#📚┃suno-school
1 messages · Page 66 of 1
I used the new Cover feature on my own Flosstradamus remix, to change it from dubstep to metalcore. I think the Suno model has been listening to Sleep Token because the riff it generated sounds exactly like The Summoning. The feature also blew out the bass in the middle of the song, I still love it tho. How has your experience been with the Cover feature?
Metalcore, djent, thall, dark, eerie, atmospheric, heavy song. Listen and make your own with Suno.
It's a longtime known-suggestion 🙏 Hopefully the devs like the idea as well!
https://discord.com/channels/1069381916492562582/1252789683176345641
In the meanwhile, if you search for posts from @mystic flume in this channel, she's made multiple albums all using the same vocalist through a great amount of trial and error. The woman doesn't rest on laurels, I'll give her that much.
I don't believe that's enabled -- or at least I didn't see the Cover option for vocal stems.
So what's the notice about giving feedback? I missed that. Not see anything in a scroll back.
I did a cover of just a guitar and vocal (folk) of 5'15" . It came out more clear, but any changes to the lyrics, as common, were not recognized and it sang some typos as before. Always stops at 3'19" after three cover runs. And minor points of singing the metering in more odd phrasing than from the original. Expected. Overall better than any previous attempts with full-instruments songs.
To clarify, you're asking how to give feedback for the covers feature?
Yes. In a nutshell.
Beyond that, I imagine the typical reporting avenues still apply -- #💡┃suggestions-web for ideas on how to improve the model, #🪲┃bugs-web for explicit errors, and support@suno.com for direct feedback
Easy enough. Oh no. I have to be decisive? It's like children: you don't want to be too critical , there is a learning curve which is a positive thing.
There's definite room for interpretation -- luckily they're not asking for detailed reports, but a thumbs up when it REALLY works, and a thumbs down when you're really dissatisfied
Once it's out of beta, it'll knock socks off
I put rap as a style and it then babble-rapped the first 10 seconds, then gunshots and proceeded into just repeating the same song. But I was entertained. Which to choose... which to choose...
*places song into a trapper keeper decorated with heart stickers*
okay
I don't understand what the cover function should do. All songs I have tried this on only returns the same song again.
I kinda wish there was a better 'get this out of my Create window' button than the down-thumb
Sounds like you need this button #😜┃suno-memes message
it's supposed to let you also make changes to lyrics in-place, but it... kinda doesn't that often, unelss it really changes things up
I will say it's generally pretty good about creating decent intros to songs that may lack them
OK, I've tested on 2 songs but it did return the same song in both cases. I think the function need some work, 😉
Good point. Mine made a very good quality sound this last go round. But didn't take from correcting just one typo. Again, that's typical, so expected. A matter of extending, deleting, etc. As usual. Fine enough. I'm used to it. But now is my efforts a thumbs down if the end result is "perfect"?
How would they review as even knowing why a thumbs down just because it doesn't re-sing a typo correction?
I want to help. Like you say, I'll just have to thumbs down the weird ones, only, from a 1-time run of "cover".
Hello, Why does the AI not take into account the requested BMP or key, always offers accelerated speech rates? Thanks for the answer.
right
that's why I want a good "Remove from Create View" button that's not that negative feedback
I keep slipping up on forgetting to type in a style with each cover of a run of the same song. So you can't click "CREATE, CREATE, . . . " 5 times to run 10 gens. It will give you 2 with the new style and 8 more with "(no style) ". One create at a time it is.
Another small point: the original is on page 93 in my library. The cover is on page 1 now. I can't hit "back button" or facsimile to go back to page 93 to edit lyrics without clicking through 92 pages each time.
Pro tip: make a Playlist called "Covers- Down on the Farm" , [your title may vary] to copy your original, and subsequent covers.
okay.. WOW. I just heard the immediate change right at the 2:00 mark on a Cover
you can type in the page number you want
it was 100% exact to the original, right up to 2:00, and then a quick change
Pro tip: write down the page number where the original sits if you can't remember the title or page number. 😁
But we need folders like for real
If I could put in a Playlist and simultaneously delete from the Library, without it then deleting from my Playlist, then I could clear the Library completely.
when you put songs in the trashbin does it remove them from a playlist?
Yes. I learned the hard way and had to go retrieve a few from trash.
Was it cool or cringe?
covers are great, but they tend to start and end at odd times in the song. I tried to extend and I think it just gets confused. I keep getting 3 seconds of silence.
Cringe. It tried to read lyrics ASAP, so didn't continue the original's flow
So, if I thumbs down, they can then listen, well they always are "listening", but they are brought forefront to my un-public song? Oh no. (A tongue in cheek portrayal of a recluse)
basically, if all it did was incorporate one tiny change I'd made into the lyrics, I'd've been happy
Bummer.
But, it shows that Covers consider a full two minutes of the song, as it was completely unchanged up to that point
Mine went for 3'19" consistently on one 5'song. Four times. So cool enough to like. But I've got to extend now and go the usual routine.
Come to think. I'll check at the 2 ' mark to see if I hear any anomalies. But seemed ok on first listens.
The best one I got removed the intro. Dang it. Oh and so it was 2'53", so that can be a clue.
It would be nice if it could do 4 minutes like a standard Generation
it could also pick a random 2 minutes, as I've had things start with the chorus
I say "best" .... but I got a new vocal never heard before. So strange but cool too. Like a Janice Joplin vibe.
yeah
I tried to get Impractical Ideals v1 resurrected, but I don't wanna blow all my credits on that
that one has plenty of lyrics updates it needs to include
listen to those breaths... what attention to detail
Oh wait. That's why it 2'53" .... she sang so fast she got it all in that time space. Haha
Yea. The breaths are funky. Wish we could "cover" vocals within a song. I pick... hmmm... Janice, this time to sing.
yeah
Wait. I can take this song and extend from 00:01, write a new song and might retain the vocalist. Worth a try...... just talking to myself. Mental note.
maybe?
I'll go for a Melanie's Candles in the Rain, type vibe. Old obscure references, likely
I used Covers to polish up two previous 'versions' of my most recent song that had promise, and one of the outtakes made me laugh so hard I kept it
for whatever silly reason, it rendered as a 1930s recording, and where there was some vulgarity, the 'recording' just cut out completely
Haha. Love it. Wonder if censored or a glitch.
I think it was a style choice
every other version rendered that quote just fine and without any censoring
can a cover of a cover help polish a song even more?
doubtful?
Doing that as we speak.
I had covers remove a lead I kinda liked
unfortunately, I can't share the song here, 'cause it's too spicy
Listening. I added [intro]... that worked..... she's still singing fast of course. But same vocal breathy quality. I little degradation, but just a bad mic battery , I think. Still didn't correct the typo though.
. I'll figure this out. Ok. Back to real world work for me.
When will Suno v4 be released?
seems pretty damned close
from the contest page, if you caught it: "We’ll be dropping some cool new features over the weeks, which you will also be able to use to put your own spin on Soundclash!"
Covers seems like just the first
and seemingly, almost a 'stopgap' solution to stuff we wanted for v4
almost like... if it were a back-implementation of v4 features
Anyone having an issue when you click get whole song on the contest songs? Mine adds a solid 2 mins of the contest track to the begining that isnt in the previous parts.
Actulaly im dumb
Just realized what the issue is
100% -- I took really poor quality Suno tracks (warbly audio) and covers have breathed life back into them 💪
well... lemme try an older one of mine
I hope they add an edit feature. Something that allows shortening/trimming of tracks would be huge
WOW, this is getting worse by the day 😢 instructions are completely ignored (IE: I say deep male baritone vocals) I give the style Darkwave, and I am getting a hip hoppish / rap version sung by a female voice
Even when I add the vocal descriptions within the tags, it's completely ignored. I used to get excellent results a few weeks back, now it seems to be always defaulting to the Rap / hip hop style, multiple re-rolls are getting worse. To be honest: my few credits are just TOO valuable to be dealing with the same type type of music, I didn't ask for.
Has anyone else's head exploded at these possibilities?
well, it's not quite as awesome as you think just yet. there's hiccups
it'll get there though
but.. the potential!!!
not quite as awesome as you think just yet
Wholeheartedly, emphatically disagree.
If you're not satisfied with the beta release of this feature, that's fair, and the team actively wants this feedback from you through the appopriate channels -- but it's super subjective to say whether or not others will enjoy the feature because it doesn't meet your expectations.
I wish you the best of luck creating, @cunning mulch 
temper expectations
see, I came into it excited too, thinking it was the perfect in-painting method
Question for anyone (not relating to covers).... i made an audio clip of my voice.. generated a song... it says i can't make it public as it might contain copyright ... it kinda .. doesn't .. Help? lol
it did rescue a couple prior takes of a song, but it also took a bit of work, and it worked best on the songs I most recently generated
Just because it didn't fit your envisioned use case doesn't diminish how ridiculously powerful the feature is.
