model, and there is no open source weights for V3?

[graceful condor]

Yeah.. This static issue is pretty bad. I hadn't genned anything in like 2 weeks and I get some really great stuff but the static in the background just wrecks it.

[broken lotus]

no... and no...

Bark is text to speech. It is open source on github

v2 and v3 are both on the website, neither are open source. They are commercial

[surreal widget]

Is there a way to see the timestamp from which I had extended a track from? Sometimes I catch a mistake later and need to go back but it takes a bit of time to figure out where I should extend from.

[wet minnow]

if you post the videos you see the switches

[mystic flume]

For now, go into your browser cache; it'll still be there probably.

[lyric marsh]

Same issue, I've wasted over 1k credits trying to play with options to fix that [slower vocals] radomly works, but sometime dramatically changes the song flow. Tweaking custom songs is like playing the slot machines. Waste a lot of credits but keep hoping the next one is a win.

I hate when the Ai gets it right but I have a typo (ARG!!) or it just pronounces things wrong like "am" (As in "I am one') half the time is pronounce it as A.M. (as in A.M, P.M, F.M, lettters instead of pronouncing it.) or the short form for Heads Up Display (HUD), it pronounces it Hude (Heavy on the UUU). Frustrating.

[balmy crown]

for HUD you can try going H-U-D
you can also fudge like if it says I A.M. you can write lyrics that would sound almost the same like
Eye Yam

by that same token for HUD if H-U-D doesn't work you can try things like
aitch/eitch (or "haitch", depending on how you want the letter to be pronounced, if you're australian or something)
"you" for u
"dee" for d

etc

innnnn other news, I'm trying to do music like for a horror game, and while I can get things that are mostly in the minor key, and have music from a music-box as the background, I can't get a creepy children's choir singing things, and that's a bummer, because children singing lullaby music is a classic for some reason

[mystic flume]

Start with a good structure so you can build several verses by extending first generation several times rather than extending the extensions. it takes longer to get something that lines up well enough that you can fix it in Audition or whatever, but the quality is noticeably better. Plus it has the added benefit of keeping the context front of mind for the model.

[sacred cobalt]

Only through song Extend is it possible to write exactly the same song but with different verses, so it’s completely unclear why Extend is only a minute, and not at least two?

many people like the exact rhythms (and voices) of a certain song, and they would like to diversify only the text (but not the accompaniment with vocals)

and if you use only Reuse Prompt, then it generates basically all sorts of bullshit for the most part - it is very difficult to find something interesting, and if you find it, reusing it again (with other verses) is problematic (since Extend is only a minute long)

to sum up the question, why Extend songs for only one minute and not at least two? extend its duration to at least two minutes

[hollow hare]

The issue is that the AI does not remember the original style used. What it does, for each extension, is study the style ~1m prior to the continuation point to learn/mimic the styling you had (which is why extensions are 1 min long as the other min was reserved for studying).

So if you had a loudness or degradation issue within that minute, it would start becoming emphasized as that is what the AI picked up on and all future extends builds up the issue as long as it exists within the 1m of continuing.

It also explains why the style can change when you vary the starting point by a few seconds

[mystic flume]

Okay huge news. I did an experiment where I compared the final song built by Suno, and one where I downloaded and edited each individual clip myeslf. The quality difference is noticeable at a glance. Whatever process they're using to combine them (which does add information to bridge the two clips, so you can't just put them in a row and expect it to work) is also modifying them.

Here, the top track is the "full song" mix downloaded directly from Suno after I was finished. Track 2 is the clips edited together. The first thing I noticed is how much louder the mix is. So they're not just stitching the clips, they're normalizing too (or just increasing the gain).

For 3 of the 4 transitions I was able to just overlap a little, but for the final one there was a small gap and some data had been added between them that I couldn't create, so I just snipped a tiny bit of the full song mp3 and used that to patch them together. Worked great.

Now, does that correspond to a decrease or increase in sound quality? Hard to say. I ran them both through the same eq settings in Audition, and they both sound okay to me, but somebody with an ear for that (or more knowledge) might be able to figure out more.

