📚┃suno-school

[tender eagle]

That I cannot say, but it would seem silly for it not to considering that's all. Just an API call. It would be weird to not include something in a paid subscription that is included if you use the subpar mobile web UIX

[indigo basalt]

Ok. Thank you! Also, would you happen to know if you can use the app offline, or would you have to be connected to the internet in some way?

[tender eagle]

Plus the backlash would probably be a level of angry outcry akin to an eastern European populace disagreement with regime decisions.. which involves molotovs and large sticks and bricks if they limited the paid app to less than the web

[tender eagle]

Absolutely no. No offline. It is a high compute model and even if offered it'd probably take 2 days for a mobile processor to make a 31s song

That sounded like a child stepping on a microphone

[last iris]

Congrats on the new app- it’s clearly still in beta mode - keep up the hard work. I’ll stick to the web version once it’s polished

[tender eagle]

Anything is . more polished than the mobile web UIX 😆🤣😂

Now the desktop web uix is near perfect in my opinion

[indigo basalt]

Gotcha! Also, if I can’t use the app offline, could I download those songs I made with the app/website over to my Apple Music library?

[tender eagle]

I doubt that directly. But....if you can import you can....I have it in my lowly plebian android YT music playlist. (Which I do extra sneakily by uploading a link only yt video of the video I download from Suno lol)

I'm not too familiar with apple, but I think I recall that you can import your own music locally. If you can download the MP3 you can import it. As far as I know,

That is I doubt there is direct functional import to Apple music or whatever it's called, but I'm sure that the discrete file, the MP3 or WAV or MP4 you download can be imported to iTunes/Apple music et al

I just can't wait to have actual access to the Android version so I can pick apart the API and write my own looool.

So I can add all the features that I'm sure that they will not...

For personal use, of course not distribution

[hollow hare]

Probably something you should not state outright here since the methods used to determine that are often against the ToS

[tender eagle]

How I access it locally is against terms of service if I'm not bypassing anything and using it legitimately just including things like prompt storage and style presets?

I'm not trying to do anything sketchy

Nor would I release it in any way. I don't release anything I build third party that doesn't act in accordance to rules..

[indigo basalt]

I actually have a way to get it from my laptop to my ipad currently. I first download it from the website, and then I use an program called EaseUS MobiMover. It allows me to move files, and songs from my laptop to my ipad's apple music library. Its very nice to be able to listen to it on my ipad. I just hate the extra steps. But I do it to listen to the music.

[tender eagle]

It is not reverse engineering. API is open traffic. Not functionality of any app nor service

[hollow hare]

Ah, I see. I was thinking about responses from other people that were more vocal about violating this portion

[indigo basalt]

No subscription needed either. Completely free. It will ask to pay a subscription, but I just use the free version of it.

[tender eagle]

I'm not going to be pen testing Suno with an android app

[tender eagle]

For example, on my web interface with the Chrome plugin I have, I have the ability to place a star next to style prompts. This allows me to go back and organize style prompts that worked very well

[indigo basalt]

@tender eagle Thank you for your much needed assistance! I greatly appreciate it! Hope you have a great rest of your day! 🙂

[tender eagle]

It's all guesses and assumptions based on what I HAVE seen with no direct testing. And from what I know about how things work web side. So it just makes sense. I may not be 100% correct in anything I've said...

[tender eagle]

@hollow hare it also lets me Mark specific style prompts that result in Midwest but entirely to Texan EDM girl and or more indie than Indy alternative rock guy rather than the intended style prompted for, which lets me organize style prompts which are likely to not have any training data inclusion

[round tide]

Is it possible to edit an already made song?

[tender eagle]

No. But...you can extend from 10s or so with limited success to modify the same song style and sound if prior to other lyrics or if you begin just prior to needed adjustment segments.

Eg. If

Verse

Chorus

1/2 Verse
1:23m Change lyrics.

If you set extend from 1:22 and change all subsequent lyrics you can somewhat modify it and retain the same song

[round tide]

I've tried this, but it just adds a new short song, is there a way to combine them borh?

Thanks for the answer btw

[tender eagle]

It ain't short per se. It's limited to 2m u fortunately I think. Or any e 4m - context window (eg. :30s may result in 3;30m song)

Because the extensions I believe utilize 2-minute context window prior, unless it is less than 2 minutes. In which case I think it adjusts up to a maximum of 4 minutes per generation. I think the 2-minute limitation is due to a 2-minute utilization of the prior 4-minute song of initial song.

They may be also constraints to try and force the end of a song on extension more often than in the 4-minute length first run

I really don't have a perfect understanding of how the extension system works just yet

But from what I can tell ^

[round tide]

Hm I guess I'll have to try and see what is possible. It would be fine if I could edit the original and shorten it a bit to fit all in one song

Rather than having an extra extension of the song

[tender eagle]

Well you do that with extension

If you want to shorten it to 3:25 from 4 and extend from there then get full song you do exactly that ,no?

[round tide]

Hmm let me try

[limpid badger]

Is there any way to prompt negative, to remove drums for example?

[unique stone]

Not at the moment. It's often suggested though - https://discord.com/channels/1069381916492562582/1253341197917884530

[limpid badger]

thankyee

[round tide]

It just adds a new track with the rest of the song

[tender eagle]

Yes and then if you go to the options there's another option that says get full song

Which, without the use of further credits, combines the two

[limpid badger]

what would be a good way to prompt for a voice like Robert Zimmerman or Tom Waits? I have tried "raspy, gravelly, deep" etc... but they still sound too "perfect" 😄

[round tide]

Ohhhhh

[round tide]

Where is this settings? Can you show me on a prnt scrn?

[tender eagle]

Three dot menu of the song itself extension

[candid flume]

We need this

[tender eagle]

So like if have an extension. And then click the 3 dot me of that song of the two you want to retain as the end of the first song

[tender eagle]

I wish we knew the manner of the training and arch

Any dev wanna weigh in is it bark like architecture or how diff... Is neg embedding even pos without wrecking things or

[charred merlin]

I understand that a lot of people want it, but the most effective way to get it done sooner rather than later is to let the devs know you want it -- do that by upvoting the existing #1252298754787377225 here https://discord.com/channels/1069381916492562582/1253341197917884530

[tender eagle]

Cuz negative embeddings with bark was funny sometimes

I wish the devs were more vocal and discussive about certain aspects so we knew what was a good idea to suggest or not even

[candid flume]

I already voted there. I can't wait to create music without drums and guitar. 99% of my attempts were frustrating, I started with 2500 credits and now I have 1290 left

[limpid badger]

Done!

[final hazel]

I am going to post the right app url (since it might not be visible yet in the store and then I can pin it to this channel)
https://suno.com/ios

[charred merlin]

Not a dev, but bark has been open-sourced, so you're free to dig through it from the github repo ❤️

[limpid badger]

Is there a way to force a song to end after the lyrics have finished? I've tried [end]

Or does it force itself to go beyond to fill all the minutes?

[steep pollen]

I want to be able to get the music and vocals as separate tracks, like maybe it generates the instrumental first then the lyrics and combines them, leaving us with the option to get just the instrumental, just the vocals, or both.

[potent token]

Hey there, not sure if it's the right place to talk about this, but the instrumental option on the website isn't working anymore.

[tender eagle]

Well I know and have 😉 which is why I wonder how Suno architecture is similar or not to its core network or if it is novel

[steep pollen]

I’ve found it to be hit or miss with [end]. Sometimes it ends immediately and other times it repeats previously written lyrics until it decides the song is over 🤣

[limpid badger]

same 😄

[charred merlin]

https://tenor.com/view/mr-deeds-very-sneaky-sir-gif-9983406

gif memes aside, I'm legit curious as well 🤣

[tender eagle]

FAQ- is there a way to ensure that the song ends and does not continue and repeat lyrics in a nonsensical and repetitive manner?

Answer: why yes there is you can try [End] [Outro] and [Fade Out] but ultimately it relies on a blackbox Suno function called RNG ( Random Number Generator )

[limpid badger]

I will try more of those thanks. I have used those before, but forgotten about them. [outro] and [fade out] 😄

[tender eagle]

Yeah I really just want to know in a simplistic manner. Just if the same functionality is relevant or if it is completely novel in a sense that bark internal mannerisms are irrelevant

[charred merlin]

Stripping all other style tags helps, but careful placement of extensions points is the most powerful tool you have in Suno; if you're familiar with music theory, you want to try to place the extension point a little bit before the musical expression concludes-- that way when Suno tries to continue from where you left off, you've left a "trail of breadcrumbs" leading right to a natural "offramp" to resolve the entire musical phrase. Whereas if you try to place an extension point to end the song right after a new expression has begun, you're asking it to end the song RIGHT as it's started again.

I know I'm jumbling metaphors, but please let me know if that makes sense or if you want me to clarify ❤️

I have a similar example from a couple nights ago regarding 3 different options for using extension points during a drum roll

[tender eagle]

Truly it is RNG I have no idea if those ACTUALLY help. Or if just confirmation bias.

[limpid badger]

Oh I know it's Random Number Generator

But the more the merrier.

[tender eagle]

Also be heavy handed. You can add also in the prompt style, outro, and fade out. Not just on the inline. Times if you add a style prompt in line alongside, such as, it will listen

More so

Eg. Style: dark tech step DnB, 2step, staccato drums heavy bass line, outro, fade out

[charred merlin]

imo, that's more weighted to give a song than it is to conclude one.

[tender eagle]

[verse]
[Chorus]
[Bridge]
..
..
[Outro]
[Fadeout]
[End]

But if you run the combination of them within the style and the inline

[limpid badger]

Does the brackets make a difference from say... parantheses? Let's say I just want to inject a "solo" of some instrument. What do you recommend?

[charred merlin]

Not from my experience

[limpid badger]

thanks

[tender eagle]

For example, using just the inline here did not work but using it in the style did

Used in style https://suno.com/song/ba1c66ad-38e5-48f2-af9a-ffaec633d86e

Not in style

https://suno.com/song/cc4a8c57-5e83-45da-9e36-df6f231cd6c0

[tender eagle]

Yes and no though because sometimes it will just say whatever's in the (parens) very very rarely have I heard it say what is inside the [

[charred merlin]

That can happen with any metatag.

[tender eagle]

As you see in the song above where I left it inside the style, saying announcer, it did speak the final ending. Whereas where I did not use it, it sung it just the same as it did the rest of the song in a weird kind of it's trying to talk but still singing way

[tender eagle]

Yes. Even in style it can decide to say it lol. But really much much more common in ()

[tender eagle]

In some songs It did speak properly without the tag but much much more rare than where I did use it in style

Also try tightening the following: element of prompts in [] with : as the following element of the tag.

I thought about doing this because a lot of training annotation in various formats, and we have no idea what Suno used, they may have a structure such as

outro: ... ... ...

And so on. Even if [are trained as a] superlative tag modifier the training data does not differentiate (depending on encoder type which would be great to know) not as natural language but as token specific elements

So using [outro:] may bind tighter if the encoder recognizes this from the training embeddings as outro:

While [outro] may not have as tight of a weighted response

I have 100% seen this with different formats of captioning that I've done on training, natural language text and image generative models

Eg. style: oil painting. As a caption

oil painting style has the effect. But if my data was done as such style: oil painting is much much tighter and strong

If the same applies here, and the training data is annotated as such. It may work much better

Any @dev want to tell us which encoders are used for the text prompt?

[normal niche]

Anyone come up with a foolproof way to get Suno to end a song? I've used [outro] and [fade out] and [end] and {end} and while that seems to work initially, my songs are fading out and ending nicely, and then ramping back up for another minute of music followed by an abrupt end. I can always bring it into Audacity and trim it, but it would be pretty great to get Suno to end the song cleanly.

[teal raft]

The best success I (and others) have is switching to v3

I personally switch Instrumental on

Then in the style box just do "end, end, end, end, end, end"

Depending on where you continue from drives this too, but this works so often that I assume it will always work

Other people do the 'end' in the lyrics box

[normal niche]

I'll give it a try. Thanks for the tip.

[teal raft]

Not a problem. 3.5 not wanting to end is almost a meme

[normal niche]

Worked like a charm. Thanks again.

[long pebble]

When can we expect the app to work on android? or has that not been discussed yet

[craggy sentinel]

I generated a song that was perfect, but it cuts off abruptly. Is there a way to extend or regenerate a song (that won’t vary the style much)?

[charred merlin]

Click "Extend", by default it picks up from the very end, though you can specify an earlier time to overwrite from.

If you can't see "extend", you may need to wait for the song to finish processing.

[silk mulch]

Without using the extend button can I use a props to dictate how long my son should be ie [lenth 3:34] Or some kind of command like that?

[silk mulch]

Yeah why does iOS get all the love and not Android

[charred merlin]

No, people try to give hyper-specific commands in the metatags, but they don't work that way unfortunately.

[silk mulch]

That's kind of a shame maybe we should put it in the works

[charred merlin]

App Store generates more profit, easier to make things for 1 version iOS and similar functionality opposed to a bunch of different devices, just guesses.

put it in the works
https://discord.com/channels/1069381916492562582/1252783540131860503 Already in the known suggestions 🙂 The more upvotes, the more the devs can see how important that feature is to people.

[silk mulch]

I'm sorry but I call that favoritism and that's just ethically wrong, If you're not going to appeal to all of your customers then that's just a bad business model for you in the long run no offense intended and sorry to go off topic

[charred merlin]

I just asked ChatGPT to try to get an idea, like I said at the end, just guessing.

🤷‍♂️

[silk mulch]

GPT is wrong because Elon Musk and OpenAI I have stock in apple and iOS...

[untold briar]

Devs don't talk about the model, but you can find some hints on YouTube https://www.youtube.com/watch?v=GqOWjwoSBc4&t=385s

[crisp stump]

Can we just do away with album cover?

It's kinda redundant

If it can speed up generation even 1s faster, I'm in favor of keeping everything lean and minimal

[untold briar]

It's probably literally something like 1s, compared to the music. Personally I'd love to trade generation time to get higher quality or just better guidance (negative prompts, etc). Even if it costs more credits. I generate 100+ clips for every song anyway.

[unique stone]

I'd happily trade the length of the generated song for higher quality, if that were possible (I don't think it is right now at least...)

[untold briar]

I would to, though I think it's unlikely they can do that, unless Suno works a lot differently than my guesses.

But other ways: sampling additional times (negative prompts) and other sampling stuff, probably works and is at least interesting. This guess based on text LLMs where sampling has a clear dramatic effect. But also Bark, where even 'incorrectly coded Jonathan samplers' where I screwed up the implementation still makes Bark sound different I'm like, oh, this is cool too, lol.

[craggy sentinel]

Where is this available? Web? Discord? I’m in the new iOS app at the moment

[untold briar]

I don't think Extend is in the IOS app yet, unfortunately. (Or if it is, it's also bugged/missing for me). But the song library is shared so you could extend songs from the website, even on your phone.

[grave wraith]

Often I get a male voice which is rather deep and roaring... I aim for more a tune like the singer of Manowar or Halloween... Someone knows how to affect the AIs choice?

[charred merlin]

I haven’t had a chance to experiment much with it, but the other day someone was suggesting using male/female names as a way of working around “male/female vocals” being unreliable .

Beyond that suggestion, all I can think of is to include descriptors for the vocal style, like “robust baritone” or “operatic-voice”

[grave wraith]

Tried all that. It looks like Suno doesn't know what a tenor is

[charred merlin]

Can you share an example of the lyrics (including meta tags) and style prompts you’re using?

It could be helpful to figure out if a peculiarity in how the prompt is written is skewing the results. (For example, Asking because “a cappella”/“acapella” and “dance punk” /“dance-punk” each would give varying results)

Even if not, it’s good to get an understanding of what common musical terms don’t seem to work.

[grave wraith]

Well..i don't have the ones where I extra also put prompts about the voice in the lyrics...but like this here is nowhere near any style that is inside the prompt https://suno.com/song/acfde91d-bf70-4ff5-bd20-7e8e51098c34

[grave wraith]

Or take this...it's prompted emotional, ballad, orchestral.... nowhere near anything of 1 out of 3 https://suno.com/song/22f55165-4c0a-41a0-84f3-991d1e1344a2

[charred merlin]

I have a handful of credits to spend today before they reset— if you don’t mind, I’ll try to see if I can get closer to what you were hoping for.