Perhaps you feel it wasn't ready to be released into beta at this point?
maybe it could use some tweaks
mostly just in UI, honestly, maybe exposing more fine control over what gets remixed
I'm trying to inject a little bit of reality
but I haven't quite had the same level of success on my older songs, as it seems like they're just too long for it to want to completely re-do
pretty sure I also dropped bumping the Cover length to 4:00 as feedback over there too did now though
Anything uploaded containing a voice can't be made public. You can still share the link etc. though. It's a bit limiting but I assume to avoid people uploading famous voices and being on the home page with their imitation
ahh ok so its normal behaviour. .wasn't aware
I think it has a message at the bottom of the upload window about it. Doesn't mean anything bad has happened at least
yup .. it did but just says something like .... can't coz might be copyright.. didnt say because of audio clip use specifically
but cool, question answerd, thank you 🙂
is there a reason why suno can't pronounce "dies" and says deeze instead? is this all a big deez nuts joke or is that word baned or something? 😂
try "dyes" instead as in dyes some cloth red
yeah, with phonetic replacements, there's not really much that can be done for moderation on English without totally ruining it for normal use
i wanted the word guise sung as in "appearance" and it refuses.. pronouncing it with a soft g.... "Guys" sorted it
yep
just makes the lyrics look balmy
I also avoid using "our" in songs, for either "hour" or "are", depending on how many syllables I want it to use
you can edit the displayed lyrics later anyway
you can? where?
From the direct song page -- you can't change them from the 3-dot interaction menu. You'll see this button at the bottom of the page -- if you're on a subscription, you can "Regenerate Video" from the 3-dot menu to make the video reflect the lyrics.
Note - this doesn't have an impact on the vocals that are sung, only what's displayed for the embedded/downloaded video.
wow .. never noticed tha before lol.. thanks
Just tried to make a pretty explicit song lyric wise and it was refused generation. Ive changed toned the lyrics down and it seems fine now.. Hoping this wont affect my account in any way... I knew i was kinda pushing the limits of what can written and didnt expect it to go through without a little bit of censoring 😅 😓
It's not documented anywhere, but you can use Ctrl and Shift while clicking to select multiple songs
@charred merlin Theharber, I was talking about the upload function limitation! which is max 60 sec length for the song you UPLOAD
explicit in what way? A lot of vulgarities?
yeah it was kinda about bdsm... i wont go into detail. Suno seems fine with heavy swearing and sexual topics.. but this was, yeah a bit vulgar.
'cause I know I haven't run into any moderation yet, and my stuff is the first to feel the effects of it
am pretty sure it would be automatic and no strike... doubt a human reviews
yeah
well if a human does, and that human is reading, im sorry, i wont do it again i promise. i know my limits now 😅
I mean, I wrote a rather... vivid song about a very recent controversy
no way it could be a re-entry into SSC5
hell, I couldn't even get any ChatGPT models to check the lyrics over for structural soundness and maybe suggesting a genre without it incessantly whining about the content
anyway... if that didn't trigger moderation, I have to wonder what you have to do in order to get bit by the electric fence
Oh none of this came from chatGPT.. its purely the lyrics written by a friend who wanted it turned into a song...
right
oh a "'friend" 😉
same with mine. I wrote 90% of it (and a friend provide that key 10% to pull it together), but I wanted a chat AI model to see if anything special might fit it
but yeah as mentioned... i know the platforms limits. I'd written some pretty full on and violent songs in the past and never hit moderation.
no legit... If it was me, i wouldnt hide it. Ive already done a couple of tracks about bdsm but, they weren't vulgar so to speak.
ah... someone who's a lot more direct with lyrics
I'm just teasing
thankyou that should fix it :)
yep. My streamer friend is like that. His chorus took a bit of polishing, but when i did, it wound up changing the name of the song
i gotta share this.. sorry guys... but Friday fun time is here:
male jaunty spoken word, rhytmically fast paced song. Listen and make your own with Suno.
went from a pun on the term 'Spring Cleaning', to the catchy phrase that leads off the chorus
yeah... certainly not my lyric style... I'll drop one of my full on tracks in #🐣┃suno-showcase . I do inline skating, and i get yelled abuse by people in cars and on electric scooters... so i had to write about and get some emotions out.
oh thats not my normal stuff .. PLEASE don't judge me by it just made me chuckle coz it wasnt expected lol
I also do fun light hearted stuff too. https://suno.com/song/759e9551-bb17-4985-996c-69d13876e8e4
Soft Voice, Pop Punk, Alt Rock song. Listen and make your own with Suno.
this one is chatGPT assisted.
I'm not all hip and up to date with all the lingo been used by gen z's lol
yeah im like.. 38... so im not quite boomer.. but yeah. i never really grew up anyways lol.
haha
But yeah a lot of the time, ill go to chatGPT, and just start a conversation. Get it to do a couple of searches. let me know what it has in its training about a particulair subject or topic.. and then from there tell it to take everything in the conversation and use it to make a song about x, from the perspective of x, do this do that use this structure bla bla bla.
ill also let it know to use [] to mark out the chorus etc. and yeah. ill take the output and then go through it line by line and just get things sounding a bit more natural. then into suno it goes. Anyways..
Just wanted to say, i hit the limits of what can be said, and im sorry.
yeah ... pause for a breath, man ! lol
i DJ online so use Suno to make all kinds of weird promotional mini tracks and oddities like the one above as well as standard songs
(ex-real-world-dj of 30 years)
I used to use GPT lyrics with a human touch, but... it gets a sound, you know?
generated lyrics are great but am noticing a lot of "formulaic" stuff come from it.. like.. "In a land down somewhere .." or... The Neon something or other.. it LOVES neon.
yep
when in a bit of pique over what I was hearing and basically formulating potential song lines in my head (that I wound up just using all of), I wrote them all down when I could, and then went to a syllable checker
i need to pay more attention to syllables lol
yes
Syllable Counter is a free online tool that helps you count the total number of syllables for any word or sentences, per line for haiku poems.
handy..cheers
mine varied in length between 6 and 12
although I think the 6-8 lines got merged into 13, 13, 12, 15
yeah .. the challenge on that "thing" about colds i posted above was it.. was an old old poem ... that has to be paced very fast for it to work .... suno was a bit slower in the delivery coz it added music when it shouldnt... but that was very hard to "pace"
but varying the number of syllables per verse and establishing (and breaking) patterns in them help make a song flow much more naturally
yup
And since you have full control over the cadence, you can do more fun things
when I originally wrote the end lines of one verse, I originally kinda spoke it with a lmerick-like cadence
with an intentional mid-line rhyme
gotya
originally, it was "so he can go and lay 'em low, with his blazing shout"
can you feel the rhythm inherent in that?
unfortunately, when it was rendered into song, I felt it didn't come out right with the previous line, so it also became a deafening shout
if it's powerful enough to put someone on their ass, it's gonna be loud
Can't create covers
hello. I have a Problem. The Ki did not follow my text and ruined the whole song😢
What membership level are you on?
Free
That's the issue then. This public beta is only accessible to paid levels
I've produced 7 decent songs for a total of... $36
and there's at least three extra variants of one that are also release-quality, IMO
And better yet, all but I think one, are completely mine
Everyone's milage may vary, but 500+ songs per month for $8usd is a mindblowing deal even without considering the inclusion of covers. The amount of songs you determine decent will vary as well, but what an incredibly exciting age 😄
@charred merlin uh, what model does Suno use for album art generation? It clearly seems to think that Sonic the Hedgehog is an acceptable subject to depict...
not mine (again), but this is the 2nd time I've heard of it happening
The covers feature is indeed very raw (Beta, as marked), but impressive.
In my case it returned he same source file in approx. 60% of all attempts, and in one case it generated very phantasmagorical hallucination (i.e., the track where neither music nor vocals match the definition).
I am eager to see how the covers will evolve and develop.
and yes, I notice it very much tried to give him side-mouth on both sides
just, uh... don't let Chris-chan see those arms
I don't understand why my songs remain in the trash folder even after permanent delete. Any guess ?
9 / 10 cover generations were the original.
Laurels? I wish. I'd be happy to get a hardy handshake.
Again another good feature (Covers) I wont ever use it for making covers, but it allows me to fix songs alot quicker, when i liked what was there but hated the mess ups by the AI, so whichever dev got that in, Good job, again keep it up.
That's mainly what I'm using it for
Yup, seems about the most appropriate use atm, helps tons
so I am having what can only be considered a weird issue
I have used all my daily points on a song and it keeps messing up verse 2 in a weird way
it keeps turning "Your poor hardcore accounts can't stand on their feet" to "your poor hardcore account is losing all night"
any way to tell suno to relook at the lyrics before starting the generation? XD
Saddly havent found any way to keep lyrics on spec yet, but you can try changing how you prompt the lyrics, sometimes prompts can cause more issues, so having it just written without any [prompt] unless required like [whisper] etc might help more, if thats already what your doing, you could try more prompts, but honestly atm its a gambit no matter what for someone trying to be conservative of points
mhm, I could try adding a promt before "can't stand on their feet" what other promts other then whisper are there? as it's bardic folk whispering won't do :p
Is that a lyric from elsewhere in the song? Or is it just random words?