[teal raft]

Super interesting stuff.
That said if you can't tell, most other people probably can't either

Especially since you are staring at the waveforms as you listen

[sacred cobalt]

if problems with volume or quality deterioration sometimes occur after the first minute of Extand (or when there), then make such a function experimental (at my own risk), but available (for a two-minute generation of a new song, the same but with different verses)

because finding attractive rhythms in a bunch of generated bullshit is much more difficult than re-using an existing song with some potential problems

if a new two-minute song turns out to be problematic, I'll just regenerate it

P.S.:

well, even if you can generate normally (a two-minute Extend) only one minute at a time (selecting and checking each individual minute for quality and absence of problems), then make it so that two consecutive Extends can be glued into a separate composition (conditionally two minutes long) - without combining with original composition (so that new songs can be generated the same (at least as similar as possible), but with different verses)

in my opinion, an excellent solution 🤔 it would be very cool

[visual shadow]

how do you guys influence the song art without having the ai sing/say it? whenever i try it, the suno ai read the description

[sacred cobalt]

@hollow hare I made part 3 Extend and it’s absolutely nothing like the original 🤔 so my proposal above hardly makes much sense

in general, you need to figure out how to save the context, at least for one or two songs (perhaps even on the user’s own disk, if the files are large)

then it will definitely be nice 💯

[sacred socket]

Using Pro sub, when you create an instrumental, where/how do you input the tune length, And, can you choose three minutes length first up

[stark niche]

it was easier to block🤣

[mystic flume]

First generation will be up to 2 minutes, but not always that long. You can try affecting that with your prompt. Subsequent extensions are 1 min max. There's no difference in how it works for Pro sub vs Free. Subscribing just gives you a lot more attempts.

[sacred socket]

tks for the reply, Tinda, but if my prompt is a jazz trumpet solo, how do I tell it I want it to go for a full two mins? not just somewhere it randomly decides in between?

[mystic flume]

I'd try extensive jazz solo or things like that. Just tell it what you want.

[sacred socket]

ok, and it may sound crazy, but has anyone tried putting in a musical score as a prompt?

[mystic flume]

long jazz trumpet solo, technical, impressive, studio recording

Just fiddle with it. You're basically doing a music midjourney.

[sacred socket]

Nice, thanks, and we're not allowed to say like 'in the style of John Coltrane' are we?

[mystic flume]

No, but get creative with it. Some people suggest asking an LLM like ChatGPT to describe Coltrane's style. Then use that.

[sacred socket]

Nice, thanks so much for your help, Tinda

[mystic flume]

no problem

[languid junco]

To add on to this, here's the version Suno created at the top, vs the version I pieced together myself.

The peaks I assume is added noise.

[mystic flume]

It's not added noise, more like amplified noise. There isn't much difference between the two waveforms except the gain has been adjusted on the auto-stitched one. As far as I can tell anyway.

It's interesting, listening to a stem vocal track created with UVR, you can hear "crowd noise" of a sort behind the vocals.

[teal raft]

And why I have a lot of an easier time getting 'clean' things when doing instrumentals. Don't have to deal with messy vocals

[languid junco]

But that's just one "Get Whole Song" so you can imagine what issues would happen of you Extended from a whole song then got that whole song, and extended from that...

[mystic flume]

Oh yes, for sure. Though I suspect they're not just unilaterally amplifying, it's probably a normalization operation. But doing that over and over probably isn't good either.

Either way it's all very interesting to me.

[languid junco]

There is a some solutions to this.

• Change how "Get Whole Song" works in that it doesn't "Add audio effects."
• Extends go longer up to 2 mins or something?
• Have the memory remember more than 1 minute.
• Or have the memory be able to remember more than just the previous part. (example: Say you're on part 3, the memory will only remember part 2, and if part 2 is only 15 seconds long, the extend will get "creative" forgetting part 1 of the song entirely.)

[graceful cliff]

good calls

[teal raft]

clipboard with a memory and let me inpaint that with a strength slider/number heck yeah

[languid junco]

Might sound a bit demanding.

[graceful cliff]

might not 🙂

[teal raft]

Everytime Suno updates it's always past whatever expectations I have so whatever the next update is, sure to please me

[graceful cliff]

We try...it is hard to meet all demands and all axis

[teal raft]

It's impossible and yeah, you guys do have it rough especially because you make yourselves known and available here in discord

That's a huge ask

and at least I appreciate it

I know lots of others do too

[languid junco]

Realistically,
I can understand that extending over 1 minute might cause issues.
I am not sure how much memory the AI can or could have or the resources it'd use for that kind of stuff.

[graceful cliff]

well, be sure to test when we update... 🙂

[languid junco]

I use NovelAI frequently. The Image generation uses credits, their solution to higher quality, more resource usage is "Use more credits."

[teal raft]

Hoping the update is such a huge jump that I look at all my v3 gens that I have waiting and go "wow these are all shit" and trash em' all and have to start over

[languid junco]

I do love that most the songs I generate pass the "Car Stereo" test.

