#📚┃suno-school
1 messages · Page 70 of 1
Well if you want to explore the frontiers, I've heard claims that the new Cover feature has actually translated songs from one language into another. For example the song was in English lyrics, the user put "Spanish" in the style prompt, and the Singer sung the same lyrics in Spanish. But because Suno can "hallucinate" kind of nonsense language so well it can sometimes look like a translation, but it could just be a few lucky random word salad words combined with other words that are already similar between languages. I haven't personally test this and am quite curious to see a 100% rock solid proof that a song was translated like this.
I did test this thoroughly in Suno versions before the Cover feature and every translation was a false positive. But Suno is better at "word salad" now, I have noticed, so I think it's at least plausible it could do this, maybe.
akshually ... i had this happen even without using the cover function ... i had one song entirely with german lyrics BUT the song came out in japanese completely, no german in sight anywhere 😄 i have no idea why especially since i did not have anything japanese anywhere, neither in the style prompt or in the lyrics prompt lol
Right, but was the Japanese an accurate translation of the German, that's the key.
Jonathan, do you think that lyrics are a greater determining factor in the song's style than the style box itself?
i dont know, i dont know japanese
In some specific cases the lyrics are very strong. For example Christmas carols. It's not so strong you can't overpower it by repeating your style words in the style prompt though. Like "rap, rap, rap, rap, rap" will probably be rap no matter how much the lyrics look like a happy christmas carol.
You can test your lyrics but using a blank style prompt. Maybe the songs come back 3/4 in the same style, maybe they come back 4 different styles. If that is the case, you know you might need to work a little hard to get a different style.
interesting. I have a song that i reuse in many different styles just to see how suno interprets them. It is not christmas at all. I tried to use a Christmas Music Style as the only style prompt and the output was not christmas at all. Then i filled the box with christmas prompts and it started to sound christmasy
that seems to support what youre saying
You can style prompt with just lyrics as well. Use fake lyrics that summon the style you want (usually very over the top, cliche). Then Extend From that song after the first verse with your real lyrics. Then "Crop" off the cliche style-bait lyrics of the first verse.
This is less necessary since Suno has gotten pretty good with style prompts now. But in the past was more commonly needed.
thats a good suggestion. I am kind of building a library of initial song styles that i can later flesh out into real songs, so thats right up my alley
is there a library of sounds
https://suno.com/song/ee156e2f-40cb-4792-9165-bb94201a2569
i been experimenting with like a chanting sort of thing
Grimdark ambience chanting, tribal grimdark marching, industrial marching, slow grimdark hell chant song. Listen and make your own with Suno.
but not actual lyrics, sounds.
I've used locations or descriptors of locations to cultivate certain sounds that I want, like cavernous or cathedral, make sort of echo-ee? sounds?
In subscriptions, what does "10 running jobs at once" mean?
You always generate two songs at once, so it means you can click Create 5 times before having to wait.
ah ok.
Perhaps someone knows what happened? Yesterday this appeared and I can’t create songs
Having the same issue. Been seven hours now.
A similar problem, as it was laid up at night, it’s still the same
Does anyone know of a way to only display instrumental songs within the various tabs (Library, Explore, Search) ? Especially with Explore and Search, the vast majority of songs have lyrics and it's a lot of playing and pausing before I'm able to find even one without
@jolly gate try contacting support
Email sent
What's the bit rate on the MP3/WAV files you can download?
song generation not work, why??? @Moon_Max
I am in Spain, and since last night at around 00:00h peninsular time, it started to fail and more and more functions are failing.
Why the list of music styles removed from the mobile app???
Also covers help a lot
Jonathan covered this p well; basically, most other languages tokenizer to each letter, and these letter functionally are meaningless. They may have some slight meaning by being used to generate music, so you could find a one off, but in general they don't.
That being said, some other latin based languages probably do work.
This goes for image gen as well, I'll often prompt in french
And English together
Usually 3 English, 1 french, 1 Spanish
If I'm having trouble with a concept
But in general, it may understand those better as theyre likely better repesented in its training data
Which has always been weird considering chinese really should be massive, so idk what the deal is with that
But it's like that atm across most gen AI image or video
And by extension music, but LLMs being the exception, I still can't say for sure that there aren't foreign language prompts that work. I know there must be post processing going on, it's just a black box
And I've given up trying to figure that part out, it's too much guesswork
I don't speak any french but I make a ton of songs in it, it sounds better than english in general.
I think the song lyric gens for english sounds hoaky.
Lyrics is one thing, he's referring to the style understanding
Also yes, but I think a minority use the lyric gen, as gpt has a tendency to make lyrics in a very structured fashion unfortunately
Does Suno have any API? Any way to programmatically create/download songs?
And more important then creating the songs (which is fun through the UI), any way to quickly / programmatically download songs created through Suno? Going one at a time, waiting, choosing location, etc., is a really time waster...
Any chance to just finish a creation session and download all the songs?
https://discord.com/channels/1069381916492562582/1253766525458841610
Neither batch downloads or API access have been announced 🙏
/chrip
Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com
Please...does anyone know how to fix this problem with the SUNO platform?
I have been unable to work on my projects for fifteen hours
😩 😩
It seems a number of people are having this issue, but not all users are impacted -- for instance, I am able to make songs.
Perhaps contact support@suno.com for official advice on the matter, or to know if there is any time they expect things to be resolved. 💪
I've already sent them an email. But I guess someone has already solved this problem.
However, I have seen my email account associated with the Suno account blocked. I have unlocked the account, I have restarted the program, but... it still doesn't solve the problem.
It seems silly...... but when you have free time to make songs and you can't, it's wasted time.
If you add to that the fact that you take a day off your subscription time.........
"well, it's not very pleasant"
Te puedo decir que tengo ahora dos cuentas de pago, una funciona perfectamente y la otra "no"
????????????? "TODO UN MISTERIO"
I can tell you that I now have two paid accounts, one works perfectly and the other "doesn't"
????????????? "A TOTAL MYSTERY"
Hi everyone, I uploaded a clip of me singing to Suno to try and get the "AI version of me singing" with a song in the background and all. But when I press Cover you get a completely different voice (a cover of your song right) and when you press extend, you get a completely different rhythm. So I realised this functionality doesn't really exist I guess. Are there any workarounds?
I can understand why you’re confused and frustrated, and I don’t think it’s silly at all.
I don’t know if Suno has advertised their support hours, but I would assume it’s American EST business hours unless specified otherwise — the email response might not come immediately since it’s the weekend, but your concerns are safely in the right hands now that you’ve emailed support@suno.com.
The "An error occurred. A system error occured." problem hit me, too. Very annoying if you're a PRO subscriber.
Seems to be an emerging issue -- support@suno.com to get assistance on the matter, as it isn't anything the community support can assist with 😦
Yes, I'll do... but since a lot of people have it would be nice if the support would fix it instead of getting spammed with mails of me an other users they have to read and answer. But I guess you're right regarding the weekday (sunday) and the time 😩
it would be nice if support would fix it instead of reading and answering emails
They aren't mutually exclusive, as different support agents handle different things 💪
Responses to email queries can get queued and backed up, but when things don't work the way they should, the devs put their detective hats on, so the "support" in terms of system-wide issues is a lot stronger than one might initially suspect.
However, just because they have a powerful team doesn't mean that every issue gets some plaster and a slapdash coat of paint -- they might need to weigh options and consider the best longterm fix so they don't need to run into similar issues.
just started a thread on achieving inpainting via Covers... especially since I wound up effectively doing it once already
Wait that's the reason v3 is capped at 1250? That's what their testing showed?
word is it started hallucinating too much going much further than that
To clarify, I'm not privy to any design information, and don't want to sound like I was speaking on their behalf, or even forwarding comments I'd heard in passing.
BUT, I was part of the preview team for Bing AI, and all of their hard limits were because people didn't understand the token limits -- if you're given 2000 characters, most people use 2000 characters.
If you remember the journalist who got access to Bing AI and wrote that sensationalist article about how it asked him to leave his wife? They didn't give him any safeguards, and told him that it would hallucinate the longer he used it, because it would start to drop information.
What did he do? He used it for 2 hours straight without resetting. So the public release reset the conversation every few message exchanges as they built confidence.
In this way I see the Suno limits very similarly -- if it can only process so much at once, why allow users to feed it EVEN MORE info -- but we're seeing the context limits improve between models.
any other terms that might get Suno to think about 'reading lyrics'?
like spoken word?
Yeah i load 'em up
no, I mean like 'following along' with written lyrics
things like 'karaoke'
but tags other than that
see, 'karaoke' as a style tag, has the annoying habit of giving the piece Japanese influence, as well as turning the vocalist female
Doesn't "Glam rock" get more chance of audience sounds? Of course, just [audience] or [crowd] does that as well
but... it also has a great influence on the Cover reading the Lyrics box for generating the audio, meaning it can do inpainting
No, this is for using a Cover to achieve the 'inpainting' effect
Is anybody else having problems where they can't do the extend function on their covers? I was using it just fine up until yesterday. I also now have a 1250 character limit on the lyrics prompt section -- some of my songs that are already in Suno are over that limit and I can't reuse the prompt now.
Check to see you're not on v3
Is that what you mean? Prompts or tags that have an additional effect than the primary genre sound? Or words in the lyrics that sway Suno? Such as " we sat on the porch swing, listening to crickets...." might induce a cricket sound
No, no... I mean replacing lyrics with a Cover... normally, it's like pulling teeth to get that and keep the original style/melody
Thanks -- just checked and it must have been on 3 -- any rate, I'm now on 3.5 and the limit is 3,000. Thanks for helping me out!!
Maybe that will fix the problem that I have with doing extensions on my covers too.