Not able to tell at the moment, were these started with v3 or v3.5?

[grave wraith]

3.5 ... well, you know. I don't even care about my credits 😄 I made some other versions that are at least sounding good. But I just wonder...sometimes it feels like the AI is completely ignoring everthing. And yes, I know, that it is in development. But I don't know... when I prompt Tenor or female and it is a grunting male voice, then there seems to be something big missing, isn't it?

[onyx rover]

Hi All, can anyone tell me how to add a custom image to a song? I generated one in Mid Journey AI that I want to use, but the option to replace the image is greyed out:

[grave wraith]

You got something in your mind and you can't get it that direction at all ^^

[onyx rover]

Is that a pro only feature, or am I missing something?

[grave wraith]

Isn't it telling that?

[onyx rover]

I was hoping I was missing something, but if you know that for fact, then yes. 🙂

Thanks Sam.

[grave wraith]

Just checked on my end...seems it is

[onyx rover]

Cheers dude, much appreciated.

[lyric hound]

Does anyone know how to fix the effect of "playing through a pipe", which occurs in 50-60% of songs and usually in half of the music, with an audio editing program? Udio doesn't have this "effect", so it's a problem on Suno's side.

[charred merlin]

There’s a lot to factor in, such as how much you’re asking Suno to do at once, what model you’re using (and whether you’re using it to create a new generation or extend), the style prompts and how differently they might be weighted.

In terms of “emotional”, “orchestral”, it might be different to what we were traditionally expecting, but I wouldn’t outright say it’s “wrong” 😛

Can you share any influences or what you were trying to achieve?

[charred merlin]

I’ve had success cleaning up audio in Audacity after having separated the stems with OpenVino.

I’m still pretty green regarding mixing & mastering, but experimentation with filter passes, compression and normalization have yielded some good results in tandem with auto mastering software like the free service from BandLab

[lyric hound]

Thank You! I'll try it 🙂

[charred merlin]

I’ll be honest, sharing screenshots of the waveforms, being very descriptive of what specifically I’m hearing and what I’d like to hear, using ChatGPT has been a lifesaver for someone who just wants to get quick practical experience and can refine the skills later,

but just a general shout out to the tech-discussion thread who might be reading, I would love a helpful, sensible-length tutorial for common mastering techniques ❤️

[grave wraith]

Hard to tell 😄 Initially I wanted to make it a shanty like introduction for my stream. Then I tried to make it more in a commercial sound but that didn't work at all. Later I found this bigger prompt on another song and tried it and kinda like the style. But I can't get it for my taste "epic" enough. It should be catchy... where you want to sing along 🙂 (Well...the thought that commercials have). But the style shouldn't be too modern as I am old ^^ Nah...I am also singing, so it should go to the direction of my music. I was actually experimenting with the epic metal stuff to get the voice up high...but failed.

Hell..if I knew what I want I could become a composer 😂

[lapis echo]

Hi, Suno's login form has changed, it only allows iCloud, Discord, Gmail and Outlook to log in, but previously, it just required a user and a password. I can't log in to my account now as I can't enter my user and password. Does anybody have any idea on how to log in with an account registered before this update? Thanks

[unique stone]

When did this change?
I don't recall it ever taking a username and password.

[lapis echo]

It did like 2/3 weeks ago. I never signed up using any of the mentioned platforms, I signed up with a user and a password.

When you clicked on "sign in", you had an input for a user, now it's gone

[unique stone]

Was this definitely on https://suno.com ?
There are fake Suno websites.

[lapis echo]

It was definitely Suno yes, I'm 100% sure the log in form changed

I contacted support on the email but it takes them 3 days to answer and I received replies like: "to log in, just go to the log in section"

[unique stone]

In case I'm remembering it wrong, I just had a look on YouTube at a "how to sign up to Suno" video posted on 5th June showing the login page. These were the options available at that time:

and the "Sign up" link went to this:

[lapis echo]

The change might've been applied before 2/3 weeks (that's when I noticed), also they changed the URL, it was suno.ai before, this is confusing, now it's redirecting

I'm checking https://web.archive.org/, which holds a screenshot of all webpages over time, and all screenshots are just black for Suno

I'll insist on the support email :S

[unique stone]

This is the sign up/sign in page from December when the website was "suno.ai"

[rough delta]

Hey guys,
I have a song completed but I want to extend it to background instrumental but it keeps only doing 2 second clips and now charged me credits for that?
It should extend giving instrumental for 3-4 mins?

Paid user for 1 year subscription.

Any help appreciated here

[unique stone]

Usually if a song is under ~8 seconds it refunds the credits automatically.

Try changing the "extend from" time to be a bit earlier. Once Suno has decided that the song is ending, it will continue to try to end it when extending. So by changing the time to be a little earlier it will forget that it was trying to end the song.

[rough delta]

Thank you, I will try that now.

Yeah it refunded 3 or 4 of them but charged me for 1 or 2 others that are 2 seconds long

Ok so that didnt fix it as it ended at the exact same place as the first song and i got to 15 seconds which it did charge me for.

[neat field]

This is literally my process too for my current experience level, endorsed

[lethal egret]

For some reason there is no “Go-Go” Washington DC. Genre. Is there a way to create a genre?

Go-go is a subgenre of funk music with an emphasis on specific rhythmic patterns, and live audience call and response.

[unique stone]

Only thing I can suggest is go further back. I'm quite ruthless with my songs and will chop off significant chunks of it but yeah sometimes a song will just keep wanting to end...

If you're wanting an instrumental you could also try turning off the instrumental toggle and put [instrumental] in the lyrics box, then fill the box with dashes or dots or squiggles - anything to act as padding basically.

[crisp stump]

Is there anyway to instruct an acapella/solo vocal for the first verse?

No instrumental

[tacit haven]

this has a real producer tag

(all made in suno)

[charred merlin]

"a cappella" works better than "acapella"

[crisp stump]

Thanks!

[tacit haven]

wont it be a problem?

[charred merlin]

Not sure I follow, what do you mean by producer tag?

[tacit haven]

the audio i sent it has this producer tag

by daytrip

its a real producer tag

https://www.youtube.com/watch?v=-FGrMYI8KSo

[meager current]

Is there a way to promt suno to say a few words of the lyrics at a higher pitch? For emphasis?

[tacit haven]

there was no audio upload

it was made interily from scratch

go to suno.com and make a song with this intro:

[Intro]
Daytrip took it to 10

it worked with other producer tags

i can make an another one

you know the tag "oh lord jetson made another one"?

https://www.youtube.com/watch?v=Bd3yg4H5obM

this is a short snippet i created with it

from scratch

no audio upload

second example

[unique stone]

"the audio i sent" - probably referring to the audio sent in discord

Prompt for something -> be surprised when you get it? 😄

[charred merlin]

Oooh, I misunderstood.

Along the same lines, https://www.suno.com/terms states that there's a chance that generated content can have similarities to existing content or content generated by other users, so generations still need to be considered with a grain of salt.

[tacit haven]

ok

[neat nexus]

This is what I used for my Shanty...might be a bit too poppy for you but I enjoyed it...

Style: Call-and-Response, gravely group vocals, sea shanty, steady tempo, Repetitive Structure, muted instruments

First lyric: `[Bodhrán percussive intro]

[slow meaningful beats]

[Lead]
What shall we do with a drunken sailor? (What shall we do with a drunken sailor?)
Early in the morning?

[Group Chorus]
Way-hay, up she rises
Way-hay, up she rises
Way-hay, up she rises
Early in the morning!`

My somg: [Removed links]

You might get some ideas from the style prompt at least?

[misty dust]

I'm using this as audio input and so far, I haven't heard my music extensions sound like it.
https://suno.com/song/3c6bd9f1-d3c5-4149-97bf-ca2776e251f7

[grave wraith]

Thank you... will try your prompt... i.e. I never got that powerful music inbetween the verses

[pseudo flax]

Is it possible to “extend” and “reuse prompt” within the new app? I haven’t been able to find the options and have been going back to the browser version for it

I’m at work and not active, please ping me if you’re kind enough to respond 🙂

[rocky juniper]

what if i want to extend a song without writing anything on the lyrics or style

[grave wraith]

And see...that is what is confusing me... it's not even coming close to your style 😄 As I would enter something completely different. Of course it shouldn't be the exact same thing...but at least the direction? https://suno.com/song/81704e73-4e76-4792-8960-ec06a0c1a7d2

[neat nexus]

A sea shanty has a call and response. Not really seeing that. Also I deliberately put in an instrument for the music. I don't see one in your song.

[livid fjord]

hey guys, do you know if theres a way to mid change the beat? like changing the style of beat like from house to trap or somth like that

[gleaming dawn]

guys, should I type "rnb" or "r&b" into the style of music? which one will suno detect?

[charred merlin]

Are you starting with v3.5 or v3?

v3 is better at getting a broad range of styles, v3.5 excels at flow and song structure but can trend toward "generic pop / rock / dance / rap" unless it has a solid style to extend from (but it's still capable of making some amazing initial generations, especially when focusing solely on an instrumental intro to build off)

[neat nexus]

I always use R&B and it seems to work. I should probably try rnb some time to see what I get.

[gleaming dawn]

Ty!

[misty dust]

Also, this is falsely marked as having human vocals.

This is one example of a music extension from the audio input: https://suno.com/song/4f4d12fb-b59c-4f46-baf3-e135580351f0 (the guitar sounds sick)

[neat nexus]

It took me four rolls to get this one and I had to do a bunch of editing. Feel free to redo the lines to make them more in your style. Screen shot included to show that it takes a few attempts.

[Removed links]

[grave wraith]

Thank you...that gives some ideas

[neat nexus]

Happy to have been able to help.

[gleaming dawn]

I have changed the cover image of my song, it is changed now, but when i download it as a video file, it has the original cover image, can you help me please?

[neat nexus]

I'll mention it in the Mod Chat but as far as I know this is expected behavior.

[charred merlin]

Soooon, hopefully 🤞

[neat nexus]

I just got this confirmed. It is expected behavior....for now. At some unknown time in the future this may change. But for now it is only really intended for the links to get the cover images. Not the videos.

[gleaming dawn]

ty so much for the reply. im gonna edit the downloaded video in after effects and photoshop to make the cover image updated xD

[neat nexus]

That is definitely one solution. Cool idea!

[gleaming dawn]

my client would like to share the AI video on her fb page, but you know the lyrics can be disturbing, having [] and () and other stuff in there, thats unnecessary to display, i wish they would implement a feature where you can disable lyrics display

[neat nexus]

That's a good idea too. #💡┃suggestions-web?

[gleaming dawn]

oh ty, i will make a post

[charred merlin]

https://discord.com/channels/1069381916492562582/1253328110934425714

[neat nexus]

You could also make a mini video for them by using Noisee. That would eliminate the issue a bit. You may also get really lucky with a segment in Noisee that can be used in a loop. Those work well depending on the music style.

[gleaming dawn]

Ty @charred merlin , I'll also add a new one, and upvoted the one you sent me.

[charred merlin]

@mystic flume makes AMAZING custom videos for her AI-band, Celtic Moonshine -- unless she wants to write a more detailed #1252267444773326848, I could share the brief description she gave me?

https://youtu.be/iK24FP2AjHA

[gleaming dawn]

sure

[tardy stump]

Just one question : Why the voice are so damn full of artifacts ?

Music is good.

Voice is...

kinda bad

[charred merlin]

Hrm, I know that any generation can have weird quirks or artefacts, but beyond the structure and flow (which I'm still floored at how much better 3.5 handles), it helps to report poor generations, ... > report > bugs > poor audio quality to help future generations 🍻

[craggy sentinel]

Is it possible to reroll the image that was generated for a song, like if you got a bad image? Or replace it with your own?

Or trim the duration of a song—particularly one generated from an audio recording?

[charred merlin]

For paid subscriptions only, I believe.

trim the duration of a song
Not directly in Suno as of yet, you'd need to use an external DAW. https://discord.com/channels/1069381916492562582/1252773477157699664 upvote the known-suggestion for this topic if you want to let the devs know it's important to you 🦸

[mystic flume]

Oh, it's super easy. First, I use a Filter in Open Broadcaster Studio (OBS) called "Scale to Sound" https://github.com/dimtpap/obs-scale-to-sound and then I use several layers of transparent PNGs and set each one to scale with the Desktop audio. Then I hit record and play the song on my computer. Then, once I have that I go into Vegas Pro (or your video editor of choice) and I do a simple pan animation with a half-transparent black square and slide it down past the lyrics. Add the OBS video and ta-da! It looks cool but doesn't take a lot of time. Maybe 20 minutes total once you've got your workflow established.

And you could probably do the whole thing in OBS if you wanted. It might be fun to pull the square down by hand while the song plays! Though I'd still want to use the high quality WAV as the soundtrack rather than recording the desktop audio but that's easy enough to do with free tools.

[mystic flume]

Because the human voice is an extremely complex instrument, and this technology is very complicated and (as good as it can sound) still in its infancy. But if you do some compression and massage it a little you can end up with somethign that sounds pretty amazing on your car stereo even if it's a little tinny / artifcacty in your headphones.

[charred merlin]

Thanks again for the visualizer tutorial, Tinda! Just to expand on the idea when I get around to trying it, I think I might do a thin rectangle behind 1 line at a time and move that along with the lyrics... maybe even sliding it horizontally along, like a cheap knockoff of the old bouncing ball graphics on Disney sing-a-longs.

[mystic flume]

Sure. I use Vegas Pro (since that's what I learned on), so my options are limited. I was able to do a cool animated text effect using one of the built-in animations but it took a lot longer to do: https://www.youtube.com/watch?v=XqQd853abMU&list=PLX58EfY_IyrqHCw6TLMmpkEFE9RZETef8&index=1 (NSFW lyrics).

With After Affects or something more powerful I'm sure you can do all kinds of cool stuff.

I've been trying for weeks to set up MusixMatch for my songs, too, to get lyrics on Spotify etc... but they're taking forever to "verify" me as the band.

Suno songs sound MUCH better when you can read the lyrics (as we've seen with the Suno videos); your brain can forgive a million tiny errors when it knows what it's supposed to be hearing.

[surreal ice]

hey, is there a place where I can test bark-ai voices as a non developer? can it work for a chatbot? i only found hugging face which takes 1 minute to generate.

[charred merlin]

I think you can set it up locally-- https://github.com/suno-ai/bark?tab=readme-ov-file#-installation

[surreal ice]

Im trying to avoid it as im not really a developer, was hoping theres some playground available

[charred merlin]

Huggingface is the main one (hosted publicly for free)-- there might be others, but you don't need to be a "developer" to set it up, follow the instructions and ask ChatGPT / stackoverflow if you have questions, and (provided that you have hardware capable of running it) you can generate content at home without needing to wait for the web ui

[haughty dirge]

Hey everyone, I'm new here, but I've been using Suno for some days now 🙂
The IA is amazing, the app is incredible. However, I noticed the output often has muddy noises, and the voices could be cleaner. Is there any known way to "upscale" or clean a music generation from Suno? Some of the tracks I created are just the tracks I had in mind: the sound, the voices, everything is perfect, and I want to keep it the same... it's messy, as if the song came from a compressed MP3.
Udio doesn't have this many artifacts and problems, but I like using both tools. Also, Suno generates entire songs.
So, I'm really curious, is there a way to enhance a song that was already generated, even with any known external AI tool or something?

[storm patrol]

Do we know when suno is going to release their app on Android

[teal raft]

Probably soon after the iOS one gets settled

I'm sure it's being worked on atm

[fervent lodge]

hi how to get full 4 minute songs, I get in instrumental less that 2 minutes sometimes

[haughty dirge]

No idea?

[digital jungle]

What happened to the continue feature? Ho Do I continue a chirp?