[teal raft]

Yep. I honestly make music I want to hear, that others like it too is cool and all but it's really for me

v3 is capable with a bit of effort to get there

[languid junco]

I am curious if Suno and NovelAI devs could work together, both seem like they'd get along.

[teal raft]

I'm not Suno but they do seem quite uhhh

"We do it our way" and it feels like from a perfectionist angle

[languid junco]

Technically they're Anlatan now.

I just know both sides are talented.

[trail fulcrum]

Has something been changed or why does Suno only generate 1 minute in Custum Mode?

[languid junco]

Always been like that.

[jade mist]

Any info on V4?

[solar cloak]

Why does Suno keep skipping lines even if they rhyme perfectly? It's really distracting to have the AI simply drop words it doesn't like.

[dull mango]

What's going on here

[low scarab]

I'm trying to figure out if there is any way to specify intended notes, with no luck. anyone had any luck or tips to try?

https://suno.com/song/827581fe-6ab1-4c18-bc14-dc972fca889d

[celest garden]

why do my generations sound like they start 0.02s into the song?

[broken lotus]

You have deliberately cut off the lyrics and prompt that would have given us the answer.

Paste your lyrics into free chatgpt and ask why it's being censored. Chatgpt is the moderator for Suno. It will give you a list.

[broken lotus]

It's AI. it responds differently to different prompts.

listen to some other people's songs (in your genre), find ones that don't start abruptly and look at their prompts and metatags. See if there is a difference in what you are doing.

[candid jay]

how if possible, caan you prompt solo instruuments ? like guitar solo

[broken lotus]

some tips for instrumental songs and instrumental breaks
https://www.suno.wiki/faq/style-and-lyrics/how-do-i-make-an-instrumental/

[candid jay]

not mucch credit to play around and finding out one-self with 50.. 🙂 thanks Wetcircuit

[broken lotus]

free credits will refresh every day...
but yeah... I ended up subscribing pretty soon after I discovered Suno. 😅
waaay too much to explore on only 5 generations a day.

[weary wharf]

No way of doing that currently. You can try minor/major in the style at most.

[weary wharf]

You can define a [guitar solo] tag or an interlude and hiope for the best. From experience if the does not already have guitar as an instrument it's really hard to get that. Still it can put different instruments.

[dull mango]

It's all fixed it was an accidental profanity

[weary wharf]

That happens from time to time. Try to add a [start] tag. This solves most of the problems, but still from time to time it happens.

[broken lotus]

I was blocked because the title of my song was 'guitar wankery'
😅
it took forever for me to figure it out...

[dull mango]

Some times the finger slips Hose and hoes are two completely different things

[celest garden]

thanks, it worked fine before with the same reused prompt but now its whacky

[weary wharf]

It also feels like the full generated song goes through an fft mastering. The bass is much stronger and vocals feel more alive then the single parts.

[plucky fjord]

Hi there, can somebody here help me with an issue?

i cant download a whole song beacuse the option doesnt exist in my page of Suno

[weary wharf]

I think it also depends on where you are in the feature space of the ai. I do not now if the seed(s?) are random or just incrementing. But sometimes you get stuck in a bad output place, and sometimes all songs feel like a miracle.

[weary wharf]

After refresh is it still like that? Download only appears if the generation is completed. And rarely the web page does not update correctly for completion.

[plucky fjord]

Yes but i can put them together

[wet minnow]

generate part 1. then extend it with a part 2. and on this part 2 you click "get whole song" to make part 1 and 2 become a new combined part.

[plucky fjord]

This is the problem , the Option "get the whole song" doesnt exist

[wet minnow]

is this a part2?

[plucky fjord]

Yes

[wet minnow]

one sec

[plucky fjord]

Tahbks

[wet minnow]

For me it works like this @plucky fjord

[celest garden]

How can you refund credits?

[wet minnow]

i have created a end part wich is just 3 seconds long. these get refunded.

[boreal python]

so, uh, how can I keep the songs from missing words from the lyrics. I'm using the whole space for lyrics, btw. When listening to the created song, and following the lyrics, some words are just skipped over.

[hasty otter]

Is there a way to influence the accent of the vocals?

[graceful condor]

I don't think there's an explicit way to do so, but you can include a nationality in your genre or style, it might give it an accent (Japanese rock, Korean electro swing, Spanish nu metal, etc...)