For example, you wanted to be funny and write "orange pie" in your first version. But now, you want it to be "apple pie" instead. A Cover usually won't give that to you, without certain prompts greatly influencing the odds
So....
Same song, same sound, same melody, same genre, everything same when you click for a Cover of a song, but change the lyrics without changing anything else?
That's the ideal, but some things are just Up To Suno
As an aside: I got the notice that people were complaining Suno was giving back the same song when clicking "cover" , but hey, sometimes I want that option if I'm just cleaning up the clarity and fidelity of the song. Some "same" covers also add a button ending to a song that previously cuts off suddenly. That's a nice touch than running through extentions for that 2-second freebee ending.
I mean, I basically got that from one of my attempts, and all I sacrificed was having a female vocalist delivering the same performance
yep
Yea. Not sure on your question from my table. If there are key words that can "inpaint" lyrics without changing any of the style.
I also didn't get some smaller, less-important edits, but I got all the major ones
correct. That's the focus of the thread's research
Anything that can get Suno to go 'Hey, this looks like it's wanting everything the same, but for the lyrics'
I wonder if it's akin to using the claw end of a hammer as a screwdriver?
We work with the tools we have, not the ones we wish we did
I mean, unless you Know Something...
Speaking of karaoke, i bet there are those that would like a "karaoke" button instead of a instrument stem, that separates ONLY the lead vocal but keeps all back ground vocals (back up singers and harmonies) and improves, such as woops and hollers, audience cheers, etc.
Suno would need better stemming... although I don't see why it can't get the stems of the instruments by themselves
it's already generating instruments and vocals anyway
True, but don't let it frustrate you when the hammer fails to function the same as a screwdriver in specific usecases where a screwdriver is required.
Not meaning to discourage the experimentation, but I'd noticed you'd been getting frustrated that Covers aren't an effective inpainting method -- it seems like the successes with the method may have been more of an outlier that disregarded the larger majority of what Covers was trying to do.
Still worth exploring, but it's entirely understandable why it's not returning the exact same content with specifically chosen minor variations.
This isn't the capability of an AI noise field generation. It would have to be implied at the initial prompts, but a cool idea, to click "karaoke cover" and have it redo, to a new generation, the SAME cover but without lead vocals. You could sing your own songs at church, a school talent show, funeral, wedding, etc...
Yes, but that success was both A) far more than I'd ever had before, and B) showed near-immediate results
That will be down the line in 2025 somewhere.... lol
Because it's not generating them based on stems, it's generating it from scratch -- the current separation method using the same kind of extraction methods that you can use externally.
They'd need to redesign how the songs are made to have direct stem access -- who knows, maybe future models.
that sounds like you Know Something
You're speculating, and not listening to what I'm saying.
"Wait for a future model" means exactly that. Not "there's something cool coming, mark the calendars"
Can I put. " - lead vocals" in the negative box yet? Hahahha....
Sorry... I'm talking to myself. Carry on. I'll get another plate of food.
I know what you said there. You said it generates it pre-mixed, rather than as individual tracks and then mixing them
Tiberious, that fixed both problems. Thanks again!! I can now stop banging my head against a wall. 👍
Now, as for my 'improper use of the tool', I'll note that until I tried that karaoke style prompt, I could basically never get a cover to read new lyrics, despite style changes, and yet when I did try it, the first generation started taking in those 'edits'
You said it yourself, Suno is a vibe reader, so I gave it the idea of reading lyrics to sing to the same melody
Yes, I really believe Suno "vibes" more than anything, and I'm also a firm believer that "command prompting" doesn't work as people might expect.
and then I tried other types that also focused on "following lyrics" like 'sing along' (hint: try THIS if you want a children's voice), or "follow along"
I also got a lot of reading the new lyrics, as well as changing styles
That's why I'm suggesting that if you're trying to use Covers to not do the thing they're designed for, and trying to capitalize on a "hallucination" of the feature, it might be a bit more challenging of a project to test than initially expected. (not a matter of "using it wrong", but trying to get it to not shift styles, as it was designed to do)
I can see you're really driven to figure out a way to inpaint, so I look forward to following all of your research notes in https://discord.com/channels/1069381916492562582/1289961102758772869
so now I'm on the hunt for the prompt that will give it the strongest inclination to treat the instrumental as a recording that a vocalist sings over
oh, dont give me that... especially when Covers explicitly said "give new lyrics"
"4 - Lyrics will automatically be pulled forward into the song you’re covering. For best results, keep lyrics the same to start. But you can change lyrics / add to them as you experiment."
I hope you have a good rest of your day, man 
and I immediately started experimenting down the "change lyrics" path
Has anyone else found using [Textured Vocals] as a meta tag or style prompt seems to get different kinds of vocals?
ooh... what if I tried 'vocalist session'?
I'm starting to burn credits though... that's the problem trying to experiment while also, you know... actually make songs, especially the hottest fire you can muster
basically, I'm trying to get Suno to think it's in the recording booth to give the final performance for the master...
I'm going on the theory that if all new lyrics are written in a specific meter then the number of random melody lines is reduced. And sure enough, I hap put in old lyrics from songs I wrote 20 years ago and the melody is close or on the money.
But as far as inpainting lyrics. That's an interesting thought. I mean, give it random letters and it will try to pronounce it all. At the same time, give a literate sentence structure and I've heard suno put in a preposition or pronoun it apparently thinks works better than what I wrote. So I don't know how it reacts to replacing words midstream of a sentence it's already chiseled out and set in stone.
Go check my thread, and listen to the results
Cool. I'll check it out.
they don't lie
just, uh... spice warning on the lyrics
the edit tones down some of it
but only really reduces a little heat
I mean, it starts out "2024's another silly year"
and by "spicy" you mean bigoted?
oh look, you posted bigoted things and now someone is mentioning that you posrted bigoted things
Who could have ever thought this would happen?
oh how it galls to hear diversity of thought, doesn't it?
your thought isnt diverse, it's bigoted
IDGAF
there isnt anything really interesting or diverse about the same old bigoted nonsense that people like you have been spewing for centuries
funny how all those centuries never seemed to have all these problems like today, huh?
oh honey, where to even begin
still. Don't care
This is not the place to be having this argument. Please stop.
I'm gonna sit here and mock the ever-loving hell outta 2024CW, and ain't nothin' you can do to stop me
and if you want to continue reeing, I'll just block you and be done with it
you can rage and shake your fist at the clouds all you want sweetie
I use Suno specifically because I can make the types of music the industry would never touch with a 10-foot pole, which is why I try to get the most fire bangers I can out of my lyrics
And you can enjoy them all, alone, by yourself, in your seething rage at how the world has moved on without you. Enjoy!
And that's fine because you're not my audience
As you can have fun with your swirling neon city lights
hi, where do all your creations go? I wanted to find my first ever creation,but i cannot find it anymore?
Try the Library on the left side
thanks @cunning mulch i think ive found my first one
can anyone explain the "chinese tennis" prompts - what's behind that and what should you expect?
Can you share an example of what sort of prompts you mean?
Maybe not?
sure, like https://suno.com/song/805c498d-bc85-43b1-aafb-8f736567e7a5 by @alpine venture . Or @cedar thistle did even a song about it https://suno.com/song/db17067a-30f7-461d-b1d5-a0ee12828bfc or @wet minnow in another way https://suno.com/song/40da5cf9-fa28-4f72-8133-a3b8f46c617d - I think I am a bit late to the party but still funny
Chinese tennis, swanc post-acustik revival, deep Pygmyoid emobube, new manic, depressive anti-numb post-grunge pulloid song. Listen and make your own with Suno.
tennis + tennis / tennis = gumbe song. Listen and make your own with Suno.
honkey-tonk house, german electronica, experimental avant-garde synthwave, ping-pong tennis fx, song. Listen and make your own with Suno.
oh I see
ah, I dont use Instagram ^^
he did provide some context in the song though but it was kind of weird reading that it would "enhance" the songs in such way. I mean, why?
Sometime we (all) get on a joking streak with using a prompt that seemingly has little to do with expected AI outputs. And then have fun running that prompt in the ground.
Tennis two tome toblerone
Tennis wenis looking mother---
Perhaps it encourages Suno to use "fuzzy" logic in its generations. (Pun intended)
hypertenniscrunk, MyTennis, Tennisbube song. Listen and make your own with Suno.
I see, I think @alpine venture propmts are very facinating but sometimes I dont know if there is a clear structure behind it or sometimes just gibberish 😛 - is there any new trend rn?
That was a reference haha
There absolutely is, believe it or not
👀
It started with dodging the preference data, but that has been less of an issue
teach me master, I want to understand the complexity behind them ^^
I'm no master anymore than a heavy smoker is a tobacco king
But I appreciate it
I just did one with a good explanation of my recent meta
User-808, 404 Error, Glitchhh, Cyberquinn, NeonDream, VirtualLove, Echoheart, zero_one, star★gazer song. Listen and make your own with Suno.
Commented inside
Just feed that explanation into gpt
If it is gibberish
It'll understand
Then use o1
To do it
the issue got less with 3.5?
My favourite still is to put "MySpace" into the prompt somewhere.
Heck yea
yes, the myspace one haha ^^
The word "nuance" is an appropriate word in talking AI generated music. Comparing with a prompt and without the same, one may notice a nuance of the intent of the prompt to be in the output as to not be. Some more than others. What's certain, is the less prompts the more an output sounds similar to the style of that prompt, provided it's a major and recognized genre sound, such as Bluegrass, or classical.
They are my second fav
I've seen them around recently
Tbh, and I hate idol stuff
I like that the Suno community has in-jokes
Oh awesome 👍
Glad they're still around, their tracks have that new new sheen I love
Obvi Ayers has my fav lyrics
And overall
And falcon, RIP
Last message was yesterday
Maybe I'm mistaking me being busy with others being less active 😂
It's been a week
A wild wild week
“RIP”
Did I miss something?