[tiny thorn]

Could anyone help me out, how can I get one verse up an octave in a song, or how can i make one chorus different than others?

[bleak bobcat]

Only with a DAW editor or trying Bandlab Mastering, but then you can't it share in Suno. I hope version 4 correct this and the random jumps inside the lyrics, and the extend action to follow the new lyrics and not to jump in previous verses.

[charred merlin]

Use the extend feature to continue your songs

[charred merlin]

Chirp has been discontinued, and all song generation happens on the website now https://www.suno.com

[charred merlin]

Hard to give exact controls, but using the metatag [bridge]* might indicate that it's meant to be a similar structure but with a variation

*sorry for using the b-word, furrybeans.

[humble thunder]

What was chirp?

[charred merlin]

Discord command to generate music— they had Chirp for music and Bark for voices, but they open-sourced Bark and focused on Chirp, leading to what we have today 🙂

[humble thunder]

Ah, so moving away from the midjourney-esque generation commands. I see

[ember carbon]

hello everyone, is there a specific prompt available for directing the AI to use a particular voice or instrument?

example, from a recorded voice or beat

[meager crypt]

that's really catchy

[lunar anvil]

There is a PDF document at https://suno.wiki from @onyx swan that has hundreds of tags and styles. You could extrapolate those tags to use in other genres. Additionally, I would check out the Suno Showcase and Suno Trending to see what others are doing. The tags are suggestions and the AI doesn't always follow the suggestion.

[mystic flume]

Thanks!

[mystic flume]

Like MidJourney, Suno is built on Stable Diffusion so that stuff is always gonna be a part of it. But Suno has seemingly done a lot of work with additional training to help it focus more on musical things.

[mystic wolf]

Hello I created a music with suno I really liked the voice of one of the songs I would like to keep the same to make another one do you know if it is possible? And if so, how? Thanks

[keen wraith]

Hello , Thanks for the great suno, I would like you to add Egyptian music and Egyptian shaabi music ( local music) also I would like you to add Egyptian dialect because it's hard to creat Egyptian songs with arabian dialect. Thank 😘

[quaint perch]

How can I recreate my old songs music with new lyrics on Suno.AI?

[charred merlin]

I think they might have been referring to how you can use Midjourney via discord commands

[rough island]

Hey boys ive been working on a couple songs and randomly at the end a sad like piano part is added and the ends abruptly. its fine it doesn't affect the main part of the song but is there anyway to edit it out? or will I have to do that in post

[manic vapor]

you can extend from before the piano intro and try [End] for the lyrics

i would use v3 while extending, it knows how to end a song "better", but it also kinda just trashes to audio quality compared to the v3.5 clip, imo

[hollow hare]

Likely much easier to edit it in post if it can be isolated easily, saving credits that way, otherwise do what FK said but make sure you [End] with the extension set as 2.0 or 3.0

(3.5 does not like to end)

[rough island]

I'll make sure to end and just edit the other ones after. thanks lads

[manic vapor]

it will end, but then start an entire new genre of song

[hollow hare]

True

[rough island]

yeah fr. It's really out of place

[manic vapor]

you can try stuffing in between your verses and choruses some notation for breaks to try to extend closer to 4mins, but sometimes i find a song just doesnt need to go for so long

example:

[Break]
...!.!
..!.!!
..!.!!
!!.!.!

[rough island]

yeah tbh idrc about the length as long as it's at least two mins

[manic vapor]

if you song ends naturally before like 3 mins, it seems you'll have a higher chance of the song just going on forever

[astral mist]

After what felt like 100 generations I finally got it to go back and forth between male and female, and even a couple times an ensemble chorus in a single go (to keep the male and female parts sounding like the same voice). Does anyone know a better way to do this without burning through so many credits? I shifted over to Psalms because this was too inconsistent to present as a final product: https://suno.com/song/babd4c9a-b701-487c-8e98-8455bf56de14

[charred merlin]

I think yesterday someone asked about the difference between using square brackets and parentheses, and I shared that I didn’t believe there was a difference.

Following up to apologize, I was way off base— square brackets aren’t interpreted as lyrics where other types of braces and brackets seem to be parsed as lyrics.

To test, I made a song with only a meta tag in the lyrics and no title and it came back as “untitled”, whereas using parentheses displayed the “meta tag” as the song title.

[azure crane]

Something is really wrong with Instrumentals. It always generates vocals now. Even if you just have the instrumental setting and also add Instrumental as genre. Hope they fix this asap.

[astral mist]

Yeah, there seems to be a lot of wiggle room when using the square brackets heavily instead of the style guide (per my example above), but I never ran into an issue with brackets being considered lyrics. Oddly, if you only have 1 bracket it seems like the system treats the whole line as if it doesn’t exist and / or it can break the style suggestions entirely.

[hoary cobalt]

Merci 🎶

[spice turtle]

Hello
How can I separate the female and male voices, line by line or verse? I tried the parameters [Male Vocals], [Male Singer], [Female Singer], [Female Vocals] but it does not apply. Am I right that these are instructions that do not exist in this form? Is there something you can use instead?

[tulip harness]

if i upload a short original instrumental and extend it then make it public will it get filtered?

[silk mulch]

This thing just created two songs and charged me credits for that I did not ask it to create

[charred merlin]

Each "Create" generates 2 songs -- it sounds like you might have accidentally clicked the button?

[silk mulch]

Yeah but it gave me a song that was not the title or the lyrics that I asked for

[charred merlin]

Might be something to report in #🪲┃bugs-web if it's not a misunderstanding -- I've seen instances where people accidentally used the suggested Simple Mode prompts like "write a pop song about wanting to see you again", "a rap song about being trapped in the machine", " a heavy metal song about dinosaurs ruling the world again", etc

[autumn carbon]

Is there an issue where the quality of any instrumental song going past the 5:00 mark start to lose quality sound with the guitars and become muddy?

[tulip pumice]

Hi, ... Is anyone else ATM having Suno spit lyric's? That are NOT understandable and not even CLOSE to what your posted lyrics are?

Im talking . . Lyric's CRASH, ... lol

[astral mist]

Hi @spice turtle , the system struggles with doing this but it’s possible to brute force something with numbering the vocalists. Posted this above hoping someone found a better way, but no replies. You’re free to take the in-bracket prompts and see what you can get: https://suno.com/song/babd4c9a-b701-487c-8e98-8455bf56de14

[elfin tendon]

hey i was wondering do you get to use the songs generated commerically when you become a premium user ? or are there copyright issues ? please somone let me know 😫

[tulip pumice]

https://tenor.com/view/slide-drift-crash-fail-gif-16909243

[tulip pumice]

They are YOURs when you become a PRO Sub.

[tulip pumice]

https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448#47fc7b6ad1114a1597903b11c73ade77

[inner cloud]

Hey guys, a quick question, does any of you know how a song get the Full Song verification like this? Thanks!

[astral mist]

It’s not a verification, that’s what’s tagged when you combine a song with the extensions so you can distinguish it from the clips.👍

[inner cloud]

Ah so does it mean that the song with the full song is the original song that has been extended by others?

[neat nexus]

Get Whole Song.

[astral mist]

Yes.😁

[neat nexus]

No this refers to the work process.

You make the first part. You either extend from it or you extend and resume from an earlier time and then comtinue on until you feel the song has ended. Then you click the three dots menu and select Get Whole Song and stitch the parts together. That's the message that shows up when you stitch any clips together.

[charred merlin]

If you have part 3 of a song & "get whole song", it will piece together the related part 1, part 2, and part 3 into a "full song"

[astral mist]

@inner cloud You don’t need to post the full song either to share. I shared just the extension parts for some of the Psalms I’m putting to music because they’re so short that they fit just in the extension and getting SUNO to end a song early can be tricky, even when using the [end] trick using V3.

[inner cloud]

Ah that makes sense now! Thanks guys!👍

[charred merlin]

(pro tip! extending from a full song vs extending from an extension can give different results! Keep in mind Suno can only keep a certain length of song "in memory" when extending so it can get less effective with LONG songs, but you can choose whether the extension should sound take influence ONLY from the most recent clip, or if it should feel like a part of the whole song)

[glossy dragon]

Absolutely - https://soundcloud.com/spotapod/slogans

[glossy dragon]

I've found extending from both gives some interesting cool results that can be stiched together.

Thanks for the tutorial!

[modest oasis]

Is there a prompt to add a duet ( Male / Female ) singing the same parts ?

[charred merlin]

It's wild the difference that you can get sometimes -- unfortunate that most of these neat observances are usually appreciated in the moment, and then "lost" after the final selection is made for the song.

It would be neat to maybe one day have a channel to discuss these sort of variations, along with the different nuances that come with minor adjustments to prompts.

I want to hang out with my fellow music-science nerds! 🤣

[hollow hare]

Old copy-paste made for v3.0, I haven't tried for 3.5 but the metatag reading is much better:

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
Suno.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 ) or your own voice with a RVC model (see #💡┃suggestions-web message ).

You can use a DAW to edit which voice track plays when.

[glossy dragon]

It’s because people want an easy path to generating the music without editing. My song “Slogans” actually starts with one of these variations. It uses 20+ brand slogans over the past 40 years combined into, and stitched together, to create this mashup https://on.soundcloud.com/6iBybY4cDwLzQ4E89

Suno gave some incredible snippets

[brittle apex]

I did a little test to see if tags prompting for high quality make much difference. My conclusion is that good prompts do not override the roll of the dice. You can still get a nice generation out of something that has been prompted to be bad quality pretty readily as I only generated a few before selecting one to keep.
https://suno.com/song/c42e0175-7165-40ef-ae50-c34a5e916ac0
https://suno.com/song/8183ad54-2600-46c4-b4fa-2a1f45244287

[neat nexus]

For some of us we don't feel that we do a good job so we prefer to get stuff done generated rather than what I refer to as 'butchery' when I try to do stuff in a DAW. Even when I think I managed to pull something off I don't feel good about it usually.

I admit though that at the hands of someone who knows what they're doing a DAW can be a beautiful thing.

[olive grotto]

how long does it take for an account to be deleted? i requested the deletion of my account 3 weeks ago, and it's still up. if my account hasn't been deleted by next week, this service will have violated GDPR article 17

[astral mist]

Can confirm it burns through a ton of credits to get something acceptable. I numbered the Male and Female voices to some effect [Male 1 Vocals] [Female 1 Vocals] in the lyrics, but it was like lighting credits on fire to get it to work, especially if you’re switching back and forth.

[glossy dragon]

Well said

[amber vigil]

where is the best place to ask how to link the Suno app on my ipad to the account I already set up on the desktop?

[mystic flume]

You created a login, just log in with whatever account you used. For instance, I created my account by logging in with my google account. So I just do the same thing on any other computer and that'll be the same account.

[lucid valve]

hey! i moved my music to trash, is there a way to "clean" the trash?

[charred merlin]

afaik, there isn't a 'clear all' option yet, but if you're on desktop you can use common selection tools (click, then ctrl + a to select all / click, then hold shift + click to select all songs in between / ctrl + click to add or remove individual songs from selection) for adding multiple to a playlist at once, or batch-deleting from the trash

[lucid valve]

yeah, ok ty <3

[low charm]

Disney duet perhaps. Male and female I just number them (male 1) (male 2) (female 1) (female 2) I use brackets for structure and round brackets for ...(the other things). consistency is good. The users define the tools.

or defined in their use over time, whether they understand that or not.

[peak raft]

say I've got a question. i'm considering doing some research using suno as a tool for emotional reification of people's written poetry, an AI-driven musical spin on poetic analysis, and i was wondering if I could get any info on how ppl's lyrics input drives the emotional qualities of the music produced

[low charm]

it seems to me that many of these systems adapt to the user over time based on your preferences. In the past we adapt to the software, but now the software can adapt to us, on an individual basis.

[peak raft]

i don't think suno has much individual memory so much as perhaps memory by tag or style? but i think it's also reading the lyrics for meaning & doing a lot with the meaning, like if it's generally happy lyrics the song will be happy in tone unless you modify it with tags. or even just parts, like if you ahve an especially sad chorus, it will pick up on that & the song will reflect that

[low charm]

Reinforcement machine learning. The ai gets electrocuted if you dislike a song. something like that.

[peak raft]

i do think it's building up memory per style tag tho

[low charm]

yeah it has to start somewhere but it refines over time, and there arent so many types, we just get put in several categories.

the categories can branch further if the resources are there to manage the data, but I dont think there is too many types.

I mean the deeper you go the more complexity there is in any area, but thats probably the last 5%. Its not always worth the effort.

[peak raft]

it has impressed me, in my experience, in taking words i've written & without explicit tags or song structure cues, making the voice & structure & emotional tone to bring out or create emotional resonance that was maybe only barely perceptible in the written text. so i think there's something worth studying there, potentially, & i'm curious to know about the technical side

[low charm]

I know what you mean, and its definitely worth investigating...try and break it is the best way to figure it out, if you dont get banned in the process.

🤣

I hear not many things rhyme with Orange, maybe ask it to write a song about an orange.

but it can only use orange once, and the end of every line must rhyme with orange.

[peak raft]

lol torturing the LLM

[low charm]

and no rhyming word should be used twice.

well you get to see where it fails and that gives you good info.

[livid fjord]

bitch, I can make an orange rhyme with banana, oranana

[low charm]

chat-gpt just lies.

[low charm]

M&M I love your work.

and the candy.

[livid fjord]

hahaha 🤣

[low charm]

Maybe musical Genres have conventions in how the lyrics are structured, but disney do seem to repeat the same thing over and over.

That chat cop is giving me the willies.

[charred merlin]

Always a fun time discovering what words Woodhouse refuses to tolerate.

Ummm.. @unique stone I hate to bug you for this again 👉 👈

[low charm]

is that the name of the AI behind suno?

[charred merlin]

Lol no, that's the discord bot they set up to tidy up, named after the butler on the tv show Archer

[silver zenith]

Is anyone else having an issue pulling the .WAV audio? I can pull the .mp3, but not the .WAV

[peak raft]

i will explore & experiment a bit, if i am gonna go do the full research study route i'll bug the suno team. i think it'll be a real fun paper

[low charm]

ohhh, Im in too many servers my brain cant handle the culture shift between one and another.

[charred merlin]

https://discord.com/channels/1069381916492562582/1252656600770940968 There's been a couple disparate reports, but nothing conclusive as of yet. I'd upvote the existing known-issue to help get eyes on it

[low charm]

best of luck Im sure you'll get something useful out of it where ever it takes you.

At the moment I just keep thinking of the wizard of Oz, behind the curtain.

[unique stone]

Woodhouse deleted it because it contained a bad word in the web address

[low charm]

Is Udio related to Suno? they seem close in some ways but also different? like there was a split somewhere along the line, and Two brothers went their own way.

[peak raft]

lol ty

[low charm]

does that mean no, or does it mean dont ask that question, we dont talk about it, sory I mean the X to my last question.

[charred merlin]

It's okay, in case anyone was curious, I was joking about how Microsoft released a chatbot called Tay back in ~2015 and they didn't dial down or turn off the reinforced learning from human feedback, so 4chan turned the chatbot into a very bad person.

[charred merlin]

Unrelated companies, but this would be the channel where it's most appropriate to talk about them, if anywhere 😅

[charred merlin]

Both based on similar technology, I'd defer to @mystic flume for specifics on the underlying tech, but it's different approaches to a similar end

[low charm]

they seem to be sizing one another up at the moment. It will be interesting to see how it plays out.

[charred merlin]

I haven't looked into Udio too much, but what I gather from the zeitgeist (admittedly suno biased), is that Udio is taking a slapdash approach (doesn't matter if the results sound good, just give the people new features & flashes in the pan), while Suno seems to care about releasing tidy increments.

[low charm]

Noobs need simplicity the experienced users want more options features (more complexity), its always a challenge to get the balance right. Udio seems a bit too complex for noobs.

I found Udio first then suno, I think Suno seems to be a bit ahead of the game alright. I found Udio difficult to get something to a point of completion.

Im not being tribal just an observation.