First, check for invisible characters in your lyrics. Completely delete and retype (don't paste) the problem section.
If the lyrics are from a different section, try differentiating the sections more purposefully. Such as [Chorus B]
The AI takes from multiple lyrical forms online, so u will have to test and see what it hooks onto, probly alot
it is a lyric from somehwere else yeah, it seems to love putting random parts of the song in verse 2 :P
I'll try retyping it aswell 👍
mhm alright thankyou both of you :) tomorrow I shall use my points and perfect this roast song XD
This indicates a structure problem in Suno's interpretation of your lyrics. Try rewording the line or moving the stanza. Is there a bridge? Any specific tags before or after the section?
Second, try removing all style tags from the immediate vicinity, including things like [Interlude] or [Guitar Solo], things of this nature. Leave structure tags alone as long as they're not overly specific.
[Verse 1]
Oh Luthien, the king of the grind,
You’ve made over twenty accounts, but success is hard to find.
“Luthien Dyes” , you named that one well,
Falling asleep at Slayer, man, how could you do that?
Beren and Luthien junior, a legendary crew,
But they all met their end, just like you knew.
Zulrah’s not the problem, it’s your clickin’ finesse,
But taking on that snake turned into quite the mess!
[Chorus]
Luthien, oh Luthien, can’t you see?
Fishing or Wintertodt with zara might be the key!
Hardcore to the core, but you’ve lost every fight,
While we’re out here gaming, you’re just losing all night!
[Verse 2]
You face off with bosses, think you’re a pro,
But once you log in, it’s a game of "who knows?"
Lagging at Zalcano right after "Seren’s" defeat,
Your poor hardcore accounts can't stand on their feet!
Logging into laggy worlds, that’s just bad luck,
But you can’t blame the server when you’re just outta luck!
Gotta hand it to you, you’re the ultimate bait,
Can’t keep a hardcore; it’s turning into fate!
[Chorus]
Luthien, oh Luthien, can’t you see?
Fishing or Wintertodt with Zara might be the key!
Hardcore to the core, but you’ve lost every fight,
While we’re out here gaming, you’re just losing all night!
[Bridge]
Just follow Zara, forget all the hype,
Wintertodt training? Now that’s more your type!
While your rivals are grinding, achieving their dreams,
You’re stuck in a loop of familiar defeat screams.
[Chorus]
Luthien, oh Luthien, can’t you see?
Fishing or Wintertodt with Zara might be the key!
Hardcore to the core, but you’ve lost every fight,
While we’re out here gaming, you’re just losing all night!
[Outro]
So here’s to your journey, to the times that you’ve tried,
With every new account, we’ll be right by your side.
But maybe it’s time to give this a rest,
Before more of your characters find sweet final rest!```
this is what I put in
godda run to the kitchen
Well, your meter is all over the place. Try to rephrase the 9 syllable lines to fit 8 instead. Such as turning "the king" into "king"
The stronger your structure at the start, the less trouble Suno will have adapting it to subsequent verses.
mhm alright I'll see what I can do :)
Did you write these or did an LLM?
what is a LLM? ^^,
I'd also suggest breaking your verses in half. So verse 1 becomes verses 1 and 2 (4 lines each).
ChatGPT.
ahhh
I used an ai to get me a base and I wrote the rest
rewritting most of the base in the process
Large Language Models suck at rhyme and meter. They're very confident but very bad at counting syllables.
so more verses with other words? can do can do
No just add a structure tag. Put a blank line after line 4, then add [Verse 2]. Same with the current verse 2. Make it 3 and 4. Give Suno smaller chunks. Then you'll see it start to move them around and it'll give you a better clue as to how it's interpreting your song.
godda run againbefore kitchen burns down XD
struicture tag as in divider?
seni seviyorum
[Verse 1]
Oh Luthien, ~~the~~ king of the grind, (8 syllables)
You’ve made over twenty accounts (8 syllables)
*And yet* ~~but~~ success is hard to find. (8 syllables)
Luthien Dyes, you named it well! (8 syllables)
[Verse 2]
You fell ~~falling~~ asleep at Slayer, man? (8, changed to fit verse 1 structure "oh Lu-thi-en" matches "you fell a")
How could you do that? (5 syllables? Fix this)
Beren and Luthien junior (8)
*They made* a legendary crew (8)
~~But they~~ All met their end, just like you knew. (8)
And so on.
Important: I'm not saying this is what you should always do.
I'm just saying if you fix these issues you will get a better idea of what Suno is seeing. It wants to help your song be awesome, so if it's messing up, try to see what you can change in your lyrics and structure to help.
It's like a little collaborative puzzle. Hope that helps.
For further insights, see my railroad analogy: <#📚┃suno-school message>
Heres some guesses: 1) You have "losing all night" in your chorus, so that has more weight than your verses.
-
You have enough words in each meter that it's harder to find the meter as Tinda says, and also, the rhyme. Even if you don't want to rhyme much, Suno likes finding order by finding rhymes.
-
Back to having lots of words. The song may be approaching the 4 minute mark or in extensions, the 2 minute mark. As a courtesy, suno tries to squeeze it all in sometimes and thus will sacrifice words, lines and whole stanzas. In this case , suno may have chopped your lines and just found a nice rhyme , from the chorus, for "Lagging at Zalcano right..."
Again. Shortening your lines, keeping them consistently similar syllables ad Tinda says , may be one fix.
-
As mentioned, go the the original from which extended. Highlight all (Select all) and ....well copy first..... then DELETE..... add one blank space and SAVE. Go back to extension and paste in just the remaining lyrics from which timestamp you are extending. Click create and see what happens. If you ever wish to go back and paste your lyrics into the original, then you can from when you copied them.
-
As a try, you can do all the above, then make an disposable instrumental extention. An extension "D" as a go-between the previous , "A" with lyrics, and the extension "B" (with lyrics) which is to follow this instrumental. Then.... extend THIS instrumental from 00:01 with your next extension "B" to try to give suno a new starting point on the lyrics.
-
Don't wear out suno that it just becomes a noise matrix by the end. You can and likely should "Get whole song" after say three extensions and begin extending again. Does this help. Conceptually yes. I do it. Seems helpful on keeping the last couple of minutes sounding cleaner.
Oh... and save you time. Take YOUR lyrics. Go to chatgpt.... paste in and ask chatgpt to rewrite the lines in your lyrics shorter, no more than 9 syllables a line, but keep the exact same meaning and intent. ... and make it rhyme with an AABB rhyme scheme (thats seems what you like) .. and watch the magic happen!
Update on layered looping vocals. Covers changes the game. If you layer cover on cover on cover you can get much more complex overlapping vocals.
I’ve been asking the same question — I wish I had an answer, but I have no idea
Recently I’ve seen pictures that are really anatomically correct, which leans towards something more novel
Is it better to use lyrics mode for instrumentals? I struggle with getting transitions between different instruments
🤝
It can help so that you can include metatags; fret-bend for example if you wanted to add a specific vibe to your songs -- like a localized or additional style prompt.
Also, if you haven't already, try generating a couple on v3 vs v3.5 if you have the credits to spare (*maybe v2? 👀 *) -- as the model dropdown suggests, v3 is more tuned to a broad range of sounds, and can be very versatile, and just because v3.5 is the newest model doesn't mean it's the best in every regard.
#🪲┃bugs-web message as one of the devs said themselves, each model has their own strengths and weaknesses.
I appreciate the advance, I would've never known that about the model 😭
What's a good way to stop vocals from appearing using this method?
The more words you try to squeeze between the [brackets] , the more likely it will be to mistake them for lyrics.
Try to give small suggestions or "vibes" instead of trying direct commanding or coding. If you tell Suno a code, it might "understand" what you're asking in the same way a human could understand pseudocode, but it doesn't mean that having heard it they are now bound to follow that -- as a friend once said, "suno is a vibe interpreter" 
2~3 words seems to work the best -- but can still be sung sometimes! Generative AI is getting better by the week, but until then, there's MANY ways to take a song that is good and make it perfect -- we're happy to help in that case 🙂
@cinder slate @tall mesa
I will try that, though the last attemp was not even 3 minutes long 🤔
also come on discord, discord emotes ar enot malicious links, do better suno discord 😂
Hello, I am not sure this is the correct chat, but anyway. The new Suno "Upload Audio" feature blocks songs with the "Uploaded audio matches existing work of art" message, even if the blocked song is in the Public Domain.
Everything is a guess, maybe educated guess, but a guess. You have enough words , at 98 bpm [the most popular songs between 98 and 118) to take up 4.9 minutes according to Google. 120-140bpm is about the top speed one can comprehend and enjoy sung lyrics. If your song is barely 3min, it's likely singing at a fast tempo to get it all in? If it skipped some lines, then it's definitely losing lines to make it fit. Again, a guess.
