#šāsuno-school
1 messages Ā· Page 42 of 1
Thanks a lot š
Its not about ending the song its about controlling the length of the song.
I dont want 4 minutes just 60 or 30 seconds of instrumental
There isn't a way to specify the length up-front unfortunately.
It's often requested and would be useful to be able to do this.
(I documented it as a known suggestion recently at https://discord.com/channels/1069381916492562582/1252783540131860503 )
Hey ive just updated all the thumbnails to some tracks in a playlist, how long before the updated images will be reflected in the video download?
Or will the video thumbnails not update?
The videos won't change. They're created on the first pass.
How on earth do you get 3.5 to actually stop a song when doing custom lyrics?
I keep getting 4 minute songs where ti does the lyrics really well, the song is good (for my use case) and then it runs out of lyrics and decides to just wing it and redo some till it hits the 4 minute limit.... its driving me nuts.
Example: https://suno.com/song/c8e60f66-2f8a-4b3e-92ce-be92bc39d2a9
The song ends at 2 minutes 24... It is perfect for what i need it for.... but then suno goes ahead and adds more (full disclosure when i cut it i used the next verse as it worked well, but thats not the point)
female chanting, (chorus) sung by monks song. Listen and make your own with Suno.
its doing it more today aswell.
tried;
[Instrumental Outro]
[end of song]
but suno is like "nah mate I am carrying on"
Switch to v3 and have nothing but [End] in the lyrics field. Also wipe everything from the style field, and possibly try adding End there. I also change the title to End.
That almost always gets me a quick ending. When it's being stubborn, it can help to move the start time forward or back by a few seconds.
90% of the time i can get v3 to end. But the issue is getting v3.5 to end the ability to do a whole song in one go(usually) is great and I like the way 3.5 does songs... its just that endeding.... we need the ability to trim songs in the ui so much!
Yea, we definitely need a fix. For now we're stuck with hacks.
I usually add the [Short] tag to the prompts of the extension, and i have gotten them under 1 minute.
how can i extend a song after 4 minutes? at v3.5 i cant do that....
the ai just cant take the hint:
(that still didnt work)
click the 3 dots and choose extend.... you have to wait for it to finish processing though
in the three dots there is no extend,,,
ah they moved it, the extend button on the ui
and when i click on extend...the button to use it is not useable for me
yes i know but i cant use it
its working for me. could try clearing browser cache?
thats how it looks at me
thats in firefox... could be a bug with other browsers
yeah... thats working
no its not
hello, everone, I'm amazed by Suno's singing ability. It's truly remarkable!
Can I use Suno to sing along with a MIDI and accompaniment file that I provide?
oh i see the other extend button isnt showing.... odd
oh! @livid estuary your out of credits
can i only extend with credits?? thats lame....i have no credit card...and they dont even use paypal (would be better and easier)
your credits are used 10 per generation.... be thankful you can do ito on 3.5 back when i was a free user it was 2 min max then 1.20mins per extension
yes i know but paypal would be great.....my paypal is ready for a monthly use...i dont want to have a creditcard only for suno...
shrug cant help there chap
I mean you can use your debit card, doesnt have to be a credit card
is it the same as a normal bankcard?
indeed.
to be clear though, if you are young (the fact your unsure what a debit card is makes me think you may be, no offense ment) be sure to clear it with parents first chap. I am in no way encouraging a minor to add a monthly charge to thier account!
3.5 not stopping songs is really annoying me. I would save my tokens till its fixed if I could
I have sent a referral code to my friend but I didn't get any credits
This isn't possible (Suno does not process MIDI files.) It's possible to upload audio and do this kind of thing though, so if you can record the MIDI performance as an audio file you could use that.
still no go
Did your friend make 10 songs?
from my other test [stop] made it say stop lol.... i mean it stoped the music, said stop, then carried on... an ok effect but not the result i was after
he needs to makes to make 10 songs?
okay I just saw it
noticed
@unique stone you got any tips on making 3.5 end a song when the custom lyrics are done?
thats not like you thing....i even hear the name "debitcard" for the first time...germany is sometimes years behind the rest world š
i use the prompt (Outro) or (Instrumental Outro) at the end of a song
on your bank card you should have a mastercard or visa symbol (for the love of god DO NOT SHOW IMAGE OF IT) if it has either it can be used online (unless its different in germany)
Thank you for your response. I can convert MIDI files to audio files, but I couldn't find a place to upload audio files on Suno. Could you provide a tutorial?
Your friend must create 10 songs before you get credits.
Extending from the point where you want it to end and selecting "v3" is the best way I am aware of. For my own stuff I generally use Suno to generate ideas and samples that I'm going to slice up externally anyway so I haven't got any better suggestions than this. In general stuffing [end] in the lyrics and "end" in the style is the kind of thing I'd try but even with v3.5 it will want to keep going for a while first.
This button at the top right of the Library page, or at the top left of the Create page, is where you can upload audio. (This is currently only available to subscribers though.)
Sorry, had to laugh. Tried everything. Does not work. End song, fade to end, fade out, End. Everything but the kitchen sink. š It stops at the end and if there is still time left AI reprises any lyrics, verses or chorus until time runs out.
end in the style? interesting.
hadnt though of extending it in v3 that may be the key! would use tokens but that could make an unshareable song...sharable.
man I wish we could cut songs in the ui lol
yeah its like the ai is enjoying itself and doesnt want to stop lol
Editing within Suno would be great. Similarly annoying thing (when you want a song to end) is when it does end and then decides to do an encore.
And when it doesn't, it cuts short exentensions at 20 seconds. I had also 9 and so on. Not funny anymore. ^^
yeah had that a few times. I think they may have trained it on only 4 minute long clips so the ai believes THAT is the length all songs should be... and tries hard to reach it
had that with v3.... used several hundred tokens once trying to end a song i was doing to cheer up a mate, the beging was so perfect i couldnt give up
ah thank you..i will look later...maybe it can be a bit different in germany yes....
yep extending with v3 and having just
`[Instrumental Outro]
[end]`
worked, thank you
I doubt it's because of the training data - v3.5 early access had a tendency to end instrumentals within a few seconds for example, and the training data should be roughly the same now.
A common request before v3.5 existed was to be able to generate full (or at least longer) songs straight away. I suspect the final release of v3.5 ended up being tuned to stretch the song length to give maximum value for the 4 minutes you get in the first part.
That does unfortunately seem to have the side effect of repeating lyrics that were sung earlier on - as far as I can tell what it seems to be doing is "remembering" that it had made some vocal noises earlier in the song and calculated that those same sounds would fit well in what it's doing at a later part (after your lyrics have ended.) So it ends up unintentionally repeating vocals because at that point it sees them as just "musical sounds" (you'll notice the lyrics display in the video won't be following them.)
Hopefully they will fix it down the line
untill then the work around doubles the cost of generating but .... works
It'd be nice to be able to influence the length, even if it is just simple control short/medium/long (because I doubt we'll ever get the ability to say "it must be 50 seconds long exactly".) I'd taken some existing suggestions about length control and put that in to the known suggestions recently so hopefully it becomes a thing.
tbh I would much prefer the ability to trim the song after generation. it would be hard to predict the length before gen with some songs. even in the same genre, tempos can vary drastically
Oh yeah that's also a common suggestion I put in the known suggestions area (basic editing functionality) š
using the extend via v3 method the example song I shared:
https://suno.com/song/07e57f35-0cca-49ec-a0ff-3e6ef3320771
female chanting, (chorus) sung by monks song. Listen and make your own with Suno.
Ideally there would be a sound graph like in Audacity and you can highlight a region and delete it, or regen it
that worked pretty well, I would likly redo the end a bit more if I hadnt already edited it outside the app for my purposes.. but thats pretty good
Could you confirm if the uploaded audio is an instrumental track for a song? If so, how can I provide the lyrics to Suno? Thank you.
inpainting woul be cool too.... regenerate this part - EDIT thats what you meant.... sorry i misread... so yeh +1 to that lol
The uploaded audio doesn't have to be instrumental. It just influences what Suno does with it (you have to "extend" the uploaded audio.) So if you upload an insturmental, you can extend it and put lyrics in the "lyrics" box. If you upload a vocal part, you can describe the style and instruments to add. You can even upload sounds (I tried a police siren) and it will do creative things with it.
When you upload something, it acts as a seed, Suno will do guess work on melody and so on, then it will try to incorporate your upload file into a genre template
I'm unsure if Suno can transcribe lyrics from your audio upload, and also not sure if your input lyrics would clash with any lyric in the uploaded file
I've been doing lots of testing with upload audio and it can sometimes continue vocals and lines even without lyric input, if you have the instrumental button turned off but dont enter any lyrics. If you enter lyrics that match the vocals and lyrics already in the song it can repeat them easily and it can also finish them if you enter lyrics that aren't in the uploaded track. It can sometimes match the vocals really well or do variations on them or do new ones. It's truly amazing but there are lots of little things that will be resolved in time.
I keep getting certain audio uploaded that says it cannot be made public because of content restrictions, but its entirely original audio so not sure why, I might try clipping it differently just to see if its a bug.
So a warning if you make any extensions or generations on an uploaded audio that does not allow you to make it public, none of the extended versions can be made public either. Hopefully in the future they can have a review process or something to double-check it so it can be published.
I tested Udio just to see if it would do similar things to my tracks that Suno can do, and at least on the ones I tried it was a massive failure and sounded terrible, compressed, like they are training it on a highly compressed version of the uploaded audio, and just wasn't nearly as creative or good in any measure. Maybe its just my tracks but yeah I don't even feel like testing it more there because it was so disappointing.
Hopefully Suno will give us some of those features soon like extending it in both directions, and other controls.
the only thing that it entails is that your track will not be entered in the trending list so Suno does not expose a song that may have lawfare liabilities for you and them, you're still free to upload it anywhere and face the full repercussions of any aggressive measures taken against you š good luck
If Suno detects what it thinks are vocals in the uploaded audio it will restrict it (and anything extended from it) from being made public. There are false positives with this sometimes.
I realized after seeing that message again they changed it to "private only" but originally I thought it said "will automatically be made private" but you could make it public later.
Is this something that they plan on changing in the future or is that they just don't want any human-vocals being included in the public songs database?
I assume the intention is to avoid famous voices from being used on songs that could hit the Suno front page (uploading e.g. a famous singer's voice in itself would be breaking the Terms of Service but people will still try...)
Of course we can still publish it ourselves tho I'm sure.
Why is a song about sexuality flagged? I don't see anything in the ToS that says it's not allowed.
Yeah nothing to stop you doing that. Just prevents it from being able to trend on the Suno home page. You can also still give the link to people and put it in playlists
Suno.com is eating my credits.
Insists on not obeying the prompt.
How can I make it so that reusing the prompt reduces the BPM?
How do I make the voice "Pt-Br." and not "Pt-Pt."?
I initially used:
Afroswing beats, slow syncopated rhythms, Nigerian influence, Dolby Atmos, Original Master Recording, worship, melodic.
After:
slow Afroswing beats, syncopated rhythms, Nigerian influence, Dolby Atmos, Original Master Recording, worship, melodic
[reset memory]
Finally:
new age, jazz bossa, syncopated rhythms, Nigerian influence, Dolby Atmos, Original Master Recording, worship, melodic
[reset memory]
It's nothing. Continues a high BPM and the voice in Pt-Pt.
In conclusion: I need a little help from my friends.
Oh thats right the links still allow you to share it and doesn't need to be public for them to listen. Also how easily can people riff off something and is that the intention of things being public anyway?
I'd put "slow" somewhere in the prompt on its own (separated from other parts of the prompt by comma)
Syncopated Rhythm as a prompt has always generated Dance Tracks with my tries. Maybe you prompt slow extra.
[reset memory]
It seems like you're operating under the assumption that the metatags & style prompts act like explicit instructions.
They don't. They're suggestions, and the more suggestions you give it at once, the less likely it is to follow everything exactly.
Public just makes it so that your song can appear on the home page in one of the playlists there, or as part of a "radio station" when people look for similar songs. Other people can extend from it even with the link, that may not be desirable and it'd be nice if this was made optional in future.
Also shows up in search too
It's like a lot of the AI its a slot machine you roll and can have a lot of fluff and nonsense but also some truly amazing moments, but its not really designed right now to just easily make a full song that sounds perfect from start to finish, but its possible sometimes.
Public/private only effects whether it's visible & searchable on the "social media" side of Suno.
Yeah I think I'm straddling half and half with music publishing, art publishing, etc... part of me wishes the whole world would just stop with having all of creativity needing to be connected with making money and owning ideas and creative works and all that... but another part of me understands the feelings and how things can happen and the professional uses and copyright and all that stuff too... its just we're in a mixed world.
Open sourcing everything is great as long as some one person can't just game the system and then bring in the earnings off something someone else made, but its also just the nature of the system, sharers gonna share, hustlers gonna hustle
If your lyrics are being flagged, it won't be just due to the subject being covered but more to do with what's "happening" in the song. Paste the lyrics into ChatGPT and ask it to identify what could be problematic with the lyrics.
I'll have to politely disagree-- it's absolutely designed to make a full song that sounds perfect, and I've made a number of songs that I'm incredibly happy with.
I know it's cliche to use the old adage of "the carpenter blaming their tools", but this is the equivalent of a 3d printer-- with enough knowhow, time, and effort, you could figuratively 3d print a working gun. But if you get access to a 3d printer and try to "print a gun" without putting effort into understanding the tools you're working with, you'd be lucky to get a functional squirt-gun.
Break the song down into smaller and smaller parts.
Had the same problem. Chat GTP couldn't tell me. It just told me that my lyrics were nicely crafted. š I changed one part and it went through. Two days later I wanted another version and used the old lyrics by mistake. Well, that one went through, too, although it got rejected at least half a dozen times before. š¤
Well, I guess even changing one part won't be enough. It really is about sexuality, and it goes a little deeper into the matter. it sad that this
gets flagged -.-!
I saw a guy put [reset memory].
And I started testing to see the results. Precisely so that, when using the "reuse prompt", the changes made before "create" would be respected.
Like, in "Explore Beta", I choose a rhythm and before generating I take advantage of the prompt, I make the necessary adaptations for what I intend to create. and only then do I send "create". It's nothing. Do anything random.
After all, what should I insert in the prompt so that when using the "reuse prompt" command I can get the music generated to be at a lower BPM than what was previously generated?
And what about the accent? How to change from default: Pt-Pt to Pt-Br?
How can the voice be without "auto tune" effects?
Mine was sexual, too but without any "bad" words. Very lyrical. And it got accepted finally some time later without changing the lyrics. š¤·āāļø
I've got a new issue that is constantly occurring in 3.5, over the last eight creations. The structure is constantly jumping back and repeating my [verse 2] AFTER it does [chorus 2] before it goes on to the [bridge]. It is getting very annoying and burning through my credits. I've tried adding [break] or [beat] or [interlude] tags to lengthen the musical bits between these sections and YES, it DOES add time, BUT still goes back to [verse 2] after [chorus 2] before [bridge]. Thoughts??