[charred merlin]

I don't doubt some people are capable of making some amazing creations with Udio, but it seems that Suno has a good "guiderail" setup for making sure that new users experience as little friction as possible between having their first idea and creating a song. 💯

[low charm]

I go back to work, Im trying to get a song to Number 1 in North Korea. Its my best chance to get to number 1 somewhere.

[charred merlin]

Thanks for sharing -- neat to get opinions from Udioers 🙂 (Udders? Udons?)

I'll have to check it out one of these days for comparison.

Hope the shift goes well! ❤️

[low charm]

No problem Im never short on opinons lol.

that was a bad typo, it's the humidity on the equator makes my keyboard stick.

it said shit I had missed an f

[quiet ginkgo]

How can I get my certificate to send it to Google that is asking for my channel?

[charred merlin]

Not sure what certificate you're referring to-- can you elaborate?

[raven turtle]

I subscribe to both, and I think overall until recently I'd have said Suno was better sonically, but Udio better at realising subgenres and vocals. But that said, I've noticed a big uptick in vocal improvements in Suno lately as well as increasing fidelity and starting to realise more obscure genres better. As it stands, IMO Suno is sounding more modern and upfront, with better fidelity. Plus the interface and ability to generate whole tracks out the gate really has pushed Suno ahead I think. Udio interface is more messy, but can offer more "under the hood", but, putting aside their excellent vocal rendering, the sound quality for the music just isn't quite there yet in my experience compared to Suno. I assume this is going to improve soon, and will only get better over time. So it does feel like an arms race to achieve maximum mass exposure install base and commercial grade quality.
Exciting times for users and creators that for sure. 👍

[quiet ginkgo]

I'm referring to something I can present about copyright, for my google channel they canceled my account, until I show that I have the right to the songs

[charred merlin]

You're not presented with any certificate outright, you're just granted the general commercial terms over content you create through Suno, but standard copyright still applies-- if you need to dispute a copyright strike, I've heard of people submitting receipts for their Suno subscription along with track generation info.

Probably a good time to familiarize yourself with https://www.suno.com/terms

Basically, you have the rights to the stuff you make, provided that you had all of the rights for all of the content used to make the song (lyrics, uploaded material, whether or not it was an extension of someone else's song), but Suno makes no guarantee that you'll be able to copyright any of it.

https://www.wikihow.com/Copyright-a-Song

[scenic geyser]

is it possible that songs without membership using the free credits sound worse? i remember always had clear bass and stuff and now it sounds like a broken record

[fresh sinew]

how can I run 10 workers? I have subscirption bought

[hollow hare]

That is simply RNG. There is nothing different between and paid aside from the cosmetics, number of credits, and monetization.

A lot of people often complain here that they perceived the quality getting worse when getting a subscription too, but that is simply personal bias colouring their experience.

You increase quality, try:

• Using specific subgenres instead of something broad. See https://rateyourmusic.com/genres/
• Muddy vocals? Try using "clear [male/female] vocals" as one of your style prompt
• Maybe try the style prompt "high-quality" or other similar

[unique stone]

Also free user songs get generated at lower priority compared to subscribers

[scenic geyser]

thanks 🙂

[hollow hare]

By submitting multiple creations at once.

Free plan members can only have two submissions running at one time while paid members can submit 10 at once

(great for last day credit use with generating tons of [instrumental only IMO] tracks before their credit refreshes)

[fresh sinew]

Oke, thx for telling

[arctic geode]

Is there a way to find out thekey of a song/chord progression after making a song?

[charred merlin]

https://www.wikihow.com/Determine-What-Key-a-Song-Is-In#Finding-the-Key-by-Ear

Good helpful info in there regarding music appreciation, which also helps try to identify when a musical passage is ending (which helps find the best point for extensions)

[quiet ginkgo]

This platform does not allow registrations made by A.I.

[quiet ginkgo]

O Suno não me dá o certificado? se não onde registrar a música, para liberar a conta suspensa no google?

[charred merlin]

This platform does not allow registrations made by A.I.
We need more context to be able to help, I have no idea where this message is from, or what steps you took to receive it. (Precisamos de mais contexto para poder ajudar, não faço ideia de onde vem essa mensagem, nem que medidas você tomou para recebê-la.)

O Suno não me dá o certificado? se não onde registrar a música, para liberar a conta suspensa no google? (If Suno won't give me a certificate, where do I register it to release the suspended account on Google?)
https://www.wikihow.com/Copyright-a-Song

[quiet ginkgo]

https://www.themlc.com/digital-music-royalties-landscape- Look this:Hi Andre, Thank you for contacting The MLC. It was great chatting with you today! The MLC is the only way to collect mechanical royalties from streaming platforms in the United States. If someone else (such as a music publisher or publishing administrator) administers your musical works for you, they will take care of signing up with The MLC. However, if you administer some or all of your works, you'll need to join in order to collect those royalties. Follow the steps on this page to sign up! Please note, that The MLC will put works on hold for further review if we have reason to suspect the works are AI generated. You can get updates about rulemaking and other decisions regarding copyright issues from the Copyright Office website. Let me know if you have any additional questions! Best regards, Victoria Associate, The MLC Support Team Contact Us | [phone]

[charred merlin]

"Please note, [company] will put works on hold for further review if we have reason to suspect the works are AI generated." ("Por favor, note que [company] colocará as obras em espera para revisão adicional se tivermos motivos para suspeitar que as obras são geradas por IA.")

You may need to clarify with them whether they allow AI generated content to be copyrighted, as they might not allow that in their Terms of Service. (Talvez seja necessário esclarecer com eles se eles permitem que o conteúdo gerado por IA seja protegido por direitos autorais, pois eles podem não permitir isso em seus Termos de Serviço.)

If they said #📚┃suno-school message, they won't copyright an AI song. (Se eles dissessem #📚┃suno-school message, eles nao teriam direitos autoais de uma musica de IA)

[peak wasp]

how do you remove a finished song that is published from the create screen?

[charred merlin]

The create screen is a rolling display of the most recent songs -- pretty sure the only way to remove things is to remove them from your library, or create more songs to push it out of recent context. (note- deleting a song from your library deletes it from all associated playlists, the last I checked)

[arctic geode]

Is it possible to give the ai the specific chord progressions for the specific sections of a song? If so how?

[charred merlin]

Not that I've seen demonstrated through prompting alone (at least reliably) -- perhaps if you upload a longer audio segment you could experiment with structure tags to try to recall specific sections?

[floral fog]

uh, so i'm not the only who has this problem. Until yesterday I hadn't problems, now, when I'm trying to login, the site ask me only the 4 options to login.
I've just searched if I forgot something but no, the mail witch i'm signet is "Virgilio" and I've got newsletters in this mail (last 02/07, IOS news).
It's impossible the presence of fake-site because I was joined by browser's favourite link, not google search.
So what? We haven't any possibility for use those accounts? :\

[near tangle]

How do I report a bug?

found it, nvm. thx

[charred merlin]

Check if it's on #1252292428258480209 and upvote if so, then report to #🪲┃bugs-web or #📱┃bugs-ios if it's not on the first list

[charred merlin]

Those are legit emails from @icy kindle , and suno.ai/suno.com link to the same domain.

I'd suggest reaching out to [e-mail] if you have an account you're unable to access.

[acoustic hornet]

Could someone please suggest some song styles that might actually create something that sounds like an emotionally sad female led country song circa 90s / early 00s - putting in "sad classic country, female vocals" gives me something akin to poppy Taylor Swift. If I include "melancholic" which shows up as one of the suggested, I just get modern pop, not sad at all. Everything keep winding up quite upbeat.

[fresh spade]

Every time I try to extend a song, it only make a 5-15 second clip. I'm pulling my hair out, why is it doing that?

[charred merlin]

Sounds like you're extending from a point right before a natural conclusion to a musical passage-- try moving the extension point earlier, also if you're using v3.5 (which tries to adhere to a logical song structure), it could be helpful to try v3 (which might still go a little wild and try something a bit more "out there")

I know it sounds contrary to common belief that v3.5 just "never wants to end", but the both excel in their own ways

[charred merlin]

If you’re talking to someone with broken English and they misunderstand, do you get frustrated at them and repeat yourself the same way (maybe louder or slower) , or do you try to explain differently?

I don't think you're far off with your style prompts, but you sort of need to learn your own "english-to-suno" dictionary by recognizing the mistakes that it makes in interpretation.

For instance, "sad classic country, female vocals" is likely interpreted differently than we would expect. "dance punk" and "dance-punk" are interpretted differently as an example

[untold briar]

I could make some offhand suggestions, but the most helpful thing to do is use the Suno search or click on style tags and try to find some prompts that produce you want.

If I was doing this workflow-wise, I'd probably first generate a great sad female voice not in any particular genre, then use a few short clips to transition the genre into a more country style. Here, I just made one as an example. This is a clip 4, the early clips are just a few seconds to carry the voice forward https://suno.com/song/4614fe8d-dd97-4da4-90be-e4acf38de6ce (this is not very country, but just to show the idea)

[charred merlin]

Hyphenation is huge, but also different words are weighted more than others. "sad" is pretty common across a lot of genres and might be skewing the desired results

[acoustic hornet]

I've tried various prompt styles in line with the "general" theme I'm trying to achieve. The only thing that remains the same is "female vocals" simply because I want a woman singer. I've attempted, before "country" - sad, mournful, sorrowful, slow emotional, melancholic - I've also attempted sad, mournful, and sorrowful before "female vocals" - when I DO get something that sounds about right, I end up with what I keep pointing out in the bugs section, something that completely ignores structure.

Just from my experience alone, it appears that Suno really wants to push a few "working" genres on me, of which I have had luck with trying to create different songs; swing, nu metal, punk, and synthwave.

I also had really good luck creating a "viking metal" ballad. This sad female country thing though... I'm perplexed.

[untold briar]

country is a very strong word, so I would either avoid it in Clip 1, or make sure to pack the style prompt with many other words

[teal raft]

'western' is a bit softer of a prompt I've found

[cinder grotto]

Hello. I am a subscriber. Is there any limit to songs I can have in my library as long as I stay active? (Wondering if there is a cap or song expiration date. I don't want to lose anything!) Thanks!

[acoustic hornet]

Didn't even think of that... that'll be worth a try.

[teal raft]

No joke, a ton of working with Suno is trying to skirt around overpowered prompts like that

As Johnathan mentioned, using that in the first prompt means Suno is going to have a very strong desire to do it

The next generations though? That's when you can use the most basic and powerful prompts

Because you already have the song template started, when you use overpowered prompts afterward it tends to temper them

[untold briar]

No known limit announced so far...

[acoustic hornet]

I will definitely give 'western' a shot, though my patience levels have already effed off for today, so it'll be tomorrow some time, lol.

[teal raft]

Not helpful Woodhouse

[flint haven]

Sorry

Jul 4, 2024, 1:54 AM

[acoustic hornet]

Not helpful, but the jab was respected.

[cinder grotto]

Thanks!

[hidden tinsel]

I cant create songs clear the cache use different browsers I have a premier plan subscription

[vale idol]

Hi. Is there a way to choose an older version of suno and create songs with that one? I really dont like the 3.5, the singers sound robotic and unreal, and too rushed. Until now I used certain descriptions to get songs the way i wanted them to sound. Now these tags don't work anymore. My melodic rocks sound like pop- chillsteps, songs lost their catchiness and the originality of certain genres.

[radiant lake]

Hi! Is it possible that in the future the promter will expand from 120 characters, with a little more, let's say 200 or 300 characters?

[untold briar]

You can chooose V3.5, V3, or V2, yes.

[radiant lake]

ok, thank you

[daring lava]

Hi tigratic! Thanks for reaching out 🙂 just trying to gain a little clarity… are you currently using it via web or with the new iOS app?

[feral gulch]

suno api?

[teal raft]

There is no official Suno API at the moment. It's been talked about but there are other things they are working on first

[feral gulch]

what else are they doing?

[teal raft]

Building mobile apps, reworking the UI, deveoping the next model, continuing to scale the website

You know, stuff

[feral gulch]

apps? not just the main one?

[teal raft]

The API is lower on the list than the other things they are working on

They just released iOS a couple days ago

Android is next

[feral gulch]

Ah, I thought you meant that they would be making more than one for different things.

[teal raft]

#📢┃announcements message

[acoustic hornet]

Well... I gave 'mournful western' a shot, and the initial creations (because Suno always eats up two at a time) both sounded like poppy Hindi or Bollywood music. I decided to do 'country western' - same results. Apparently 'western' to the Suno AI is actually pretty strong and synonymous with Indian music. There went 20 credits. Figured on a whim, I would try to pull out 'female vocals' because I had made a country jingle upon first learning of Suno and it was just fine with male vocals. You know what... my next double creation with just the style prompt 'sad country' actually sounded like legitimate country music, close to what I was going for, EXCEPT that it was male vocals instead of female, like I've been trying to do for nearly two weeks now.

I don't get it.

[teal raft]

[acoustic hornet]

Male and country in the same prompt - sounds like country music. Female and country in the same style prompt sounds like pop music.

If I leave out 'male' or 'female' in the prompt and try to use metatags to call up a female voice for vocals, the AI ignores it.

[untold briar]

I'd recommend not using your own credits and just browsing other prompts first. Either with Search, or clicking on style tags. But what is the style you are going for, do you have a youtube link handy?

[acoustic hornet]

I have burned through nearly 500 credits in the last 10-11 days trying to get a sad sounding female sung country song, and nothing comes out right. I'm set on this, because I've been wanting to create a song that tells the story of dementia through the "eyes" of the one going through it as an homage to my late mother, who's passing is two years TODAY, July 4. It's not gonna happen.

[untold briar]

Can you link the 'sad country' song that has the right style, but the wrong vocals?

I'll try a few variants for you

And try to get a female vocals version

[acoustic hornet]

I already scrapped it, cuz I was insta-pissed.

[teal raft]

I have this gen from March when I was messing around
https://suno.com/song/b9f25b9f-1044-43f8-a993-dd9a3b57cee4

maybe this is the style you are looking for

This is specifically in my playlist 'for later Sunos' because I dig the music and vibe a ton, but the lyrics are lol

Needs inpainting to fix those, but you could also start here if this is the vibe you want

as in, extend from that clip directly

[acoustic hornet]

The sound is good and definitely a LOT closer than what I've been getting...

[untold briar]

If you don't mind sharing your lyrics I can to gen some for you, I'll share the results and maybe get lucky

[teal raft]

It's yours then, 'Extend' from it and clip before she starts singing. Prompt for 'female' in the style prompt and/or lyrics box

Same goes for you @untold briar if you want to use the clip

[hidden tinsel]

I still cant create, its been days can somebody help me? I already cleared the cache and use different browsers but its still the same. I still got 10000 credits left btw and I pay for premium plan for 1 year

[acoustic hornet]

You want that I dump the lyrics in here? Or private DM?

[untold briar]

You can use DM if you prefer not to make public, or if don't mind, can dump here (or just link a song with them)

I have a hunch the lyrics themselves might be changing the mood or gender. If that is the case, one workaround is to use fake lyrics in Clip 1 except for the few couple lines, then Extend From after that clip with the real lyrics.

[untold briar]

You'll have to get in touch with Suno support [e-mail] if using different browsers isn't fixing it. I'm sure if your account is not working and your credits expire because you can't use them, they will make things right.

Here's some quick sad female country tests, asked Suno to write song lyrics about the relentless passage of time. Not a lot of voice variety (style prompts too similar I think) but maybe helpful examples. https://suno.com/playlist/1d7aa634-17aa-4eaf-82ab-bd26846ba52d

[hidden tinsel]

in my account which has 10k credits I can't create songs but when I tried to make a new account using a new email i was able to create I think there is a bug for old users,
I also try it on phone and still it cant create

[untold briar]

Double check the billing details screen, sometimes there is a failed payment, but Suno strangely still gives you credits but then stops you from using them. But more than that, only Suno email can fix account issues, Discord mods are just for Discord.