Yea. Try what you have time for and get a feel for what seems to work.
I think I have a problem with my songs that, even if public, are not listed in suno search. Can someone from the suno staff tell me why this happens? The song title is "What a pain To my buttock" . Same result if I search with text "SearingElectronic291" that is inside title text
https://suno.com/song/291c40d4-4d7c-4672-9249-ea2632ff79dd
electronic rap song. Listen and make your own with Suno.
oh I scrap any song that doesn't reach atleast 130bpm XD I usually have it as a requirmenet set in the music type
@unique stone Nvm, its also good at reverse generation, so i guess ill be makin some cover-esk versions of some songs, since it seems to do a good reversal for downbeat to upbeat for instance with most of the stuff intact to some degree, so some fun to be had.
it actually adds extra lines and re-adds the part thats missing later on in the song 😂
learning this ai is confusing sometimes :p
can time stamps be used?
It's certainly got a lot of utility to it. I managed to get some percussion "stems" by prompting for only percussion stuff on a song that I'd uploaded
Ahh, ill see if i can do some break apart stuff later then, might be interesting to see what i can rip off, then rebuild using multiple cover parts and same or altered lyrics
Yes. It gets funky. Many simply accept their "funkified" song, if it makes 90% sense and entertains their tastes. Not me. And likely not you. Go for perfection and try what you can to get your results.
Anything you find that works and is helpful, I'm(and we) is/are always wanting to know. Good luck.
thankyou and you got me spot on, perfection is the standard
@tall mesa respect
Should the shared link embedded video update after updating the lyrics and regenerating the video? I am noticing it doesn't.
wonderin if theres an option to get the instrumental version of a song that ive just generated. The stem functionality affects the quality of the music 😩
Within Suno the stem functionality is the only option (because the music is generated as one from the start, not separately, so splitting is "messy") There are other ways to do the split that might give better results. For example, Ultimate Vocal Remover is a free piece of software that you can use if you have a good GPU in your computer. Or there are online services that do stem splitting.
Thanks for coming back to me
This is one of the experiments I'll be working on with Covers. Can I convince Suno to both omit the vocals and keep the same instrumental Melody? Certainly worth pursuing. And it seems possible, maybe.
Alternately, split the stems and manually upload them 60s at a time to see if Suno will reproduce it at better quality. Also very likely to be difficult and time / credit consuming.
I recorded just me speaking and ran an extension with Disco, polka style... it gave a spoken voice with a Swedish accent. So go figure.
Oops... *ran a cover.
Ooo.... wait... I ran another cover and it got my real speaking voice over disco music... progress!
Love the covers... awesome feature, folks, thanks!
I do have a question; wondering if anyone can give any pointers.
At some point in my song, I need it to sing exactly this: LA-la-la LA la-LA, full stop (capitalized LAs are stressed). But of course Suno will do all kinds of lalalalas, just not the one I need. Any way to coax it into the right lane, or am I asking too much?
The covers is fun... It's hard to "not" get the same song. But from time to time, it amazes me.
Original vs cover... Male / female change.
Try using ' instead.
LA'la'la LA la'LA
to indicate stress?
Yep. It works better. - tends add breath only. Or splits the word.
hmm, OK, let me try. Thanks!
the best way to try to replicate something exactly is to upload someone singing it that way. Otherwise, you're just gonna have to take a Suno interpretation
I can't sing for sh*, lol
Well, it gets closer (just a few extra lalas, instead of three miles long previously), but still no cigar.
I guess I'm burning all my cover credits on this🤣
At the end Suno does what Suno wants 🤷♂️
It's really hard to do those things.
Yeah, I figured. Thanks for the tip anyway, didn't know aphostrophes did anything.
How's this possible?
https://suno.com/song/5e27b199-126d-40f5-87dc-bb45a7cf2393
How one does generate this kind of sick music?
Is it luck? Is it a combination of taste and an infinite amount of Extends?
That prompt is completely experimental and lacks any kind of structure, so, how the models outputs such an AMAZING track?
France-wave, post-post-vibes, cassette, deep, deep, deep, intricate song. Listen and make your own with Suno.
Maybe @teal raft can give some insight 😄
@teal raft this is a compliment for you
Actually, thank you! Uploading audio where I'm humming the melody and singing the la-las correctly (however badly), occasionally produces exactly what's needed, whereas the last fifty attempts didn't land even once. Trois morceaux de robot pour toi :)
Glad it worked!
Interesting!
Try to punctuate it phonetically, like hmmm.. oh, if I wanted to do the "da da I'm loving it!" jingle from macdonald's I'd type it out like "da-da, da, da daaa, I'm lovin' it!"
Yeah, variations of that is what I did for the first 50 tries 🙂 Audio did the trick tho
I'm confused. I can't type the lyrics "You're dead" because it's a producer tag? What does that even mean? It's in a sentence like, twice for the refrain. These are my own lyrics btw
producer tags are a brief sound-bite that play at the start or end of a piece of music, kind of like a "signature" of a particular artist. It's likely doing this to stop people from trying to replicate known tags.
Try mis-spelling it in a way that would sound the same "Your ded" for example
thanks that worked
You can give this a try: upload pure silence, say, 6 seconds
Absolute pure silence, not just quiet
Then, extend from that silence from 00:01
See what you get
After that yeah it's just blasting through extends and seeing what you get, picking and choosing timing
JOOOOOO! Just wanted to give you information that you can get your metal tracks a better quality by using the Cover feature! 
You can nearly eliminate the "echo" and "orchester" in the Guitars
What's a good style prompt to get an orchestral song with winds, strings, etc.?
A comprehensive one beyond "orchestral"
If you take a song you created while on an non-subscribed plan, and now that you are subscribed use the "Cover" function, does that finally get you around the issue of licensing?
Do you stick to the initial lyrics and style prompt? Or do you tweak them as you extend?
In average, how many tries/time/credits you spend to be satisfied with the results?
Dang, is such a wizardry!
To my knowledge, any generation created during a subscription should belong to you, but wait for a moderator to confirm
Truly I just wing it, typically when I start I don't have a real objective when I'm ~working~
I spent a lot of credits, but it isn't like I really need to
I'm playing
So there is no 'waste'
Some songs pop out almost fully formed and just need a bit. Others I have to chase down and spend all my time on them
Others like Cool France Doritos is just messing around
https://suno.com/@yolkhead
https://suno.com/@alexayers
These are two people that are really blazing the trails
https://linktr.ee/yolkhead | Autistic. Previously published poet. I make badgood music.
I'll be back around September 16th. :-) Until then... www.alexayers.com
If you want to see weird prompting techniques it's these two. Others as well, I just know these two people
Changing up lyrics can be a bit of a no-no for Covers it seems?
I tried changing some mistakes in a song but the model just sings the original lyrics 🥴
what if you remove the original lyrics from the original song? (as in edit them and delete them from the song page)
Does that work? I'll give it a try
Nope, it seems the Cover insists on singing the previous lyrics. I guess the model gets confused if I only change up a word or two
I could just be patient for inpainting
I dont know, I've just seen the suggestion on here before
its crazy that if you make an instrumental song but put [something here] in brackets like [High Definition] as an example. it will sing that but that doesn't happen if you make a song with lyrics and put that. why is this I wonder.
Alex is amazing. I'm aspiring to get that good
Oh I will stalk further your prompt and music guys... Thank you for sharing some perspective with me!
Do it like me: See a prompt Alex or Yolk put out, then slam it whatever and hit generate a whole bunch 
On my way to giving this a try
I wanna post stuff here but I don't want to post the "raw files" that Suno spits out. I want to post the final stuff i've mastered. I got a Kpop album out for my AI band Luna Flare (its on spotify now and other places) my death metal AI band Goliath drops on Oct 31 and my personal stuff releases next week Geezah - Folding Space. I wish I could upload mastered stuff
Question: why is my [Bridge] getting cut when I record? It replaced it with my last [Pre-Chorus]?
Is it possible to tell Suno to not play any melody at all? For example if I have a track where there is a pad is it possible to tell Suno to not play any melody on top of that?
Maybe try "nonmelodic" or "minimalist" in the prompts?
I don't know how they do their stuff, specifically, but I've gotten good but strange tunes out of everything from ASCII art to the file contents of hyper-compressed jpgs.
This one was an experiment with ASCII symbols, https://suno.com/song/7fac6528-d22e-44b3-a98a-c60494913443 This one's JPG data: https://suno.com/song/18019a99-4db3-4bf2-ad02-8c3d1e46a94e this is ASCII art and an ancient unsolved cryptogram: https://suno.com/song/4eb80d70-4a82-4ec7-a90d-8bd8878e6e9b
INSTRUMENTAL haunting pop-rock score, 1968, KEYBOARD, GUITAR, BASS GUITAR, TAMBORINE, COWBELL
song. Listen and make your own with Suno.