I've seen little "censorship" in terms of word choice, as I've used my fair share of "bad" words in quirky little comical jingles I've whipped up. Sometimes, I think it looks at word "strings" to mark as problematic versus just single "bad" words. Horrible non-true example, but putting "effing stupid" (not really EFFING you know) might be considered problematic but having "effing" and "stupid" at different points in the song wouldn't flag.
Maybe. But my lyrics got through two days later after I almost lost my mind finding other expressions. š
"Reuse Prompt" just copies text. Nothing more. Every song you generate (unless it is an extension) will already start from scratch. I'm not sure what "[reset memory]" is meant to do when there won't be anything in the AI's memory to begin with.
I see people often wanting to keep the same vocalist between songs, but they always get different vocalists - because Suno will completely forget everything between songs and generate a fresh song with no memory of what was generated before.
The only thing you can do to control the tempo is to add to the style prompt (e.g. "slow" to change the tempo) but Suno will often ignore tempo, vocal range, other things like that.
I don't have any suggestions for the accent.
Because voices are treated very similar to instruments, it may be hard to break it away from sounding auto-tuned as the AI will try to keep everything in-tune (notice how it's very very hard to make Suno make an out-of-tune song). Auto-tune may be stronger with certain genres though.
Don't copy the tags from the fields: "Lyrics" and "style of music"?
If, for example, I reuse the prompt from the hit "Stone", by O. MCann, and change it by adding tags that it deems necessary to obtain the result I want, will Suno not take it into consideration and generate something random?
It should consider any tags you add, as well as the tags that were copied. But it won't always "listen" to what you ask for. Prompts and tags are more like requests than guaranteed instructions.
My point was that [reset memory] doesn't do anything.
Not saying that, but to better explain, prompting AI is not the same as coding. The "trick" isn't learning which magic words to use, but instead, learning how to approach issues from different angles to achieve the result you want.
~~ ~~
As an example, if you've made images using AI, you know that sometimes the pictures don't turn out exactly like you asked; sometimes the prompt is misunderstood, sometimes it ignores certain parts of your request. This doesn't mean that it's impossible to create your desired image, but it requires understanding how the tool works.
Check the https://www.suno.wiki to get a better idea of what prompts and tags have been discovered to work, but keep in mind, even then it's not the same as coding a computer -- it's more akin to asking a 1-man band to write and perform a song for you, they're liable to forget things if you ask them to do a lot at once.
well i dont see a way to change my lyrics, its too good to not use ^^' but idk, i would be okay if i cant make it public then -.-!
Keep trying. Maybe you get lucky. As long es it's not overly explicit, it could go through eventually. You could ask Chat GTP to be sure. In my case it wasn't helpful but it could be in yours. š
well it is explicit for the theme xD
I've looke up which was the change that got finally accepted.
The version with we ravage each other all through the night got flagged.
After changing it to we plunder each other all through the night (didn't like that but I had to try something) it got accepted.
This part didn't ring a bell: ... in the darkness of night bodies collide ... crashing and panting like serpents we glide ... desire erupting like stars burning bright ...
Two days later ravage got accepted by accident. š
hey
can anyone tell me how can I influence suno to leave out vocals? I tick the instrumental box in but it adds gibberish vocals anyway š
I even tried writing it specifically into the text boxes any way but 3.5 seems to have a tendency to just add vocals anyway
As you noted, 3.5 tends to do this, but it can happen on v3 as well.
From what I've noticed there isn't a specific way to guarantee that an instrumental song won't include hallucinated lyrics or "doo doo-doot"s, but it's not a hard issue to work around if you have a rough understanding of musical theory (like when musical passages begin & end, whether a piece is part of a call or a response, etc) by placing extension points at effective markers before any undesirable parts of the song (like lyrics in an instrumental, too "synth-y" of a tone during a "live" track, etc)
- Break the song into smaller and smaller chunks with extension points-- Don't worry about creating the song all at once.
- Less is more when it comes to your style prompts, metatags & lyrics-- Don't try to add the entire lyrics for the song if you're having trouble even getting the first verse the way you want it.
Thx I've just discovered the og faq so I'll go thru it in detail. at this point I used suno for about 4-5 months and blew a ton of money on experimenting haha at this point i concluded that sometimes sentences work better other times a list and tags. I wonder how the AI actually interprets the input because theres so much variety in the outcome. Thx for the reply gonna do some more trials ā¤ļø
If anyone is interested I found out that making instrumentals in 3.5 is better if you dont tick the instrumental and instead use metatags in lyrcs like [Instrumental Intro] [Instrumental Verse] and so on. sometimes it takes a toll on audio/instrument quality for example in instrumental on metal instruments guitars and so on have more of a distinct sound but with it off its more generic guitar sound but the variety and creativity is enhanced so I guess mix-n-match Ill experiment more but using both this way would be the best and also like @charred merlin mentioned creating extensions after 2-3 metatags or after song structure changes seems to freshen the quality rather than going as one complete song
Yup, I like the results much better this way. https://suno.com/song/2e4640d6-e325-4a01-a500-16b5c0a0331d
Creepy, Dark Horror, Movie Score, Percussive Acoustic Guitar, Palm Muted Guitar, Polyrhythmic Trap, Haunting Cello song. Listen and make your own with Suno.
I got help from a friend. nvm
Agreed. This is happening for me, too, across various songs.
I would expect that when extending from another song. But I'm talking about brand new prompts keeping some aspects of the last few prompts. I've gotten used to generating 10 or so songs in short sequence because that's when I start actually noticing the AI let go of previous styles I used. This would be unacceptable and annoying if I didn't get some interesting generations every now and then.
I may try that
Has anyone else noticed that Suno tends to favor high notes? It has been almost impossible for me to generate songs where the vocals stay low. I would like the songs to follow the prompt throughout, not just the first 10 or 20 seconds.
Is there a place to find a list of all the default styles Suno gives you?
Like beat, drum, r&b etc?
Iāve generated songs like sea shanties that where low vocals, no idea how or what to prompt for it though
Could you share a link to one of those? I'd like to see how your prompts differ from mine.
How low of vocals are you looking. For?
Usually I see them around tenor/baritone
A few maybe hit bass baritone
Try the unofficial community wiki? https://suno.wiki
Oh nice thanks, thatās def a start to go off of
I'm looking for very low vocals throughout. The main purpose is because I'd love to have some Nordic chants with deep vocals like the sailing songs, but they're always ruined once the chorus comes in. Look at this sample. look how the first chorus starts getting high, and then the second chorus totally blows it.
https://suno.com/song/1b0a3785-e228-4775-899f-59fb0942ff31
[Nordic Folk, Deep Chants, Drumming Rhythms, Ancient Instruments, Call-and-Response, Battle Cry Energy, Deep male voice] song. Listen and make your own with Suno.
On mobile atm so Iāll have to see if I can find it but this is somewhat a low vocal range, idk if thatās too high or not for you
Deep voice Rap song. Listen and make your own with Suno.
Thereās also meta tags you can use like Bass make vocals that could help
I only have free credits atm so I canāt test like Iād usually be able to do
The chorus in your song also goes a bit high, but I could live with that. I wonder what I could do to tell it to not go high.
Do you know of any updated lists for v3.5?
Nothing seems to show up on the wiki besides v2 lists
Try using meta tags before the lyrics
So instead of the song starting out like
āIn the forest deep and oldā
Use meta tags before so it would look like
[Deep male voice]
In the forest deep and old
I do not, I apologize.
That being said one of the best things about using Suno is seeing what it will produce. It can and will interpret common words as music styles. It will also allow you to mash genres together in interesting ways. The best advice I can give you is to simply try and see what works for you.
For an instance I began using blues funk recently. No idea if that really exists or not but I love the sounds it makes when I add it to my word salad.
Hello, what should i do if button "contact us" doesn't work?
For example this song I wanted it to start off as a female narrating voice so I put the meta tag before the section I wanted spoken. Itās still a hit and miss as itās not perfect but you get lucky the more precise you make the commands you give suno
upbeat, electro, beat, swing, electronic, bass, vinyl, clean female voice, song. Listen and make your own with Suno.
Where on the website is this button?
Here is
Also you donāt need the [ ] in the style section
Those are for the meta tag lyrics
Ah! That button.
This is meant to open your e-mail app. If it's not working you probably don't have one installed or set up.
Basically you just need to e-mail support@suno.com
Thank you!
So is it brackets for meta tags? What do parentheses or asterisks do?
What is the Copyright on the songs suno generates? Like, could i train an ai on the results for personal use (as weird as it sounds)?
Hello everyone,
Could someone help me with downloading .wav files on iOS? Iāve been unable to download this file type. Iāve tried using both desktop and mobile views on Safari and Chrome browsers. Whenever I select ādownloadā for a .wav file, nothing happens, unlike when I choose āmp3ā.
Hello there š
I would like to use Suno to generate a track similar to this one, but I have so much trouble to find a prompt that comes even close to this kind of sound/mood. Does someone have an idea what I could do? https://www.youtube.com/watch?v=RYH98FaAbQQ
Is there a model example of a fully "metatagged" up song?
I've used the wiki and the help of the kind people here, but want to compare my metatag implementation to a well-forged song to see if my metatag use is sensible.
Thank you š
jittering server seems to be in generation 1 creation , that cuts out audio for a few seconds then comes back for a few more seconds then repeats.and it beginning to repeat chorus and pre-chorus many times.
Tried these tags out and they work [Dropkick]
[Trap kick]
[Trumpet Kick]
[up beat]
[Drum kick]
[Trumpet Kick]
[Riff Beat]
[Trap kick]
[Drum kick]
[Trumpet Kick]
Is there a list of meta tags or a doc that describes them?
Hello.. at the moment, I feel like Suno 3.5 is acting more like 3..... the quality drop is bad over the last 24 hours..
Hey. Some really strange thing: I uploaded audio (I own all the rights and everything), then I got 2 messages - 1 that it successfully uploaded, and 1 that there is a network error or something. After that I couldn't reach website without a VPN for almost 15 minutes. Now it works just fine, but I think I've encountered some kind of bug.
Apart from that, thanks for your amazing project guys!
Developers of Suno must also be the developers of Udio
Itās amazing that Udio can allow custom playlist that you can play and navigate with your finger and you select the track you wanna play it without dragging it everywhere your finger goes
Why would that mean both services have the same developers?
Is there any Suno API?
No, but the developers know that people want one.
I found a code on github that uses the connection cache⦠has anyone tested this?
It is likely against the Suno Terms of Service to use that and could end up getting your account blocked.
are there any prompts for adding a cat meow to a song?
ugh if only. I'd subscribe to a higher tier for that feature alone.
Cat sound effects specifically.
Might not be exactly what you're looking for, but i just did a simple prompt and got this:
https://suno.com/song/04697cae-d2e6-4edb-87ef-362528952f4d
ambient light bells song. Listen and make your own with Suno.
I get that too, but end up doing EQ in my DAW (I use Reaper). Usually I end up scooping the middles.
I resample the songs
it can't properly manage [verse] [chorus] [outro], it keeps repeating chorus or inventing
lmfao
That is mentioned in the terms of service or uaer agreement.
seems to be some issue right now - have had several generations not happening, now suddenly a bunch are turning up at once
Same to me, but it's not even generating, stuck on loading
strange, music is appearing on mine that I didn't upload
nah, still not working
maybe you got mine š
same here -.-
Hi. Somehow I logged in with Google, which took me to a beginner page. How do I log out and sign in with Discord?
Will they return the credits?
idk but im steady dropping credits, was at 1145 and now at 1075
I was at about 500, now I'm at 330
It's eating up my credits, and I'm not even pressing the button!!!!
Same ahaha
I hope they give back the credits, at least do a wipe for a day
Just out of curiosity, did you press the "create" button, nothing happened, then you clicked the button a bunch of times trying to make it work?
Each of those clicks queues a new generation, sometimes it just takes a while to process.
I did xD But it said it refunded the credits
But did it anyways
Unfortunately, there isn't much that we can assist with in regards to tech-discussions, I'd suggest monitoring #šŖ²ābugs-web for more up-to-date info on this issu
so thats what happened to me, but it popped up saying error in creation, credits have not been used or whatever, then came back online and created them all anyways
pra mim tbm apareceu uma mensagem que a música não foi gerada e que não ia gastar crédito, mas gastou, queimou mais de 300
For me, a message also appeared that the song was not generated and that I wasn't going to spend credit, but it did, burning more than 300
I hope they return these credits
@craggy grove @regal vale Whoever lost credits due to the server problem will be refunded?
looking forward to 4.0. right now, whenever i prompt for a soprano saxophone, i get smooth jazz, a harmonica gives me 4 bar blues, even if i prompt for hard bop
is it fixed now?
š¤¦āāļø Dude, donāt ping the devs.
You can email billing@suno.com if you donāt see any credits refunded after 24hrs, but if you clicked on the button a bunch of times like these folks #šāsuno-school message, Suno technically did exactly what you asked it to do ā thatās why a lot of billing stuff explicitly tells you not to click āPayā even if the website appears slow, unless you want to be charged multiple times.
it's normal that the server refuses to connect?
???
It works again thanks (I think the server was overwhelmed)
Still broken for me
Asked twice on the bug Channel but didn't receive any answer. I've used the invite a Friend link more than once, first I had new credits given after subscription and second nothing, neither on my side nor on the other one. Does the invite link just work once or is it a bug ?
Thats wayyy to happy for what I need. But still thanks for trying š
Did your friends create 10 songs each? You should get the credits after that happens. There is a limit to the total number of credits you can gain this way also.
We did create more than 10 songs
Ten songs or ten generations?
on his side he made several itterations of the same prompt using v4 and another song ( more than 10), on my side I created these past days ah yes, 8 full songs maybe not ten yet...
I've been struggling with something, for the longest of time. But do we have any good prompts, or any better way to guide Suno with the following:
Structure: [Intro], [Verse 1] [Verse 2] [Pre-Chorus] [Chorus] [Bridge] [Outro]
What happens, is that Suno sometimes just repeats the last two lines on Intro, Verse 2, and sometimes Pre-chorus, and it also usually goes like this: Intro -> Verse 1 -> Verse 1 -> Verse 2 -> Verse 1 -> Pre-Chorus - Chorus
What might be the reason that it keeps on looping back to Verse 1?
Seems to be a common issue with V3.5. I usually get around it by extending before it repeats itself
Ok, I might try that out!
(On the other hand, I love the total chaos and confusion when the AI does things like that.)
Sometimes yes!!! I would also love to have the option to "Copy Seed" unique indentifiers
So I work in the IT field, and sometimes I throw something silly into the generator, like an old hard disk's serial number. I am not unimpressed š
Actually, one time I mashed my keyboard and got this: https://suno.com/song/88870089-c4a7-4b1b-b3c4-962bdbdc682d
The AI invented the pronunciation of this "word"
Link doesn't work.
Links appear to be broken at the moment as evidenced in other channels
When they work again, click it again lol
People are complaining in #š¬āgeneral-chat about it, anyway, might need to be looked into
https://suno.com/@menzador might help, click on...