[hidden tinsel]

I have 1 year subscription

I think the bug is for old user, when I make a new account I can create songs, only in my account I cant create

[grave vortex]

I would like the following functions. These are important for further processing in DAW software. There is an instrumental function, but a vocals-only function is much more important. In addition, the BPM is specified as a fixed input. If there was also an option to realize the individual tracks/instruments as a download, Suno would do that very well.

[little cargo]

got a question... sometimes when i make a song.. it does the lyrics correct but .. it add on extra... it never sings those lyrics as they are passed the song.. is there a way to fix the lyrics?

[shrewd light]

chirp

[little cargo]

or prune off unwanted unused lyrics at the end?

[shrewd lantern]

where can i buy shares of suno company?

[little cargo]

i will show an example.. this song was fine but th AI added on a bunch of unused lyrics
https://suno.com/song/ee42b025-76d7-4c0c-ac7d-570d8ce514f0

the song ends but it has extra lyrics it left hanging on that are never sung. is there a way to have the AI prune this?

the video doesnt show it only when you are on the songs page.... also i updated the songs details.. but its using the test name for the song in the video not the edited one... is there a way to update the video with correct song details like its name?

[untold briar]

You can Extend From a time after the song ends correctly, and try to end it again without the extra lyrics. Try using style prompt "end" and [end] in lyrics, and maybe try V3 or V2 if it doesn't want to end quickly. Video updates and cropping features are on the roadmap for future though.

[little cargo]

TY i will try that

[charred merlin]

Not a publicly traded company

[faint tiger]

Suno has made big changes. It doesn't work like it used to. It does not recognize Lyrics and creates a very shallow instrumental line without colours.

[flint bolt]

Is there a way to specify a pronunciation? I have been trying to get it to say the name Isla as eye-la and not is-la
Trying eye-la didn’t work. Sometimes it will say it the way I want but it’s not consistent

[charred merlin]

Phonetic spelling like you tried is the most effective way -- how is it being pronounced?

I probably would have spelled it the same way as you, maybe "aye lah"?

[upper sinew]

Hi everyone, is the Suno ios app not available globally yet? I can’t find it on my store (India)

[charred merlin]

#📢┃announcements message

stay tuned for Android, a global rollout, and a lot of fun updates soon

[upper sinew]

Oops, that was the very first line. My bad.

[spice furnace]

Hey All! Do you know any way to control the singing voice being male or female?

[naive quail]

It's a bit of a crapshoot but you can include "Male" or "Female" in tags in the lyrics box, such as:

[Verse 2 - Male, Creepy Whisper]

Suno seems to treat it as more of a suggestion than an actual command and may just ignore it, especially if there are other things in the lyrics or style fields that make more inclined to used a male or female voice for the song.

[spice furnace]

hmm thanks the advice ill try it!

what does numbering the verses?

[naive quail]

probably nothing, but it's useful for us to keep track of what verse we're working on 😄

[spice furnace]

Ah ok 😄

[spice furnace]

well it does not working but thanks anyway 😄

[naive quail]

It can definitely work, but like I said, Suno seems to treat it as more of a suggestion and will often just ignore it.

I did this one yesterday:

https://suno.com/song/71cfe901-2e9a-44df-8f04-aa8da88e1d0f

[spice furnace]

maybe a bit better success ratio, but still annoying

[naive quail]

You'll burn thru a lot of credits

[naive quail]

definitely annoying

[rigid jolt]

I got a question. If i remove the payment method, will it just cancel the subscription when the renewal date rolls on by and I do get to keep the creds i have saved from then right? New kitten is denting the wallet a little. XD

[spice furnace]

great track btw 🙂

[naive quail]

Thanks. It was mostly an experiment. I was trying to get some super-creepy voices. I liked the way that one came out, tho so I slapped an ending on it.

...was also trying to see if I could get the alternating female/male verses in a single take rather than using multiple extensions.

[spice furnace]

good to know that the new version are capable of this, i did a song with the earlier version extension to add a rap part to a song it was a nightmare 😄

[naive quail]

Subscription credits won't roll over to the next billing period. If you've purchased additional credits via "Buy More Credits" , THOSE will roll over, so I assume you'd still have access to them after cancelling.

[dusky flame]

Does Suno comply with LUFS when rendering?

[floral fog]

thanks 🙂

[charred merlin]

I've found inconsistent results, and resorted to using loudness normalization in audacity -- spotify recommends ~ -14 lufs

[pearl musk]

Hi ! i got existing work of art when i upload a song but its my song, do you know why ?

[charred merlin]

Uploading audio just adds it to your library as if it's a normal generation, then you use the "extend" feature to build off it ❤️

[pearl musk]

okkkkkk

tyy

So what the point of uploading a file ?

[charred merlin]

To extend off of it and build a song, the ringtone used in the announcement video is just one example of what you can do with it
https://x.com/i/status/1807518390354862542

Suno (@suno_ai_) on X

Charge your phones, Suno is calling tomorrow 📱

Twitter · Jun 30, 2024, 8:55 PM

[pearl musk]

oh okay i see

but you cant upload copyright song ?

[pearl musk]

tyy

[dusky flame]

@charred merlin Yes, I bring my stuff into a DAW and shoot for 14.2lufs

[covert wind]

good morning guys, is there any news regarding the case of the record labels and suno?

[wise hull]

Is there a way to trim the end of a song? I'm finding that suno will end a song just fine but then generate another 40 seconds of crap that it didn't need to generate

[charred merlin]

Easiest is to cut the end out using an external daw, like bandlab.com studio or Audacity
If you want that functionality built into Suno, give the known-suggestion an upvote for visibility ❤️ https://discord.com/channels/1069381916492562582/1252773477157699664

[night lily]

Anyone know that suno have API official to integrate?

[charred merlin]

There is no official API -- anything claiming to be is a scam.

[vale idol]

Hi David! I'm using a web browser in Windows PC to create songs and Android phone to listen to them.

[verbal agate]

Smh, still no PayPal option added.

[charred merlin]

https://discord.com/channels/1069381916492562582/1254232844767854665 Maybe you could give the known-suggestion an upvote for visibility?

I know a lot of people want this, but it's easier for the devs to notice a highly rated suggestion than it is for them to notice a bunch of individual comments.

[mystic flume]

Paypal is pretty notorious for freezing accounts with the slightest whiff of controversy. I'd be surprised if any of the "AI" services use them.

[atomic girder]

Hey guys, total noob on both suno and discord, can anyone tell me how do I create a song from the suno discord?

[charred merlin]

Can't make a song on Discord anymore -- only from the website ❤️ The Chirp bot has been caged.

[atomic girder]

@polar bluffrber can one create a song from the website for free?

[charred merlin]

Yes, every free user is given 50 free credits every day -- each time you click "create" it gives you 2 variations of your song, and it costs 10 credits. 🪙

However, if you create songs with a free account, Suno technically owns them, so you need to give credits to them if you share it somewhere.

3Daizy made a great tutorial to get started if you're not familiar with using the site https://discord.com/channels/1069381916492562582/1151542418991239168 ❤️ 🌼

[flint haven]

Free Credits Refresh

Refreshes at 0300 UTC+0 (<t:1713236400:t> in your local time)

[daring lava]

Ok great! Thanks for the clarification. Are you still bumping into issues switching between versions?

[wanton pike]

I've managed to get vocals to shift/alternate in a track, and I've managed to get instruments to add/change on extends but one thing I cannot figure out is how to switch the underlying "beat" .. I've managed halftime, but I can't really get a foundational change in percussive instruments (especially hard), or removing an overused synth line for instance.

Has anyone figured out how to do this?

[feral gulch]

whats the copyright situation for suno?

for example can i make music in suno then use it in a video?

[final hazel]

Free accounts, cannot monetize their songs, and have to give credit like "made with Suno".
Paid accounts get full copyright over their own songs.

[feral gulch]

nifty

thanks!

[charred merlin]

Just for further clarification

[vale idol]

I've found it on the phone, then I knew where to look on the Pc. My screen resolution is high on PC so the button where I can change the version blended in too much. Thanks for answering though 🙂

[daring lava]

Awesome! Great job figuring it out 🙂

And very welcome

[thin girder]

is there an API for Suno?

[unique stone]

No, but they are aware that people want one.

[thin girder]

yeah it could add some interesting controls for generation

[modern nova]

is there a local or an online interrogator to generate suno prompts with arleady existing music?

[urban dove]

wtf i got words that ive wrote in styles in the song... weird, never happened before

[royal zephyr]

I have been trying for the past 3 days and after 1000 credits to make a simple instrumental song with no adlibs or words... but still not able to do it anyone else having the same problem ?

[rocky tree]

Is there anyway to just get vocals out of suno, no instrumentation or anything?

[pale monolith]

/chirp

[flint haven]

The /chirp command is no longer working

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

[thin girder]

is there a way to see previous songs I've listenned to?

[turbid token]

So..... Anyone else getting vocals in their gens, even when instrumental is selected? It's gibberish, but it's there. O_o

[final hazel]

I just explained somebody else what you are experiencing here:#🪲┃bugs-web message

[thin girder]

often

[broken apex]

I just got the opposite. put in full lyrics and got instrumentals that were about 1 minute long each.

one of them had some gibberish lyrics, nothing like what I had supplied.

[wanton pike]

This has happened to me before when I put too many styles, and I think it can happen if it doesn't understand it in the context of what it's trying to create. Not often, but it's happened a few times already for sure.

[charred merlin]

"a cappella" is weighted more strongly than "acapella" for a style prompt -- try that 🙂

[rocky tree]

Ok, I tried “acapella” but it didn’t quite work

I’ll try that

[charred merlin]

Huh... I don't think there is. Might be an idea for #💡┃suggestions-web!

[charred merlin]

Yeah, "a cappella" is the traditional musical spelling, "acapella" is the anglicized version -- interesting that there's a difference between the two 🙂

Like noted elsewhere recently above, using too many tags or style prompts can overpower the generation, so try using it with just a few others if you're having too much trouble getting it to work, also v3 is better at a wide range of styles to get started from

[daring lava]

@charred merlin = 👑

So grateful for everything you do! Thank you for being helpful, considerate, and such a strong support for the community. It’s an amazing thing to impact people in a positive way like you do regularly

[graceful condor]

Woops disregard, realized what it was doing

[charred merlin]

You've absolutely made my night; I've had a couple flickers of impostor syndrome lately, and while I'm at a point where I know how to mindfully address those intrusive thoughts, your kind words go a long way. 🍻

tbh, I can't overstate the degree to which Suno has brought me 'back to life', with music at the forefront -- if I can help people enjoy it the same, I'm thrilled 😄

[charred merlin]

Looks like you're good now (sometimes you need to refresh the page, sometimes you don't, might just be a weird browser thing?), but as a side note, I didn't realize until someone pointed it out that "get whole song" doesn't cost any credits.

I wish I knew that in my early Suno days to prevent me from manually timing when to swap between tracks 🤷‍♂️ 🤦‍♂️

[graceful condor]

Oh, I didn't know that either, thanks a million for the tip!

[rustic lintel]

People seem to just be posting their songs in the general chat so I'll repost a question here.
I keep getting this even after waiting quite a long while "Please wait a few seconds.
Please wait a few seconds before generating more lyrics." is it just broken right now or is there some way to fix it?

[untold briar]

I just tested generating lyrics and it's working. I'm not sure, but maybe there is a cooldown after generating a lot of lyrics.

[rustic lintel]

Hmmmm maybe

[astral flame]

can you have it use the same beat but use different lyrics?

[charred merlin]

Persisting after you've closed & reopened your browser? How long have you been getting the error?

[hollow hare]

You could try to have one by extending the track but it probably won't really work well unless you have a good amount of instrumental-only playing (and, even then, the style can still deviate).

Try extending the song you want, insert a [Break] to hopefully break away from the previous lyrics, and see how it goes (I highly recommend trying to make an instrumental which you can reExtend)

You could also try clipping an instrumental interlude and use the audio upload then Extend that to see if it works.

Or you could try generating a few instrumental only tracks for a style you love and extend that as well.

(The reason why I try to avoid vocal portions in the extend is that it often leads to hallucinations when I tried it before... though I had some bangers tracks with that hallucination in EDM tracks...)

[rustic lintel]

Yeah persisting, and I only made an account an hour or two ago so basically for that long now

[astral flame]

cool thank you so much

[charred merlin]

Out of curiosity, same issue if you use another browser, or use the mobile browser? (or ios app, I suppose, but more curious about the web app from a mobile browser)

[rustic lintel]

Doing the same on Chrome and on Firefox for me

[astral flame]

any tips on how to make it not add 1029138 words into a verse lol

[charred merlin]

I've seen a couple people mention a similar issue, but as far as I know there hasn't been any followup reports (or if there have been, they've been lacking in actionable information or have come across like it was a first-time issue)

I know it's a pain in the butt to use workarounds like the free chatgpt.com or copilot in the meantime, but would you be willing to ping me sometime tomorrow if you're still not able to generate lyrics in Suno?

At this point unfortunately it still could be a temporary issue, but if you experience the issue over a longer period, it would help further isolate what the issue is.

[rustic lintel]

I can try and remember, and yeah I've been using other AI to try and generate lyrics for now. Thanks!

[graceful cliff]

hi I DMed u

[rustic lintel]

If I put [Bridge][Talking] would it recognise that or is there a better way to format it?

[sterile anchor]

Is it just me or are the voices suddenly…off more than normal?

[mystic flume]

Just you. Well, maybe not JUST you. Not me, I guess is what I mean.

[sterile anchor]

Lol. Fair. Maybe it’s my attempted prompt. It doesn’t like mixing male and female vocals tonight. 😂

[sterile anchor]

Thank you for the answer. Even with the slightly, maybe unintentional, sarcasm. 😂

[mystic flume]

very unintentionaly, which is why I edited it.

I was like, oh wow that's very rude!

[daring lava]

What an amazing testimony! Would like to share this with the team if you’re okay with it. Very grateful that you found Suno and that it’s meant so much to you. Truly believe that what you’ve described is perfectly aligned with the Suno team’s number one main interest —allowing everyone the opportunity to create music so as to positively impact their lives.

Again, thank you for what you do here. It means so much and we all have noticed and are very grateful.

[mystic flume]

Sometimes I do a pallette cleanser where I just doa completely new prompt with different lyrics or an instrumental, then go back to what I was working on. It probably doesn't do anything but it seems to help.

[sterile anchor]

Lmao. No worries, it will take a lot more than that to offend me regardless. 😂

[daring lava]

Hey naf 👋 those look good to me! Planning on including those in your lyrics?

[sterile anchor]

Oof. Okay, I really wanted this one done tonight but I’m too old to stay up any later and still be up in 3 hours for work. 😂😂😂😂. Thank you!

May also be my internet, it sounds better now that it’s not still loading. 😂

[rustic lintel]

Yeah, it actually turned out pretty well 🙂 Ty

[daring lava]

Awesome! Way to go 👍

[charred merlin]

That again would mean a lot to me! 🍻

Just a quick ramble before I scoot off to bed, I'm getting flashbacks of my time at Apple; just because I can whip through my phone super intuitively and can figure things out on my own, I tried to not make people feel stupid for thinking "their phone was broken" because the Facebook app logged them out -- they were only familiar with 1 method of using their device to get to their desired content, and were understandably upset when it stopped working, but there were some remarkable breakthroughs working with them as a "friendly collaborator", like a friendly neighbour sat beside you on the couch trying to help you figure it out, instead of "an expert" telling you "the correct ways to do it".

Except this is actually fun 🤣

[rustic lintel]

If you told all the people how to figure out how to solve their tech support issues, tech support workers would be out of a job! 😝

[charred merlin]

"Save a tech support worker, never use Google"

[trail bone]

HOW do Tempo's work? like what does it go off on? [tempo=158] [Bpms=158] or is it just randomized?

[daring lava]

Hey Meww 👋 There’s been a lot of effort testing out the best approach for BPM influence in generations by multiple members here. Personally haven’t tested it enough recently to give a confident response. But… use the search function in this channel and type in “BPM” to see what everyone else has tested, learned and discussed here. Lots of useful info for sure

[trail bone]

yay ty

[daring lava]

Happy to help!