COY SEDUCTIVE FEMALE SINGER 1980S new wave rock, electric lute, synth, bass, guitar. female chorus, mysterious & technoy song. Listen and make your own with Suno.
1978, indie rock electronic rock, France, synth-pop, new wave, disco, male and female vocals, synth, fire alarm, guitar song. Listen and make your own with Suno.
Interesting… yeah I’ve seen quite a few weird tracks using ascii characters
Checking it now
Sorry, didn't think it'd expand the links out to full length, generally it keeps 'em short.
You can use ASCII to produce recurrent muscial motiffs too, if you use symbols that the system generally refers to as notes. Usually ><,.'-=`~^/ are good options. A whole lot of my other songs use that method.
I can't recall what I did differently, but once I did a couplenof cover of an original until I got something I liked, I then did a cover of that cover and changed lyrics. It worked.
One cover was just babble, but I liked the harmony, so I did a cover , with real (test) words of the babble cover song and it was fine. Then had to do extensions and a "full" song generation. But no problem.
It will hopefully be a bit more straightforward with inpainting 😅
I got a lot of (cover)(cover)(cover)(cover) , then a few extensions, but it worked. https://suno.com/song/7ad1d0e9-d52b-4670-8c81-fe201e362d2d
Male vocals, emotional, barbershop surf, 60s rock song. Listen and make your own with Suno.
It was a test. So the song lyrics is useless. But the final test came out well enough to see I can get good results using cover. https://suno.com/song/8f09695d-c5f8-410a-80fb-21adc6a79069
Male vocals, emotional, barbershop surf, 60s rock song. Listen and make your own with Suno.
Out of curiosity, and as a rookie who’s getting started… is there a guide that curates this kind of knowledge, like the ascii chars being referred to as notes by the system
Not that I’ve found. Most of my tricks are from my own experiments
If someone had an ear for that sort of thing and made a chart it would be sweet. Assuming it’s stable enough
Even if I change all existing lyrics (aiming to get the original melody), Suno ignores that 😕
It just sings in Bulgarian (original lyrics) and not at all in English which I want to use the Cover feature for at the moment
Any English version I do get sounds far off the original
Hmm... Can I use the original's Stem and cover that?
The instrumental stem
Did you run your first cover with random lyrics to just get a melody and vocal style from your original. Then cover your cover with the lyrics you want to pull in that style and vocals. I'm not saying that works, but it's what I did . So. Maybe.
Actually, one of the latest tests is very close:
https://suno.com/song/a3cd2e42-ce22-4b85-9b52-a7e0d70c73c4
There's definitely RNG
Yes. You can cover stems. I think so.
I don't know for what I'm listening , to listen to your song? But it's nice.
I have never used stems because I didn't want to personally, but any idea if instrumentals somehow keep any vocal context in mind for Covers to use?
Because AI is about 80% RANDOM.
Does a vocal fingerprint remain in the stem even if the vocals are stripped out? Digitally, it's a possibility. What AI does eith the instrumental though, might not retain the vocal melody you would need. Blah... blah... yadda yadda. Just have to try and see what happens. 🙂
The best way to find out is the empirical method for sure
Yeah but for something like this, it actually should be a bit more structured. Ai is only scattered if it doesn't have guide rails. That's the problem we need to sort out with AI now. It should be able to regenerate on something it "hears" or rather reads when you extend or reuse a prompt. We're not there yet. Another year and we will be
Hi Suno, I am using the download video feature and if I change the image on the song details it doesn't update on the video. I have tried regenerating video several times but no luck https://suno.com/song/171d7180-4e60-40e0-a2d9-145338a9091f
deep funk, ska, rock song. Listen and make your own with Suno.
AI starts with a noise field and removes anything that doesn't fit the criteria of your prompts. But stem stripping is more of an analog process, in my mind. Removing certain frequencies represented in this case, digitally. Again, to reiterate, I'm mostly stupid, so guessing.
Upon testing, I feel like Covering an Instrumental does get me closer. Either a "fingerprint" as you put it might be there, or it could be because instrumental stems are bad at isolating vocals so you can hear feint vocals
Think it might reach behind the stems and still track off of the original? As it seems to do with extensions, often singing lyrics from the original despite changing lyrics in the extensions.
https://suno.com/song/b61db336-2e4c-4380-9a11-4b818d130863
Almost the best result in terms of closeness but I feel like the lyrics are paced in a bizarre way at the beginning
Perhaps
Well you feel you're going in a forward direction. That helps.
For sure. So bottom line, covering stems is a little easier when changing languages as opposed to covering the original
It's interesting
I get the pace thing. All I can do is run off about 8 gens and see if suno starts on a different beat or inflects differently.
Well it worked good with babble vocals that have no language or word equivalent. So I'm going to try humming or scatting a melody and see how clean that works.
I did spoken word recording. Failed. No melody, so ....useless.
I wish there was a parameter (slider) which would let you control the likeness, in other words something line denoising strength
Right now it's random and some results can feel like 0.1 denoising
Others like 1
It's a mysterious error at this point. I hit it with the phrase "what do you want me to do?" But that's way easier to rewrite than "you're dead"
Ur dead maybe?
My railroad analogy might help. <#📚┃suno-school message>
Thank you
That seems to be a Discord thing. Once you post a Suno link, it embeds the current video and caches it somewhere. If you regenerate the video and post the same link again, it always uses the older embedded/cached video.
Hello all, I'm not sure where to post this, maybe here.
Cover functionality: I took a rather lengthy song I made (7:59) and tried generating covers for it.
Suno made a... 2-second song.
Original song: https://suno.com/song/9a4eea02-eb1d-439f-b250-8804c3c237b3
Cover song: https://suno.com/song/f490b4ed-3f0f-403b-9c97-eb2dbdafcf01
While I commend its impressive summarization skill, I would have expected a slightly longer song 🙂
doom metal, slow, aggressive, melodic, piano, drums song. Listen and make your own with Suno.
Ska, humorous, fast beat, vocaloid, brass instruments song. Listen and make your own with Suno.
grunge metal, sludge metal, low tuned guitars, 70's metal., slow tempo song. Listen and make your own with Suno.
I am absolutely loving the Covers feature. It doesn't always work, but when it does ... 😍
I'm liking the cover feature. I know it's in BETA. Just providing feedback that 6 of 10 covers are producing nothing new, same content.
Hearing a song you made months ago as a stripped back acoustic version breathes new life into it. Makes it feel more like a real song, Plus 'covering' a song you made in v3a in 3.5 can give you a cleaner sound.
do u guys know a tool for only get the beat from a suno song?
Yeah, it's hit and miss per song for me, with some songs it's more like 8-9/10 are just the same song!
Heyy, As im experimenting with the cover feature, I have noticed that sometimes it comes up with interesting stuff.
For example: I am experimenting with trying to "upscale" the song with the cover feature, maybe bring in some now sounds to the same song or "clean it up" a bit with prompts. And most come back as the same song with slight differences (As expected) But I got this one and it kinda "scared" me.
Because it's a completely different song and the lyrics are not mine.
It sounds very good (for me) the vocals and everything is cool, but for me it kinda sounds too "real".
What do you guys think could this be?
Random suno weirdness?
Or maybe suno is copying a song?
[Indie pop rap], [ alternative r&b] song. Listen and make your own with Suno.
Yea, every so often I'll get some lyrics that are not mine. Sometimes you get these outlier outputs that are just nothing to do with the prompt
I've had a few like this. The lyrics are basically gibberish with the occasional real word thrown in as the AI tries to figure out what sound to make next.
it's one reason I wish we had control of the strength of the cover because mine vary from close-to-original, to completely different song
Yeah, I have gotten the mumble lyrics too but these ones are a bit to o "clear" I guess. Maybe it's just me. 😅 I even have a playlist with songs that sound nice with those gibberish lyrics, but yeah this one caught me off guard.
I guess suno is getting better at mubling too! 😁😁
"vocal stem" as a cover style with nothing else gives some curious results. Complete silence apart from singing, but the singing is usually nothing to do with the original song. I had one singing "I love you, don't know how" in two places and then tailed off into gibberish.
(sometimes it just gives random noises like an intermittent snare drum or a single bass note)
Damn, what a fascinating tool! 😂 Cant get enough of suno 🥰
I have a feeling that there may be some kind of stems involved in what powers the covers feature which is why I was poking around with this prompt... it's interesting to see what else it can do besides just straightforward covers of a song. So far I've managed to clean up some previously noisy songs, convince it to give me some isolated vocal, drum and guitar parts of different songs, and then improvise over those (very hit/miss though, this all took quite a few tries.)
I think you have a point, I've been testing for few hours, and it does some interesting stuff, need to keep trying!