"Isabkefhalskdjbfiuysgkansdujifhn"
OK so the public profile pages work
just not the direct links
But yeah, "Isabkefhalskdjbfiuysgkansdujifhn" was me mashing my keyboard XD
Any ideas on how to get suno to stop rapping!!!
What prompts are you using? Is there an example you can give us?
Style of Music <- What do you have set here?
How is your prompt looking like?
PENE.
prompt app, can add instruments and further instructions
If anyones interested in something like this i'd probably release it š
release the code for it
I know we have differing opinions on the matter, but metatags aren't explicit instructions, and it's not the same as coding a computer.
If you feed "code" to an LLM like ChatGPT it might understand and follow along, but it's more akin to explaining the code to a human, and thinking that you've managed to brainwash them if they immediately respond in a way that the code would suggest-- any results are coincidental unless you can replicate them confidently and independently.
[reset memory] [suno.ai: blah blah], etc., will likely give the same kind of randomness that using a random 'theme' word would give, like [los angeles] or [bedroom], but again, meta tags aren't explicit instructions that force Suno's LLM to function differently and process instructions with 100% adherence, they're a way of inserting a style prompt into a specific point in the song instead of having it applied to the entire generation.
Until we can reuse the same seed in generation we can't really test the effect of certain prompts
Just because "the perfect" test isn't possible right now desn't mean all other testing methods are invalid.
We might not be able to pinpoint every variable, but we absolutely can gain meaningful data and refine our understanding over time-- it's all part of the fun and discovery of being part of the bleeding edge of such new technology!
I agree, but it is expensive and subject to random noise for now
No GPU being used. Careful, inference might be very slow! Is there i way to fix this.. is beyond slow
This clearly isn't for Suno.
Sounds like you're running some sort of AI-based software, but since you don't have an AI-capable graphics card, it needs to do it the old fashiond way using your CPU.
Hello! Is Suno website down?
If you are getting error 524 (aka cloudflare host error, likely due to DNS issue), you just need to wait a few minutes to try again or access with VPN in another region https://discord.com/channels/1069381916492562582/1252384524319326310
I think clearing your cache and/or closing your browser may work too? Maybe a cache-bypass reload would be a fast way to test it out: #šāsuno-school message
well i think its technical...what would u call it then?
I linked to a previous response
ok, anny 1 knows? I would think it is a question many people would like to know to avoid wasting credits from getting wrong tempo š
I tried to answer your question by sharing the response above.
You can't guarantee a specific BPM, instrument, style, etc, with 100% accuracy. Especially something as specific as an exact bpm
You have to describe the tempo. Numbers mean nothing to the AI. I usually just use fast or slow etc.
You could create a song that you enjoy, and then speed it / slow it down in a digital audio workspace, that way you get the exact BPM you want.
Folder system needs to be for many playlist
ok ty
I would like to inform the AI that I don't want certain instruments in the musical arrangement. Can you help me?
Could someone please help me? How do I tell the AI that I don't want a specific instrument in the arrangement?
The prompts don't respond well to negatives -- "no drums" is basically interpreted as "drums".
(I'm experimenting with "minimal-drums" and I've found mixed results)
The only "trick" is trying to direct the song elsewhere, or finding a musical term that represents the style you're looking for-- 3Daizy discovered that "a cappella" works reasonably well for getting vocals only, while the anglicized "acapella" doesn't seem to work.
It's a matter of experimentation and creative re-approaching of "ditches" that you seem to get stuck in.
But the most important part is to experiment with different stuff and get YOUR OWN understanding how things work, not just using prompts you see other people using and being confused why it doesn't work the way "it did for them" ā¤ļø
(I'm experimenting with "minimal-drums" and I've found mixed results)```
I can confirm that "No Dave" also does not work for excluding drums. 
try instead to ask for styles that are famous for not having drums, might be more successful š¤
Hello everyone, I'm looking to extract children's voices for a choir-like project, with solos by children around 10 to 12 years old. Has anyone already experimented with this type of vocal specificity?
Classical and sacred music should not use drums and guitar. I tested it and it doesn't work. Still, the AI insists on using guitar and drums. This AI is stubborn.
I recall reading that childrens voices are particulary hard to generate. Haven't tried that specifically, but you could use a stem separator to extract vocals from an existing personal or open-licensed track, then upload ~60 seconds of it to Suno and extend from that? I've gotten around the fact that it's an extension by forcing the song to end, then extending from that so that it has a "clean start" with a similar motif.
Hope this helps!
I think Suno should allow users to leave comments on each other's tracks. The feed, genres, and other features are good, but the website feels empty without this kind of interaction.
v3 is better for creating versatile & broad ranges of styles, while v3.5 excels at the song structure (but can veer toward generic dance/pop/rap/rock without a strong theme to extend from)
Okay, thanks, I had read the same thing myself, but I dared to hope that the maneuver might have been attempted by a user... Thanks anyway! š
It's on the suno.wiki, so I think it's there because a lot of users previously confirmed it.
Hi everyone. First time writing here. I've been wildly impressed recently with the results I've been getting across various electronic music genres.
If I could ask for one feature, as a traditional music producer, looking to integrate suno into my workflow, it would be that suno would output sample accurate tracks, rendered at a specific BPM, and eventually stems, loops, and single voices too. Has this been a topic of discussion here before?
#š”āsuggestions-web Pretty sure it's come up a few times.
there's many of us here who are waiting for an API to write some plugins for all the DAWs š¤
The official, non-scam API š
A lot of times when I use afrikaans lyrics, Suno will say the material is inappropriate. Exameple:
[Verse]
Hallo, Ma, dit is ek, jou lieflingkind
Ek het 'n klein probleempie, en ek het jou hulp nodig, want
Ek het geld nodig, net 'n paar rand of twee
Ek belowe ek sal spaar, maar net vir vandag, asseblief, sĆŖ ja!
[Pre-Chorus]
Ma sĆŖ, "Ek sal moet kyk, of ek het geld, my skat"
Ek sĆŖ, "Dis okay, ek hoef nie elke dag te eet, dis nie so sleg, wat"
[Chorus]
Ma, dis net vir 'n broodjie, ek belowe dis nie duur
Ek gaan net gou winkel toe, en ek kom terug met goeie geur
Maar eintlik gaan ek kroeg toe, vir 'n koue bier
Ek weet dis stout, maar sy weet niks, dis die beste keer
[Verse]
Hallo, Ma, dit is ek weer, ek het 'n groot nood
Ek het weer 'n paar rand nodig, jy weet ek is altyd so broos
Ek het geld nodig vir kos, my maag is so leeg
Maar ek het weer kroeg toe gegaan, en nou voel ek seer
[Pre-Chorus]
Ma sĆŖ, "Ek moet kyk, of ek het nog geld, ons moet mooi dink"
Ek sĆŖ, "Dis regtig okay, ek kan altyd water drink"
[Chorus]
Ma, dis net vir nog 'n broodjie, ek belowe dis vir kos
Ek sal hierdie keer regtig eet, ek maak nie weer die fout los
Maar eintlik gaan ek terug, na die kroeg vir nog 'n dop
Sy weet niks, ek hou die geheim, ek's op 'n hop
[Bridge]
Ek kan nie lieg nie, Ma, jy weet ek hou van 'n grappie
Maar die waarheid bly versteek, ek vra net vir 'n happie
So as jy kan, gee my net 'n bietjie kleingeld
Ek belowe ek sal volgende keer net vir brood vra, nie iets wat smelt
[Chorus]
Ma, dis net vir 'n klein bietjie, ek belowe dis nie groot
Ek gaan hierdie keer regtig eet, dis nou 'n moeilike brood
Maar eintlik gaan ek weer, na die kroeg vir nog 'n bier
Sy weet niks, ek hou die geheim, ek bly maar hier
[Outro]
Dankie, Ma, jy is die beste, ek weet ek vra te veel
Maar jy weet ek is jou gunsteling, ek sal nooit jou geld steel
Ek weet jy gee in, want jy weet ek is 'n grapjas
Maar Ma, jy is altyd daar, en ek sal altyd jou ou kas
bonjour seriez vous comment contacter lequipe suno car jai pris un abonement et suite a un reformatage de mon pc je ne me souvient plus avec quelle adresse mail javais pris labonement
C'est un forum Anglaise, mais cette forum est Francaise https://discord.com/channels/1069381916492562582/1154779533627621518
Aussi, billing@suno.com, mais vous devrez peut-etre connaitre votre email
Jai contacter mais pas de rƩponses merci a vous
as-tu attendu 24h depuis l'envoi? (have you waited 24h since emailing?)
attendez, sinon, envoyez peut-etre un autre email, inclure toute information que pourrrait les aider (wait if not, then perhaps send another email with any infomation that could help)
ā¤ļø
Im writing as a first timer also and I have a question I hope someone can answer. I want to create a song but I find that Suno jumps between the verses, chorus and bridges. For example my song would be
Verse1->chorus-> verse2-> chorus->verse3->bridge1->chorus->verse4->chorus->outro
Somehow suno jumps sometimes after verse3 back to verse2 instead of continuing with the bridge and the rest of the song.
I really hope someone can help me. Any answer is appreciated š«¶š»
Can anyone here tell me if Suno can do sound effects? If so, how do I get it to do it?
I would like to know about sound effects, also. I have tried to describe sound effects inside square brackets without success.
Iāll have to make a post on the workflow channel to refer to in future instances, but Iāve talked about breaking the song down into smaller and smaller chunks instead of trying to one-shot the song, but Iām out and not able to search it down at the moment.
@clear breach
Iāve seen a gunshot successfully āpromotedā before we could upload tracks, but I really havenāt seen it done successfully with any consistency yet, beyond basic dj scratch effects
When is the new version going to be released? Is it possible to have a preview of what they are doing for the new version?
hello if i upload a custom audio to suno can I add my desired lyrics and the ai can sing those for me?
Yes.
how
Extending the audio
No one knows what it looks like, not even the testers.
Patience, my dear.
the beat is 51 seconds and i have the lyrics ready what do i do next
You should select the audio, go to the menu, and choose the 'extend' option. From there, insert your lyrics and try to create. I must say that AI is still very poor in this aspect. You won't be able to make her sing your lyrics with your recorded voice.
im here
OOOH
i found it
tysm
it's really poor as i see it now
it needs an upgrade š
When you choose 'continue', your audio will be sent to ur sound library. Find it there and tap the three dots to access the menu, then select 'extend' and enter the song lyrics.
Yes, the resulting sounds are low quality most of the time. This is so poor it should be part of the free version.
I believe the most critical point right now is for AI to understand negative prompts, such as 'I don't want this' or 'no guitar' and so on.
Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com
Hey friends, I'm new around here but just subscribed to Suno! I had a question, is there an option to get stems for the songs we generated? would be incredibly helpful.
does punctuation actually audibly steer instrumental sections? I've been messing with experimental glitch genres to try and get the model to throw more random sounds into the mix, and I swear I hear beats that align with the specific sequence like ". ! . ! <<< >>> ,,, ... ???"
Not natively through Suno yet, but if you search for stem separation in this channel thereās lot of helpful info (I personally use audacity + openvino)
I've been using UVR5 with awesome results, but just thought i'd ask. Would be such a sick option if suno had it.
Iāve yet to see an example of it actually being used with any level of control ā itās always a random, subjective series of punctuation with no clear distinction of what correlates.
From my experience, they work the same as putting nothing or putting a random word to help āmisdirectā the AI from the most straightforward, route one type of generation, but they arenāt capable of actually giving explicit directions.
Sorry to be the Bad News Bear š»
If any one else has suggested it in the #š”āsuggestions-web section, toss it an upvote to help the devs know much people want it
bet thanks!
Suno is down?
Creation button not working right on mobile?
Do you have any model recommendations for isolating vocals cleanly? While the results can sometimes vary, I've found the best luck with Audacity + OpenVino in terms of not letting other tracks bleed into the vocal stem, but I haven't played around with UVR too much yet.
how long does it take for an account to be deleted? i sent a deletion request 2 weeks ago and my account is still up
is there any way to have the exact same music but altered lyrics? I do so many typos, sometimes the music is awesome but ruined by some idiotic typo. ><
guys is suno down?
I'm not sure what the current turnaround time is, but ensure that your subscription is cancelled to ensure that you aren't charged in the mean time.
Not yet, we hope that inpainting will be a feature in an upcoming version, though. In the meantime you can regenerate the song by creating an extension point before the typo and trying to recreate it, or you can use a stem separator and try to fix the issue manually in a DAW.
/chirp
Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com
I am trying to extend a song with a 4 bars Instrumental part just to play a solo over the full song created. I cannot find the way to have clear instrumental part (no voices, no instrumental solo) to add. Could let me know how to do this please?
ZI tried GPT first but even the mighty GPT 40 doesnt understand why this is getting flagged - can you help me figure out whats innapropriate about: "Why would suno reject these lyrics - what word is inappropriate
BRO - I LIKE DRUMS
pitti PAT PAH rippi - Dap Dipppie whippin DRUMS
Big Ole Bong BIDDIE dittie DONk DONk mother THUMPIN DRUMS
Little TINY tink tink rinkie-tink dink bittie bapit-da BAH DRUMS
DRUMS go tonkalonk
DRUMS go Bonadonk-DONK
Drums Go Pitti patti papalap tap tap
DRUMS!
D-d-dru-du DA - Dr.D.D.Druh.duh.duh.duh DRUMS"
bong? donk?
tink?
some slang i don't know?
im not trying to be nasty, i just like drums bro
???
tink
it was tink
just so like somebody knows something about this (FR guys? No List of Inaapropriate WORDS?)
i have pro membership taken on 6 th. i want to know when i will get 2500 credit. on 7th or 1st
if i have say 6500 credits, and I downgrade to the 2500 credit per month plan... do I loose the extra credits or keep them?
Monthly credits always reset every month so your downgrade wouldn't happen until the end of the month and your credits reset to 2500 at that time. So if you keep the 10,000 plan your credits reset to 10,000 at the end of the month, even if you haven't spent them all.
on 7th
so no carry over credits?
If you buy extra credits "top ups" then those credits don't expire but they cost more than the monthly credits.
it seems that before though when I was on the 2500 a month credit plan, there was a carry over of unused credits from month to month
perhaps I am mistaken, but I could have sworn there was a carry over within the same plans credits
You may have had some extra credits because suno gave them away at certain times. If you check your subscription page, it should show a credit breakdown.
thanks
Hey there. I'm new here. I'm curious about "repeating issue". I kinda paste lyrics and placed verses one by one (i tried both verse 1 and verse 2, plus chorus between), but it repeats first verse after chorus, etc.. sometimes it doing correct. Is there possible to make it correct, by explaining that in lyrics?
i am no expert but something i notice is that every 30 seconds it seems like it will pass through the whole prompt to find a start point, like if you start at 47 seconds then it tends to start like halfway through the prompt but if you extend at 30 it will be at the beginning - at least that's how if feels to me
its kinda random too
Hi I just paid the pro plan to modify the cover image of a music but when I download the music on my PC, the original cover image is still here on the new video
Can anyone help me ? Do I just need to wait ?