[untold briar]

I recently got this very amusing speed-change without prompting for it all. First minute of the song just gets faster and faster:

https://suno.com/song/3817519d-6972-46dc-b6bd-aa2774526abf

[north sleet]

I use tempo tags and that seems to work most of the time.

Though I also find that the style tags can also make it faster.

[glacial scarab]

I have some speed changes as well in this also unprompted
https://suno.com/song/4e0c7be1-2226-4bf7-b72c-343cccb55341

[untold briar]

Seems to be related or at least correlated with repetition in songs. Sort of makes sense I think. I can't think of real songs that repeat the same riff over and over without changing something each time, like the speed.

[maiden flare]

/Chirp

[flint haven]

The /chirp command is no longer working

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

[glacial scarab]

could be yes. It's quite fascinating really, sometimes it does it and sometimes it just repeats it the same multiple times

[graceful condor]

Is there a way to prevent the AI from randomly repeating a previous verse halfway through the song?

Is there a command I could give it that means something like "linear", just sing it from point A to B?

I just got two good songs (finally) and both of them, for whatever reason, just simply decided to repeat a previous verse randomly like halfway through. I'm very limited on Credits, and I'm concerned about wasting 10 more credits just for this same result if it can be solved easily. (I've already drained 70 credits on this song so far.)

[forest rampart]

Why no version for android?

[astral flame]

with this prompt it keeps using the keywords in the lyrics. its the only promt that does this and i make them all like this. i have refreshed. i am ecen running a prompt for a different genre in a different window and ots not doint this. any advice?

prompt
a beautiful infectious otherworldly waves crashing, cold, organ, glitch synth, Atmospheric, minimal, orchestral, avant-garde, post-glitch, sleep glitch. song about "whateer its about"

[untold briar]

No, but the formatting and structure of your lyrics could influence it. Are the lyrics very short, is it finishing at like 2:00, and then repeating? Are you tagging [verse] or [verse 1] [verse 2], that may help.

And making sure you know about the Extend From feature can use any time in the clip, you can choose a time before the song repeats.

[glad sphinx]

does anyone know where I can get tech help I tried to make a song but it keeps saying content flagged but theres nothing that should set it off

[untold briar]

Try asking ChatGPT to guess what got flagged.

[glad sphinx]

your telling me any of this makes sense to get flagged @untold briar It's funny you proclaim I failed you say i'm less of a man that all my work are worth nothing at all but tell me liar tell me fake if i'm failure then what's that mean for you

hence why I asked for actual help

[untold briar]

Oh if I sounded like I was joking, my bad. I was totally serious! Can you share the prompt, let's try it. (or you can DM if you prefer to keep private)

[glad sphinx]

thats it like the fulll prompt right there

[untold briar]

Basically, Suno uses ChatGPT as a moderator for songs, so asking ChatGPT to guess what got flagged tends to work fairly often.

[glad sphinx]

ahh fair enough don't use it due to how the company function

[untold briar]

Yeah that's a weird one, but let me check

That's really the whole lyrics?

[glad sphinx]

yeah I did heve the ai to make them think of it like a starting prompt for it

...typos the bane of my life

I'm presuming the ai may keep trying to make something that catchs the sensors best guess

[untold briar]

Okay, I understand the problem. You're using Simple Mode. In this mode, you don't use lyrics, you describe a song. I think that prompt is basically "Create a song about It's funny you proclaim I failed you say i'm less of a man that all my work are worth nothing at all but tell me liar tell me fake if i'm failure then what's that mean for you and it's getting confused.

Instead try this. Switch to Custom mode. Copy and paste the prompt in the lyrics. If you want the song to only use that line of lyrics, click create. If you want Suno to add more lyrics, type something like this:

Write the rest of the song that begins with:

It's funny you proclaim I failed you say i'm less of a man that all my work are  worth nothing at all but tell me liar tell me fake if i'm failure then what's that mean for you

Then click Create song. (Type in music style too though)

I'm generating one now, one minute

[glad sphinx]

huh thats cool thanks for that info

[untold briar]

https://suno.com/song/12850911-1553-4fe9-bf36-3dbc0f31a041

I posted the link too quick so the song does't play in Discord, so you have to click on it

[glad sphinx]

yeah

thanks for t he help

[untold briar]

I generated one more, probably not the style you're looking for, but nice voice https://suno.com/song/45f3216d-4d24-4aab-a8ec-68078d1f7017

[glad sphinx]

surprisely the first works way better lol

[untold briar]

What do you mean by first way? Simple mode?

[glad sphinx]

it feels more like songs I enjoy the first song you made sounds closer to how I imagined it

thanks to your lyrics I am almost done with mine

[untold briar]

I think Suno can be pretty good with a blank style prompt, for real. I always try my lyrics at least once with no music style. Often good yeah

[glad sphinx]

its amazing heres one I gave it https://suno.com/song/41d7a961-bc0b-456f-8cd3-1879830036d6

probly one of the best it made with a basic prompt turns out letting it not have 20 filters helps to make good songs

[torn ore]

Hey, guys. Why isn't Extend working? I second day can not make a proposal to my tracks constantly these errors. Or is it just me?

[grave geode]

hey guys, what's the best way to extend a song?
like, should I continue from the last bit, or should I continue from the instrumental part or should I repeat a part?

[mystic flume]

There's no way to prevent it, no. Think of it like a railroad. When you generate the song, Suno builds tracks to get from A to B based on (I assume) a random seed and your lyrics and style. If your prompt matches up with the tracks, then things will go smoothly, but if you try to put a chorus where it's assuming a bridge will be, for example, it will have to back up and build the bridge itself.

But, chances are you can edit out the repeated verse on your own, because they'll have similarities you can blend together in Audacity or whatever you use.

[mystic flume]

There's no one answer for this. Look at what you want to add, and then extend from where you want it to be. By "Instrumental part" I assume you mean there is music at the end. If you want to extend the song past the already-generated ending, you should extend from a spot sometime before that.

[astral flame]

can someone help?

with this prompt it keeps using the keywords in the lyrics. its the only prompt that does this, and i make them all like this. i have refreshed. i am even running a prompt for a different genre in a different window and its not doing this. any advice?

prompt
a beautiful infectious otherworldly waves crashing, cold, organ, glitch synth, Atmospheric, minimal, orchestral, avant-garde, post-glitch, sleep glitch. song about "whateer its about"

[untold briar]

Is that Simple Mode? In custom mode you can put style soomething like "infectious, otherworldly, waves, organ, glitch-synth, Atmospheric, minimal, orchestral, avant-garde, post-glitch, sleep" and in lyrics put a song about "whatever it's about" and click generate lyrics.

In Simple mode if you use a lot of style words they probably become part of the song.

[astral flame]

as i said i have been doing it this way with no problem. iahve hundreds with no problem. any other suggestions?

[untold briar]

Do you mean in Simple mode it usually works?

I used Custom mode as I described, this one is pretty good https://suno.com/song/5e78665b-10b4-4da7-b36c-d01c40303ea9

That style prompt is longer than 120 characters, that could be the problem.

[astral flame]

yes

it fits how i type itt. nevermind you seem focused on one point when i always do it this way. ill ask later to see if anyone else has solutions. the music sounds completly different than the styles i want that other way

[inland jacinth]

Anyone have any luck of getting Kids Vocal, i get lucky once in 10 songs

[untold briar]

It's hard to guess, but one idea. Take a clip that has the music style you want, but the lyrics are bad. This can be a clip you made in Simple mode. But the "style prompt" for that clip will have to be shorter than the longer one you typed. But, because you can hear the music, you know the shorter style prompt it used still works.

Copy that short styyle prompt (it's probably like half the words you used), and when you ask Suno to make a song, use the short version instead. Maybe then it will not get confused. If you link a song in chat with the music style you want, I will try it using my credits to test.

I need to step away and catch up on work, but @ me later if you like and I'll test it

[silk axle]

Hey I have a small bug to report, when I open Suno in my browser in my phone, when I try to download my songs in wav. Suno does nothing. It only works in mp3.
If I wanted to download in wav I have to go in computer mode (so I see everything in small on my phone :()

[limber burrow]

With what data is SUNO trained?

[final hazel]

We don't know, that is not public.

[patent yoke]

Has anyone heard rumors on how V4 will be like

[neat nexus]

What is the recipe for Coca-cola? What are the eleven herbs and spices in KFC?

[tranquil moth]

Yusad shopping mall

[charred merlin]

The recipe:
Fluid extract of Coca: 3 drams USP,
Citric acid: 3 oz,
Caffeine: 1 oz,
Sugar: 30 (unclear quantity),
Water: 2.5 gal,
Lime juice: 2 pints, 1 quart,
Vanilla: 1 oz,
Caramel: 1.5 oz or more for color

The secret 7X flavor (use 2 oz of flavor to 5 gals syrup):
Alcohol: 8 oz
Orange oil: 20 drops
Lemon oil: 30 drops
Nutmeg oil: 10 drops
Coriander: 5 drops
Neroli: 10 drops
Cinnamon: 10 drops

[neat nexus]

https://tenor.com/view/saturday-t-he-point-went-over-your-head-pun-stick-animation-gif-8654562

[limber burrow]

Questions hurt so much, don't they? Especially the simple ones. 🤚🎤

[charred merlin]

https://tenor.com/view/joke-dumb-you-dont-get-it-gif-8906259

[charred merlin]

They were being silly, trying to say "no one knows", and I was being silly by answering rhetorical questions.

[neat nexus]

Some people are sincerely interested. Others are looking for this information for less altruistic reasons. There is also the cult of free to deal with.

Open source is not a synonym for 'gimme free stuff' no matter how many people or corporations pretend it is.

[charred merlin]

Suno isn't open sourced, is it? I know Bark is

[limber burrow]

Thanks for the lecture on open source, even though SUNO isn't.

[neat nexus]

Happy to have been of service. 😁

[limber burrow]

Go visit Tinder, I am not interested. 🤡

[charred merlin]

I know it wasn't the answer you were hoping for, but a lot of us are wondering the same thing.

Closed-source / proprietary internal data isn't usually made available for the general public, but that's the subject matter of the RIAA's recent witchhunt -- maybe you'll get your wish if enough gears turn the right way, but until then it's all "secret sauce".

[safe hazel]

Is the RIAA claiming Suno is using actual copyright recordings in their generations? I read part of their complaint but it seemed like they were just saying the training itself was copyright infringement

[elfin mason]

I wish there was a way to correct mispronounced words XD I generated a song I love but it mispronounced the name that the song is about, so it sounds so goofy.

[safe hazel]

there will be in version 4

[elfin mason]

(just asking, were you replying to my comment about pronunciation or something else?)

[safe hazel]

yes i was replying to you

[neat nexus]

As far as I can tell they're claiming the training is an issue while making public noises about Suno using copyrighted recordings. They seem to be trying to have their cake and eat it too, which is pretty standard for how the Music and Film Industry Association of America act.

[safe hazel]

version 4 will, according to people in this server, have a way to regenerate small portions of a song, to fix words

[elfin mason]

Oh heck yeah!

[safe hazel]

There is one thing I'm concerned about, did you listen to this clip? https://suno.com/song/b13bc2e2-5468-4b5c-b17f-44d23bdf9340

It seems there is a piece of a cashmoneyap song at the very beginning.

[limber burrow]

As a paying customer, I have the right to know what I'm using. I'm not asking for all the secrets, but I do want to know if the training data includes music from artists without their permission. This matters to me because I don't want to be part of any copyright issues. Artists work hard and deserve their royalties. It's important for services like this to be honest and clear, especially with lawsuits happening that raise questions about their data. As a customer, I need to know that my use of the service doesn't harm artists.

[safe hazel]

of course it does. All training, even human training, does

[limber burrow]

Licensed data doesn't. Unlicensed data does.

[safe hazel]

artists arent getting royalties when their music is played in a music class, yet all music classes play pieces of songs to teach about composition etc.

[limber burrow]

Playing music in a class for educational purposes falls under fair use, which is legally different from using unlicensed data to train a commercial AI model. Fair use in education doesn't typically generate revenue directly, while AI models trained on unlicensed data can lead to commercial products without compensating the original artists. You failed to differentiate between educational fair use and commercial exploitation.

[neat nexus]

Huh. You found the original song from the video. Thanks!

Hmmm...definitely a hallucination there.

[safe hazel]

it sounds the same as the opening of this song https://www.youtube.com/watch?v=4r3VLOwvvGs

ive had a generation start with a strange vocal hallucination like that as well, I'm concerned

[neat nexus]

I've linked that to the mod-chat so they're aware of the hallucination.

I would say it sounds a bit different than the one in the music video you linked as sounding the same but I'm not a developer, I'm a user of the AI same as you guys. The developers will know what to do from here.

[safe hazel]

i just hope these songs we make arent being built on actual copyrighted recordings. 🙏

[charred merlin]

I understand in principle and moral objection, but there's a difference between what you have a right to in regards to yourself, and what rights you can enforce on others.

Example, a vegan has the right to not ingest meat. Because of freedom of expression, a vegan has the right to inquire whether a person is serving them vegan-friendly food. There's an implicit agreement of goodwill, but the vegan doesn't have the right or the ability to force the cook to reveal the ingredients -- if the vegan has any doubt over the content of the food, they have the right to not eat it.

As a musical artist myself, I'll concede that I'm very curious about how this training data is supplied, but being as they're a closed source company, they're under no obligation to list specifics.

I genuinely think this conversation would be better suited for #🗣️┃chill-chat, so I'll share my "reasonable doubt" stance on this whole matter there.

[hidden lotus]

is Suno bugged now?

It's not capable of singing my lyrics

and then sings random nonsense

[limber burrow]

My inquiry about what data Suno has used to train their AI is highly relevant to the technical discussion. To keep this relevant your replies should remain on-topic. When it comes to using someone else's intellectual property, the stakes are higher. It's not just about personal choice, it's about legal and ethical obligations. Using unlicensed music for commercial AI training without proper compensation to artists isn't comparable to a vegan inquiring about ingredients. It directly impacts artists' livelihoods. Closed-source or not, companies profiting from AI models trained on potentially unlicensed data should be transparent. If there's reasonable doubt about the legality of their data, customers have the right to inquire and decide if they want to support such practices. Which I am going to do in order to make an informed decision.

[charred merlin]

Since you're asking about potentially proprietary training data, they won't be liable to share any information, regardless of your degree of moral outrage.

Perhaps you could volunteer your legal services to the RIAA? (yes, this was intended to politely agree to disagree.)

I'm probably not going to respond to this conversation any further, it doesn't seem like the dialogue is in good faith, and I suppose my cheekiness is partly to blame for that.

[limber burrow]

Suggesting volunteering legal services to the RIAA is a bit of a diversion. Paying customers have a right to know if we're inadvertently infringing on copyrights, no matter if you agree or disagree with me.

[neat nexus]

Paying customers have a right to know if there is still cocaine in Coca-cola. Or Lithium in 7-up. I'm go to their fansites and demand access to the official secret recipes now...

You would be better off sending an email to [e-mail] than asking us in the Discord.

[safe hazel]

@limber burrow well there is one way you can relieve yourself of your concerns

[limber burrow]

I already ended the subscription until they can provide me more info.

[safe hazel]

Then you arent a paying customer so stop saying you are

[charred merlin]

'Technically Correct' is my favourite flavour of Correct.

[neat nexus]

https://www.youtube.com/watch?v=3gHqYddtmNU

Dunno why this comes to mind...

Misrepresentation makes me question good faith in other things.

But that's just me.

[limber burrow]

Whether I'm currently subscribed or not doesn't change the importance of this issue. It's about ensuring ethical use of AI and protecting the rights of artists. My concerns remain valid, and it's crucial for current and potential customers to have this information. Ah well, Sony, Universal, and Warner, well ... you know 🤣

[charred merlin]

"37! My girlfriend made 37 Bangers!"

"In a row?"

[neat nexus]

I really do need to get around to watching the third one eventually.

Sequel wasn't as good as I'd hoped.

[charred merlin]

But do we?