I tried but after some generations I observed that Suno tends to produce a melody most of the times (this is an example of what I was diving into: https://suno.com/song/8ea96b96-cd17-4672-badf-4d15aeaee9be) but man, it was damn hard to achieve consistent results (Maybe it's because I'm bad at prompting) but thx anyway
In the Suno Remix Contest (feat. Flosstradamus), how can one extend a track to add outro and get full song without including the uploaded original?
Im just a humble member here, so this is my best answer with limited knowledge.
First, i assume "uploaded original" means ones of the choice of stems offered to be remixed by extensions?
I think it's built into the definition of "full song" that it includes the original. So, a trimming feature to trim initial , mid or end sections are not available in suno at this time that i can find.
https://www.bandlab.com/splitter
This free tool will separate your songs into vocals / drums / bass / and all remaining instruments grouped as "other". (there is a paid subscription to add tracks for guitar and piano, but the free tool is revolutionary to have for free 💪)
If you have a fairly good GPU in your computer there's also Ultimate Vocal Remover which is free (I just use this since it saves having to upload/download audio to another service.)
Thought of a good way to describe why things like audio2audio and image2image are so powerful in ML. It's very similar to the difference between a finite state machine and a turing machine.
There are limits to where I can end up in a finite state machine or a particular model. But once I can iterate using my own outputs, I can functionally get anything I want (this being analogous to the tape in a turing machine)
This may be a bad analogy, but I love turing machines lol
So I'm making it
Hy, 😦 not working [Verse 1 Male] [ Verse 2 Female] how to select Voice Gender ?
For a brief moment I was able to extend and cover at the same time and it was great, but the fixed the bug and im sad lol
I don't understand the concept of covers, the generations I get are different tonality and different everything, it's as if I generated the same lyrics with a different prompt. There's no comparison to believe that it was meant to be a cover. Is there a way to make the tonality, tempo or vocal character match the original, to make it at least minimally similar?
I'm having the same issue. I've even had whole lines changed but it still sings the original line.
https://suno.com/song/e7e24f57-df68-4d0a-817e-f5bd6a1a8da3
Beethoven in here
anti-Microsoft Windows PC anti-sample insight wall, error XP 95 windows err boot samples, harsh drone wall song. Listen and make your own with Suno.
Literally, it's that one that's p easy to play the opening from
Fur elise
I would agree with that. Basically same for me.
From 00:13
It starts a portion
It's a variation, but it's distinct
V cool
However, does indicate slight overtraining
I'm beginning to wonder if the safety training involved classical
I assumed previously there wasn't any
But now I'm thinking there was
Has anyone experimented with case use in prompts? upper, lower and title case? Saw a guide that mentioned upper case is given highest priority/representation when following a prompt, followed by title case and then lower case.
Anybody else having trouble renewing their Suno subscription? I got an email saying the payment didn’t go through, so I went directly to my Suno account and tried different credit cards, all declined
Has anyone had any luck getting their older v2 songs to be able to be covered? I just get an error that says "Your clip has been stuck"
Hi, who can I contact to resolve the issue of scam transactions through your service?
My card used to purchase subscription on your AI I guess?
There were 2 transactions for 30$ from my card, Which I didn't do
Why the songs so short when on 3.5? Very annoying
then you should probably contact the company that you have the credit card with
they cannot revoke an already confirmed transaction
Yes, it worked for me. Try deleting cache maybe?
If there are transactions you didn't do, have you considered that someone else could have done them?
if someone is charging stuff to your credit card (and it's not you) I would be concerned about my identity being stolen
no one has access to my card except me)
Yes it is, I'll block my card after refund
@charred merlin Trying to follow your guide on terminators and pivots. Finding it hard to do in practice. How is it actually done? I try to do one style with prompts for intro, chorus, end and then extend it with a new style prompt, but it just continues the old style.
Well Suno as a service doesn't scam people nor absorb money from wallets
I have never been charged unknowingly nor multiple times
Then this makes it more likely your card could've been stolen
I know, someone used my card for sub on Keno, to, I guess sell the account
you should contact the company and cancel the card now, before someone charges something really expensive to it
on krea and midjourney I already got the refund
I've set limits to 0
Already
Oh, tyvm!
I'm actually free at the moment 😄 Normally spinning a couple plates & get distracted on the follow-up, but I'd be happy to take a look at the song you're working on & see what I can help with 🙂
Let's go in PM. Thanks
Hello everyone. I’m having trouble making the second part of a song have an ending when extending it. I’ve tried using the end tags, but almost all extensions last 1:59 minutes without reaching a proper song ending. Any solution?
find where you want the song to end. switch to model 3, put [End] in the lyrics, and 'Ending, Short" in the prompt box.
How can I tell suno to do like:
" lyrics
(Put the uploaded audio clip here)
Back to my lyrics"?
Only thing I've been able to do is have it start with the audio and extend it.
What's the promt box? Style of music box? 🤔
Yes. Sorry.
Thanks a lot!
No way to do that now, I don't think. Would be a nice to place it where you want and build around it.
Ah gotcha gotcha, ty for responding.
I'm thinking then best way now is gen first half, extend it from 0:00 to gen the second half and use audacity to chop it up so it goes
1st half-audio-2nd half?
And just hope it lines up ig?
Is there currently a way to keep or reuse a certain voice you like for addtional song? There was a voice for a song I generated and it was absoultely banging! If there is currently no way to reuse/save particular voices, is that something that Suno intends to implement in a future update?
I have a question, when you put the lyrics in the box and use [Start], [Melodic Interlude] most of the time the song start with direct lyrics no music in the starting any idea
Someone please help me with this question - I love to create bootleg versions of the top songs on SUNO ai - 1. can I post them publicly on SUNO only ? 2. I'm just adding a verse to the origanal song, is that ok?
At the end of the Flosstradamus Soundclash master there are samples that are not reflected in the provided lyrics. Has anyone figured out what the words are? At ~2:53 there is a sample that says something like "Take it up but don't break it up". I could be totally wrong on that though.
I tried a few things today that did just that. I took the clip from a voice I liked from one of my Suno songs and I uploaded it, and then created a song from that clip using extension. It worked with two different clips and two different voices. Give it a try!
mine is doing this too and it's really annoying. Mostly seems to be on v2 and v3 songs...I really wish I could get it to work. I tried taking a part of it and uploading it, but I really want to re-cover the full v2 song, but its to long to upload. :\
I found out that it did it because I tried to use an audio i uploaded before it existed
I had to reupload the audio
i hope at one point they will let us upload a full 3 or 4 minutes of audio for the Cover feature, not just 60 seconds
i hope for that too
I just extended the instrumental “stems” or at least I hope I did .. 😂
I was unaware vocals were required for this latest contest
how long do I need to make the clip to extend from?
I might be unable to pay...but I would find a way if it gave me the ability to make it redo only a part of the song it created ^^,
IT WORKED!! Thanks!!
just a curious question, but is suno capable of making sound effects if you use the V2 version of suno at all? Just thought this might be useful for creating custom sound effects lol (like not songs but sound effects)
Eleven Labs actually has a feature for making sound effects, might be better tool
i notice sometimes the covers just create an entire new song which stinks for taking up my credits lol
how do you extend the instrumental period between verses and prevent the ai from repeating stuff?
that's why they gave the 200 free ones
where the hell am I going wrong?! https://suno.com/song/e880d9a2-1332-4a54-acf3-979188e073ce
Cybernetic Metal, Roland Bass Drones, defensive, but shifts genres song. Listen and make your own with Suno.
hi all.. im playing around with the old bark software and it seems the weights for: mi-test.pth are nowhere to be found any longer. Any ideas how find it? thnx
Just came back to Suno. I see some cool updates and I'm glad about it. Anyway, I'm attempting to create complex rhythms in my songs and curious about some better prompt engineering on my end. Typically I use custom so I can toss my own lyrics and genres in. Any technical advice for this?
I have a question about a song I've written. I've made some changes to the lyrics, but they've affected the beat. Is there a way I can keep the beat while making the necessary changes to the lyrics?
Not really. Your best bet is to go "Create -> Get stem" and then rework from there.
My way when I find an intro or intro+vers I like is to download the song, cut it with Audacity, upload it as audio file and extending it.
Gauh, the cover feature is just broken for me now. It doesn't even do anything close to the source audio it's a bit annoying xd
i am free user, i dont think i have access to Get stem
Think I need some help with tags, is this not where they go or is this incorrect? i am not getting a female voice or acoustic
i usually do this: [Verse 1 - Acoustic Guitar, Female Voice] and it works for me
alrighty, lets give it a shot
hmmm about 50% female, and 100% ignoring the instrument, weird, ill keep fiddling
I think it would be great if there was a way to set the style prompt for a specific verse, chorus or bridge
Also a feature where if you input audio you can select a percent of how similar you want it to the source audio. Sometimes results end up in an entire different genre.
@charred merlin and I have been doing some hard trials with that the last few hours. Try this: Part 1 is the style you want for the song. Part 2, extend from the point the verse/chorus/bridge will start AND THIS PART IS IMPORTANT - Delete ALL your previous prompts and write the new style without any of the earlier prompts. The song will continue but the style shift will take place where you extend.