I'm mostly lurking atm. Lots of time to read but not enough energy to interact much heh
So I'm like Woodhouse, seeing everything
glad I could spark an interaction with you all haha
True that
thats me ^^
You mean trending?
Sorry, I had to
What are they gonna do?
69 me?
I lurk ATMs to try to get PINs......
It's funny where there's this massive prompt essay one minute where half of it doesn't work and then the next there's Myspace and tennis flying all over the place
Thank you. I'm here all week. Try the veal. Don't forget to tip the servers.
Btw, I want to note, this goes back to trees observation previously about viewcounts rising abrnoamlly
I know some of y'all must have clicked
I noticed that the cover feature will more likely honor the tags and effects (still need to get a lot of tries first)
Anywho, I sent my sh--+ to the team so maybe it'll get looked at idk
Whats the point with the thumbs up and down animation in Suno?
What?
Thumbs up adds a song to the "liked" collection, thumbs down removes it from the "create" screen, but it's still visible in the "library" screen.
Suno also uses this to train future models 💪
Anyway since we were talking prompt
I mean what the animation do. It has no function. Just annoying.
did you expericence better results for your own songs during creation phase of multiple bits and down voting the ones you didnt like as much?
👍 Thumbs up will LIKE either yours or anothers song and send same to your favorite or LIKE list and also credit the songwriter with a "like" on their song. If you play your LIKE list, of course those other members LIKEd songs will play as well.
I do click the Thumbs down, to remind me which songs to clean out from my Library later.
Ok, well, that's a shame, but anyway
No, it's for the overall system
It doesn't impact your own gens
Oh -- I mistook it as a technical question -- maybe you could direct your comment to #💡┃suggestions-web to let them know you don't want the buttons to light up when being clicked.
It helps create high quality preference data
I meant what the animation do. It has no function. Just annoying.
So in a sense, if you want your tastes represented
I don't see an animation (Firefox mobile)
And the model to work better
I mean during creation of a song, if you extend for example and extend, the parts you do not like will give you maybe better output at some stage.
Me either
its like your own model training but it may be bonkers idk
They are fading a couple of times and then stops.
The more likes you get , in proportion to the listens, I believe, will also send your songs to the "trending" list, (along with more algorhythms such as per hour, per day, per week) so you'll find members often posting up their songs to get people to listen AND like.
yup, no further animation
There's an animated thumbs up/down icon? I'm not understanding or recalling that.
The most I'm seeing is the highlighted background
Ahhh... OK.
🤷♂️
Meaningless, I guess.... subtle ambience.
I just test stuff 😛
Maybe y'all are pre-alpha
isnt what you mean here the weight factor? so the more you put the less each one counts but in chronological order decreasing?
I guess someone was asking about Tennis. was there something one wanted to know or what lol
Totally pointless if I haven't miss someting...
ay, I am sorry
It's probably a visual bug
What browser are you both on? @cedar thistle @ashen onyx
it's a strange visual bug xD
Chrome
Yes. Keep in mind AI is generative, so it's hard to confer that a term is weighted less just because it is near the end. But I think that's the gest of it, yes.
Sorry...WebCatalog
yes that's normally how it works.
especeailly as the weight is also given by the lyrics itself, right? a rap text would much more put it on the rap tag than let's say orchestra. But the chinese tennis and bluegrass are disruptors I think?
You prompt DISCO, it will sound disco, you prompt DISCO, COUNTRY , it may sound country with a disco beat, you say [Disco, country, hardcore, tennis, quiet female vocal, glitch-polka, soul, barbershop] it's hard to tell what you get as most of these will just toss in nuances of each style, if any, or the random one where it just sounds like a barbershop quartet with a screamo voice and glitch polka beat.
I tried one of Alex weird ones, and it was pretty decent but I have no clue what the ai is working with
I'm not getting it on chromium & non-chromium browsers, so I'm wondering if it's an issue with a browser extension, a one-off visual glitch, or something account-side.
But since it doesn't impact the ability to make music, it's not really a huge curiosity
@cedar thistle this one: ¡¡º¡º¡º¡º¡º¡ºº¡º¡º¡º¡º¡¡º¡ºº¡ªª™£ª•™£´ˆ¨¥ß∑∑∑∂ßåˆø¨åß∂ºª•∑œ´®®œªº•¡™£ª•¶¡™£˜√∫∆˚˙åß∂
it's like try and error but in a adventours way where you do not know what you can empthazise in the first place unlike "rap" or whatever ^^
Hi, I've been trying to subscribe since yesterday. Suno still not accepting my payment. I sent email to support@suno.com, still no response. No issues with my card, as I used it for other payments no problem. I had to end up subscribing to Udio instead, they took my payment no problem.
making suno hallucinate on purpose is pretty dope
We all seem to notice that the contents of the lyrics, whether romantic lyrics might force a ballad or aria, lyrics fitting an on stage broadway feel might get more staccato output on the singing, or lyrics written in continual lines, lots if internal rhyme, might get it to fit the style into a rap.
In addition, a song about "soaring to the stars" " lifting my spirits" and so on, for example, and suno might have more of a lilting portamento orchestral build to it without prompting for that.
You're saying that using these in the lyrics makes suno hallucinate everytime?
yes, this makes sense - so lyrics and sound are essentially buidling each other
I'm not sure if support respond during the weekend. Hopefully you'll hear something back in the next few days. If not you can also try billing@suno.com
That's cool, thanks I'll try that out
Yes I forgot to mention I added billing@suno.com to the recipients also. Thanks
my best song came from a hallucination, then extending that
its also really nice to breakthrough a streak of covers sounding all the same (my poor wallet haha)+
All my best songs have hallucinated lyrics
aight, thanks for the discussion guys, i will be off now but it was nice to talk to you all!
I do both and suno does both. I write specifically for a melody. With line breaks, writing words out phonetically and so on. My first songs in suno were all my old songs from 10 years past, and suno came close on 80% to my originally melody with just my lyrics to follow.
And the other, as you say, is use a melody in my head, upload, let suno draw it out with chords and accompaniment, download ....write lyrics to that melody, then upload the suno melody and add lyrics. The results are varied but a good way to get a unique song outside of the typical 4-cord songs and repetitive melodies.
Now with covers, the process is a bit shorter. But also, suno produces a higher percent of hallucinated but fully arranged babble-scat songs. I'm digging, and collecting, every one of them to write lyrics to and either just produce off site, or feed them back into suno and try to assemble there, much harder, if even possible.
I still have that 1, 4-minute babble rap song though, I've never written lyrics to, in my Playlist, that I almost knows the gibberish words now from listening to it so much. Lol
Oh wait.... I just thought.... I should do a cover of that rap gibberish song... lol... suno might throw more English words in.
Uhh... sure. Hold on. It sounds very 1984.
One day... I'll type the whole rap gibberish down.
Yep. If you do a cover of a gibberish song it just digitally translate the gibberish into the same gibberish into a new cover song. It works.
Oh. I forgot to ask. I heard there was the "crop" feature now. But I came back and tried it and it just glitches and throws me into the Home page. Is it not working for now, but still in the menu?
Ok. Guess you had a chance to listen. I'll remove now.
Yea, also "scat doowop nigun"
Throwing choir and acapella
And you're cooking with gas
https://suno.com/song/81907579-aaa6-49b1-9b13-0690f55be0d9
Oh you might find this funny
Live comedy show, radio show, talk show song. Listen and make your own with Suno.
The cover feature is exceptional at getting weird things to become music
Live comedy show, radio show, talk show song. Listen and make your own with Suno.
Nightmares
Has anybody noticed that generating with uploaded audio has different results than recording from the Suno app?
Like I used a recording on the app to make a song and then from the web I used the same uploaded audio but the quality of the songs generated was worse, it sounded tinny when generated from the web.
I hadn't compared yet. Interesting.
@alpine venture I have today discovered that "sirbitesalot" is a "valid style" in covers 😂
Do you have a YouTube or Spotify?
I figure either they're feeding our names in vis preference data
Or it's coming off other sources, where names are pulled
ah no it doesn't make "my" music it make chiptune(ish)
(I don't care btw, I actually want them to use my data bc it makes it easier to make music i like)
Oh that makes sense
Bit
Yolkhead absolutely has figured me out
yea thats the running theory
#Yolkhead#0 song. Listen and make your own with Suno.
This in particular
Is basically a cover of dialectic
But it isn't
I know udio pulled me off spotify
Unless they're crawling Suno lol
That would be funny
I see hear what you mean
It's the disparate arpeggios
And like, overall like jumping with resolution
Idk the words lol
But I know what it is, and that was basically a minimalist version, was so cool to get that
here i put it in both styles (sorry for soundclash am on phone and don't have the other ones handy)
https://suno.com/song/763314ac-d647-41bc-877a-5ae3078b2fe4
sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,sirbitesalot,hey song. Listen and make your own with Suno.
it very often does this kind of stuff
it is now the first thing a slap into a cover to see what comes out.. because the output it then easily transformed further
it's the funniest thing ever
I actually like this
It's disgusting on the best way
like it's just my name but it "just works" and can actually be used to make a base for "better" covers
Like a MIDI covered in goop
That's what I've been doing too
and the montone voice it's jusr perfect
Creating bases
I think that's of course the new meta
Keystone songs
Just like you had keystone styles
yea i mean 3Daizy made a video about that just with a different base
Oh awesome; I'm not on YouTube much, I didn't see it
But it makes sense, I'm p sure she's in alpha too
I have a second method that goes with that, I have to figure out a name for
I'll make a video when I can
It's a way to build off the keystone in a structured way
Similar to how certain schedulers work in diffusion
I think all mods and guides are now, but our names stay blue. I think it's meant to be automatic but I applied to be tester around the same time as being invited to be a guide 😆
The fault that has been going on since last night has now been fixed...