[hasty otter]

yeah I tried that but didn't find much influence. I did however just try translating some of the lyrics into french and that helps guide the whole tone and accent of the song 😄

[manic vapor]

the album art seems to have been tweaked recently. the generated results have been much nicer

[lucid quail]

How can I keep using the same voice for new music prompts?

Can someone please help me with that

[final hazel]

Suno is not designed for that. Other songs have totally different settings/seeds.

[weary wharf]

Is that a part? You need to click on the "Part X" ones to get a whole song.

[low scarab]

you can try adapting the style of writing a bit, and also suggest an accent in the style. Here's one I made requesting a Jamaican accent, and adapting the lyrics a bit to get that Jamaican tone.
https://suno.com/song/094743ac-52b7-4dfd-8f1f-a217b9175737

[sacred socket]

took me a while to find which version I needed and where to llook for 'get whole song' multitasking for the over 45s.

[sacred cobalt]

is there a chance that this will change in the future? Let's say I would agree to store (on my disk) even 50-100 GB for each context song

if it is expensive to move so much data over the network (to the main SUNO server), then perhaps there would be some sense in a local application on the user’s computer (for such cases) - which could also be controlled from the main web application (for versatility)

those. if the user wants to store the context, then he downloads the local SUNO AI application, and checks the appropriate box in the web application, and then all music generation occurs on the user’s local computer (until he removes the checkbox)

What do you think about this solution to the problem of reusing musical rhythms and voices from the best AI music tracks?

[prisma osprey]

how this extend work?

[hollow hare]

https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

Sorry for the 2x ping, wrong pasted item earlier

[final hazel]

This is more a question for the developers. I am just a moderator. But anything is possible in the future 😉 🌼

[lyric hare]

you are being censored

i write songs about the neurodivergant experience. and its nearly impossible to sing about any of the darker experiences of the neurodivergant. the allistics are threatened by the harm they cause, and so its "wrongspeak" which chatgpt has been prompted to censor. so you are probbly writing lyrics that threaten the status que,

if i want to sing about the scary things that challenge the world view of allistics, I find I need to use a different service. Suno is only for happy, non threatening lyrics/topics that dont challenge the establishment in my experience. You cant express deep feelings or you well be censored.

[teal raft]

[dull mango]

It was poor spelling on my behalf as I said Hose and hoes are two different things

[lyric hare]

lol yea im pretty sure the gpt prompt is "censor anything that is not christian rock safe lyrics"

[dull mango]

Is jesus technically a curse word 🤔

[lyric hare]

LOL!

yea this is why you dont let an LLM you have not fine tuned, do moderation. Its a wild ride if you do.

[dull mango]

Is that a commandment though shall not take the Lords name in vain and what if you song is about gardening or teaching kids gardening

[lyric hare]

we work with ElevenLabs for voice reconstruction (disability app) so we get early access, and i can tell you from what i have access to right now, eleven labs music will not be censored, I wrote a song consisting of chain swearing in french, and it did not GAF.

[dull mango]

This is a fork this is a shovel this is a nope you can't say that that's a no no word

[lyric hare]

and the quality, is amazing, not this quantized noise we started getting mid april. I largely suspect Suno was getting "the hug" so they adjusted the quantization to reduce the need to add capacity at the cost of quality

[coral kayak]

hello guys, can someone help me with ableton live 12?

[lyric hare]

(looks at what server im in).. uuuuuu no,

[willow wren]

Same Buzz was about udio but the Mood or Power is Not comparable to Suno. So i generate with suno stem with htdemucs and use midi predictions. After that i use Synthies do Patches so it sounds great. Only the voice Could be better After Stemming but Vocal effects in the DAW are ok

[glacial urchin]

Okay so if I have a suno subscription can I monetize it Spotify Apple Music and YouTube or would it be against the Eula or tos

[mystic flume]

I'd suggest writing the first verse in that language, and then a second verse in english, and see what happens. That might give you a starting point.

[solid pivot]

Youtube Music (different from normal Youtube) at least has rule where it says "no AI generated music"

[mystic flume]

If you pay for Suno you 100% own the creations you generate during your paid subscription period.

[glacial urchin]

So if I were to post it and get monetized, would I be able to keep it monetized if I cancel my monthly subscription

[teal raft]

Yeah

You own the songs fully

[mystic flume]

Yes, correct.

[teal raft]

It's tied to when you generate them

[mystic flume]

"in perpetuity" as the saying goes.