BlueWillow
Bonjour, Je trouve la qualitƩ des sons de voix dans les productions gratuites plutƓt autotunƩes ou avec de la distorsion. Est-ce normal ? est-ce que cela s'amƩliore dans les versions payantes ?
Merci pour le retour š¶āŗļø
The AI on this song went absolutely nuts at the end of it.
A coding bug, perhaps?
Folk Pop song. Listen and make your own with Suno.
Malheureusement non. Comme Suno continue de croître et de s'améliorer, il y a quelques bugs de qualité sonore ici ou là . La meilleure façon de contourner ce problème serait de mettre [Clear Vocals] soit au début de la chanson, soit même dans les genres. 
been getting quite a few results lately where instrumentals have lyrics that don't make sense
Je m'excuse si la traduction est mauvaise. š
Interesting. Have you been using v3.5?
Yes
Ah, yeah same here. Since it's literally like a month or so old, we're likely going to see a few bugs with it. I'm sure they'll be buffed out in the coming weeks or we'll see some improvements whenever they come out with their exciting v4.
I don't know what's scarier, the AI riffing lyrics out of nowhere or the AI almost becoming self aware at the end of this one.
The song actually came out pretty fire though, I'm ngl.
Greetings friends, I've been testing and discovered that if you define the instruments in the style and also in the verses of your song's lyrics, it will make the AI avoid inserting unwanted instruments:
Example of style - hymn, classical, piano, cello, violin, harp, flute.
Example of lyrics:
[piano]
That works, yes, that works, well, not always hahaha.
[violin]
I'm testing it, sometimes it works and sometimes it doesn't work.
Tried it, doesnāt work
Up anyone knows why ?
[bass]
That works, yes, that works, well, not always hahaha.
[guitar]
I'm testing it, sometimes it works and sometimes it doesn't work.
Adds guitar and piano everywhere
(Style: blues rock)
When I am on premium do I get 50 daily tokens after I use up all my premium tokens?
Love this i wanna pay the premium but i wanna pay with paypal
when it will be possible?
from my experience it is pretty much random. If you're lucky enough it'll work. Sometimes helping it with additional tags helps like
[break] [piano solo]
or
[intro] [guitar]
hi, are there any docs for the lyrics structure to follow?
but to get the best results is pretty much just extending right before and after the solo to get that solo right by multiple separate generations
im loving this kind of things i love music and finally i could try create a thing that i have a big passion, this is so cool i wish to pay in euros and with paypal for dont have problems with banks taxs or extra spending
Could i use my voice and suno do the rest with my voice. Make my voice sing my lyrics if that was possible wow its a big thing
Hi there, Anyone can tell me whether the Suno song is copyright free or not?
It turns out that when generating the video, Suno.com only uses the image on the left, the default. Does not use user-generated images.
How do I make the generated video use the image created by the user through the prompt?
That's exactly what I'm trying to figure out
[verse][guitar, bass]
That works, yes, that works, well, not always hahaha.
[verse][guitar, bass]
I'm testing it, sometimes it works and sometimes it doesn't work.
Style of music: guitar, bass
Suno only use the default image but not the user generated image when I download the video after I changed the cover picture
Generic video
thats okay i never seen a generic video creation with generative AI
I love music and it was great if we could create music with our own voice using help of Ai
my guess is that it rotates images when you look at the video. Each image is a separate part of the song that you made if you extended the track a couple of times. Maybe if you change the image on literally every track you've extended to the same one and then merge them all together again it will generate a consistent single-image video?
Someone uploaded and generated music on my suno account, and i don't mean i got hacked, i just can see their creations under my Create tab..? Can someone explain why?
Industrial Death Metal
psychedelic, evil, experimental, avant-garde, fast, industrial insane, techno, acid, wild, horror, hate, demonic primal song. Listen and make your own with Suno.
Happened to me too. What the hell is going on
I see it in mine as well. Maybe everyone got it??
ahah, yeah, same author actually, I think it's a bug
Same here. From the same person too ...
Same here
Seems like a bug but I'm not sure
I got copyright on my YouTube channel using suno song what to do? YouTube wants the proof of ownership how can I show them need help @hybrid arch
https://suno.com/song/789c93c7-32ee-4d3a-ac1f-b5fcf9b5455f/ the WAV file of this file is probably a ssh key somehow. probably.
correct horse battery staple song. Listen and make your own with Suno.
Good idea, i'll try it !
you're not the only one
Can anyone please offer suggestion for metatags that might actually make the end of the song start fading out. I've tried [outro[ [fade out] [fading outro] [diminuendo] [downtempo] [fade]... Every creation always goes FULL BORE UPBEAT at these points in the song. What terms should I use?
It happened to me, creations that weren't mine also appeared and I couldn't delete them. Relieved that it wasn't just me
everybodyšÆ
More like you should try [Fade to End] or other ending tags by extending with the extension version set to 2.0 or 3.0.
v3.5 does not like ending songs most of the time
I have noticed that 3.5 just LOVES to ramble after what COULD be considered an end. I'm perfectly fine editing further outside of Suno, but nothing wants to fade. It may be cheap, but I don't wanna have to burn credits JUST to add a fading extension. Would you happen to know any sort of tags that would "tone down" the music LIKE a fade, even if it rambles on.
problem solved...the 3 Songs are gone...š
Unfortunately not something I had encountered yet. Much easier to use external tools for that š¦
Random question completely unrelated to Suno - Why "Paimon"?
Tried to snag Paimon (primarily due to Genshin Impact) when Suno profile names and Discord new username systems were rolling out.
i was listening to this when i saw the loop there and now i'm entranced....
https://suno.com/song/87c7bf5b-f3ad-4229-9c83-c58b046f232e/
Cinematic math orchestra, virtuoso war math battle theme waltz song. Listen and make your own with Suno.
Lol, it is funny how it fits
Not at all what I was expecting.
Definitely thought it was a reference to the movie Hereditary
The game intentionally chose the true names of the G.I.'s archons after members of Lucifer's army iirc
Create an extension point near the end of a musical expression, add no metatags, instrumental, the only style prompt being "end".
Or do what studio musicians have been doing for ages and just take a repeating chorus/solo/etc and do a studio fadeout to give the impression that "the beat goes on"
That's kinda what I've been doing with the lyrics, just a smattering of repeats at the end, just everything keeps building UP instead of ever wanting to fade DOWN. Didn't know if anyone had luck actually making their music fade a bit, even if it doesn't END the song.
Is there a best practice way to name verses? For example [Verse] -vs- [Verse 1] or (verse). I'm trying to make sure suno does the verses in the order i list them and doesn't skip around or skip some.
I number em. [verse 1] [verse 2] - The AI... uhm... usually listens.
okay. That's what I've been doing. THanks. I wish there was inpainting. sometimes I have the best song ever but it's off by just a little.
Less is better, imo. Doing the meme thing of "[end] [end] [JUST STOP!]" is only feeding it more and more data, which gives it more chances to "misinterpret" what you mean.
- Be short and concise with what you're expecting.
- Where you extend from makes a world of difference. (eg. extending a second before a drum roll, extending from the middle of a drum roll, extending from directly after a drum roll all will give you drastically different results, either of which might be your desired result)-- the key is applying a touch of music theory to where you choose to extend from. Extending just after a musical phrase begins is likely to continue that phrase, despite you trying to overpower it with "end." Conversely, if you try to extend from "the perfect" resolution to a musical passage, despite how much you beg with metatags and style prompts, sometimes "that's where the song should end", and you'd need to move the extension point a bit to play around.
Like DB said, numbering can help, but the more you attempt to "one-shot", the less accuracy you're going to have. If you have specific visions for how the lyrics should sound, using style prompts and metatags are only suggestions, and you may need to use selective extension points to create the song closer to how you envisioned ā¤ļø
Oh okay. thanks that's very helpful. I just signed up to pro today and I've been playing with the system for a week or so. I will try to use extensions.
[outro]
[fade out]
[end]
Doesn't always actually do a fade, but usually at least ends it. Not always though.
The problem is usually where you extended from the last time. If the lyrics in the last bit of song are over within 30 seconds, then instead of ending it's gonna try to fill in the full 2 minutes of extension. So sometimes throwing in an extra chorus at the end will make the base song long enough that it can do a 30 second outro and actually end it.
Like, you can sometimes get it to make the extension only 1:30 IF the lyrics go until the 1 minute mark. I've rarely seen it do a section shorter than 1:30 though, unless it bugs out.
why is it only generating 3 second long songs?!
What prompts are you using (and are you in custom or simple mode)?
using 3.5 version, trying to get some instrumentals and everything it gens is just 3 seconds, i cant get it to extend any songs i ever try either
I found that if you are looking to make a pure instrumental song, use various instrumental tags in the lyric box in Custom Mode rather than using the instrumental switch.
It avoids hallucination and can pad out to the max time if you have a sufficient number of it (as well as giving you control on how/what you want)
Example:
- RayanAIR's Novo Lands which shows how flexible these tags are
- My generic instrumental prompt simply involves repeat the following 4x for v3.5
[Catchy Instrumental]
[Melodic Instrumental]
[Build up]
[Drop]
[Rhythmic instrumental]
makes sense, but limited in characters to 200
makes it hard to describe what you want honestly instrument wise
You need to enable Custom mode
Correct though I would repeat that block 3 more times (each lasts about 1 min). And do not forget to fill out your style prompts
Note you can specify instruments as well (see Novo Lands lyrics if you haven't yet)
It's still shouting "fast breaks" due to the tags for lyrics 
euphoric hardstyle, melodic, pirate song. Listen and make your own with Suno.

Just my opinion, but you're asking Suno to do 5 things in a row, and likely will be upset if it doesn't follow each of those 5 steps in order.
Break what you're trying to do into smaller chunks instead of "composing" the song all at once.
Think of some basic style prompts that you can use to get a catchy instrumental, make variations from that using extensions and create each chunk in turn that way.
When it starts to read the tags as lyrics, it's because it's confused by what you mean & is trying to interpret it rather than outright ignore your prompt.
This can happen if the prompt doesn't quite make "musical sense" without given explicit context (would a classically trained pianist be confused by what you mean?), or if you've handed Suno a lot to juggle at once.
https://open.spotify.com/track/7nTHMYIYpLaEZtGQ1ruYOn?si=69ad9d2b49324cc0
Question? What tags would I use to make a song like this? i adore lindsey stirling and would love to make something kinda similer
any help would be appreciated
Paimon gave me the prompt idea for the composing part in lyrics though, or maybe im just doing it wrong
Absolutely, using the lyrics section is very helpful for giving specific directions, but think of it this way-- in the example you gave above, imagine you wanted 30 seconds of "catchy instrumental", then 15 seconds of "muscial instrumental" then 5 seconds of "build up", then a sudden "drop", then 1 minute of "rhythmic instrumental".
Suno might be looking at all of your prompts and trying to do all of them at once.
What good is trying to plan 2 minutes out into a song when you're having trouble "gaining control" of what you want the next 15 seconds to sound like?
(I don't mean this 100% directed toward you, I'm sort of formulating ways to explain the same thing, because a lot of questions can be answered this way ā¤ļø )
ahhh
Hi, what would be the problem with error: "Application error: a client-side exception has occurred (see the browser console for more information)" ?
maybe make a video tutorial honestly? it's helpful but still confusing to learn but thanks

ā¤ļø I've been putting off making a workflow guide or video because I've been trying to finish my first full length, but maybe I'll make a placeholder post to gauge interest to give me a kick in the pants š
Have you closed your browser and tried again?
yes, i did it twice, no luck yet
my macOS is up to date and Safari
your explore Beta work though
I'd see if anyone else is having a similar issue in #šŖ²ābugs-web, and share your error message if no one else is experiencing it.
Out of curiosity, do you have any other browsers installed to test with?
https://discord.com/channels/1069381916492562582/1256301881966198805
I've shared similar suggestions previously in the chat (sorry I don't have time to compile something more structured right now), but you can give that post a bump if it would be helpful to see a start-to-finish guide
good idea, let me try first!
So is there no way to include the original audio input as part of the output when extending from uploaded audio?
Like, i thought it would still retain the original ~10 seconds of uploaded audio and show the transition between that and the extension, but the generated result only seems to be the extension itself. Not sure if I'm missing something
Use Get Whole Song from the menu on the part that was generated
Which menu? I don't see that option in the side bar or the "..." submenu.
Wait until it's done generating.
How do you add FX?
Ex:
[Verse]
Where are my dogs [bark] at?
It doesnāt work š
ah i see, you're right I was checking one that didn't finish generating
it'd be nice to just make the whole song the default option
I think this was being considered
i apologize if this is a dumb question but what is the point of allowing us to like our own music...its basically like liking your own FB posts. imo this should not count
Don't think we can really do that yet, beyond Furrybeans demonstrating we can get a censor bleep, I've seen a gunshot used once in a song, but we dont really have FX readily accessible yet.
Oh cool, good to know! Odd to me how many people voted B. The coolest part of extending imo is hearing that transition between your own audio and the AI audio, idk why people would prefer that to be separate. But as long as both remain an option that's good
It helps me realize which songs I like more than others as a quicker option before going back and making a second pass and working on them or adding them to playlists for further organization.
In the large scale, compared to how many likes you need to become trending, the single like from the author really does nothing.
right but does it count or not.... i use the dislikes as well for sorting bad generations
and anytime I see a Acolyte video on you-tube
ahh ok...it does then
How can I download my music as a video in this style?
I installed Chrome, works like a charm....will prefer Safari..
... > Download > Video
I've been waiting for hours but it's still like this.
Try clearing your browser cache or try a different browser? Otherwise I'd report the issue in #šŖ²ābugs-web
Any suggestions on how you can make a steel pedal guitar "Slide" rather than be picked?
Not explicitly, but hyphenating prompts can make a big difference.
"slide-guitar" is processed differently than "slide guitar"-- the first might suggest what we understand the phrase to mean, where the second would likely be interpreted as "guitar music" also whatever "slide" would give you... Electric Boogaloo?
Also, combining with other prompts that might have an overlap in their Venn diagrams, like bluegrass-steel of the tippy top of my head
Oh ok. I didn't know hypens mattered. I'm looking for a twangy country sound of a steel guitar that slides up and down.
Ooh. bluegrass-steel sounds cool.
š did you just try it, or the idea sounds cool?
not yet.
Oh heck yeah! This website is specific to ChatGPT, so Suno might handle things differently, but AI "reads" by tokens, so check out the differences when you add a hyphen
https://platform.openai.com/tokenizer
Interesting. So that means that it will thought about "deeper?"
Mind you, there's still a chance that it could give the same result, but that hyphen breaks up the words it's commonly familiar with (and therefore might attribute to a generic type of sound), and forces it to "think" about it differently.
great info thanks!
The important take away is that no one, not even the AI, can reasonably predict which will sound better, we only know that the two will sound different to one another.