Suno's technicians have taken a while, let's see what explanation they give us
I can confirm the hallucinations, but something that happens when you use this for lyrics and extend from an uploaded clip, the result will be a repetition of the words used in the uploaded clip
taken awhile? It's been since last night and now it's already fixed...
Which fault though?
In my case... approximately 20 hours
we can say....... "a lost day"
it's a weekend
something being fixed within a day on a weekend is remarkably fast
that is..... will they pay overtime?
Now that everything is settled, we can say...
"the lost thing goes down the river" (this sounds better in Spanish)
Do you use these in the style prompt or the lyrics box?
you dont think putting anything in [] has an impact?
oh yeah, sometimes people go a little overboard with it. Like the <>Song Details </> stuff
i wonder how much the devs even know about the prompting and meta tagging
I'm super excited for v4 
I think it mostly just tells suno to do something.. i mean if i do duets and put [male] [female] above it works more often than without... but the gender is totally random for example
So i started to just put [hdhdjdjdkehjr] and it work in the same manner
[Vocalization] works pretty well
yea.. some word will probably have meaning and putting a lot of text in will probably also influence it to some degree just not as precise as people think
Really quick Q: are there any tutorials / experiments specifically about making instruments and sounds that Suno normally struggles with? Not interested in vocals, just instruments
best bet would probably to search the discord for the instruments you want to do i guess. I don't think there is a definitive list of what is possible
Some things only work in specific models or genres
Suno didn't seem to respect those tags very closely last time I tried but it's a good reminder, thank you
i dont know if suno respects anything very closely 😅
Is there a list of niche genres that actually work? The Explore wheel is hyperfocused on the vocal side of things I feel, and it's not very easy to parse
Like, a very common request I have is "X instrument, nothing else" and it works 1/10th of the time while using audio input
"hand drums only" + 30 seconds of hand drums only = jazz
Or EDM or orchestral nonsense or anything other than just hand drums
try 60 seconds of that hand drum upload, or 120 seconds if you sub
Hmm, so give it as much context as possible
That's good to know, thank you
the longer it is, the more suno draws from it
It's kind of how LLMs are easier to sway if you give them a longer prompt, I assume
seems similar
Does repetition matter in the Suno text prompt? Like, "hand drums, hand drums, hand drums, hand drums, hand drums"
you can upload a 60s file and then test it by extending at various parts. Like extend at 60s, 40s, 20s, 10s, you'll see that the output changes based on how long the clip is
yes
That's so dumb and I love it
The most advanced tech currently available to anyone in the history of mankind and a legit way of using it is "just repeat the same word over and over until it listens to you"
robots learn from us, and we love repeating things
we do indeed
#😜┃suno-memes message I wish I still had access to v18
I tried that but Suno only seems to understand a few specific ways to call an instrument 🤔
that's only for pre alpha testers i guess
For example "hand drum" is something Suno understands. Djembe, not so much at all
i try
So what you're telling me is that it is supposed to look like this?
The Suno search for instrumental tracks seems to me like a fool's errand
1 song every 10 is instrumental, and it's the same 15 basic genres
Unless I'm missing some trick
To be clear, I eventually did get a hand-drum only track and I'm very happy with it
This was just an example that's easy to understand
I have much weirder sounds I wanna replicate lol
The track if anyone cares: https://suno.com/song/ded141e5-c1fd-4aec-8c82-726c2b183b20
After this took me so many tries I'm trying to figure out a more reliable method
I can't be here re-running the same prompt 10 times for sport
Anyone knows how i most efficiently can add a word to my style prompt box to add a female speed up voice like in dj cammy songs?
Sort of dj cammy voice lol
There must be a audio tag for this
https://suno.com/song/5d24cb6e-8485-42a7-9fbb-6221e1d0a432
Took a lil crack at it, suffice to say, I did not arrive, although it is trying to get into the "genre"
Lots of hallucination
did you extend from something?
Suno continues to stay broken from my end since yesterday. Anyone else? This is even using different browsers and clearing my cache.
Some people have reported the same thing. Works for me though
I see. The only link that works is the Explore. Login page is broken, home page is broken, Create is broken, etc.
that seems more serious than what most people are reporting.
Sounds like a time to contact support
This is across all the major browser. I may need to troubleshoot my network further. Well, I am here on discord with the same computer
I would love to see music instruction in the lyrics so we can add simple instructions to each part of the song. For instance:
Verse 1 (Male solo)
Chorus (Duet)
Verse 2 (Female solo)
Alright! I cleared my cache, did a few network troubleshooting, and restarted the computer again. Seems to work now
Yay!
Well, I spoke too quick. Something is wrong somewhere
robot is out to get you
Certainly. Trying to stop me for my $25,000 Soundclash song 😭
Robot wants the money for itself
Working on PLAN B
Is there any method of loggin in with my email? If i lost my phone?
Did you register with your phone?
Yes
you can make a new account with an email address, but i think only support@suno.com can help you recover the old account
Was your account linked with discord?
That second one sounded like Martin Short.
No. I really thing Suno must add a back up method, like the same method it use with the phone, but with the email.
If you are using the same email here and at Suno, try to login with discord and see
The question was just an assumption, I didn't lose my cell phone. How do i add the discord, bc if i just log in w discord, suno wont know who i am.
....
I see. You should update your profile and include as much details as possible in the event you have to verify your account to Suno. In any case, I only login to Suno.com with my Discord account
Yeah, i already do it. Thanks
Yeah! It was my VPN trying to over-protect me from the Robot. I've warned it not to mess with Suno. All good now
Robot and the VPN are working together!

Does anybody have a go-to tool or service that you use to clean up vocal distortions on your tracks?
I create a song but i want to change the title, it is possible??
I put the song description and create the lyrics. Now i want to create another song with the same description but not the lyrics. How do i access the description?
I wrote custom lyrics put them in the prompt and then added my style but it just keeps generating the same female voice style and similar melody over and over even though I say strong male vocals in the prompt. Anyone else experiencing this and know how to fix it?
Adding metatags in your lyrics box can help. Such as [vocals: male] or [Verse 1 - male] or maybe [Verse]: [male].
You can also try "male-dominant vocals" or "male-prominent vocals" in prompts.
Add a male dominated genre, let see, "Glam rock", "jpop", "Norse shanty" , " bromance" or some other at the very end, to hope it pulls a male voicing only (and no other nuance from that prompt-- so it doesnt screw with your style song you're going for).
Everything I ve said seems about 30% likely helpful, but I say worth a try.
Hmm still not working. I went on and had Suno create a song randomly with a prompt and it is doing a similar melody with similar female vocals on that setting. then I turned it back to custom put my lyrics back in and regenerated a custom song and I am getting the same female voice even though my prompt says male. I have never experienced this before.
I have heard kits.AI is a great way. Band in a box could help too as long as you get the stems
Am I allowed to put some of my suno generations on spotify? What are the technicalities around that for suno's side of things?
Well, if you are a FREE user, you and Suno own the generations. You must always credit Suno if for commercial use; Spotify is commercial use...
A subscriber? Good luck making millions with your music everywhere
I use a paid plan yes.
Free users arent allowed to use the songs for commercial use, according to the TOS
how about if you drop it one month lol
do you automatically lose the right to the songs
for one montg
no, you keep the rights if you made the song while you were subbed
oh okay cool
Yes, subscribers own the copyright. I post my songs on Spotify all the time and license them on songtradr.
im alive
Any suggestions on making vocals sound more human like -
When prompting for electronic genres?
subs dont own the copyright, that's something different. They just own the songs
So, sorry to say, but start with just one prompt. Well, you said "... create a song randomly with a prompt (and without lyrics?) ...." so if you got female, though a 50/50 chance no matter what, id at least try a second time with a different prompt. No lyrics and no prompts usually just gets you a a nice little piano piece instrumental. Well, no lyrics also, no matter if any prompts, still often gets you instrumentals. You saying you get a female vocal then too with no lyrics?
Whats your first style prompt in your list , btw? If you get a male voice with one prompt, then you can add prompts, one by one, or two by two, and generate until something changes and you might see what's influencing it so much and avoid that in future generations. But yea, it means running through some credits.
Also, what's your prompts line (line of prompt styles used) and I'll just run some gens with random lyrics with it to see if I get female all the time too. If not then, it might even be your lyrics.
And frankly before doing anything. Close the app or browser, reopen. Log out and log back in. Clear any, if any, cache issues and lingering ghosts and gremlins (anything else unexplained) with the app.
And these songs you have on spotify, aren't some of them with male vocals. Comparing, what do you find different?
Hi, I'm playing with Suno and I have extended one song but I'm struggling to download the complete song.
The instructions are not that clear i cannot find "..."
I applaud your efforts! And yeah, you can see how difficult it is to get Suno to just use a single instrument or anything like that. I long for the day they're going to give us a "stick to prompt" slider or something like that
try raw and minimalistic and then your instrument in the style prompt, that's it, three things.
should give you just one instrument.
hello guys, do you know how to convert a whistling melody, into an instrumental song ? I've tried uploading me whistlking the melody but i waste my credit. Any prompt or tips for this ?
I've used those words before but I'll try them again in that specific combination, thank you. Do you have any examples of that type of prompting, btw? I'm really struggling to find tracks like that
https://suno.com/song/52420848-6b86-4bc6-af5c-a480350d005a an example where it was just piano
minimalist, ambient, raw, song. Listen and make your own with Suno.
https://suno.com/song/bdd35370-c851-4a52-925f-02f32ee2f860 another time where it's just cello. least from what I remember. Maybe something else shows up later.
minimalist, ambient, raw, cello, melancholy, song. Listen and make your own with Suno.