There are no sure things. Just like ChatGPT still lies, so will Suno not always deliver what you ask.
No thats fine and I get that, I just need to know if I am in the ballpark, with formatting and placement
https://suno.com/song/0cb11bff-6063-474e-a91b-68ddb42eb156
This took plenty of trials, about 800 credits. But finally I got both the style shifts and pace shifts I wanted to fit the lyrics.
melodic death metal, tight, heavy drums, powerful growls, dynamic, piano, female vocalist, guitar harmonies, operatic song. Listen and make your own with Suno.
well i know thats possible, but it would consume too much credit since im a free user
A more randomized but still effective way than extending is to include prompts that naturally have style shifts. I like "dynamic" for the music and "duet / operatic / symfonic / musical / duel" for vocals, then I try to tag it where I want it to happen, see the video above.
Would the input in the style of music field effect the tags at all?
My experience is that the prompts (style of music) is giving the ingredients to Suno, the tags are the recipe on how to cook it. If you don't give it chicken in the prompt, it does not matter if you write fish in the tags.
(I have no idea on what I did with that metaphor)
lol confused the living hell out of me is what you did with that, But I am assuming one is just as important as the other its just no matter what if the style of music is country, I cant get an acoustic, I cant get a harmonica, The plan is acoustic for verses and full band in chorus
Suno is not fact based. It is emotion based. See my video from before. I have more or less zero luck just writing an instrument, but if I write the instrument and an emotion, Suno MIGHT just pick it up. So tag your verses, chorus and bridges with emotions.
Try "[Verse 1: calm guitar]" + "[Chorus: intense country]" or something in that way.
I see i see I was going through that from the suno docs, but yea ill take a look at your video and keep playing I appreciate the help
Hello, I am trying to use a sound bite in a song? It's from a recent debate that I am sure no one watched. When I go to upload it. It says this work is already claimed or something like that.
Do you have the exclusive rights to it? If not, you would be violating Suno's TOS by uploading it.
You likely got a message from Suno saying it cannot upload the clip because there was a hit against the ContentID system, which makes sense, as the clip is copyrighted.
Even if you could use copyrighted material on Suno (with having the appropriare rights/licenses), which you cannot with Suno's strict TOS, I would not risk dealing with that particular material because you are dealing with Big players that can sue you and put you through court for several years if they wanted to.
Even if Fair Use would be legible somehow, it would have to be determined through the court system, which will become very expensive
Hello!
How to prevent Suno from speeding up the music with long lyrics? I noticed that when I add long song lyrics (5 verses, a bridge, and a chorus, each with 4 lines), Suno forces a faster pace in the music, finishing 'singing the entire lyrics' before the 2-minute mark, even though there's no reason for it to do that.
Is this a known issue?
Is there a way to stop it from speeding up the tempo just because the lyrics are long?
I mean
you put the text from a REAL song
it returns an error lmfaooo
"how to DESTROY the concept of a cover"
It's meant to cover your OWN songs and uploads
Just to point out, knowing you are using the free service. Even the same lyrics, no changes, if running , say, 6 generations back to bak,will get different beats, metering, styles. I assume the idea is , given there are 8000 variations of how a melody can go, Suno is geared to randomize as much possible. Since I'm on a higher account, I run several then pick the best one and extend from that. Then, from that one, listen to where sunonstruggles with meter -- such as I'm too wordy-- then I'm a bit safer, assured, that making small changes should only helps. Of course extensions mostly keep the same meter and beat, but even then, suno can change up .
Now for your question, you changed lyrics so suno changed meter. Yes. Makes sense. In an extension, for suno, I think it tries best to maintain whatever is in the origin, but if sang 4 quarter notes on a line which you added a three syllable word, it might not figure how to sing in quarter time and sing 2 words in 2/4 time or many words in 7/8 to get back to the 4/4 rhythm. IOW, lose beat.
So yea, tricks to fix. Well last things first, with the new cover feature, that might help emulate whatever beat you prefer, but that's down the line for free service.
Next, you can, when changing lyrics, try to visually match your new words nea to syllables and length as the old.
Next, you can look where it starts to get wonky and find a rhyming ending where you can break the line , line-space down to put remaining words on a new line.... suno light work this out to add the last words on beat.
it still destroy the concept of a real cover.
great dont use it
pointing out something isn't hating bro
Ran out of space... Oops.
And, you can focus on being a bit structured with your verbiage, so that the lines have similar syllables and length. This can be boring. But you can, and I think it helps, change up the metering (lengths) per stanza. So, if you write a long 12 word line, and then 3 more. OK. Maybe one more stanza of 4-lines 12 words per line. But space down and throw in some anthemic lines, short anthems, shouted words, or just a stanza of 4 lines but at 8 syllables or less.
But point being. Prep your lyrics carefully before running the first original generation and try to appease the gods with what looks like structured lyrics , whatever that means, more than what looks like just lines of prose or poetry. Suno is trained on lyrics of previous and popular songs. So the best means is to pull up lyrics of songs you like and see how they are structured. You may be writing that way already. I don't know.
So those are a few suggestions for your tool box. Use at will or toss in the fire.
First question is: Is it the same fast pace, speed, beats per minute, throughout the song. Or you mean it gradually increase tempo from beginning (slower) to end (faster) in the one song generation.
In my case, the question is:
The song already starts at a faster pace.
It doesn't gradually build up; it starts off in a faster style.
When I add fewer verses, I can create the same style but without the accelerated pace. It seems like Suno feels obligated to create something faster to 'sing' the entire song, even though it's unnecessary.
Sometimes it even sings the entire lyrics twice, meaning there's no reason for the pace to be so fast 😐
what are your prompts? Try putting "slow" in the Style box
Also try breaking up the lyrics.
`[Verse]
My first line is this one
My ssecond line here
And this is the third line
Do you see what I'm doing?`
and put periods at the end of the lines
Last few days I feel like I’m back on v 3.0 even it says 3.5. All the old problems back again - wrong stresses in words, messing up lyrics. Did anyone else notice any issues? I’m running 15-20 variations just to get all words pronounced correctly 😩
Covers are not working for me. I've tried 10 times with different recordings. It always attempts it and then just gives up.
Beta
We're all experiencing the same
I know it's " beta ". Just reporting that it's down at the moment.
You'll notice that certain genres work better, for me I've found that these work more often:
banjo
country
disco
funk
indie
can one do a cover of a cover? I have a cover that I love, but it's a little bit too fast. Can I cover that, and tell it "Medium Tempo"? Or will the results likely be garbled and not understandable? I know I can always try, but I wanted to ask if someone else had any success.
It's not that it's not working the way I want it to.. it's that it's not working at all. It won't generate anything, it just loads and then stops trying.
yes
ok thanks I will try
Results depend per genre and what you want it to do. And of course, it's beta 😉
Was this your original layout of lyrics or this was your attempt to fix the speed issue on a later attempt?
If former, we would need a control that has your lyrics as the singlespaced line stanzas to see how suno treats it. By its own example, suno writes its on lyrics in singlespaced, often 4-lined , stanzas.
If the latter, then the typical ballad or non-prompt tempo is likely 98 - 100 bpm. A total guess for me, but even the average suno created song, say for rock, pop, kpop, is 120 bpm, and glitchy EDM synthcore up to 130 or more.
Honestly I have NOT YET looked for a suno-created lyrics song in a hardcore genre to see how many words suno might use compared to the typical 8 syllable , line pop ballad stanzas.
So, you're saying you have few enough words to stretch out to 3-4bminutes but suno sings it fast and cuts out early in the pre-3-minute mark? And I guess either ending time there or adding in odd hallucinations or repetitive verses or piano music or something as a filler?
All these questions is to flush out what might be happening.
Here's what suno does when I try to do a cover:
What Browser are you using?
Do you have enough credits?
Yes, I have 2500 credits. And the free cover credits. I'm using google chrome. Never had an issue before.
Personally, I had to move completely over to Opera GX a few weeks back, it just stopped working for me
Ok, I'm going to try this on opera, I'll report back in a couple of minutes
Do your covers just sound the same, but maybe noisier, static-y? Or they sound ok, but with glitching, fade in and outs , warbled singing/instruments and such.
Are you switching up with single prompt genres to really make it easy for suno to get the point to give you something different?
So..... can you just post your suno library version?
I do different things dependent on what I'm trying to get and what style I'm using. You have to try things out and see what works for you.
There is not a way to get BPM so we do all sorts of things to work around it.
I'm about to make dinner so I'll be back later.
I get the same result on Opera. I'm just feeding it a short piano melody I played. And trying different genres. It never goes through. Also tried singing. Same result.
I'm using suno 3.5..... where do I see which library version I'm using?
Ok. Enjoy dinner.