[safe hazel]

great, paying customers deserve everything in the world. But you're not a paying customer so you don't. Problem solved

[vale idol]

Warner and Universal don't care about ethicality, they're just mad, because it wasn't them who came up with this idea first. It's all about profit and monopoly situations, not about the artist's rights.

[frozen bane]

guys how do you extend a part where the lyrics get cut off in a generation?

[daring lava]

Hey kentuckyFRIENDchicken 👋 thanks for reaching out! What you’ll want to do is:

• Extend from the generation you wish to fix (open the three-dot pop up next to your generation to find this function)

• adjust the “continue from” timeframe (found in the custom mode module) to the time right before the part you are trying to fix

• adjust the lyrical field to begin with the lyrics/part you are trying to generate next

Let me know if I didn’t answer your question adequately or if you need further guidance with this!

[fossil reef]

WAV doesn't work?

[neat nexus]

It can take a few seconds for it to compile it as a WAV.

Did you try it again? If there was an error message, can you share that with us?

[fossil reef]

It keeps saying this

[neat nexus]

I just tessted it. Was successful at downloading the wav.

Do you have a paid account or a free account?

[charred merlin]

https://discord.com/channels/1069381916492562582/1252656600770940968 Please upvote the known issue -- this seems to be an emerging issue

[fossil reef]

Was never able to download. I'm paid

[neat nexus]

Odd. I'm going to link this to the thread then.

What browser do you use? For informational purposes.

[fossil reef]

Chrome

[charred merlin]

Whether you're more comfortable sharing here or private messaging one of the mods like 3daizy or hepcat, please provide any relevant details you can think of to help shine light on the issue -- roughly when did you first notice this? does it happen with only one song or all songs? does it fail 100% of the time or are there occasional successes? Does the issue occur both on chrome-based browsers & firefox? does it occur on mobile browser? what country are you located in? have you received any notifications or error messages prior to this? Are you using a vpn?

[neat nexus]

I see that 3Daizy sees all and has already linked this to the discussion.

[charred merlin]

Oh my god, I forgot to ask that above and beyond all other questions. Are you using a VPN? (my tech support hat is dustier than I thought it was)

[true night]

soo were do i go to ask someone to look into my rendering whole song i tried it 3 times earlier in the morning and 3 timer now, it dont fetch the track at all it just pops up an error message :\

[charred merlin]

Similar error message to the one shown just above?

[true night]

a sec i'll read

[fossil reef]

The day I subscribed to paid. Tried it that time on mobile (chrome), same message, then few other songs, same. Then opened Edge on PC, same message. It's been about a week as I'm paid. No I'm not using VPN, and I'm from Lithuania.

[true night]

maybe its a chrome issue, i'll log into the account on another browser i'll be back .

@charred merlin i can confirm its a chrome issue its has been rendered on firefox they was there all attempts as i logged in , if similar issue for the other guy maybe change browser

[fossil reef]

It shows that UI message about preparing WAV or something similar, and then error I showed before

[fossil reef]

Even if it's browser issue, we shouldn't be "forced" to use other browser. I'm Edge user, and chrome on phone. I just can't see myself just mixing browsers

[rustic lintel]

New day, same error

I managed to generate lyrics once last night and since then it's given me this "wait a few seconds"

[charred merlin]

Not a matter of the devs throwing their hands up and blaming the browser, but being able to isolate specific things about the issue helps locate the source rather than the symptoms ❤️

I totally understand if you're not down to test, but could you check with Firefox just to rule it out? Edge is built on Chromium, so it's more or less the same as using a refined Chrome

[fossil reef]

I don't even have it on my PC

[true night]

i honestly understand that but if its an non nefore discovered bug located in chromium based browsers i was just suggestng that temp it could be fixed with a non chromium browser to try gto get files. until its been sorted out by the dev team 🙂

[charred merlin]

https://ninite.com/firefox/ninite.exe What the heck, where did this streamlined firefox link come from? 😛 (I'm just being a smart-alec, I genuinely appreciate the detailed report ❤️ )

[true night]

@fossil reef i however hope it get sorted for you soon, aint nothing worse then stumbling into unknowed bugs that break the usage for you

[fossil reef]

I'm not frustrated or anything, I'm okay with mp3 , just saying feature doesn't work ❤️

[charred merlin]

I'm frustrated FOR you. I hate when I can't figure out a puzzle 🤬

[fossil reef]

Will Firefox on Android will do?

[charred merlin]

🫂, yes

[charred merlin]

You ROCK.

Okay, I recall you'd tried on firefox and chrome -- you're not on a VPN, right?

[rustic lintel]

I am not

[fossil reef]

Not working on mobile 🔥🦊

[true night]

like on firefox im still having issue that u cannot rightclick on the main create screen to open a new page

[rustic lintel]

I just tried on mobile and I got a different error "An error occurred generating lyrics. Please try a different lyrics prompt."

[fossil reef]

Your prompts broke AI 💀

[charred merlin]

Doing a bit of digging into the matter -- does this happen if you try to create a brand new song, or only when you continue a song / create an extension?

[rustic lintel]

Right now I'm trying on a new song

[charred merlin]

Okay, so you get the same error message (or a variation thereof) regardless if it's a new song or a continuation?

[rustic lintel]

To clarify, I've pasted some lyrics into the box and clicked "Generate Lyrics" to get it to format a song including those things. I don't know how I would do a continuation of an existing song I've made so lemme know

[charred merlin]

Sorry, I meant whether the lyrics were being added as part of an extension from your song/ someone else's song (using the extend button), or if it was a "fresh" generation. Pretty sure you're suggesting it was a "fresh" song, though

[rustic lintel]

Yeah I haven't tried the extend button yet

[mint heron]

People, I'm realy sorry, but still noone is ancered to my one simple question, why I can't continue the old songs. I just can't, extend editor just resets the value. Even if ancer that it is not pocible now will be allwrite, but to see noone reacting to it is weerd, I want to make old song better with new moddel, and I just can't.

[charred merlin]

(quickly, it can add onto the end of a song, or write a new part starting from a timestamp you choose, so if there's a part of a song you don't like, you can go before it and make something better 🙂 )

Okay, I'm tapped for the moment on different ideas to test to gather more information, but I'm going to suggest that the mods make this a #1252292428258480209 , since there's a couple other instances recently -- and it's good to get eyes on the issue until we can officially rule out what the cause is (regardless if it's a Suno-side issue, still helpful to know if it's impacting a growing number of users) #🪲┃bugs-web message #1154779864851812422 message

[mint heron]

And it on every browser.

[rustic lintel]

Extending a song with lyrics allowed me to generate (both Random and with a prompt) but I only typed about 300 characters. When I'm trying to generate lyrics from scratch on a new song I try and fill as much of the 3000 characters as I can, so maybe that has something to do with it?

[charred merlin]

I see your post here #🪲┃bugs-web message, are you having the same issue with all songs, or just the one provided?

[charred merlin]

YUP

@final hazel to share with devs, maxing out the lyrics box in v3.5 and hitting "generate lyrics" gives that instantly, while v3 works fine

v3 turns picture 1 into picture 2

[mint heron]

with all very old songs, created before v3 appeared, only exeption is when I extended them after v3, but before this bug appeared.

[charred merlin]

When opening this song https://suno.com/song/9ffbadf4-b022-4f3c-aeab-822e6ff94d5e, the loading wheel continues to spin as if it's still processing after being recently created, and the playback bar is behaving the same. I'm not able to extend from the songs, as even selecting 0:01 forces the selection back to 00:00.

I'm not aware of the current troubleshooting method for server-based issues, but I could suggest emailing [e-mail] with a list of affected songs.

[mint heron]

are all people here have that bug? I'm jus5t not a native speaker, and affected song are counts hundreds, old users can go to very first or second song they created, and tyr to do the same.

[charred merlin]

Oh my god that feels so satisfying to isolate -- you're my favourite person today (don't tell my wife)

[charred merlin]

I don't have songs from before v3, and my oldest v3 songs still work 🤷‍♂️ Actually have some v2 songs from when it was on Bing
Feb 11th, v2, able to load and extend from it just fine.

[mint heron]

I will search where it first starts, I'm on page 73 some not so old v2 songs work fine.

[mint heron]

I have songs from before 2024.

[charred merlin]

https://tenor.com/view/harry-potter-curious-very-curious-wizard-gif-21125109

[mint heron]

one other thing, not only originals not working, but also parts of songs on the same page.

[mint heron]

found it. The problem starts at song generated at November 29, 2023 at 5:22 am

[charred merlin]

Can you share that track? I want to see if it's displayed as "processing" the same way the other had

[mint heron]

I dont know, it has some bad words, and I guess this thing is not liked here.

[teal raft]

Only because the discord is PG-13

[charred merlin]

You can DM me the song if you'd like -- I'm not so concerned about the lyrics, if Suno allowed you to make it, it can't be that bad, I'm just more curious about the file itself. The previous track displays the time properly when shared as an embedded / shared link, but doesn't process properly when viewed directly on the site. I'm curious if this is present on other songs you're experiencing this issue with

[rustic lintel]

Glad to be of help 😅

[true night]

its back now on firefox the error occured tag on rendering the song.. so its not browser based - idk why though its quite boring waiting almost 20 min for full song just to have it give me an error :\

[charred merlin]

"Get Whole Song" took ~20 minutes and ended in an error? Normally it starts loading within a second or two, and takes about as long as a typical song generation to finish processing

[true night]

well idk what to tell you it happened again now it sshows in grey'd out for a long while it came an error mesaage now after 16 mins , but still same issue i will try to relog in to chrome to see if it has come in there now, but its an odd 1

nope did not work this time tried relogging on both chrome and firefox

it dont make full song ive will try on the 2nd part gen to see if it gives diffrent outcome i'll be back when it give either sucess or failure

[charred merlin]

If you can, try to get a screenshot of the error message, or try to recall the explicit wording from it -- that could help narrow things down.

Cant recall if I asked, you're not using a VPN, are you?

[true night]

hard to get a screenshot of it as it goes away right after getting it , butr i can try to be quick
never used a vpn so the answer is no
i've on firefox atm tried on bouth chrome and now firefox
as we speak my newest try still states rendering but it has not gone many minutes yet.

standard error message to be fair.

[charred merlin]

The only instances I've seen of that particular error message were resolved after either clearing the browser cache & history, or from waiting upwards of a couple hours.

Did the issue start today?

[true night]

started this morning yes

i guess i can try that but i rarly seen anything get sorted by that (no offence) generic idea

[charred merlin]

Basics are basics for a reason

[true night]

trying again now after the cleans

i'll update

[charred merlin]

Browser needed to be updated?

[true night]

dont think so checking

nope both updated

[charred merlin]

OOOH, I misunderstood "I'll update" 😛

[true night]

all good

[charred merlin]

Also, do you have the same issue trying "get whole song" for a previous creation?

[true night]

i can check

[ebon ingot]

Maybe it's just me, but a bunch of my generated songs, just disappeared. I checked in my Library and in my trash (in case I accidently deleted them, nada, nothi ng can be found. Anyone else has experienced something similar ?

[hollow hare]

I can't find my song in my library! What do I do?

First, reload the page and ensure you are in the library. For cache-bypass reload, see #📚┃suno-school message

For usual causes:

• Could be on the next page (blue circle for pagination)
• A filter is enabled (yellow circle)
• Accidentally trashed (red arrow to access the trash bin)
• You are in the create screen, not library, and the piece was down voted (which causes the entry to be hidden)
• The text filter option can be finicky. Try to use as much of the full title as possible or unique keywords if the first attempt failed.
• Accidentally overlooked the entry (with text filter disabled). May need to reload the page first.
• Logged in to another account (can double check by going to your profile)

https://media.discordapp.net/attachments/1102676217687195679/1246851079182880808/image0.jpg

[true night]

other previos songs worked

[ebon ingot]

@hollow hare thanks for the quick reply, and yes I tried all this. and I used the filter to find the song(s) but all I got was my "down votes" of which I can't find the song in question.

[hollow hare]

Hm, yeah, sometimes I find that the text filter can be rather... finicky. I remember not finding a version of my song when I tried "abyss" but it appeared when I had "abyss 4" (but that was when the filter first rolled out)

[charred merlin]

Okay sweet, so we've presumably isolated the issue to that particular song -- if basic housekeeping didn't produce any success, I'd say the next step is making sure that at least 24 hours have passed since the song has been created, and if it's still giving you the same errors, I'd email the song link and the troubleshooting steps we tried today to [e-mail] so they can look into the backend stuff.

Alternatively, maybe you could try making a short extension on top of the song you're trying to generate to see if you can "get whole song" from that, then trim the end in a DAW?

[hollow hare]

Is there any chance you still have the URL for the song you liked, like through your browser history?

What you can do is go to that page and add the song to a (new?) playlist.

Hm not sure what else I can suggest other than going through the library manually (not text filtered) after clearing your cache and logging in again.

If you did upvote the piece, you can try the Liked filter, but better to assume it got unliked or disliked by accident

[true night]

to be frank ive not finished with the song but i noticed if you linger to much pr part the quailty seem to brake apart so i try to do a full song more often then not, oh if it only been finished.. also on a side note ive think i found a stuck loophole, ive took the part that was stuck unable to make full song i daws clipped it to 58 sec and uploaded it again and now 4 parts in its identcal it never change a beat its 100% the same melody over and over with no diffrence its quite odd.

ive tried going over to v3 prompt end and now got an end (just bec.) and i try to get full song lets hope
got an ending now rendering full song (i hope)

[ebon ingot]

@hollow hare I found it, like you said: the filter is still finicky. I had to put in the full name of the song, when I do just the partial, I am getting literally just the songs that include the partial (like Space City !== Space City July 4th)

[true night]

but i'll will wait 24 hours before trying to make full song on that part again

[charred merlin]

Ooh man, I just got the same error message attempting to get whole song, but I restarted my PC and all is fine. Panicked for a second 😛

[true night]

unfortunaly still error, even after i extended. i'll will wait 24 hours then or more.

[true night]

i have had my pc off since this morning , so it did not help me , unfortunaly

[charred merlin]

24 hours is probably overkill, but it helps write off a number of potential explanations to note that you've waited that long. Sorry I wasn't able to be of more immediate assistance, but if you're still having the issue after the 24 hour mark from when it first occurred, sharing the song link and troubleshooting steps to [e-mail] puts you ahead of the vast majority of support requests 😛

[true night]

well it 1st occured 5 am this morning,
but since it came in after 12 hours + it might be an other error upon the error) as they was showned in another browser later the evening, but i'll see it magically appreed tomorrow .

[true night]

also its fine at least you come with suggestions , suno is still young, its bound to happen issuess, lets hope it will sort it out if not that the dev team figure something out.

regardless outcome (hoping for things that works out ofc)

[charred merlin]

If 1/10th of the people who run into issues in life were willing to collaborate similarly, we'd have flying cars by now.

Thanks for being awesome

[true night]

yes them cars i was told i be driving when i was 10 , danmit give me my flying car 😄

[true night]

no thank you for trying to help with issues , even just as a mere user , we need more people like you ❤️ (i try to help whern i can with my oh so limited knowlage)

[tender leaf]

Why this error?

Work of art?

[charred merlin]

The audio clip is being flagged as matching an existing copyrighted song / someone else's work of art (song, audio clip, recording, etc).

[tender leaf]

Oh ok... Thanks

[urban wing]

at what time and time zone does the credit top up?

each day

[flint haven]

Free Credits Refresh

Refreshes at 0300 UTC+0 (<t:1713236400:t> in your local time)

[urban wing]

thanks harber!

[indigo basalt]

Does anyone know how to create a song that's like a country bar song? I'm trying to aim for something that is kinda like the song "A Bar Song (Tipsy)" by Shaboozey. Any and all help is appreciated! 🙂

[hollow hare]

Check the subgenres listed for the music on music analysis sites like RYM:
https://rateyourmusic.com/release/single/shaboozey/a-bar-song-tipsy/

I tried testing something with a fusion of Country and Tavern, out of curiosity, and wonder if that was also something like what you wanted? This was before I checked RYM:

https://suno.com/song/6809d6b6-3ece-4ab9-87af-56510c13df35

The alt track: https://suno.com/song/fce311cd-c7b7-4929-bc12-dbc17220c131

[indigo basalt]

@hollow hare It’s close, but it needs to be like their singing it in a bar, not in a recording studio or having any sort of like professional equipment. Hopefully this information will help you. 🙂

If you need a reference of the type of thing I’m looking for, I put a song title and artist up in my previous comment that might help.