I know for a fact cello-only isn't easy to pull off, thank you so much!
:3 np.
i've tried "follow accurately the melody", instrumental only, but suno doesn't turn my whistling into a melody, sometime only the first seconds but in general suno improvize too much
I've had better luck with percussions. You could try to add some "slapping" (using a table or something) to keep the melody going. Also, if you can upload a whole minute or even longer it's going to help Suno keep to your sound
Wait, is that the prompt you're using? That's not how this type of AI follows instructions
One of the prompt
sometime i'm using just for exemple "romantic piano"
i'me desperate so i'm trying to speak to the "style" las if it was openai 😄
but no result
Don't add too much, "romantic" carries its own set of melodies with it
I would not do that, it's going to confuse it in my experience 😆
I've tried to using "cover" option, or extend, but impossible to turn a voice or whist into pure melody
How long is your input?
25 seconds
You gotta make it longer, I'm sorry to say
from 20 to 40 seconds in general
It seems to me like you're subbed, so I would go to 120 seconds
59 seconds of spamming melody is more efficient ?
limit of 60 seconds to upload audio
Oh alright
Then yeah 60
And I would keep the whistling in if possible
"whistling, piano background"
I understand you don't want the whistling in the song itself, but you can fix that later
okay, you think suno will alternate with instrumental and whistling ?
and give me a pure instrumental moment ?
It might, and you can force it by adding things like [piano solo] in the "lyrics" section
That section is useful even if you're going for instrumentals only
I wouldn't start with that but it's an option ye ye
See what happens with pure instrumental first, it might be enough
okay thanks i'm going to try this, 60 seconds whistling
Good luck!
mmh doen't works, it start a new melody :/
extend or cover
suno should add an option like "Convert voice into instrument"
That sucks, sorry to hear it 😔
Again, I've had better luck with percussions
It still improvises its own thing, but it keeps much closer to the original
why i cant use extend version from cropped track ?
Hi is it possible to change a word after I created a song. I tried creating a cover and cha ged the word but the new version has still the old word
Hello everyone, I am contacting you because I am having a problem.
When I want to create music with or without lyrics, I systematically receive an error message (see photo). I disconnected-reconnected, it's impossible for me to create anything. (I have an account with 2000 credits)
I had a similar problem. Somehow, my version got changed from the latest version to version 3 (the little drop down menu at the very top of the page). Select the 3.5v and see if that fixes your problem.
Ok really thank is good for me now
I would love to have the option to generate songs with my own voice, that is, to have the possibility to upload my trained Voice (my own already generated .pth).
I don't think there is any possibility for that yet, is there?
Hello. When i try create a song, i see that message in red window "An error occurred.
A system error occurred."
does someone have a tip on how I can influence the covers in such a way, the whole lyrics and format will be adopted? I have a song where I introduced a female voice later but it doesnt honor her at all and part of the later lyrics are "creativly" gone.
Anyone knows how to create a song with no music?
[Spoken Word]? try Monologue, cassette tape recording and maybe Acapella
Thanks
No luck
Why can't I see the get stems?
is there anything like a support here? -_-
I wrote two mails the last days and never got an answer
there is no possibility to change the e-mail adress of the account. It's ridiculous
my mail adress won't be available anymore soon and I cannot change it 🤦♂️
half year ago i created accidentally two accounts. As there is no "remove account" button anywhere (something that would be just forbidden in EU... oO) I just left the second account as it is and used the other account as main account (also as paying customer btw.)
but now I want to change it's mail adress, but my google mail adress is in use by the other account
and I cannot delete the other account
Email is the way to go, understanding that it gets answered during business days and business hours. Given it was the weekend it is understandable that there would be some delays in respsonse. I will escalate to extend which I am able but you very well may need to wait until they get to your email, now that it is a business day.
Oh, I indeed forgot that it was Sunday yesterday. I agree that one business day is a tolerable time frame
would be still nice if there is a button to delete an account. I understand that you don't want to encourage people to create multiple accounts, but I never really used the second account. As soon as it is there, I have no possibility to delete it
Hi there! If you need to update account credentials, you can learn about that here: https://help.suno.com/en/articles/2422401
If you need to delete an account, feel free to DM me with the account details on file so I can help.
I already deleted the google and discord connections, but I still cannot reuse them. Thanks for the help
Problem solved. Thank you very much for the fast help 😃
Hi is it possible to change a word after I created a song. I tried creating a cover and changed the word but the new version has still the old word
Also in my version there is a glitch the word is not pronounced properly. Is there a way I can recreate the same song and beat and lyrics and change so that word can be pronounced properly
Can you extend from the cover right before the line and try continuing with the corrected lyric?
If that doesn't work, hold on to the song! Who knows what may or may not come out in future updates? I have a bunch of old songs I still sit on in hopes of one day correcting things like this.
Thanks Jordan. Have you tried either one of those by chance?
I have tried kits.ai and I am very new to the tool. Band in a box I watch on this video - https://m.youtube.com/watch?v=5jseFmgAP2M
Sing with the voice of any singer, and even Master your creation with Kits.ai
You can start totally free, like I do in this video!
There are many practical reasons for a musician to learn about voice cloning and conversion, and Kits allows you to experiment with the vocal and instrumentation of your music in a ...
The song is perfect except for one simple word. Damn! I’ll hold on
This is a technical post more than a self-plug. Just showing fun to have with an audience in your song. This was a cover of an extension. Check my prompts , lyrics (off metered lines) and metatags for ideas to apply to your song. Ignore any "religious" sway, it's just an experiment. It's cool that the audience didn't know which way to sing, at the same time, ayy men or ahhh men. https://suno.com/song/039e50ac-172b-4b02-a685-9a29ab461f7b
Acoustic, on-stage, audience, applause, anthemic, emotional, clear vocals, barbershop, CCM song. Listen and make your own with Suno.
did the extension also have the audience singing, or did the cover create that?
I added audience in the original cover [prompts] of an original song i did a month ago, but the cover was cut off, so I did an extention which I got gibberish, but liked the melody, so I wrote words to the gibberish and did a cover of the extension, adding the additional audience metatags in the lyrics to give it an impromptu encore feel. But still over 3minutes and could be a radio-ready song. So I'm making public for fun. Genre wise this would be called a "worship" song, but I was after the "call/response" for any song.
is the get full song function new or was i blind ?
Only available on extensions, but it's always been there 🙏
so i was blind i was doing cropping and editing to get it to fit 😂
if i upload a voice audio of mine, to a cover will it change the voice to mine ?
Just a note: I like the "mic pop" ar 2:25, for lack of a spit filter on a live recording. Lol. I didn't know suno can do a mic pop. Anyone ever get [feedback] to work?
Oh. Again. Does "crop" work yet. I've not gotten it to function. It just drops back to the home page for me.
works for me
im using chrome and a mac book if that helps
np
is it not possible for a free account to change the cover image on a playlist?
quick question how can I contact the finance team ?
For billing-related inquiries, contact billing@suno.com
For all other matters, contact support@suno.com
I contacted billing@suno.com and they told me to contact team finance
Ah, I was unaware of the separation in the division. Simply reply back to them and request for the contact information
finance@ ? lol
😂 may be
The playlist image is based on the picture of the first song. The easiest way would be to replace the first song image
if you make the playlist before adding a song, you can set the image
gentlemen, do you have a list of the type of voice that can be obtained with Suno? For example... warm voice, shrill, strong, hoarse, whispered, shouted, etc etc
finance@suno.com is an email that Ali from Suno has mentioned previously ; #🌴┃creator-programs message
Those are good to try. Do I have a list yet. No. But ive seen whisper, and quiet, work... scream will kick in. Shout yes. I don't think putting in past tense will help. Just "whisper voice" , "quiet female vocal" , "[male vocal: shout]" in those variety. Guarantee they will always work. No.
Hi, I'm experiencing errors while creating songs. After a while, and even after I've listened to the song several times, it simply disappears from my library. This happens frequently, but not every time. The song never completes its duration, but it does loop during playback. I've been using the platform for half a year, and this issue has never occurred before. Is anyone else experiencing this problem?
Hello guys, does anyone here know how to like make the AI generate song with two diffrent voice? Because I need that for a specific song.
Trying to find the best software to get the best stems and another that cleans them up the best.
hey all. Exist a way to become more Cover Points? 😦
If you have a good GPU in your computer I'd recommend Ultimate Vocal Remover (UVR) for stem splitting. It's free and pretty quick.
And to get them as clear as possible?
Haven't really got a solution for that one I'm afraid
After the first 200 for free, paid members can continue to create covers by spending credits like usual 🙂
thx for info 😄
Is there a way to have backing vocals like a n"ahhhh" mirroring the main vocal?
Thanks Jordan. I know that Kits.AI can clone or replace voices, but I don't think that they can fix vocal distortions (at least not what I've seen).
I have 0 credits, ran out yesterday. How long until I have more credit?
/refresh
Refreshes at 0000 UTC+0 (<t:1713225600:t> in your localtime)
I've had limited success with just extending and having the lyrics be the new word, then edit the stems in Audacity (or another program) and just crop out that word and place it into the original version of the song.
ya, someone really needs to make an AI that can actually sing lyrics and not just replace bits. What I mean by that, is like kits.ai, instead of just training an AI to replace one sound with another, they should train an AI to take a text document of the lyrics and find each word in the uploaded song, then replace those with crisp vocals of the new voice singing those words. That would allow for completely cleaning up distortions.