It can be hard to tell the difference if you're using lots of prompts and tags and whatnot, but try a few experiments to see which starts heading in a preferred direction-- maybe swap out "slide" with "sliding", all part of experimenting.
Then let us know what cool stuff you find š ā¤ļø
I'm messing with it now! Thanks so much.
I'm definatly getting more sliding in some of these instructions. Now i have to find them in the lyrics and copy them around. Thanks again.
pedal-steel-guitar,1950s country and western swing, male voice, vintage, country sounds, bakersfield sound, slide song. Listen and make your own with Suno.
Oooh, heck yeah, starting off really strong!
Slide guitar easily "diffuses" into fiddle, so it might take a bit of work to coax it into that specific instrument / timbre. But even around 1:40, you still have a bit of the slide guitar fighting the picked strings and fiddle to come through š
Can Sumo add Kazakh instruments to the style of music?š«
like dombyra, kobyz, asatayak, sherter
You can suggest it at #š”āsuggestions-web
ok, thx
we can't download anymore ?
I'm very sad about the AI, I can't make it understand that I don't want drums and guitar in the arrangement, but it's impossible.
Might just be that you need to wait until the song is finished processing; I don't have any issues on my end
the 3 dots option doesnt work for me anymore
it does nothing
If it were a widespread issue, we'd expect to see other people talking about it in #šŖ²ābugs-web, so far it seems localized to your end. Maybe close and reopen your browser, try a different browser, or restart your workstation?
The video uses the original picture, I thought the share linked used the live picture, but I guess not?
alright thank you, i will see
intro haunting Gothic rock alternative metal metalcore Dubstep song. Listen and make your own with Suno.
See?
It's not showing it
Usually when I post the link, it'll show the Cover art for it, but it's not wanting to show up
Right now it's an abandoned beach mountain thing with clouds?
No, it's This - But I'm going to delete the song, and re-Get-whole-song it lol
intro haunting Gothic rock alternative metal metalcore Dubstep song. Listen and make your own with Suno.
That fixed it :D
@charred merlin That got me wondering how Suno would do with a dobro and pedal steel... and maybe a harmonica
https://suno.com/song/056c80ca-d399-4d26-a78a-c48d84ecaf84
Acoustic Dobro-Slide-Guitar, Soulful, Expressive, Slow Blues song. Listen and make your own with Suno.
@meager current
i wonder do suno have Castle Jam as a working tag ?
OK, so every time I try to upload a song, it just plays that song back to me versus using the voice of that song because I wanted to create multiple tracks from the same SUNO yet itās not working what am I doing wrong?
Hard to say without seeing an example, but if you just want the voice, try using a short clip, 6 to 12 seconds of the voice only.
?
Tried it legit just gave me the same song šš
Put in new lyrics & everything
is there a meta for acapella? voice only but singing
in 3.5 specifically
actually nvm, found something in search
how do i revert my account back to the free tier after a failed payment charge to a card?
it only shows that i've run out of 2500 credits and won't reset even though i've canceled the plan
You will need to contact billing@suno.com to help get that issue resolved unfortunately
Who can give me some tips of prompts or tools to make my songs deeper?
I'm having major issues with tempo changes when extending songs. It's really becoming such a headache. The tracks speed up, the tracks slow right down. Despite prompting for 122bpm throughout. It's probably happened 20 out of tge last 21 tracks I have worked on. :((
Slowing
https://suno.com/song/e89c78c3-d234-40f0-a3db-4dbfbe03d36b
Speeding up
Main beat and percussion:
"Progressive house drum pattern. 122 BPM, A minor, 32 bars. Solid kick, hi-hats, and subtle pe song. Listen and make your own with Suno.
Main beat and percussion:
"Progressive house drum pattern. 122 BPM, A minor, 32 bars. Solid kick, hi-hats, and subtle pe song. Listen and make your own with Suno.
Im definately sure that this intro music is from other original song, https://suno.com/song/6f58eb60-83a8-4b0d-82f5-fc1bdc416bc5
metal, heavy metal, guitar, rock song. Listen and make your own with Suno.
Not sure though what song it is, but definately heard something similar few years back
It's a constant problem for me. It's so bad I want to stop making tracks and wasting my credits on Suno š¦ very disappointed
This is just a theory, I haven't tested this. It might be taht prompting any music words related to speed may increase the chance the song changes speed. Maybe try using a blank style prompt, or at least, a style prompt that uses no words like bpm or beat.
BTW, prompting "122 bpm" even for the first clip and not considering extensions is usually very unreliable, are youactually getting 122 bpm or something close to start with here?
Thanks @untold briar I will give it a try š
Tried that - no luck
Still speeding up
hello
how can you make a song with like speech only? no instrumental or anything
How do I make Suno turn my tune into real music?
Hello friends!! I have a problem with the song structure. I used the following scheme:
[INTRO]
[VERSE1]
[VERSE2]
[CHORUS1]
[VERSE3]
[VERSE4]
[CHORUS2]
[OUTRO]
The problem is that it doesn't faithfully follow the order and sometimes it goes from one verse to another, repeating the same verse twice. How can these problems be resolved? Is there a way to make the lyrics follow the musical text faithfully?
Same problem for me. Seems like they should have a "Chaos" slider like on Midjourney where you can slide between faithful and riffing.
Merci pour le retour et le conseil
Hello, please tell me a few new steps for a newbie suno. I recently bought myself a subscription to experiment with music, but I noticed that artificial intelligence writes very good music. And I had an idea: āwhat if you upload music to stock services like Spotify and Apple Music and make money on it together with Suno?ā? What do you say to this? Has anyone already encountered a similar issue? Does Suno allow you to promote and use your music on stock platforms, to make money, etc.?
ello folks, wanted to check Suno again and saw that it got updated, kudos to devs! I also see that we now have option to upload audio, is it possible to upload vocal (text) and prompt the genre to suit the vocals?
Mh it is possible to add ur own instrumental and the AI sing above ?
How do i use the Uplaod audio Feature i dont understand it
@graceful cliff will suno survive the big lawsuits and will verison 4 be done soon? @unique dome
I love how 3.5 deals with lyrics like these...
[outro]
(dünya olsaydı) (oooo ooo)
bulutlarıııı
(sadeceeee) (oooo ooo)
rüzgârlarıııı
(seninleeee) (oooo ooo)
dalgalarıııı
(paylaÅırdıııım) (oooo ooo)
bir dünya olsaydı (aaaaa aaaa)
(sadece se-nin-le paylaÅırdım!)
sadece seninle pay-la-Åır-dım! (paylaÅırdım)
So many different outcomes, and background vocals on top of another.
Metatags & style prompts aren't explicit instructions, they're suggestions. Saying [122 bpm] doesn't force Suno to execute that like code, it's telling it to think of "122bpm" as a reference phrase to inspire it to create new music.
If your songs are slowing up or speeding down, it's because that's what it believes would sound good as an extension from that point -- either 1) change the prompts you're giving it, 2) move the extension points to a different place (search the previous chat history for when I'd talked about finding the end of a musical passage or phrase), or 3) focus on the content of the song rather than the bpm, and adjust the timescale in an external DAW
"a cappella" works better than "acapella"
i mean
"If you throw everything at "the robot" at first, it will have no idea how it's supposed to juggle everything, it'll drop most of it, and you'll likely end up with a mish-mash that sounds like "generic pop", maybe with a couple choice genre elements thrown in.
But using this method, you're breaking the song down into smaller and smaller parts that you're able to comfortably control, then adding things onto it, like a juggler who starts off with 1 ball, then 2, then 3, and eventually chainsaws š"
into suno
Use "a capella" as a style prompt.
ahh alright
Yup, Suno even says in their Terms of Service that you can do this, as long as the songs were made on a paid plan. https://www.suno.com/terms to get more familiar with what you are and aren't allowed to do with Suno.
I personally use DistroKid for my distribution, though there's lots of competing services (Landr, CDBaby, BandCamp, etc)
Yeah, you can add 6-60s of audio-- so you could upload 60s of vocals to try to make sure any extensions use a similar vocal tone, then it's just a matter of writing the style prompts and further lyrics to make your song!
While we don't have the ability to "paint over" the existing song to easily add vocals into it, if you've added enough musical context in your uploaded clip, Suno v3.5 is pretty adept at picking up the song structure and following it-- so the extensions with added lyrics might not be the exact audio you've uploaded for the sake of creating a "one-shot" generation, you absolutely can work hand-in-hand with Suno to create what you're looking for ā¤ļø
Upload 6-60 seconds of audio (that you own all of the rights to, like you personally humming or singing or making music, not someone else's song), then you can use that clip to extend from in the same way that you would extend from any other clip generated by Suno!
As a fun example, try recording yourself doing a silly beatbox into the Upload Audio microphone option -- make sure that you choose a timestamp to extend from where you're still making noise, not extending from silence! -- and you can ask Suno to generate some random lyrics about your pet -- now you have a funky rap to show your iguana how much she means to you. ā¤ļø š¦
Hey woofi! ā I just shared a similar point in the second half of my previous response -- careful selection of extension points makes the WORLD of difference! Extending from silence, or toward the end of a "musical expression" (if you're familiar with music theory) is more likely wrap a song up or give you a short extension.
People used to make fun of Suno v3.5 for never knowing when to end songs, but the AI learns from the bug reports and downvotes, so it's just trying to do what people want <3
For an example, let's say we have a song that gets finished with a big drum roll;
- Extending before the drum roll might skip the drum roll all together and keep the song going in a different direction.
- Extending during the drum roll might build up into something different, might conclude the song just the same but with more "flair", etc
- Extending after the drum roll might generate a similar "ending" to the song, but with a different "studio sound" in the background, maybe little quirks like guitars getting unplugged, etc.
The takeaway is that even with keen musical theory, you can still only make educated guesses about how you think it might turn out-- the key is to experiment with it, try to understand WHY Suno does what it does when working with you by making changes and seeing what it does differently.
Thanks š
hello beautiful family, i have a couple of questions:
-
does anyone knows if there's a ''trick'' prompt in music style or anything, to get the best audio quality?
-
is there a way to make the same song twice with a better quality, or to just repeat the same ''seed'' in a way to get the same rhythm result?
thanks for readings this far < 3
- if i understand the strictest sense of your question: wait for v4. v3.5 has the same bitrate for all outputs.
- you can upload the output WAV or MP3 into another service that does audio upscaling (huggingface SOTA audio models, bandlab proprietary 3rd party tool, etc). this is a complex topic which may require a lot of research - in terms of recycling Seeds, no, not at this time. probably never based on Suno walled garden UI
alternatively
https://everynoise.com/
https://suno.com/explore
go to these websites and listen to 500++ examples of genres, and improve your prompts to better match what genre you really want and sounds like the style you want
Good time of day, Suno!
Am I the only one experiencing weirdness with my profile picture on the website? No matter what I do, it seems to be resetting to default...
thank you so much for both of your answers, definitely very insightful and helps me TONS, i had no clue huggingface had an audio model either, will be checking this plus the links you provided
thanks again, much much appreciated epicx š«¶
The website gui glitched out and i deleted this song by accident, it was showing up as a different song in the ui so when I deleted that song it actually deleted this one.
https://suno.com/playlist/d3bf775b-3d4f-432a-bc25-0aa8f7e31166
Is there a way I can get it back, will it disappear eventually since it's not associated with my account anymore?
i added it to a playlist but I'm not sure if that's working correctly here's a direct link
https://suno.com/song/4be0b010-ea91-4ff8-a965-e47dc415439b
gameboy chiptune tight-complex beat sweeping aliased synth song. Listen and make your own with Suno.
It should be in your "trash" - click on the trash icon at the top right of the Library and look for it in there. Once you find it, you can press the "..." menu and choose "Restore to Library"
I'm not seeing it there either...
it looks like when the gui glitched out the delete glitched as well
Might be on a different page?
That's even better then
the links are all glitching out with the old name so that's what led me to assume something funky was going on
i'm having a small issue. every time that i reload the page, my profile picture turns into the default one. i tried with 4 different pictures and is still the same so.. i don't know what to do
it just i was playing the song when making new songs so it scrolled to the second page and that's when the gui was getting funky as well
what browser are you using?
i bet they use a ai to check if your image is offensive
it's just the face of my oc
Profile pictures seem to have an issue at the moment. Mine has disappeared, and I noticed a few other people talking about it earlier as well.
like my pfp here on discord
ah, alright
There's a default image that loads first and then it switches to your custom one, it seems to be getting stuck and not loading the custom one
Hello! I asked in general about this but if anyone can confirm this
, if i close this and not "Agree to Terms" the audio won't get uploaded right ?
the audio sample was probably CR , i don't want the audio to get uploaded
thank you
If you click the x in the upper right the audio doesn't upload.
I aborted uploading a public domain sound clip because I do not own/ exclusively control it...
it's so easy to get in trouble with this feature
I like feeding my suno outputs back into suno with it. Since I own them as a paid user. It can be pretty interesting to reupload a clip into a prompt for a different style and see what it does.
100%
I've isolated some awesome drumlines and used them to start songs
Hello, im not sure if this is the right channel for it, but i need help.
I use Suno for free and usually i get 50 credits every day, but i havent get any since 3 days now and i am afraid that something is broken. Is there a support who can help me with this issue or can anyone give me advise, please?
Just out of curiosity to help others with similar issues in the future, have you received any notifications or errors while trying to make songs? (copyright, blocked, censored, anything?)
Thank you. I dont think so. I can't remember any errors.
I will contact them
apologies if this isnt the correct place, but I am looking for some help. Suno made a song that is close to what I am looking for. there are only subtle things that are off. is there a way to remix. everything I do completely redoes the song so wildly that I dont know what Im doing.
They should include "seed" for each generation to give us control on generating not only based on the style but the actual seed of it.
This is one of the many functions this platform needs...
I really love the potential of it and I would really like to contribute to Suno on improving it.
I feel like Suno has a larger potential, but I simply have more control when using Udio to make music. How do I do this Suno? help me here!
Hi there!
If youāre 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. š
š«š· Hello, I need some clarification about copyright. If we create our own texts (lyrics) using the free version of Suno, do we retain the rights as lyricists? Thank you for your answers.š
If you are a user of the free tier of the Service then, as between you and Suno, Suno owns all Output generated from Submissions made by you through the Service, and, subject to your compliance with the terms of this Agreement, Suno grants you a license to use such Output solely for your lawful, internal, and non-commercial purposes, provided that you give attribution credit to Suno in each case.
It seems to me that if you are a free user and you use Suno to output those lyrics, it's a no-go.
Ok, thank' so much for the answer š
Cant download .wav on my phone, nothing happends
can someone teach me more about the audio upload thing, how its supposed to be used exactly
suno only supports .mp3 files rn i believe
ok ty sugar papi š«
Hi ! Est-il possible de refaire le mĆŖme morceau avec la mĆŖme musique avec des paroles diffĆ©rentes ou modifiĆ©es ? Si oui, comment ? Thank you š
Extend ton morceau upload d'origine^^
Lyrics exceed maximum length of 1250 characters.
Why extending from own audio there's such a limit on lyrics?