[glacial urchin]

Ok thank you

Jus making sure I don’t get sued

lol

[mystic flume]

Smort

[glacial urchin]

Ok another question if I generated a song on the free plan but then extended it on the paid plan would I be able to monetize it

[final hazel]

No. it needs to be 100% created on the paid plan for that, There are no loopholes 😉

[glacial urchin]

Ok

[warm palm]

I used all my credits to try to make some songs in the same style as freaks from the surf course group, without result, any ideas?

[thick kettle]

Hello guys, I pay for a $10 subscription, and I don't know how to make a song longer than 2 minutes. I also can't add multiple songs to the queue (2max). Additionally, my music has issues when it starts. Does anyone have a solution?

[neat nexus]

[EXTEND] button for the one you like the most.

(Don't forget to adjust your lyrics.)

[thick kettle]

Oh right, something like
[Fast Violin intro]
[Verse 1]

works ?

And what about bugs when music start + i can only genereate 2 musics and not "10"

Or it's like "you can run 2 per 2 and max loading 10 ? maybe ? x)

[neat nexus]

You're allowed to do as many as ten generations at the same time.

If you're running from the first part, these are five submissions which creates a total of ten starts. But they will all be individual tracks.

If you create the first part and then extend from that, you can try for five submissions extending and have ten different options to choose from for which part you wish to continue to.

[neat nexus]

That should work.

There is a lot of good information on the Unofficial Community wiki here: https://suno.wiki

[thick kettle]

Ooh thank you so much !

[thick kettle]

Thanks ! 😄

[vivid geyser]

Hello, quick question, I'm experiencing some slow-downs currently. Is it due to something specific ?

[final hazel]

Suno is working on the system. Suno is experiencing heavy load atm

[vivid geyser]

Okay, that's what I thought

Thank you very much!

[pallid leaf]

The lyrics created by Suno blow my mind every time. Is it based on some specific AI model? I've tried to generate some lyrics using ChatGPT 3.5 for example, but it usually fails badly.

[weary wharf]

ChatGPT 4 I guess. They might be using turbo curently. They should switch to 4o if not already.

[neat nexus]

So I was reminded of this old thing from when I first started on v1 Chirp back in the day...

It was way too long and I didn't notice the length of it until I tried to stitch it together and it took forever and a day to do so. Back then your first generation was 40 seconds and subsequent ones were like only twenty seconds. So since I kept going with it I never really noticed the length until I got done.

I decided to do a remake and post the original again with the remake for comparison of how far we've come since then..

I'd say we've come a long long way...

[broken lotus]

we use to make 8-12 minute songs on v1 🔥

[neat nexus]

We did. Took a bit more work but it was possible.

[thick kettle]

Question, can we change the picture of the music on suno ? (like can we import or re-generate the picture)

[neat nexus]

There is supposed to be a way by ending the last part with:

[End] [End] (text prompt describing the image you wish for.)

[thick kettle]

Omg really ? 🤌🏻

[thick kettle]

Is that working on "instrumental" mode ? x)

[neat nexus]

I wouldn't know. I never use instrumental mode. You'd have to ask someone else.

[thick kettle]

No problem, thank you so much for the info ! For the "classic mode" the way is

(song)
[end]
(A girl with long dress)

Something like that ?

[neat nexus]

That is my understanding. Except I think you have to use [End] twice maybe? I've only done it once before.

[thick kettle]

I have to try it x)

[neat nexus]

Yes! Try..

https://tenor.com/view/all-the-things-hyperbole-and-a-half-do-all-the-things-excited-gif-16018689

[thick kettle]

It works i guess, i just ask : (A chinese girl, long dress, flowers)
It's pretty correct for me

Thank you ! 😍

[hollow hare]

Using End once is sufficient. I just use it twice or thrice because I had the AI singing the tag a few times in the past

[neat nexus]

Thanks. I wasn't really sure.

[tall dew]

is the bot down??

none of the cmds are working for me

[final hazel]

Chirp is retired. You can make music on https://suno.com

[modern belfry]

Why would Suno be saying the words I have in brackets [ ]. I typed [Emotional Guitar Solo]

[alpine venture]

when you get the vocals too clear

https://suno.com/song/91fbb411-a495-45cd-bb23-89322a19af00

everything else is.... gone

Anyway, to those who find this interesting, it seems to me that if you want to smooth out overfitting in the model, a quick and dirty way is go load in tokens which are a lot more generalized than your overfit tokens, but which aren't necessarily "in a different direction" ie could themselves be found on musical tags related to music similar to yours

this is similar to how fooocus works, although I'm not sure if they know this intentionally or not (presumably they do). Basicslly it loads sdxl with tokens. Since the model is MOSTLY good, the more tokens you load it up with that are unlikely to be strongly distant from your desired style, the more likely you are go have "smoothed out" the overfitting

[exotic sail]

How do people get anime pictures for their songs, is there like an extension for that?