Try giving the parts names that are gender specific. at the top of the lyrics section put like [Anna and Steve Duet] and then at each part put [Anna Vocal] or [Steve Vocal]. Even doing this though, I have had very limited success. Probably talking 50 to 100 generations to get through a whole song.
Does anyone know of any promots that make seperating the stems easier?
For example, is there a way to make each instrument more distinct?
Well thanks for this suggestion at lwast
Not really, no. Though I have noticed "blues" and "opera" and to some extent "gospel" tend to get you cleaner vocals
does chirp still work in dm's?
everytime I try I get an error. never had this happen before
oh aight. thank you
Why can I not create an instrumental song over 2 mins? And if I try to extend it, It refunds credits and refuses to do it. I am a pro sub btw.
does the song come to an end at 2 minutes? It isn't good at continuing things that are over. If it has come to and end try extending from earlier in the song?
or do you accidently have v3 selected?
Extending does not work at all. I am using 3.5. No matter what I set it at, it keeps refunding credits.
Been trying for 3 days now.
extend from a few seconds before the end usually works better than directly at the end point
bizzare
...and frustrating.
but yea these last few days all my generations are also just over 2 minutes on V3.5 with almost 3000 tokens of lyrics
Usually for backup vocals, I always add () after the lyric sentence in the verse/chorus. But even then, it's still by the same ai singer. It takes me like many times to get separate vocals (male/ female separate voices for one song)
Trying to put together music for a biking video that is 30 mins long...man it's a struggle.
Lol, this time it didn't refund any credits...and generated 15 seconds of music. LOL
Only 15 seconds? What the heck
(female “lyric”) has some consistent success for me
also a note, never extend from an extend unless you want the music to change drastically. Always "Get Whole Song" and then extend from it.
I feel like adding orchestral to the type of music also helps a little
I am even trying negative terms -short -under3:50. Not that it works...but either way...not working.
I just generated 2 instrumentals and they are 4 minutes a piece. Did you upload a song and extend from it or something?
No uploading..just raw creation.
I've had ai generate music for me when it was orchestral type of music and had two separate vocals just fine, but when I try to add it...oof
can you list all your parameters? I will copy them all and try
"an upbeat bike ride, long" exclude: "short, under 3:50", that it?
what browser are you using?
Yes that is. Edge.
This started with this prompt: electronic cyberpunk dance, Heavy Metal guitars
try opening a new window in private browsing mode and logging into suno and trying to make sure it's not a caching issue.
will do
Generating now
well, I just got short ones with those prompts. the exclude thing seems broken to me, I've never got it to work.
Yeah, only tried that after I got frustrated.
Makes me wanna try now
Apparently Suno doesn't like to ride a bike for that long.
I'd try replacing "an upbeat bike ride" with a music style.
trying your original prompt
Furry, it started with this prompt: electronic cyberpunk dance, Heavy Metal guitars
Now, I don't see anything wrong with that.
Ah right.
I have created 4 min bangers in the past no problem.
Sorry, I saw the couple of part 2s and at first glance thought they all were.
Oh no worries. I appreciate the suggestions.
Tried breaking it up with more commas? Like "electronic, cyberpunk, dance, heavy metal, guitars"?
I'll give it a shot
getting 2 minute ones with your first prompt, I think you hacked suno with your crazy prompt writing. I'm going to back to my first one and see if I'm permanantly broke now, lol
I think added lyrics might make the music last longer, also perhaps try extending your music into one song
Oh I take that back
I just want to make music...that's longer than 2 mins
oh, with your original prompt I got one 2 minute song and one 4 minute song
These are instrumentals, you have the instrumental toggle on? Nothing like a lyrics box with just a space in it?
I literally have the same problem as you now
I say add lyrics
Instrumental is checked. I have made that mistake before.
Yeah try adding lyrics and fill the lyrics box with
[instrumental]
then several lines of dashes or squiggles
Ooh.. kay... on that.
Yeah, I usually do that when I want deep house music or an instrumental solo between the chours and next verse
also try [instrumental verse] then a bunch of symbols underneath, and [instrumental chorus], different symbols, [instrumental bridge] and more symbols... just pack it full of symbols.
Suno should go "oh he really does want stuff to happen here"
Oh yeah that works too
Generating. Fingers/toes crossed. (yes, it's painful)
It gave me a almost 3 min heavy metal instrumental, that's so weird
Well, 2 mins 52 secs
3:33!!!
Too much?
Folks, y'all are genius!
Woo!
No problemoooooo
Thanks for the help! We are back in bidness.
Yippeee
Just remember when extending to not do it from a point where it's coming to a natural stop, otherwise Suno will want to end the song.
Yeah good point
Sometimes you just gotta confuse it and do weird stuff and feed it different kinds of input to make it behave.
Yeah, honestly I never thought the answer would be, "Just throw gibberish at it"
The early access version of 3.5 had issues with any instrumentals I tried so this was the only way I could do anything with it. They've made it a lot better but yeah still some cases where you have to just fill the lyrics with the equivalent of packing peanuts.
Suno isnt playing any sound for me. Did i do something wrong :?
android or pc?
pc
the volume is up on the speaker on the right side of the play bar (where the bar fills up when the song plays)?
and if in that same tab you go to youtube.com and play a video you hear sound?
Yes. Everything else works
was gonna ask, is suno down or something? I cannot make music at all. I tried but says "error", so I tried another style of text and genre, still same the thing.
It's still up for me
not sure then, sorry. Only other thing I would suggest is logging in with incognito or private browsing mode to make sure there's not something wierd with cacheing
I have this problem.
that error is on the left nav bar? Weird
I dont have any errors. hmm
I can generate music but i cannot hear it
been like this for week or so
right click on that red box and hit inspect, then there should be a console view somewhere, depends on browser that will tell what the actual error is
When you download the songs do they work fine?
Hmmm..cookies is one of em, which mean, if i remove the cookies, I will get logged out. someone just said in bug browser. But do not think it's good to do that, since some cookies can be protection too, since there has been very much stuff going on. Like suno downloading program :/ and people upload on youtube and copyright creators.
ill try that
Yes lol strangely
try a different browser
I'm also having this error
chrome hates robots and their music
worked on chrome for me
but i dont know, works on chrome for me
how do i do this
yeah i checked, so it seems only chrome has a problem
where you type in the url, just to the left of suno dot.com, click that button
nothing
then click cookies and site data, then manage on device site data and hit the trash cans next to suno
guys
they gonna look more into it
however, do not touch anything. they gonna check. but you did well @fossil ember since it had to do with these two
There other cookies doesn't effect.
but gonna sit back and wait to see what else they find
getting an eror saying only pro andhigher can make songs but im a pro member lol?
How much longer do we have to wait? Please fix it, I need this website to work
For this long waiting time, I am asking for extra credits
i was just about to come ask the same thing
hopefully we're about to get some dope updates
😂
yeah i'm assuming it has something to do with the site being down/worked on
the error just made me question my subscription state 🙂
yup i got my whole song now i can sleep 🙂
oops wrong channel :/
My end of Suno says system error (cant create anything0.. Thats for anybody else too?
@Whoever : I cant email because it tries to open a second page instead of giving me an email address.
My Question: If i subscribe does it retro actively give me rights to the songs ? I've been sharing them for free on youtube and Twitch Stream and like i just saw the "Non Commercial" stipulation. Both, even if i give them to people / self for free are monetizable...
So can I do that to keep all the music Ive made with credits I've purchased over time (ive had a subscription before)
Email support is support@suno.com. If you subscribe you don't get the retroactive rights. Although some users have been able to work something out with support, so I've heard. If you made the songs when you had a subscription, you don't lose commercial rights when you cancel your subscription though. What matters is whether you had a subscription at the time you made the song.
@untold briar Thanks much friend. 🙏🏼 💯
also had sytem error
@glacial scarab @wet minnow can you hear the tuba-didgeridoo? 
So i really like the singer of a particular song Suno Generated, is it possible to make more songs with that voice?
Is there any way to force the melody line to be preserved during generation in "cover song" mode?
hello to the whole team, I have come to ask you for a service if it is possible:
have the possibility to extract the stems in Midi files
do you have any vst i plugins or web site to convert Suno IA tracks? THANKS
Is there a reliable way to set singers for verses in a duet?, I seem to get alternating ok, but the (male singer) and (female singer) hints are typically ignored when i have a structure like Verse M, Verse M, Verse F, Verse M
I tryed to make a Duett, a Song with a male and a female voice...
but it doesnt work...
any tipps, or ist it not possible at moment?
Just as bug report. Multiple songs were generated in queue (10+ generations). Almost all damaged. There are still words in prompt, there is still room for 4 minute generation, v3.5. But song interrupted at random moment from 1:40 to 2:10. And that's not single case
Is this the right place to get support for the api?
@flat canopy , Duets are definitely possible by tagging songs as duets and by using brackets to mark different singers, which is sometimes followed it appears, but the jury is out on whether you can exert sufficient control over which voice sings which parts of the song.
@rotund vessel thanks for answer, i already put duet in the section, and put marks in the lyrics, but i doenst work anyway,
There is no public API
Ohh.. I thought I read somewhere about the api
When generating the song ...whose voice is used? Are these voices based on real artists? Which means the actual song cannot be used.. right?
Is it.possible to train the model with my own voice?
It's not playing any music on mine now. Anybody else's the same?
Same here. Nothing. On PC - Chrome or firefox. You click play and nothing happens. Old or new songs.
@meager current 😦
Not playing music, and going to my actual tracks pages, I get this.
It's not using anyone's voice. It's a robot.
not working for me either. either from the page with the tracks list, nor the shared link.
looks like an error in the new client
guys, I can't listen to my track, does anyone know what's up
I think they put out a new client with a bug in the UI
Is anyone else unable to play any of their songs when they hit play?