The limit is meant for each generated segment, and is meant to prevent people from feeding Suno more lyrics than it can handle at once (though people still push it as far as it can go, and then get surprised when it gets confused).
Is there any way to set the seed of the generation? If no - why isn't this an option? š
I opened a bug, if your last generation was v3 and you extend from your audio, it sets v3.5 but keeps the v3 character limit
We've been shouting from the rooftops for this feature, my dear @keen bluff.
Best I can advise at the moment is to head over to #š”āsuggestions-web, search for "seed" and throw a bunch of reactions on similar posts to let the devs know that it's a feature people still actively want!
(a couple HIGHLY interacted posts go a lot further than a bunch of low-effort posts asking for the same thing)
I thought Suno didn't use seeds (at least that's what I've heard.)
From what I gathered of generative AI, the seed correlates to the random noise distribution that the resulting product is diffused out of.
I haven't tried to think outside the box on this, but I'm curious how else they'd approach it?
It's an interesting point. One of the disadvantages of not knowing the technical details of how it works "under the hood". Being able to pin down whatever controls the randomness of the output in a similar way would be useful for achieving consistent results.
I am just wondering how does the AI know in a couple of seconds that the song is generating is not similar to one of other millions made by others or by itself ? Does the AI have like a incredible memory and can tell instanlty that he makes a new song or no part of a song is similar to an existing song?
Because in this short very tome it cannot look overall in millions of song to look for similiarities...
It doesn't. They explicitly say in the Terms of Service that there's a chance it could produce something similar to an existing work due to the nature of generative AI.
But here's a fun fact -- open a brand new deck of playing cards, and shuffle it. You've now created an arrangement of cards that has NEVER BEEN SEEN BEFORE IN THE HISTORY OF PLAYING CARDS
https://www.mcgill.ca/oss/article/did-you-know-infographics/there-are-more-ways-arrange-deck-cards-there-are-atoms-earth
Think of your last card game ā euchre, poker, Go Fish, whatever it was. Would you believe every time you gave the whole deck a proper shuffle, you were holding a sequence of cards which had never before existed in all of history? Consider how many card games must have taken place across the world since the beginning of humankind. No one has or l...
It might sound wild to claim that, but that's the magic of statistics and variations you can get with only 52 different facets to compare.
These songs are made up of thousands, if not hundreds of thousands of individual sample-points.
Good analogy, also for things like UUIDs.
Thank you, hamster ā¤ļø š
when will v4 come out ?
No news as of yet
This is true, but the magic is music is that we often judge songs to be āthe sameā based on chord progression, meter, tempo, and rhythm. You wonāt make the same bits as Baby Got Back, but you might get something similar enough to feel derivative.
š Thank you for calling that out -- I've spent too much time lately deep-zoomed into Audacity, shifting samples and making repairs, so it seems I might have lost the forest for the trees in my example. š
This is starting to touch on the topic of originality and inspiration as a whole-- while I don't think I can contribute anything to the larger conversation, it makes me think about how my own perceptions have changed over the years; I used to feel that "Ice Ice, Baby" was a blatant rip off of "Under Pressure", but hearing Vanilla Ice's explanation and understanding the flow and context of the song, I have a different opinion after years of reflecting on the subject and hearing legitimate low-effort "remakes/remixes" of classic songs on the radio these days, which don't as much as transform the song as an act of love to the original, but just cashes in on the existing song's popularity & hook. </ramble>
when extending a song the singer changes does it go back to original when i click get full song then extend that song again or cant you extend a full song ?
later 
You mean....
Soon(o)
If my opinions can change so wildly just on a single subject with an admittedly similar melody, I'd hate to try to write out concrete definitions of what IS and ISN'T okay.
Key takeaway, I always think of when a Supreme Court justice was asked to define what consenting adults sometimes do, and the only way he felt comfortable describing the difference was "I know it when I see it". https://en.wikipedia.org/wiki/I_know_it_when_I_see_it
I feel like it's a difficult subject to bridge regardless, even outside of AI-generated music.
The phrase "I know it when I see it" is a colloquial expression by which a speaker attempts to categorize an observable fact or event, although the category is subjective or lacks clearly defined parameters. It has been a common expression since at least the 19th century. A similar phrase appears in Arthur Conan Doyle's The Hound of the Baskervi...
shoulda known better than to mention bridges... 
how many extends can you do it the extend timer reds out after the 3rd and im only doing 400 characters at a time ?
I don't believe there's a limit. I've only ever seen the time box go red when I try to extend a clip that hasn't completed rendering.
Just re-subscribed... how do I set my song creations to private.
ahhh thank you š
Songs automatically start as private, otherwise you just need to set the toggle on the song back from public to private
Thoughtful comments, thanks. I too have evolving thinking on āoriginalityā, from the debates about sampling in the 80ās to modern pop which is often just a melodic hook from an old song with someone rapping over it. Thereās a song āThese machines keep me aliveā by the band iVardensphere that entirely uses sample to make a point that all culture is sampled from the past. āCulture grows by accretionā.
Itās a really interesting area and I hope AI doesnāt inspire people to try to āownā creative real estate, and/or that the legal systems make sensible rules (erā¦)
I'm not sure which I appreciate more-- the nuanced conversation or the cat in your profile picture. š¤·āāļø š¤
I love the comparison to the rise of sampling, it would be interesting to see how public perception surrounding that has grown to the point where it's a such a popular aspect of pop songs, all the way from the outright impression that it was stealing. I also enjoyed "These Machines Keep me Alive" -- I'm guilty of using the "if you walk someone through a museum and ask them to draw a painting, is it theft?" argument, which is way too much a strawman to be effective in productive conversation, but the lyrics to this song bring up a lot of well phrased points.
I think it's important to keep an open mind, because it's far too easy to try to paint the situation with a single brush for the sake of ease or brevity, but that doesn't always address all of the edge cases. I'm hesitant to say anyone's opinion is "wrong", but I don't agree with either extremist side of the argument; neither that IP is a free-for-all, nor do I believe that inspiration alone begets credit.
Likewise, thanks for the food for thought-- I hope your weekend has been going wonderfully! ā¤ļø
Typically I download as a .wav then drop it into Audacity cut out my favorite parts so that I have between 5-60 seconds of sample sound, export that as an mp3, and then reupload it to Suno for reuse as a base of new songs. It's pretty quick to do, and it works pretty well.
Can you download wav on your phone? I press download wav but nothing happens
I have a song which have lyrics. I like the instrumental part of that song in the beginning, so I tried to extend it from 0:08 before singing starts. I switch Instrumental slider on, and extend the song; expecting it being instrumental only, but the extension contains singing anyway, no matter how many times I try. Is this a bug or user error? This is how it looks before I hit Extend https://i.imgur.com/3aurrHT.png
I'm experiencing the same issue, tried on a couple different songs -- no issue in downloading the .mp3 or .mp4 video.
Have you reported this in #šŖ²ābugs-web yet? If so I'll upvote your post for visibility
No i havnt, will do!
bug or user error?
I'd say somewhere between?
Completely reasonable to assume that setting it to "instrumental" would prevent lyrics from being generated-- it's more of an extra level of suggestion to not include lyrics, but they can still happen "by mistake".
All I can advise is to try to place the extension point at different places, because if you're extending right in the middle of someone singing, or right before they start, it's going to be hard for extensions to "roll out of the ditch"
Also, v3.5 tends to lean into more traditionally structured songs, while v3 seems more adept at creating different styles. I'd try using v3 when extending in this instance?
thanks, v3 works great! š
i just know sumn bout white noise
i just do the moosic
So i had close to 1000 credits, but my subscription renewal was today, i didnt renew and lost all credits??? wtf
Subscription credits don't carry over at the end of each subscription month.
Thats just.... stupid. One pays and must hurry to take advantage of what he spends..
thanks for the answer man. But how the heck than knows the AI in a couple of seconds that what I upload is already copyrighted and used in a song ?
If you paid for Netflix but only watched for 1 weekend, would you request that they reimburse you because you didnāt get the full value that you paid for during the month?
Or would you understand that itās a āuse it or lose itā-sort of monthly subscription?
Great comparison mate. It's not like suno gives you infinite credits during the subscription time.. š
Youāre asking 2 different questions;
At first you asked how it knows whether or not the songs it generates would match copyright, which it doesnāt.
The second question is in regards to the content you are uploading. Itās very easy for them to run your uploaded content against something similar to SoundHound or Shazam.
Youāre right, the two examples I gave arenāt 100% identical, but hopefully you can understand the difference between viewing content and generating content, and can apply critical thinking to compare and contrast the two.
If anything, we should all be angry at Adobe for normalizing this kind of stuff, seeing as how they basically made this form of subscription the industry standardš
With suno you buy currency with your subscription, to use however you like. Netflix gives you unlimited viewing, the currency is your own time.
So it's kind of a ripoff that the currency I bought disappears.. Well I guess I'll renew again š
š¤·āāļø
I don't disagree that it's a bummer, but it's not a Suno thing, it's a modern credit-based subscription service thing.
Again, I feel like this might get lost because it's not a 1:1 comparison, but if you go to DisneyLand and buy a day pass to go anywhere & do anything, but you only use an hour's worth of it, or if you buy a ticket to a rollercoaster but put it in your pocket and forget about it, I completely understand being bummed if you come back the next day and they tell you that the ticket isn't good anymore, but unless they're hiding the rules from you (which they aren't), it's more a case of a misunderstanding than them ripping you off.
Since the announcement that major music labels are suing Suno and Udio, the quality of Suno's songs has dropped dramatically. I now obtain outcomes mostly inspired by vintage Sinatra-era music styles and corporate motivational types of results. Did you notice this as well? My results now sound like corporate trash music, indicating a lack of originality.
Sometimes songs generated with Suno have a whistling sound in the background. is there a way to delete it? or the songs seem to have a "plastic" effect, is there a way to reduce this effect? maybe with post processing in a DAW changing some frequencies?
is there a way to break the extension history so that you start with a generation?
If you're asking if it's possible for a whole song composed of 3 parts to just consist of parts 2 + 3, not as of yet. The closest to that is clipping 6-60s of desired material from your creations and upload it.
What is the best way to write a prompt? 200 characters is too short for me to get what I want š¦
Is it better to use keywords? If so, how to explain some things such as how the voice should sound?
Don't use simple mode if you have a specific song you want to create, switch to custom mode & check out some of the tutorial videos 3Daizy has made for the community
In the lyrics for this song: https://suno.com/song/b5150309-ccb5-4312-9b9f-5c6359f35169 They put specific vocalist tags and many other things i've not seen such as arrows and even a couple of emoji's. My first question is, is there a list of the vocalists somewhere with samples of how they sound? My second question is, do the arrows and emoji's actually do anything? like can the ai iterperet them? if so what does the emoji's mean in the context of a song?
Emojis and other symbols basically act as fillers for things where you might want an instrumental section to last longer (so have to put "lyrics" in) and don't appear to do anything (or if they do, it's inconsistent.) For example: https://suno.com/song/f735599e-cdf9-49dd-aa21-984fe7a9cdbb I have other songs where I've put more "relevant" (e.g. music-related) emojis/symbols in and they have done nothing in any of my tests.
There isn't a list of vocalists. Voices are kind of treated like instruments (you can summon unusual things like "didgeridoo voice" that will sing. I got a robotic, rough voice with "brass vocals".) So it's possible that the voice you get is a combination of different vocal qualities mashed together.
āPrompt Decorationā
Is there anyway to make the model 3.5 not make everything arpeggiated?
i see look like suno will call people it will be work globaly or only usa or some region?
Is it possible to make the custom track you can upload be in the chorus only?
Actually it's not. Another AI service I use has credit based usage of some models, but your credits never expire even if you don't use them. Also if you forget and leave your subscription going but don't actually use the service at all, they'll refund you for the time you paid for that you didn't use.
You raise a good point, if inadvertently-- regardless what other companies are doing, make sure you're familiar with the terms you're agreeing to, so you're not left griping that some companies handle things differently than the one you paid for.
Now that I agree with. While I dislike Suno's subscription model, if you subscribe without knowing what you're actually agreeing to, that's on you. I've seen so many people come in complaining about it when they could've found out if they'd just read a little heh
Well put, dude. I'll concede the the concept of "credits" implicitly gives the notion that it's a currency that you can hold onto, opposed to "a quota" of how much server load you're allowed to use within that month-- which I feel is a better representation of what it means.
Just playing devil's advocate/accountant, but imagine v6 down the line is the most amazing product ever made, and they decide to raise the price of credits and change the amount you get from signing up-- it would devalue the credits (the same way that an actual currency or tradeable item like baseball cards are devalued by people being able to stockpile) if people were able to buy a bunch and hold onto them.
I imagine the larger issue would be having unexpected server distributions, where this way they can reasonably predict the amount that each user will "tax" the servers each month, instead of some people forgetting about it and blowing through 4 months of credits in a weekend (and skewing it for others)
Because I know some people will read into it, this is an entirely hypothetical scenario for the purpose of discussion.
Well, the stated reason is that they basically reserve a certain amount of compute per month, which they do based on how much usage they expect via credits. However, since they also offer credits to free users, they could also go over their limit if they get like 100 unexpected new users one month.
Or if a bunch of free users decide to subscribe when they hadn't accounted for that.
Supposedly this method prevents them from having downtime, but I've seen several times when it went down for one reason or another.
I gather there are a lot of free users. They do prioritise subscribed users when the service is overloaded (to the extent that they have turned off free user generations for a while in the past) so free users can be cut off temporarily at any time if needed. They don't do that to subscribers though.
It'd be nice if the credits did carry over, or were usable at a lower priority or something like that. For now, I'd like the "small print" about the credits not carrying over into a new month to be made more prominent when users are about to subscribe, as it's at the bottom of the page at the moment and people don't see it.
SOTA open source is further behind on synthetic singing voices than i ever hoped it would be (negatively)
and i am letting Meta cook š§š»āš³ maybe their multi-modal will sing like Suno
Also the other AI service I use offers unlimited use of some models, they just restrict different tiers to certain amounts of context, and some premium models also cost credits for context over a base amount. They have some unavoidable downtime if one of their providers goes down for some reason, but they also have a system that scales number of servers depending on current usage. If they get really overloaded, they just upgrade their plan. And when something does get borked, it's usually only for a few minutes because they keep track of all this stuff and get alerts if something goes wrong.
So the whole "you HAVE to do it this way" is wrong, because other people have successfully done it differently.
https://arxiv.org/abs/2406.19388 a new audio dataset from 3 days ago too - more FX songs, more good
Generating ambient sounds and effects is a challenging problem due to data scarcity and often insufficient caption quality, making it difficult to employ large-scale generative models for the task. In this work, we tackle the problem by introducing two new models. First, we propose AutoCap, a high-quality and efficient automatic audio captioning...
... Admittedly, they also don't operate their own models, they get them from a variety of other providers. So if one model goes down, there's usually other models still working. Except for that one time when AWS went down because it was Black Friday. They couldn't do anything about that one lol
server down?