[alpine venture]

you're basicslly playing a game with averages

so if you want the gens to have clarity, it seems to me broadly used things between multiple genres and tike periods like
post- or vocal or

years

[fluid robin]

maybe they have anime girl picture description in the song ?

[alpine venture]

which is where I spent my last 100 tokens targeting

I was kinda holding off on messing with this bc only 100 tokens

but it seems to have worked, at least, compared with my other methods

which also do work, but color the style

[fluid robin]

(song)
[end]
(A girl with long dress) or try this maybe ?

[exotic sail]

ohhhh

Got it now

[alpine venture]

both of them tho

https://suno.com/song/ad1b09df-b6f9-4f25-a2a9-6e17cb8328af

on my final gen

have clear male vocals

which I know can be tough

the logic being that years may be in the caption data

in which case double zero puts extra weight onto modern music

and the rest diversified

but ya, i need to so more tests

and I don't know how sunos tokenizer works

[thin flint]

I haven't visited and/or created with Suno for a bit. I can get amazing [Outro]s no problem. Is there a tag for intro that works as good now? I guess just use the tried and true technique of make an instrumental and do a continue where you want vocals to start. That's how I used to do it.

What I did to test was generated lyrics, add [Intro] and it didn't work, and then add [Outro]. Outros were perfect.

[final hazel]

It's very dependent from genre to genre. So a "sure thing" is not always there. What I do though (if it is not giving me a decent intro after several tries and metatags) is start it as an instrumental, then extend in the instrumental, and fill in the timestamp when I want the singing to start.

In my video I show some more tips: https://youtu.be/lj8utBd_IUM

[weak moss]

Hey, I have a question. Is it possible to make a song with 2 voices?

[final hazel]

You mean a duet? Yes but it's not easy.

https://www.suno.wiki/faq/style-and-lyrics/duets/ has information and you can use mine as example to learn from.. this is a duet I was working on: https://app.suno.ai/song/5045e5f8-d213-4cac-9d87-99b370e89adb/

[weak moss]

Yes I mean a duet. But I want to know if there is a command or something like that to put in the song more than 1 voices

[final hazel]

You can look at the examples on the website and song I provided 🌼

[glacial urchin]

Is it possible to change the voice of the AI?

[teal raft]

Not within Suno, no, but there are a myriad other ways to do it outside of Suno

[final hazel]

RVC Replay comes to mind.. but elevenlabs can do it too

[teal raft]

Oh can it now? Neat

I knew about RVC but Elevenlabs is new to me

Then again they are moving toward music too so why not

[final hazel]

Yes you can upload your voice sample to recreate it with another voice so I assume for singing too.. not sure fully though

[teal raft]

ah

[final hazel]

I was too busy training my own voice talking hehe

[teal raft]

Still: There are ways

[final hazel]

100%

[teal raft]

IN Suno though, no. Not yet

[glacial urchin]

Yea I thought so

[final hazel]

Not yet.. who knows in the future.. anything is possible

[glacial urchin]

Seems like all the good stuff you can generate is on the free plan

Like every time I make something really good. It’s always on the free plan.

[teal raft]

tony has been posting suddenly after a few weeks of not

[glacial urchin]

Then I buy the monthly subscription then it’s all terrible

[teal raft]

Makes me think that v4 might be sooner rather than later

[final hazel]

Yes 4V is near

[teal raft]

[glacial urchin]

Ai is Moving really fast

[final hazel]

https://tenor.com/view/happy-gif-27563888

[teal raft]

That isn't a thing though, there is zero difference between free and paid plans in so much as how then generate music

Besides the rights you get with those generations

[glacial urchin]

Idk

[teal raft]

Confirmation bias

[final hazel]

Sometimes your lyrics just don't flow with the music you choose..

[glacial urchin]

Yea

[final hazel]

And I see it just like poker.. If I poker for free, I am much more loose.. but if I play for money, I play totally different and making the wrong decisions

[glacial urchin]

I generated the song with my lyrics once it was on the free plan I was like not gonna sound very good, then It sounded amazing

[glacial urchin]

True

[final hazel]

I am past that point thought.. but it makes me more undecisive with so many credits.. I often have several endings

[glacial urchin]

Yea

[final hazel]

But.. looking forward to inpainting.. LET'S GO!