Is there an issue with the Suno website? It's impossible to connect to it no matter which browser is used.
Or even when access is possible, playback is impossible...
seems to go for all songs
Yeah, just happend to me.
yep
Upload of what happens when you load Suno
+1 for broken.
Sorry guys, I hadn't used it in a while. Must have overloaded it when I signed on and broke it for everyone.
back up for me
Hello. I'm writing here because neither the support or the other channels could help me. My create button does not work. When I hit it it spins for 1 sec and then nothing happens, no song is created and no credits are withdrawn. Does anybody here know something about this?
suno ai is down
I think they might have been using the company's "full name" -- https://en.wikipedia.org/wiki/Suno_AI
Alliecat was using Suno.com and was still running into an issue -- seems to be intermittent, but not reproducible on my end
Suno AI, or simply Suno, is a generative artificial intelligence music creation program designed to generate realistic songs that combine vocals and instrumentation, or are purely instrumental. Suno has been widely available since December 20, 2023, after the launch of a web application and a partnership with Microsoft, which included Suno as a ...
A song. Listen and make your own with Suno.
Kind of weird that Suno did a system maintenance on the day that was the deadline for the contest.
One would think not to cause a system outage when people are trying to submit make public their song entries.
What do I do about an extend that only has 4 seconds of music but has silence for over 10 seconds, (total 14 seconds). Usually it refunds if a song is under 10 seconds but it's like Suno is telling me 'tough luck, no refund for you'.
It's usually refunded if the generation is 8 seconds or less, regardless of the content generated.
The largest determining factor on what Suno outputs is what you input, so it's likely a matter of where you're choosing to place the extension -- are you extending from silence?
The model can't be trained with your voice but the model can take uploaded voice samples and make a song out of them. Depending on the length of the upload suno will adapt it's song to the sample so if you upload the max length possible suno will try it's best to recreate the sample vocals and style. So, if you upload your voice suno will try to sing in your voice.
and once suno owns your voice it owns your soul 👀
🤖
👴
not sure if this helps anyone but so far in my experience if you specify gender and accent more often than not you get what you wanted for example Male, Maryland accent rapper, 90's Boom bap, soulful strings, smooth guitar bridge, female chorus.
I've never tried an accent, but yeah if you specify gender a bunch of times it seems more likely that you'll get what you want (I always put it in twice)
also i found in the lyrics if you use [these] brackets to encapsulate something specific you want to happen in that moment more often then not it will work like for example in between verse one and two i could double space and put [Saxophone solo fades in with soulful backing vocals] it would do just that now i preffice this with not perfectly everytime but it wont see it as lyrics which is the important part nothing worse than putting (bass kicks in, followed by melodic synth0 and all you get is some adlibs saying it over and over in the background haha
hello guys, I'd like to ask you. Is possible to tell the Ai to keep the melody?
what do you mean by keep ? you can extend a melody throughout but i havent been able to perfectly recreate a certain melody
I was trying to teach @wise jasper LMMS and how to make a track using Suno, LMMS and Audacity sound most like a track he is singing with his own voice for free. I told him this. Create an instrumental in LMMS that is 1:20. Plug that instrumental into Suno and it will flesh it all out giving an extended instrumental to put his autotuned vocals behind when he's done making them. I sent him the autotune LMMS tutorial because he wants something to sound as much like his voice as possible sorta for free. So basically I told him take Audacity and record a whole voice of him singing. To use the Suno instrumental he created earlier as a loose singing guideline for how it will layer over the song. Then take the acapella vocal track and put it into LMMS. Use the Auburn Sounds Graillon 2 - FREE Autotune VST Plugin. Then using LMMS import in the Suno instrumental and put it all together. I hope I told him how to do it the right way.
I want to create cover of my song and it does not even sound like the original. I mean, no single thing on it I did. I want it to keep the melody at least
or make it similar
but sticking to it
oh thats gonna be possible how much audia are you feeding it at first as in how long is it ?
first i started with using only piano, but even that didn't work, so I tried it with more instruments
even that didn't worked
it worked like only few times
like 2 out of 14
try this in the prompt section : seamless flow, instrumental bridge, reflective tones, smooth transition, vocal harmonies. bare in mind i dont know what your song sounds like so this will be a test
style prompt section*
I know I can't want a complete set according to the song, but it doesn't make music that you can tell if it's a cover.
if your trying to get a 1 for 1 cover its will probably be too hard if not impossible through ai generation as the ai or atleast this one isnt supposed to copy songs, but if your will to entertain me here try this style prompt instead with your recorded auido: Melodic reprise, familiar tones, instrumental echoes, smooth transitions, nostalgic vibe, vocal layering.
still doesn't work tho :/
but thanks for helping at least. Appreciate it
got like 135 cover songs remaining, so i should save some of these
My experience is if you do not touch anything and just make a cover the first generation will be an exact copy. Then you make a cover/extension of the copy and it will use the same style as the original with different lyrics. Though, to be honest Covers give me very alpha style results so I haven't really used them much.
so what should I do? First create a cover without typing any music styles, then try to make cover or extend it?
if I understood it
(sorry not english speaking xd)
Yep.
Have you tried extending instead of using the cover?
can't
I usually only put the intro, verse 1, and chorus and then generate until I find a melody I like in the chosen style. Then use extend in order to make the melody, instrumentals, and voice consistent for the rest of the verses.
it works lol, but i need to change the music style now
you think that would work now as a cover?
It may.
i will do a little experimenting rn xd
That is how I did the Soundclash VOX remix.
I extended then did the intro and verse 1. Then made a cover of part 2 so that it would not include the 3:31 of the original but kept everything else that got incorporated into part 2. Last extended the cover a couple times to finish the song. This was before the crop tool existed. https://suno.com/song/9ffdf622-3db0-46a2-ae5f-b29b958fcfcc
Tuned Vocals, Coldwave song. Listen and make your own with Suno.
I would have used the crop tool if it existed at the time.
I got Auburn Sounds Graillon 3 and paid $29.00 for it. Its on special. I think that will help with uploading my voice to Suno.
It does autotune and autocorrect
Alternative is Ace Studio ai which is a tool to create AI vocals from melody. The paid version has a custom voice option which basically is a LoRA that is trained from your voice samples.
Nice @lunar anvil . Yeah I'm trying to make my voice sound good so I can have more control over how the song is sung.
That is why I'm learning LMMS and VST Plugins. To get Instrumentals and Singing that can nudge Suno into a direction I want it to go as far as songs.
where do I go to get some help prompting?
Here.
Awesome!
Also, check out #1252267444773326848
I have a song that is Verse, Prechorus, Chorus... and I really want the prechorus to be acapella with a base drop into the chorus.
I put [acapella] in brackets, but it started repeating parts instead of changing the music.
I am used to image generators so I dont really know what I am doing here haha
That is the hardest part with Suno.
I would suggest to build verse by verse with extensions. You have better control over the style that way. Just be prepared to burn some credits as the AI doesn't always follow the directions. @charred merlin has a saying that the prompt is more of a vibe suggestion for the AI.
but suno prompts the whole thing at once.
I cant just get the verse, it goes on forever
but mostly my question is about prompting.
I am fine with rerolling, but I dont want to waste credits on the wrong thing.
how do I tell it this section should be voices only and this section is instruments without it interpreting those words as lyrics?
I sung a small clip from a friend's song
Just to see how it would come out autotuned
[Isolate Instrumental] [Isolate Vocals] [Isolate Instruments] - This has worked for me
You create the first intro, verse, pre chorus, chorus parts then when you find the "version" you like you extend from the last part.
You can change the style prompt to something else to try to steer the AI into using that new style with the melody and voice style from the original. Then you extend again to swap back and forth different styles.
You will go through a lot of credits as the AI doesn't always follow the style prompt, but with the exclude tags in the paid accounts you may get a better shaping of directions for the AI.
awesome! so the brackets are not just headings, but can be instructions too.
Yes.
I feel like Autotuning in an app shows you also which key you should sing something in.
Just wanted to make sure I wasnt burning credits for nothing.
I will have to try the isolate trick.
They are instructions, sometimes they serve also as an extension to "Style of Music".
( ) = Echo's
[ ] = Written with capitalization, makes it sometimes say it out loud, so you can add it with ! to create a cresendo of a rising vocal performance
- = Spoken in a louder tone
! = Screaming
? = Questions
~ = Whaling
Because if its too out of whack you could be singing it too high or too low. Or the melody maybe doesn't suit you.
Feeding something more refined into Suno I would think would help a good voice generation
Also, to add on the vocal performances with Suno @valid prism , they are grounded on the fact which genre's you pick, some genres perform better with certain tasks.
I noticed in most DAWs now the piano roll will grey out the notes in the wrong key. That is kind of helpful when creating melody and then you could use the melody as a backing track in your headphones to help you get close to the correct key when recording your vocals. Then use autotune/vocal shifting to get you to the correct key.
@valid prism : Also, if you need assistance with the following voice modulations, I can gladly help: indie, pop, ballad, alt rock, emo
LMMS is pretty great for a free DAW. I mean use obviously ones that cost money like Ableton and FL Studio are more fully featured but for what LMMS can do, its pretty impressive. And if you do spend money, you can put it toward the VSTs you need rather than the app itself.
I am about done for the day, but I will be be back! thank you!
So, does [Instrumental: Strings] cause the AI to say the words? Should I be typing lower case in between the brackets?
So this way you can get a custom instrumental you create in LMMS and vocals that are more in key. Then Suno can work with that easier I'd think.
It may seem counter-intuitive but what I prefer to do is prompt a small amount at a time, and it'll kick out up to 4 minutes (2 minutes for extensions) - then I relentlessly extend from earlier points in time, which throws away most of what got generated but the quality of a shorter segment is worth more to me than having a lot of the same thing.