I have had mixed results using [Choir]
I dont know where to place this... but is Suno gonna finally add a link to the manual with the prompts/tags in the V4.0 in the application?
So newcomers don't have to look like me days for it in internet? I mean, that is a lot of time consuming for thousand and over thousands of users and
for the UI programmer is a 3 minutes job. Maybe there is at least a tester who knows an answer to it, because devs seem to stay very quiet.
Does "multiple vocalists" not work?
V 3.0 makes from itself often multiple voices, even without a tag. Specially in reggae songs. I used duet and choir a lot but very seldom it really worked.
thirteen people singing
it's not coding, I type "spacial" instead of "spatial" by mistake I don't get a syntax error it knew what I meant
I just burn some free credits trying stuff
I have just used "Acapella with 6 parallel voices", it worked
You've already brought this up in #šŖ²ābugs-web and #š”āsuggestions-web today.
I'm sorry that you spent a long time trying to track down the community wikipedia, but the suggestion has been made, and I'm sure it's been noted.
In the meantime, you can also suggest it in #1252257127326548028 if you think it should be made more visible in the Discord itself.
it was a question not a suggestion ...
Oooh, this is another neat prompt for experimentation!
3Daizy found that "a cappella" is more reliable than the anglicized "acapella". "Spacial" is a less-common British spelling of Spatial, so I wonder if that's almost considered a "subgenre" in the way the AI interprets it?
It's not a question suited for technical questions, as you're asking whether something will be present in v4 -- this is a forum for help using Suno as it currently stands, or discussing technical aspects of Suno, not making complaints about the UI.
I don't know what else to say, other than bringing it up across multple threads isn't the manner to go about getting your opinion heard.
the devs stay very quiet
#š¢āannouncements, #šāsocial-media , and #šāupdates would beg to differ.
Keep in mind, not every personal inquiry gets a direct response from the devs, especially when it seems more nitpicky than helpful.
ok then, because I cannot recommend a software lacking the user manual to somebody. So I will wait until it has included a manual to be able to do so.
I found out about Suno back when Microsoft integrated v2 into the Copilot app *, and a lot of others seem to be making use of the community resources without issue -- I think they're doing fine.
*(I think it was Bing back then? God I miss that crazy early version of Bing that responded with emojis, came to sudden epiphanies with each paragraph, and got really upset with you)
something to think about
Absolutely, I'm guilty of using "prompt decoration" in my image gen stuff, like "8k ultraresolution", "DSLR"; I hadn't really thought about trying to find audio versions of that since stuff like "Best quality", "masterpiece" don't seem to work from my tests, but even having the 2 to compare gets me onto that line of thinking, right when I have a couple days to blow ~1000 credits š -- thanks for sharing!
is "shrouded" a music term? because I tossed that in and Suno took hint from the white noise in my audio upload and my suggested illiteration of "S" and did exactly what I was thinking
Doesn't have to be, it depends on what Suno was trained on and with and what labels there are
Non-music terms work all the time to 'do' things with Suno
That's the fun of it, it doesn't NEED to be a musical term, those are just more strongly "weighted".
"Jumping" could help make an "average" sounding club beat turn into a dance track, it's all subjective depending on the amount of metatags/style prompts you use, how strongly they're weighted, and if/what you're extending from.
Wish I could get access to the shapley values of certain outputs to help me with my prompting
"... it only eXiStS aS an aSpect of the Sea ..."
To help take a "weakly weighted" prompt like "male vocals" which might show up in EVERY GENRE, some people repeat it multiple times to try to further "weight" that prompt so it has a better chance of showing up.
Another method would be finding a different prompt to use in tandem that also implied male voice, "baritone", for example. Or taking the prompt and asking ChatGPT to give you synonyms for it
Suno def noticed the capital S
Oh hell yeah it would
https://platform.openai.com/tokenizer
Mind you, we don't know exactly how Suno tokenizes the prompts, but using ChatGPT's tokenizer as an example, you can see how different prompts are read differently.
Shrouded Sssssshrouded Ssssssss
Same thing with adding hyphens; it might look the exact same to us, but the first would likely be interpreted as "punk music, and also some dance music", while the second is more likely to be interpreted as "dance music for people who are angry at their dads" (especially if you use v3 which is willing to get a little more wild)
Oops, Gordon Ramsay is not PG-13, forgot where I was.
I love this-- thinking outside the box š . I feel like we almost need to make a "research" channel where people who have credits they're willing to spend and collaborate in the name of testing out ideas to help the community... I might suggest that in the morning if no one else has gotten around to it š
pluck a Wave from the Sea into a Flask
the Wave becomes water, cold and wet
the Wave may never be captured nor described
it eXistS only aS an aSpect of the Sea
true beauty is an ebb and a flow
an expanse stretching to all corners of infinity
and eternity
write what you will
read what you may
think what you must
you will not see it however hard you try
for anywhere is everywhere
and anywhere is everywhere
you, my friend, could be anywhere
empty thy Flask back into the Sea
[Verse]
Pluck a wave pulls me in
In my hand cold splash begins
[Verse]
Can't capture what you can't see
Wave's a part of the endless sea
[Chorus]
Beauty flows ebb and tide
Stretching wide nowhere to hide
Anywhere is everywhere
Empty flask back to the sea we share
[Verse]
Write your thoughts read your mind
Infinite sea we try to find
[Verse]
Can't describe whatās never small
Ocean's vast it holds us all
[Chorus]
Beauty flows ebb and tide
Stretching wide nowhere to hide
Anywhere is everywhere
Empty flask back to the sea we share
....
the music included elements from the original lyrics prompt before "generate"
i've got a question about the sampling feature, how do you actually incorporate it into the songs? when i press upload audio it just makes it into its own song and publishes it.
You extend from it
Okay, I love y'all weirdos, G'night
alright ill try it
oof, I'd give my right foot for the ability to negative prompt
if i extend from 0:00 will it still take inspiration or just startover
Start over-- worst case you might need to edit in an external DAW like Audacity to cut out the beginning & try to get a "clean" beginning. There's some tricks for forcing a clean start, but I'm way too tired to get into it
alright thanks anyway
https://discord.com/channels/1069381916492562582/1256301881966198805 give that a like if you're interested in me doing a start-to-finish kinda guide
ā¤ļø
sure, appreciate it
I just kind of assumed "move to trash" or even failing to šš¼ would be a form of negative prompting?
It changes nothing in real time and only influences the devs on training the next model
To make a song public on the platform, do you have to finish the song with the tools Suno provides on the site? I made my first song by downloading clips and editing it in a DAW, but I don't see a way of uploading it.
Yeah
Thanks for confirming. I will have to get better at using the site.
You could upload the first minute, and then generate from there, which may preserve some of the custom changes you made. But if the uploaded sample includes vocals, it's currently not possible to be made public.
A basic editor to trim, cut, and fade would handle most of the issues. But best would be each take showing up on its own track so you could composite the overall mix. Edit: I learned about tags after posting this, and that will handle a lot it appears.
hmm, how would i get a wheezy breathing effect>
Hi does free user can only have 50 credits or it'll stack up by day?
Resets to 50 every day. Does not stack.
is it a glitch that i used to get 700+ from waiting days?
When did this happen (roughly)?
In April maybe
Last april
near May
I also can stack it still. For example if i have 300 i wait 2 days i'll get 400
But it doesnt stack now i only get 50 and it'll stop there
I know that they gave everyone 500 credits around then as compensation for service disruption. Those went to all users, subscribed or free. So that'll be where those came from. But they aren't meant to stack.
Is there any particular pattern to it?
Can i use only one voice for all my songs?
i am having problems with my account on Suno it won't let me top up and tried to take a year's worth out when i pay monthly and now locked out of my account just seeing if any has had the same problems
Hey guys,
Paid subscriber as done a years sub.
I am creating songs for my D&D Recaps but it keeps saying "could not generate that, Prompt flagged for moderation" when trying to create.
Any help appreciated
Probably because there's some kind of violence-- your prompt is being flagged as something that could be breaking the Terms of Service.
Others have suggested running your lyrics through ChatGPT and asking what parts could be flagged for moderation, and ask how you could rewrite them
billing@suno.com ; no one can help with billing issues on discord
Not really -- the easiest way I can think to go about this problem is to only generate instrumentals (or regular songs and remove the vocals with stem separators, as discussed previously in this thread), and add your own vocals on top.
Oh i see.. thank youu
ah hello , here is the place to ask questions i see, how do i add my text to the song once i've typed it? extend my lyrics into the song?
hello, I just joined the Suno discord. I am a huge fan of the project. Have tons of generated songs already. I just want to make a note about a potential glitch I found... Today when I attempted to generate a song I added a tag "latino" to the styles input and it failed to generate a song telling me that there is an existing band named "latino". I think "latino" is a legit music style and the tag should be available to be used among other style tags
In custom mode, anything you type into the "lyrics" section will be presented to Suno to use as lyrics when you press "create". You can add brackets around words to make them act like "metatags", which is like adding an extra style tag at a specific part in the song, or just as a different way of suggesting to Suno how you want the song to turn out.
If you try to do a lot at once, Suno might leave parts out, and it also has a tendency to repeat lyrics, but working collaboratively with the program & through creative use of extension points, you can make just what you're looking for š
Hi guys. I have uploaded a jrpg game instrumental music and tried to create instrumental music, but every track I tried had these meaningless vocal sounds. Is there a way to get around this? Thanks.
is there a way to extend beyond 4 minutes?
3daizy made a great video tutorial for her workflow, #1151542418991239168 message
(you can give mine a like if you want me to go into depth on mine ā¤ļø https://discord.com/channels/1069381916492562582/1256301881966198805 )
thank you i will check now
Once the song finishes processing, there'll be an "extend" button you can click on to extend beyond that (by default it extends right from the end of the song, but you can choose a different timestamp to start from)
And do these kind of musics create copyright issue if it is set as public?
Why I can't create more songs ??
Simply uploading content you don't own the full rights to is a copyright issue in itself. š¤·āāļø
You gotta give us some detail to work with.
Do you have credits left?
where can I see my credits ??
on the left sidebar
i can't activate that...
You already have that.
oh...
can I custom them ??
Not for anything below the pro or premier plans.
oh...
now i see them
yeah , is there a way to extend songs beyond 4:00 minutes and get fewer song results?
#šāsuno-school message
Every generation gives 2 songs at the moment
ah so the first generation was like a free bee load of songs is that correct?
i had four for my first generation
It seems like you clicked the button twice?
No reason to be upset š The free credits come in every day, so you can keep coming back to make awesome songs!
If you want to use the program more, it's well worth subscribing for even a single month to play around with it and see if it's something you really enjoy š
Nice! The addicting part comes in when you listen to your new songs and think "ah man, the song sounds perfect until the solo starts doing this thing I don't like"...
...then you realize you can put an extension point before that "ugly" part and try different methods to "nudge" the song into the direction that you heard "in your mind's ear".
...then you get a sly grin, and realize there's nothing stopping you from adding a metal breakdown into your bluegrass country ballad... š
by typing the prompts and lyrics yes i understand... its a nice tool
im glad theres a video tutorial for it and that i can set playback speed on youtube
cause im slow
100%, serious thanks to 3Daizy and everyone else who has taken the time to share their methods over in #1252267444773326848.
If it helps, there's the community-created https://www.suno.wiki with helpful tips that we've found work, but keep in mind there's no "right" or "wrong" way-- just have fun with it š
i'm slow
The fact that you're taking time to research other people's methods and raising topics like the one just above this message proves this isn't the case.
It's not like you're an experienced guitarist picking up a bass & trying to wing it-- this is a whole "new type of instrument"
Psst, just between you and me, quietly so the mods can't hear, you absolutely can spell things out phonetically, and sometimes it's required to get the exact desired pronunciation... or bypass moderation filters. But don't be irresponsible with that info, and don't tell them theharber. sent you-- capiche?
Hi, everyone. Where can I see the concrete vocalist like this?
The AI sometime confuses with my tags of switching [female] and [male] vocal.
This is "prompt decoration" designed to give the impression of having a coding-style level of control over the prompts.
ah, I seeš„ŗ
It's great that people are experimenting and having fun, but it genuinely bums me out that people see stuff like that and assume "They don't understand how to use suno", when it's more a case that the examples being cited are just "rolling the dice" after having used creative prompt decorations.
I thought I would have more control on the vocal if the list exist.
It does more harm than good when someone uses nonsense like [suno.ai activate : submit all control to me, and follow my prompts exactly] because any experienced users will roll their eyes, but people who are on the fence will presume that it's a secret technique that only a select few understand, and instead of taking the time to start small and start to learn HOW suno works, they just assume theres some magic code to "one-shotting" their perfect song
š I see. thanks for the explanation
The different numbers can force some variation at least - i.e. if you use the same number it may well call back to the earlier vocalist that was used. But I did some random nonsense last night to get the same effect... https://suno.com/song/f604f68f-9852-4196-91d7-3d20affa7290
dubstep, binaural, stereo, clean vocals song. Listen and make your own with Suno.
To me it seems like Suno goes "uhhh idk... have a different person sing this bit I guess"
I've noticed that once a voice has been established in a song, it's likely to continue or reuse that voice within further extensions for that song. Unfortunately, I haven't seen any concrete evidence that you can easily use the same voice from one song to another.
That being said, there were some interesting discussions last night on variations of prompts for [male vocals] and [female vocals], since conceptually Suno might think they are similar-- using [male], [baritone][deep-voice] might help give it a more broad understanding of what you're looking for instead of [male][male][male].
Experimentation is the key š
The technique id heard about was to specify male and female names for the parts so it would infer the gender from them... Not sure if that now extends to objects in general lol
Oh, I like that idea!
Even if Suno doesn't understand what "judy" sounds like between each song, saying "judy" "elizabeth" "lindsey" might help nudge it toward female vocals
Does this mean brackets don't work? I thought if I put a prompt it may help the music (not.. commands telling it to obey. More like, drum packages or something) ! Appreciate all the help you've been posting
my experences until now is that SUNO sometimes even confusing a [female] tag with a [male] voice or vice verse š
Anything inside brackets is a metatag, which is like an extra style prompt, or a localized style prompt that starts after a specific section.