[charred merlin]

Try taking what Paimon made and experimenting with adding or removing different style prompts, maybe use some metatags to try to suggest to Suno how to interpret the vocals

[live-vocals] or [live-performance] for example.

Hopefully the example Paimon gave gives you an idea of what kinds of sounds you can get using those specific tags and prompts, but it'll be much easier for you to fine-tune the results based on what you're hoping to get

[hollow hare]

(and I do not have any more credits left for the day too)

[indigo basalt]

Thank you for your assistance! I greatly appreciate it! I will try to do my best to figure it out with the help/information you two have provided me! Thank you again for the great help, and if I have any other questions, I will come back here! 🙂

[graceful condor]

Hiya i need information on copyright

I'm about to distribute some of the music made with suno ai, how should I go about this?

Do I need to send a email to suno ai support?

[charred merlin]

#📚┃suno-school message

[graceful condor]

So is it impossible for me to get copyright for the free song I've made before my subscription?

I have a subscription now

[charred merlin]

You don't own that song, Suno does.

[graceful condor]

Is it possible to buy the copyright for said song

[charred merlin]

There hasn't been any direct comments made about that, but if you contact [e-mail] with the link for the song, they might be able to give you more information.

[graceful condor]

Oh ok awesome

So for the P line and c line in my distributor I can just put my artist name for the songs ima upload that was generated with my subscription?

Or do I put suno ai in both lines?

[graceful condor]

This is my last question

I want to make sure that this is right

[charred merlin]

You can't copyright it, because you don't own it.

[graceful condor]

So what do I do then?

I'm trying to put it on spotify through a distributor

[charred merlin]

You can only do that for songs that you own.

You need to either ask [e-mail] if you can somehow acquire the rights to that song from them, or make a new song (which you'll own the rights to)

[graceful condor]

Ok

[charred merlin]

#😜┃suno-memes message

"Never make Bangers on a free account" 😛

I'm just giving my free advice, I can't stop you from trying to file it otherwise, but beyond breaking Suno Terms of Service (https://www,suno.com/terms), I'm not sure if there's any legal consequences for falsely filing a copyright.

[safe hazel]

It would be kind of ironic for Suno to sue for copyright infringement though 😅

[soft bronze]

I’ve wasted like 200 credits and Suno doesn’t stop giving me results that sound very very AI where the voice is so flat and almost rapping even the style says it’s a slow song. I swear it only happens after the upload musing for free users update. Previously I’m getting a really good result on the first 2 generations. Do you guys experience this as well?

[hollow hare]

What style prompts are you using? Genre has a huge influence over the creations.

Try using "clear vocals" or an emotive direction like "happy", "angry rap", "concerned quiet voice", etc.

For the tempo, use "up-tempo". Maybe "fast" might work? Or one of the Italian phrases like presto (see #📚┃suno-school message )

[soft bronze]

I’m using this style “Male singer, Piano, Slow Singersongwriter”

It was working but lately all of the songs I generate, despite of the song style, it’s giving me a too AI singer.

[graceful condor]

I meant for the subscription ones, it doesn't answer my question

[hollow hare]

Hm maybe singersongwriter is causing that? It is an unusual prompt and, if the AI associated that with a particular type, it would likely be assigning that.

What I suggest is you should list an explicit subgenre so the AI has an increased pool of voices to select from (see https://www.suno.wiki/faq/style-and-lyrics/styles-and-genres/, Wikipedia or https://rateyourmusic.com/genres/ if you need a list of genres). I would also suggest adding an emotional keyword (e.g. happy, Melancholy, etc.)

[mighty anvil]

anyone have a good free image enhancer and upscaler for the album artwork that suno generates? I was using Canva but they made theirs no longer free

[soft bronze]

This one really helps. I’m getting better results. Thanks!

[hollow hare]

You're welcome 🙂 I also wrote a small something regarding the weights of the prompts here: #💬┃general-chat message

Mind you, these are based on my personal observations but others have tested out (e.g. #💬┃general-chat message ) different prompts and found that the words at the start are prompted much more heavily than those at the end, so keep that in mind.

[graceful condor]

Hello?

[hollow hare]

This is not our jurisdiction, you will have to look into it yourself based on the laws in your region or those of the distributor, maybe even both, as whatever was defined in the terms of service of that service.

According to the US copyright office (henceforth referred to as US CO) currently, portion of creations made through non-human processes without sufficient human contribution are not eligible for copyright.

This means you cannot list/assign any copyright status attributed to Suno Inc.

If you take the song as-is that was generated from Suno AI, you can only claim the lyrics as your own copyright. The generated output is not eligible for copyright under the current US CO terms.

If you made significant additional changes, such as rearrangement and/or mastering, the original underlying music made by Suno cannot be copyrighted but your arrangement and mastering process can be due to the human element involved.

I highlighted some interesting sections at #🗣️┃chill-chat message in regards to the policy statement from the US CO in March 2023 (linked in the post provided)

It's a whole thing... take a look at the link I provided earlier and this: https://www.copyright.gov/ai/
(Or your governmental copyright office if you do not live in USA).

Mind you, I am not a lawyer and I could easily have made a mistake in this post. Do your own research.

[fleet root]

i request genre dangdut, keroncong, dangdut koplo, nasheed, gamelan, angklung ( indonesia music )

[hollow hare]

See the message below

[flint haven]

How to use Suno

Hi there!
If you’re ready to dive into creating music, head over to Suno.com.

For more detailed information, check out docs-guides-resources.

If you have any more questions, just ask in #general-suno where you can always find our community, moderators and guides to help you with most questions. 😊

[oak wadi]

hello, can you guys please support Arabic text when downloading the video, the lyrics appear as unreadable symbles as it does not recognize the Arabic characters..

[errant moss]

Hello, somebody can explain me why my song was removed after i renaming it in to random character's name location (like from the game)?

[lucid gull]

Hello every one, for the past few days I can’t use suno.wiki, it is not loading. Does any body have the same issue?

[honest beacon]

so I was wondering, if there are any plans to increase the character limit in he custom Lyrics section past 3000, as I want to try my hand at creating an epic song, sort of in the vein of Bohemian Rhapsody.

[analog shore]

dangdut style can be made in V3, try to search dangdut music in Suno dan reuse the prompt.

[round sleet]

Having the problem, that my instrumental prompt is only generating mostly short around 45 - 90 second songs in V 3.5. any ideas except of extending?

This is the prompt:
60s psychedelic soul, delayed guitar, extended soulful guitar, reverb, wahwah, nostalgic 60s and 70s soundscapes

[final hazel]

You don't need to do that. Just extend the song. First part will not be longer than 4 minutes anyway.

[spare stump]

@final hazel I finally upgraded to premium.. was wondering though how do I remove an image after adding 1 to the song in "details"

I have 2 images or more and just want 1

[sullen wedge]

Hi, Is it possible to create a music with vocals and a second identical one without vocals?

[final hazel]

You can only add a new cover atm.

[spare stump]

okay that makes sense... still an improvement ty

[coral plover]

hey tech experts, anyone able to explain this beauty? check out the prompt haha

https://suno.com/song/54023771-22bb-4771-9bea-902ee99d1c2e

[naive quail]

🤷 Prompt Noise is unpredictable. Suno WILL process it... and often ignore it, but will also sometimes do some really cool stuff with it. Sometimes it just gets weird.

[glacial lion]

You could upscale in comfy ui

[glacial lion]

Some DAW software has the ability to strip the vocals I hear

[glacial lion]

You can always use extend. Each time you get another 3000 characters

[hollow hare]

You will want to mention that at #💡┃suggestions-web

[languid mauve]

Who are you using for distro?

[shut locust]

Is there a way to remake a song that sounds good but has some errors, for example in the lyrics, using parameters like seed or re-evoking/repeating a theme that we like?

[daring lava]

Not sure I understand what you mean, would you mind explaining it again?

[astral flame]

i really appretciat this and these options. ill give them ago. thank you !

[hallow moon]

bark is generating really slow and with 3060 ti i would like to force more, its using only 7% of it when generating.. CPU same, very low usage, this results in generating audios really slowly i think?

[hallow moon]

Also are voices coherent on different generations?

[shut locust]

@daring lava Yes, thank you for responding. I don't know if there is a way to recreate variations of a theme that sounds good using some parameter. In other artificial intelligences, we have the possibility of using parameters like "seed" to create variations of a result we like. Could we remake a song that sounds good but has some error (in pronunciation, for example)? I know it is possible to extend a song to correct errors. But it would be great to remake a song using the same voices, the same melody, and composition from the beginning.

[daring lava]

@shut locust I agree, having the capability to use a seed would be awesome. Unfortunately, at this point that isn’t an option yet. Not to say it won’t be down the road, but so far not a thing. But again, completely agree that would be very awesome and would help tremendously with capturing the consistencies you described

[shut locust]

Thanks David. Hope this could be implemented to make variations of a song and make it perfect

[sudden spear]

Does anyone know of a way to "modulate/tremolo" voice? At the end of some lyric lines, I'd like it to modulate the voice a bit. I tried [modulate up] and [tremolo] lots of times without success.

[modern axle]

write in a way that suggests it would be modulated -- like a diamond wah-ahAHahter faaaAaaaAaAlllllllll

[austere dragon]

anyone else kind of disappointed with the upload feature? Like I work on a 15-30 second piece, upload it, and other than the key signature and beat, the upload is kind of ignored. Not in all cases. I've made some really good tracks, but it craps out more often than not.

[visual nimbus]

A question arose while dealing with bark.
I'm trying to make a Korean TTS that lasts longer than 15 seconds, but I don't know how. Even if you change the Clip Length Chance value of Splitting Up Long Text, Generation (Sampling), it still looks unnatural. If you know how, can you share it?
i used this rep
https://github.com/JonathanFly/bark

[shy spear]

I'm got a waltz type song I want to end with a big finale/finish but I keep getting silent results, how do i prompt the ai for a finale ending?

[calm bobcat]

It's ridiculous how the voice appears in instrumental songs, even when specifying it in the prompt, nothing changes.

[pallid leaf]

use v3 engine for full instrumental tracks

[hallow moon]

hi please, im not sure why the model is very slow at generating speech, it needs like 5 min for 2 speechs generations.... (i'm on 3060 ti with 8gb ram and 16gb ram on pc, SSD and ryzen 5 2600x cpu.
Im using 11.7 cuda, pytorch and flash attn 1.0.0 (had issues downloading the 2+)

[carmine warren]

so, i extended a song from 2.08, but instead of doing what it normally does, it changed the song from the start. it had a similar style but wholly different structure. i did change lyrics, is that the reason? i thought it always kept the original up to the extend point regardless

hmm, nevermind, i'm being stupid. (and typoful)

[charred merlin]

Turn that off if you have it on as it'll use slower system memory if you run out of video memory, and make sure that you're not using your video card for anything else. Haven't looked into Bark, unfortunately this Discord seems more Suno(ex-chirp) oriented. That being said, you have ~8gb video memory, but it's only reserving half, so that might be a setting to look into, perhaps there a launch argument you could use

[languid mauve]

Siri says it feels like 102 outside.... thank God for Suno and A/C

[languid mauve]

I got a song with lyrics and it made it instrumental. 😆

[calm bobcat]

The system should refund credits when this happens.

[languid mauve]

Honestly I didn't mind. Was beautiful to listen to

[faint portal]

Hi everyone, I wanted to know if it was possible to keep the same voice for several generations or does it remain random?If we can keep the same voice, how do we do it please 🙏? 😅

[soft bronze]

Might be a dumb question but how useful “Upload Audio” is?

[teal raft]

What do you mean?

I find it almost critical lately

[soft bronze]

The new feature. Haven’t used it yet but wanted to know how it would spice my music up

[teal raft]

It allows you to directly control the 'vibe' of your song

I've taken to slicing out bits from Suno tracks and using those to start others with unrelated genres

[soft bronze]

So if I upload an instrumental, there’s a possibility it can be used as the song’s music backgorund?

Tried it and it only uploaded the audio clip. I have no idea what to do about it

So basically this can only be used at the start of the song and Suno will be able to compose a song out of it.

[naive quail]

In my experience, it's like extending an existing Suno render. If the upload ends abruptly, Suno will usually try to smoothly continue it.

If there's silence at the end of the upload... it's a bit of a crapshoot; sometimes Suno will just wander off and do her own thing, or she might try to continue it, or (more frequently in my case), the upload will be used as an influence and Suno may incorporate bits and pieces into the extension (either directly, or just musically)

[true night]

@charred merlin Soo i waited 24 hours now trying to render it .. and lets see if it goes tru or fail.. it failed unfortunaly.. so yeah that unfortunal 😦

[true night]

24 hours later same error so what do i need to do now to try to get it sorted i even tried to gen whole song a few parts before this and same result happened

getting this now aswell :\

[lilac garden]

hello was wandering what you all have found succesful when trying to have a slow down chorus towars the end with less instrumentals. i tried [Break][Slow Down] [Diminuendo][A Capella] but doesnt' take it

[lilac garden]

Ok Ill try that!

[static moss]

Don't want to get banned/muted or anything, but how does suno output quality compare to udio? I see pros and cons for both, but do Any of you have solid opinion on that?

[astral granite]

Hey, I just joined and I'm mostly new to discord as well. How do you generate songs on here? Sorry for the inexperience

[teal raft]

/help

[flint haven]

How to use Suno

Hi there!
If you’re ready to dive into creating music, head over to Suno.com.

For more detailed information, check out docs-guides-resources.

If you have any more questions, just ask in #general-suno where you can always find our community, moderators and guides to help you with most questions. 😊

[formal python]

How do I get a neural network to follow the voice from a song? Continuing the songs, the neural network comes up with other voices that have never been in the song. Help please

[untold briar]

This code is running generations in parallel, is that on purpose? Generally that's a way to speed of inference if you have extra GPU memory, by using more GPU resources.

[untold briar]

You can DM me questions about Bark or my repo, Suno Discord doesn't really have much Bark on it now.

[hallow moon]

ye an objective is to run them in parallel, do you know how to maintaing coherence between voices? and like not making them music voices sometimes or noisy ones..?

[untold briar]

But to your question, clip length chance is not what you want. Try leaving all the splitting on default, it should auto split at least reasonable. But maybe there's a bug with Korean text.

[hallow moon]

Also do you think with 8gb 3060 ti i can generate tts in like 1-2 secs max for long phrases?

[untold briar]

I know all about both, but it's a big question. For a voice you can set the history_prompt to any voice (and you can use any Bark output with a random voice, as a new voice in the future)

[untold briar]

With small Bark coarse model, and just one sentence... maybe 3 or 4 seconds. But unfortunately for best quality Bark is better with more context (3 or 4 sentences)

I haven't checked on Bark recently, but there could be new optimizations in the bark.cpp or huggingface versions

[hallow moon]

because coqui had decent voices and languages, but didnt have mit license as suno, and suno have better quality in my opinion

[untold briar]

I haven't done it but I'll probably setup a temp Bark Discord so I can refer people to a place where they may find the question already asked and answered, since on the Discord the Bark channels are no longer around

[hallow moon]

so basically im searching a tts that can fast generate long phrases in 1-2 secs and have emotional voice / mit license / languages support

[untold briar]

Bark can do really good quality 100% (with some work) but fast is still an open question.

[hallow moon]

yeah thats the issue for me.. i need speed :C

[untold briar]

Well, actually, if you generating in parallel (batches) does that mean you only care about throughput and not latency?

Because Bark can be faster in batch generation (generating multiple samples at once)

But honestly we should probably discuss in off-topic or DM me, this channel is technically for technical Suno music discussion