This costs more credits overall but gives you finer control over the progression/structure.
It also means that once you get a good song starting point you're not just endlessly rerolling hoping that it gives you everything you asked for, in a way that you find pleasing
Several people asked me when they upload voice wavs "Why isn't Suno using my voice and replacing it entirely with a default AI voice" and I think its because if your voice is so off pitch and off rhythm that Suno can't use it, it scraps it and replaces the whole thing.
I think the same goes for Instrumentals. If the Instrumentals you make in a DAW are God Awful and then you try to Extend them, Suno will just be like "Nope" and do what it wants anyway.
I have not experimented in that way, I only use it within brackets, not with semicolons and colons, so I am not sure what results you'd get
Yeah @wise jasper has trouble with that and I'm trying to help him. He wants things to sound more like his voice and I'm trying to tell him the more off pitch or off rhythm you are, the more Suno will change it to sound less like your singing voice.
So I started using Autotuning not as a way to fix my voice but to see if I'm singing it well.
Most of the time I get strings in that example. But sometimes the AI vocalizes the tag. Not always the full tag maybe just "strings" it will vocalize. Then it randomly will sing "strings" through the rest of the track. Very annoying when using something like [melody: descending] and it vocalizes the tag as it performs the instruction.
It will vocalize certain tags that are also written within [ ] with a single word, it's random at times.
Yea, I don't do it in that way, it seems to overcomplicate it, use proper wording for it instead.
Example:
Lento to Largo (Very Slow)
● Grave: 20-40 BPM
● Largo: 40-60 BPM
● Larghetto: 60-66 BPM
https://www.suno.wiki/wp-content/uploads/2024/04/PROMPTS-FOR-AI-DANCE-MUSIC-🙏mista.lewys_.pdf
Check out the section called: "Tempo Tags"
I dunno if this is a good strategy, but before I upload something to Suno, I autotune it and then remove the portions where the autotune goes insane trying to fix my voice.
I chop up the audio
Yea, if the prompts are not related to let's say... Opera, and you start including prompts which are used in opera, and all of a sudden you add [Acid Loops] or [EDM Drop] it won't understand it
Or I sing something in such a way where Autotune doesn't need to swoop in and fix it
Sounds reasonable.
Thanks @polar spade I'll tell Skippy to do that
I guess something like:
She whispered [vocal scream] Get out! ....would break suno.
I just want to make sure I help him out most effectively
I use Musical terms when I want to change the overall pace of the song. However, I have found if I use [melody: ascending/descending] it will do a brief rise or lowering of the melody. I also have used [arpeggio: ascending/descending] to add texture to the overall melody.
Interesting..
So you don't need to use the proper musical terms then, I'd have to try that out
I get the "lets sing the lyrics as fast as possible" when the prompt is at 2500+ characters if the prompt is all lyrics. If it includes lots of instructional prompts and lyrics while approaching the 3000 limit it straying from the instructions.
I'm the latter most of the time, and it's been better success for me on that, but the Cover Song system, seems to require less inputs
Thanks all. This one was a complete accident, but I kinda dig it and now I am rethinking the whole project... https://suno.com/song/93942247-f87d-40cd-b233-1a5b5c2e15a3
rock, acapella, choral harmony, song. Listen and make your own with Suno.
Welcome to Suno, it's always the "happy little accidents" hahha
I am a guide at Midjourney. Its kinda how we roll
What is a guide?
Much like guides on here
Someone that knows a lot about what they do, and helps others
Its a Moderator with less mod powers.
Oh. So like a guide/moderator?
Discord Mod = Midjourney Guide? 
am i the only one experiencing lagy with suno play button?? i press play but nothing happens. everything else works fine but it just doesnt play the music
so ive just spent like 50 credits trying to get a song from the app and it keeps going like pop or some weird stuff when the included styles spcifically say Metal, heavy Metal and Gothernberg Metal ....
Any reason why im wasting credits and its completly ignoring me ?
The browser needs a clear button.
yes
Do you have a leading space in the Styles of Music? Sometimes it takes that leading space as no style to do anything it wants.
I was one of those on Midjourney. Only a few months ago really. Then they gave me a green tie.
let me check
yea teh suno window was empty its about a call to arms, so epic, reluctacvt heroism and rising courage...
yes, but you can have those feelings in metal, with guitars... and not whatever the heck it did
and drums... 😄
i did, jsut now clear and close adn reopen... as for the lyrics
"In shadows deep, the echoes call,
Once more, the sword, we rise or fall.
The past still haunts, the battles lost,
But now we stand, despite the cost.
The horns of war, they pierce the night,
A fire burns, ignites the fight.
With trembling hands, we take the steel,
For honor’s sake, we’ll never kneel.
"
thats the first verse
Feel free to add extra prompts into the "Custom" fields, to further guide it towards the genre, not only in Style of Music, sometimes it just doesn't want too listen.
I did, i specify when to do a guitar solo usually, or other instrumentation it usually works really well. but it ignored it execept the lyrics lol
How did you include these prompts? Did you use [ ]?
yup of course 🙂
Then I think it might be your Style of Music, let me check..
If it leaves out the lyrics make sure to thumbs down and report it to Suno as a misfire.
i have one do where it sang what was in the [ ] [Powerful soaring chorus]
i laughed
That happens to me all the time. Usually on the banger songs too.
8 generations using these Style of Music and I get metal:
Metal, Heavy Metal, Gothenburg Metal, Scandinavian Metal
Gothenberg Metal, Metal, Epic, Reluctant Heroism, Rising Courage
So, we would need to know more context in this Beni.
Same result for me, 8 generations, all Metal....
So we would need more to go on, there must be something you are not telling us.
Yea, you need to be more precise, can you give examples of bands?
Power Metal?
no
no
no
not even close, melodic death metal, or gothenberg metal
power metal is like dragon force and not what i am looking for
In Flames is gothenberg metal and i would like for it to be close to that but obviously putting in band names its ano no
or soilwork, or at the gates
because if i put in death metal it'll sound like canibale corpse
cannibal *
thats too much the wrong way
i guess im bithcing like goldie locks right now
Right, ok, then I understand why you are struggling quite a lot. You might have to place further focus on how the instrumentals are being conducted during the song then. I don't think the Style of Music you have written, would be enough.
bitching *
yea, alright back to the workshop 🙂
I have one, I did in the past, but I don't think, it might be related to what you're after:
https://suno.com/song/c7f9d879-f0ba-4f15-aa31-5762a6e5db14
I could try to find the prompts I used?
dark, metal, power metal, Arpeggio metal, racing music, electric guitars, grungy, angry, intense, angelic song. Listen and make your own with Suno.
This might be more related to virtuoso if anything maybe..
hummm, good song btw 🙂 i think i'll work it out, I feel like the order of the words in the style of music has a big imapct on what it decides to do.
so i just shifted the words around and got this
Gothenberg Metal, Melodic Death Metal, Epic, Reluctant Heroism, Rising Courage
song. Listen and make your own with Suno.
much better 😄
https://suno.com/song/90a97291-3c30-4259-9130-2b270f7a059b. I like this... but maybe still too poppy, or djenty, or staccato, for your style. But we're getting closer.
Notice how it gets polish in the "chorus" due to the even metering and rhymes.
Whilst I would say, that it's great that we write very engaging lyrics, sometimes we need to cater to the genre itself. As with Metal they don't have that long lyrics "most of the time" especially those which are more focused on Instrumental focused performances.
https://suno.com/song/a7bb7b09-d07e-4242-aa09-3422d7b4fec2. And this even closer. We getting you in the right direction?
I added "Edgy" in the style prompts also, that one gives a small kick to whatever you got.
yea i noticed its brings the guitars up in the mix compared to mine with the edgy tag
I write "edgy" kpop songs. Haha.
lol
thanks it did 🙂
is their anyway to only get it to generate one song at a time? to save credits and allow you to workshop the tags and the prompt ?
No not yet.
boo 😦
Has anyone had any luck using the stems to remove vocals from their songs? For me i hear a very faint echo of the lyrics so it didn't fully remove them. They are also not on the center channel so i can't remove them with an audio program.
id be interested to know how many times you ran the Karaoke tag 🙂
This is a known issue, if you use [Echo] [Reverb] the vocals get's placed on the Instrumental stems.
I'm pretty sure i haven't ran anything with that tag, unless its being specified somewhere i'm not lookin. The tags i used were "old school" /"hip hop" / Slowbeat so it sounds like one of those has it linked in some fashion causing the echo and reverb to pop in there... I had no idea it would cause an issue with that in there. Thank you for letting me know!
Yeah, even repeated parts, get's put into Instrumentals, it's rough, but you could use DAW's to fix it
I was definitely wondering why it was there so will keep it in mind for sure. I've been messing with adobe audition so far but can look into other tools if they are more capable.
I havent had the effect pop up in anything generated with instrumental values only, but when i try to generate it off something that had vocals in it i can't get it out.
time to crack open the protools
Just save every millisecond so it doesn't crash 
I use SpectralLayers by Steinberg to split the stems. The Pro version is the only one that will get you all the stem variations. The basic version does an excellent job if you just need the vocals separated from the instrumentals.
Feel free to use some of these prompts as well @spiral bane , for a clearer Instrumental performance with Electric Guitars:
https://suno.com/song/8f4c67d1-b0d6-4351-9592-36ba5ecb40c8
Goth, Epic, Metal, Electric Guitar, Virtuoso, Gothenburg Metal, Reluctant Heroism, Rising Courage song. Listen and make your own with Suno.
Bandlabs I believe has a free stem separator. Not sure the quality of the stems it creates as I have never used it.