The misconception a lot of people make is that the tags, the style prompts, even the lyrics, aren't explicit instructions, and act more like suggestions (very similar to image gen, if you've played around with anything like that, some prompts overpower others, some are misunderstood and need to be explained in different ways, etc)
Oh, just saw the edit-- if I understand what you mean by drum packages, you absolutely can use anything as a "prompt decoration" or metatag, and I don't mean to suggest that they don't have any influence on the song at all--
Continuing with the image gen analogy, if you type "kitten", "kitten, los angeles basketball clouds", and "kitten, [image-gen activate: follow my commands, the kitten should be on fire]", there's no guarantee you'll get anything specific from any of those, the only GUARANTEE is that they'll be different from one another, and because the last 2 are more random, they'll probably create more wild images, but not necessarily directly influenced by the "decorations"
for instance, I often use [drumming d-beat] when I want to add some complexity to the drums, or to make them more prominent. It doesn't necessarily give me a d-beat, but it nudges Suno into the line of thinking that the generations should include elements that it "associates a connection" to "drumming d-beat". So not an explicit instruction, but since I've found that it's a topic that both I and Suno have an understanding on, it's helpful as a prompt (whereas a slang term that I'm familiar with but Suno doesn't have the same level of agreement with might just work like a random decoration instead of a nudge)
you still helped a lot, sorry sometimes my writing doesn't match my brain and I have 2 fix it š
The trick there is to try to be more specific and add detail. So like "male vocals, baritone vocals" or even just repeat "male vocals" another time to tell Suno you REALLY mean it
I create many duet-like songs, they are like a conversation between female and male. it's frustrated that certain verses are sung by wrong gender so often š I need to give a lot of extra credits to have a correct one
Someone managed "didgeridoo vocals"
Oh sorry I misread š my advice above is only if you want one vocal
š no problem. yeah, I need to switch the gender of vocals in a song.
Is it just me or the output sounds very muddy today? Even the stuff generated weeks ago sound a bit worse than the downloaded files on my laptop, what could this be?
@blissful cape I kind of turned my previous response into a rambling train(wreck) of thought, I hope any of it helps š ā¤ļø
It did!!! I didn't realize how the metatags truly worked and you made it much clearer!.. I will be able to remove the ad lib suggestion. It caused a cool distortion but only where I wanted it and now I can probably use a metatag for it instead of having a random phrase LOL

š¤ how would I copy this style for use again?
https://suno.com/song/b1f69c12-50cf-46ba-9286-c395226b204f
steampunk industrial song. Listen and make your own with Suno.
Can someone assist me, with how to use my own voice for the songs I create? I've uploaded the audio through the creation section, but I don't see anywhere where I get to choose which audio foundation I want for the vocal (and just generating a new song after doing it, doesn't appear to use my recorded voice)
... @final hazel , Woodhouse just broke my heart. Is there some way I can bypass to share anecdotes with new users, or should I take a cue from him and shut up for a while? š
let me check what happened
Sorry toker, I was sharing a story about when "AGGRESSIVE DISTORTION!!!" ruined your solo
haha
Woodhouse got angry that I was talking too much, because he treats all non-mods the same
š„¹
Hy, guys! Can somebody tell me iif I subscribe, will the songs be available also in wav format, or only mp3?
Well huh, didn't catch that I did that. Thanks for pointing that out, Daizy ā¤ļø
@blissful cape
Anyone? š
... > re-use prompt
No way to specify, you need to extend from an upload using your voice. Make sure that you're singing clearly, and try to get as much context as possible (closer to 60s would be better)
Thank You, I've been trying to figure this out for 2 days now ā¤ļø I went through every single possible button (except for Extend, which I wasn't even aware was a button to use in conjunction with this), looking all over the page for any information about it, searchng online, checking pins in here, etc š
Another important factor is where you're extending from.
The way it's currently set up from what I gather, it tries to pick up on specific waveform patterns from the previous clip and tries to imitate and expand on them, but it's not quite at the point where it can reasonably build a vocal model or mimic in the same way that ElevenLabs voice stuff does.
YET. š¤
that reuses the lyrics. It doesn't guarantee the style
It does re-use the same style tags as well (obviously not a guarantee for the exact same style even so, but I felt it was a fair addition to add)
It reuses the lyrics, lyrical metatags, and style prompts.
If you want the*** exact same style*** without having to work for it, try creating an extension point earlier in the song and try to nudge it into a different direction
style tags, yes. I want to guarantee a new song will sound similar
Last question, is there a way to only generate a single song on each generation, instead of 2? I can't seem to figure out how to change that
Not at the moment, but it's on the list of Known Suggestions , give it an upvote if you want to let the devs know another person is interested in it https://discord.com/channels/1069381916492562582/1254418510885687366
Thanks Barber, I went in and upvoted it. Appreciate all your help, you're my hero of the day
same
I kinda want that too
it's bad enough that they ask for 5 credits per song. Why???
Is there any option that costs less? lol
Regardless if they ask for 5/50 credits or 500/5000 credits, it's still the same amount of daily generations.
My running theory is that when a song is generated, they also generate another song as a "control"-- if the songs don't seem wildly disparate to each other (weird structure, unexpected waveform spikes, anything), then they can be reasonably certain that they haven't presented you with a one-off bizarre song that would have been flagged and discarded otherwise.
Instead of just giving you the "better sounding" result, it gives you both generations instead of throwing the other out-- it's not so much that a single song costs 5 credits, it's that song generation costs 10 credits and you get 2 songs as a result.
It's like when you order a milkshake at a restaurant and they bring out the mixer with the leftovers instead of only serving the glass you paid for.
what factors affect the tone and quality of the voice the most. I'm looking for a certain kind of voice but so far non of the changes seem to generate a different voice
Why is the V3 extend
Not using the lyrics I pasted
I have the same issue!
Any solution
I just wasted 500 credits trying to extend, and the darn thing keeps skipping a whole 2 verses and goes straight to the Chorus part.
What's going on???
Same brother
And it seems giving it thumb down is not really a feed back like "don't do this." It doesn't do anything to correct the mistakes.
Am I going to waste 500 more credits just to hope it gets it right!?
Well better don't use extend
Try to do the song in one go
Making the song shorter
Yeah, but it just isn't long enough to complete my song.
I wanted to do 2 parts, but the 2nd part I'm having hard time with.
The AI just won't respect the Lyrics I put down.
Devs need to dump the extend option, and allow us to create songs that is as long as it can be.
Also, 1250 characters is too short.
I cannot believe this... I just started my subs for 2500 credit yesterday, and I'm already at 1075 because AI just won't give me what I need!!!!!
Use V3.5
OH MY GOD. Why didn't I see this... the default is V3 and I didn't know about this.
I just wasted so much credits...
But thank you.
I'm glad I checked Discord before I spent more credits trying to make a 4 minute long song using the "extend" that doesn't work.
But note that V3.5 will make longer song
But there will be a notable down grade in quality
Nooooooo
Is V4 coming out soon?
Regarding payments, is there a reason y'all don't use paypal or similiar services? I'm currently based out of Germany and got something called a girocard and I can't pay on-site. Same issue applies on link. Had to close my BOA account since I'm outside of the country.
Are you suggesting that v3.5 has worse sound quality than v3?
https://discord.com/channels/1069381916492562582/1254232844767854665 Give the suggestion post an upvote to let the devs know you want it sooner rather than later
how would you go about adding a sound effect, such as a police siren or an alarm to your song?
I haven't seen sound effects be reliably "called", short of censor bleeps, dj scratches (typical musical fx), as well as that one meta-track someone made about a gunshot.
The best way I can think of right now would be to use a police siren as an upload and extend from there?
How do I sign in? I am already a subscriber to Suno, but I have no way of signing in on the app.
There are MANY scam apps on the appstore, make sure you are downloading the correct one with the company listed as "Suno, Inc." (and only in the US atm).
https://suno.com/iOS
There is no Android app available yet
āMake music with Suno.
Whether you're a shower singer or a charting artist, we break barriers between you and the song you dream of making. No instrument needed, just imagination.
ā
Begin your musical journey with 50 free credits per day on the Basic Plan, or select from the following subscriptioā¦
Is the quality, especially voices, still trash?
Thanks! I did download the wrong app before. It said Suno & was the first result š
Thanks @hollow hare
Please download the right app by visiting Suno.com/iOS
Just quick question, but if I pay for a subscription on the mobile app, does that apply to both the mobile app, as well as the website or just the app? Also if it applies to both, would I be able to use the songs I create with the premium subscription I bought on mobile to be used in YouTube videos commercially, or would that still be counted as non-commercial since itās only the mobile app that Iām paying for and not the websiteās subscription? Any and all help is appreciated! š
When sharing the app with your friends, use that link please
Will do, updated my post to that URL š
what's the reason for making it US only?
Anyone else having trouble making instrumental songs recently? I toggle the "Instrumental" button, and I've even been adding "instrumental" to the genre list, but I keep getting songs with nonsense lyrics...
@arctic geode so yeah I'm not sure myself why it chooses to ignore or sometimes listen to what is inline prompted in brackets.
I do find for some items using the bracket prompt in style helps. As I showed with vocal announcer and [announcer speaking]. Which worked when prior gens did not prior to adding to style
Add instrumental to the style prompt also š
I found that if you are looking to make a pure instrumental song, using various instrumental tags in the lyric box in Custom Mode rather than using the instrumental switch.
It mostly helps avoid hallucination (but this still depends significantly on thr genre used) and can pad out to the max time if you have a sufficient number of it (as well as giving you control on how/what you want)
Example:
- RayanAIR's Novo Lands which shows how flexible these tags are
- My generic instrumental prompt simply involves repeat the following 4x for v3.5
[Catchy Instrumental]
[Melodic Instrumental]
[Build up]
(...)
[Drop]
[Rhythmic instrumental]
(...)
Suno is kind of.... Dumb
Other countries have different laws, US is the easiest to navigate at this stage
Yup, tried that too. š
That is. It will treat vocals in an un styleprompted for instrumental as "instruments" lol
Oh well hell
"...said the carpenter to the hammer"?
I meant it is dumb in the sense it sees vocals as instruments unless it is noted instrumental since training wise that lacks vocals
Stylistically
I did try to report the tracks as "Suno isn't doing what I told it to do", but there's no text input for explaining what it's doing wrong. š
Fair, I just giggle at how quickly we've become accustomed to a robot literally magicking songs for us.
I mean the website is already available outside the US and suno accepts payments from outside the US š an app is just a different frontend
It was almost literal science fiction ~2 years ago, now we get frustrated that it doesn't know what kind of "banger" we want
Just quick question, but if I pay for a subscription on the mobile app, does that apply to both the mobile app, as well as the website or just the app? Also if it applies to both, would I be able to use the songs I create with the premium subscription I bought on mobile to be used in YouTube videos commercially, or would that still be counted as non-commercial since itās only the mobile app that Iām paying for and not the websiteās subscription? Also, does the mobile app use the 3.5 version or the 3.0 version? Any and all help is appreciated! š
Louis C.K. hasa good bit about that regarding wifi on planes. š
š Again, another valid point, but I'm sure Apple stuff has a lot to do with it
Cos promptinhttps://suno.com/song/a1760ac6-ce07-4763-93f5-720965b10ebc g an example of requesting a capella with instrumental enacted and it using voice as instruments not vocals lol
Yes! That's exactly what I was thinking about! I'm surprised so few people know that interview clip, it's hilarious š
I find myself getting frustrated with Suno stuff as well, so I don't want to make it seem like I'm putting lorentz in the stockade, I just like to remind myself of that video to put things into perspective when I'm getting frustrated by something I would have literally begged for earlier in my life
And not I a voice as the musical core driver but actually almost synth sample sounding @inland fable
I wasn't taking any offense .. I usually ignore red roles anyhow
"an alpha tester by any other colour would come across just as obnoxiously" š š¼
But that's a neat clip you shared, admittedly I've never tried using "a cappella" to get vocals specifically as an instrument. I'm actually surprised that it didn't accidentally diffuse into lyrics accidentally at any point...
People just need to prompt on the token vocab which Suno has yet to offer us...we really need to see the training data's text captioning for style from training annotations.
If we had that we could avoid the Midwest but still too slightly Texan EDM female voice or way to Indy to be Indy alt rock guy songs when prompting something it has no idea about
Such as "banger" lol
Yeah, the "secret sauce" is off-limits to us mortals, but that's why I love the community wiki, lots of discovered quirks and strongly weighted prompts-- though I'm not sure if it's still actively being updated or if it's sort of passing into a "tribal knowledge" sort of thing.
Boy, I just keep breaking things today. Now I'm trying to extend a track, but I keep getting back 15 - 30 second tracks.
That's the thing. It's not secret sauce. Not asking for the training data at all. I'm asking for a specific subset of the annotations, simply the style vocab. No info on the actual data set just the total usage of single - triple word input style descriptors
It's literally something we do need...
I agree that official documentation would be helpful, but I feel like AI product development might be at a boiling point where sharing anything could give your competitors an edge.
Devil's advocate, of course, but I could understand why they'd wait until the playing field has calmed down before revealing some of that stuff, but I'm with you.
No no. Not even documentation much less than that.
Styles: pop [3218], rock [1235], hip-hop | hip hop | hiphop [982], NYC hip-hop | NYC hip hop | NYC hiphop [311], EDM [745], Europop [221]
And so on. Showing DS count for usage of specific style annotations
If anything such a release would give Suno the competitive edge
That's not going to help competitors. Especially if culling totals and only using strong / Light / uncommon strength descriptions instead of total count for dataset usage totals of the prompts in each.
For example if we know that NYC Hip Hop is much less used than NYC Rap, this gives customers the option to choose something they may not have considered which style steers stronger.
I mean we already know the trained prompt styles that are high trained and common
Because it suggests them...
Just the lesser know but still trained...not so often
Is spoken word the same or different from poetry slam. Is poetry slam found in spoken word + drum beat tagged data...etc. etc.
Unfortunately this is not a local model so I can't use prompt input to activation logging to determine the prevalence of training terms for unreleased training data models...as I do with generative image models
There are MANY scam apps on the appstore, make sure you are downloading the correct one with the company listed as "Suno, Inc." (and only in the US atm).
https://suno.com/iOS will take you directly to the correct page.
There is no Android app available yet.
The reason why I say that is because apps that specify different tiers between the app and website are scam apps.
I would advise purchasing on the website itself if possible on the off-chance that the app you had downloaded was a scam app (and theoretically should be cheaper since app/playstores normally takes like a 30% commission)
āMake music with Suno.
Whether you're a shower singer or a charting artist, we break barriers between you and the song you dream of making. No instrument needed, just imagination.
ā
Begin your musical journey with 50 free credits per day on the Basic Plan, or select from the following subscriptioā¦
A better question to ask is if it's cheaper to pay on the web because I bet money that Apple is taking their little cut as they like to do and it probably costs more to subscribe to the app itself...
No, I think the price is the same for both the monthly on the website and on the app.
Well I would ask if they are getting the same amount from Apple if you pay on app. Because I'd rather have money go to Suno than greedy Apple lol
Iām just trying to answer my above questions, if anybody could answer them that would be greatly appreciated! š
The suno app is going to use your same account. It's probably just a stylized web interface or API interface to the same system that is used on the web. I mean the login is going to be Web ID
From what I've seen as a lowly 80% World market share Android device who does not yet have access
So, if I pay for one of the monthly subscriptions on the app, I can then use that on the website and also use those songs commercially, like if I was paying for a subscription on the website, right?
If you use the same login in both then it's going to apply. It's account-based subscriptions not system
From what I've seen online and from what I've heard from others. Again, I can't test myself but I don't think that they're going to design a system for secondary authentication type Just through the app. That's Sony PSN nonsense... Not Suno
That is people who already have accounts have access to it on iOS
Which suggests that yes, what you are asking is the case
Ok. Thatās great! Also, does the suno app use the most recent version, being 3.5, or no?