For when you get back. You are saying your "different" attempts , be that they may be different genres and style, yet ALL are sung at a fast pace? Can we take one and re-structure it to look like the text of a "typical suno created song", use "pop" as the style and see if it outputs the normal Suno <City of neon lights > ballad? That would be a good starting point for this discussion to have a control song.
Hi friends does 3.5 support “tags” Or music prompts? If so where can I find that list?
there is no official list but you can find an unofficial wiki in the docs-guides-resources channel
None of the recordings I've uploaded before work for covers. But if I upload the same file again, I got it to work (at least at this moment).
.... and of these discussion I've delved into, i might need to continue later. Only have 7% left on my pacemaker. Will need to recharge. Lol
Are you asking me to a share a link to a song or recording, on here?
But anyway, I got this to work by re-uploading the files I had uploaded. That seems to do the trick for now.
When you posted " here's what suno does when I try to do a cover" and posted a gyazo.com link, I felt a direct link of the actual suno output might be better. But also, I don't click on outside links in here, so theres that. 😄
But, you seem to be happy with a solution so.... Hazzah!! Good job!
Ask questions anytime. I learn more that way.
I understand, it was just a video of it loading and then nothing happening.
Thanks
I'd say... what we consider the main/major/top 12-20 genres we might see in a "song contest categories" list, are likely distinct enough to show a wider diverse sound from suno. Yep. Something like vaporcore or hypnogogic, not so much.
Oh cool. Glad you spent the time to make the vid. Other on here are far more knowledgeable and so something like that can really help.
Hey, I hope your solution stays consistent. Let us know.
Will do, thank you 🙂
Thank you again for your help looking into what was and wasn't working regarding Terminator Extensions -- I'd always focused so heavily on the v3x series that I really didn't consider v2 for technical issues, but Lindormen's Skatt was really exciting -- I'll reach out when our timezones overlap & we'll figure out getting the song onto your suno account (extensions with permission, etc), or I'll send a mastered finished version 💪
Wait are you trying to sslow down or go faster?
COVERCEPTION? Yes. It doesn't really do what you'd hope. More like playing a game of telephone as it slowly wanders away from the original.
Not a song I'll be using but it's a good example of Covers having clearer vocals and instruments (this is 99% identical to the original especially the first parts). It's interesting, and it makes me wonder what powers Covers. Is it v4, or is the quality better because there's also an audio input?
https://suno.com/song/6bf7c163-5587-4d7d-91d5-8d61a47e3ae0
Arabesque vibes, Middle Eastern ornamentation, female vocals, oriental strings, Turkish vibe, infectious, fast song. Listen and make your own with Suno.
The only purpose of this song is to be an example so I won't publish it
The original definitely sounds less clean
I think it was @alpine venture who described it as Covers having an Udio-esque quality. My bad for the mention if I misremembered ✌️
🤣
Try replacing your style prompt with "end" and adding [Outro] and [End] tags
Change to V2 or V3, style prompt "end, end" and [end] in lyrics. If you just want to get a quick ending like a fade out. Suno Tweeted that better Editing functions are in the works, so hang on.
Good to know! Hope that also includes inpainting. I'm no fortune teller but hope is free of charge 😂
Yes; I can't remember when, but that sounds right
Udio has had superior sound for a while, but their composition needs a lot of work
It's possible, but Suno is like a genius at composing
It's like the deep blue of progression
And ironically, I think unlike chess, the old fashioned way won't survive
Since there was a deeply embedded gaming culture surrounding chess, whereas nobody is like "hey, let's all go to the composing tournament and go make some compositions." Bands will obvi still be a thing, but video games, movies, etc will all switch.
But yea, udio is a lot more finicky; I was mostly using it to "upscale" and denoise Suno gens
But with the tighter control from the cover feature, the outputs are a lot more polished
I agree on the comment about Udio. I just don't quite like its outputs nor extensions. Suno is very well-rounded. The only critique is the robotic vocals, but it looks like they may go away soon.
I made a mistake in a word but I didn't realize it until I had a good music. Is there a way to correct this word without the music being completely destroyed/replaced by something else because of the extend button?
Oh, def, you can actually get clean vocals now w the audio2audio
I'll make a vid Abt it when I can
https://suno.com/song/032235e0-353e-4cf6-b599-3dac588bb75f
Most trad example
If your song is under 3 minutes, try Covering it. It may fix your lyric of choice
What😭😂
3:03
It will work, Covers can be up to 3:25ish long
I mean, Covering will cover the entire song
But I'm not sure if the lyric update will be easy
goth trap, lo-fi indietronica, underproduced, nostalgic furrycore, watery guitar solosolosolosolosolo*->1010101010111111 song. Listen and make your own with Suno.
And this
how to covert a song? is it the upload option?
Extreme ckean
It should be possible. Post the link
I'll take a listen in a bit ✌️
oh i see it now
What about cutting the end with external tools and sending that?
ty
Sure; you can follow the cover chain to see the process
No
after try i admit i don't understand what cover do
Cover covers
Oh I'm in the wrong channel
does anyone know about how to limit the track generation to as specfic length of time? For example, like generate a song that is less than 30s
or just anyway to make the output quicker and shorter
There is no direct way to do this. It's probably hard to do as a Clip 1. You may need to Extend a song from very early using model V2 or V3 to end it quick, within 30 seconds. Alternatively repeat your short song a bunch of times in the clip and maybe you get 3:00 of song, but it's 5 complete 30 second seconds.
i have try to covert but that only give a different song...
https://suno.com/song/1fbaf433-c760-4878-8cba-456f2dfa2e56 here the original song (the things i want to change is in 1:05 i have write "un" but should have write "une"
epic, disco song. Listen and make your own with Suno.
with the covert that give ... somethings else https://suno.com/song/8cf0f666-485e-4f7c-b51c-ba34287ee50f
epic, disco song. Listen and make your own with Suno.
I am a Premiere Plan user and have been trying for two hours to use the new cover feature. I get nothing but this message "clip got stuck while generating". I am trying to revive a V2 song that I have been unable to extend. Any suggestions?
Today, after multiple efforts, I haven't been able to use Cover option succesfully. Either it gives me the exact same song (or I can't tell the difference at all) or it give me something totally different (when I use live version as a prompt, it litterally improvises something that is not remotely using the lyrics of musical scheme of the original)
how i solve this?
Try extending the V2 song with V3.5 and then Covering it maybe. But it could be an old buggy V2 clip.
I'll tackle it again tomorrow. Need to sleeeeep zzzz...
As general advice for anything in Suno, if something isn't working, it might just be a very unlucky prompt or song. Don't retry too many times. Take a break and try Covering some other songs. If everything is failing, maybe there is something in your prompts causing problems. But probably other songs work better, and it's just an unlucky song.
No suggestions much, but from my experience, the error has been always premature. I wait. Back out of what I'm doing and go to create page and click refresh. Wait. And soon the gens pop up and start generating. Apparently "stuck" is not permanent?
New method:
https://www.reddit.com/r/SunoAI/s/rVjegw2qH8
Same two things happened to me. So it's an ongoing theme for the beta. I used the "different" song to my advantage as it had a melody. I just "covered the cover" . Wrotes some lyrics to test and covered the "babble-scat song" into a "real" song. Maybe useless as a whole, but from that experience I've gotten better cover gens by taking any cover that I like from the list of "bad" or cloned results, and just re-cover the cover with the new prompts. That seemed to help get a cleaner result and follow the prompts. A workaround for sure and done more to diagnose what's going on compared to what should be happening on a 1-time cover attempt.
OK. I'm lost at this point. I thought we were discussing your attempt to stop Suno from outputting fast tempo songs when it was not your intent, nor had you prompted that way. That seemed your original issue.
I then was trying to get an idea of what you felt was a tempo you wanted as a result. And perhaps discern whether its the structure of the lyrics or the prompts that might be making suno generate fast tempo-ed songs. I suggested we create a control song (of anything) that you felt was the right speed - to compare to see what differences there were.
You had to leave for a while. I left to return this morning. Then you asked if i were trying to slow down or go faster. That confused me. But again, I'm trying to define what you think is a tempo/speed you like and use that to compare lyric structure and prompts against all the song that are outputted too fast.
Have you resolved your issue in the meantime? It's always poor to have just one person's ideas, especially mine, I humbly interject, as I am making a best guess from my experience so far. Until others jump in and concur my ideas to compare your fast songs with what you think is a "regular" song created by you, then we could be going down a rabbit hole.
So, again. Still having issues? Im happy to keep going.
Hi Team SUNO @hybrid arch - a new feature called Covers is for saying a song into different style...It is not working at all I have tried some of My Songs but it is giving the same results each time for different songs & I lost My most of the credits. I have send the bug & inappropriate reports in the app...Please check this issue..
Also work on the upload option because it is also not working as a proper option it changes the upload voice vocals into it's inbuilt Voice Modals after song generations...
I should clarify the issue: On the latter, you are saying, if you upload an audio of, let's say, your singing a song while strumming a guitar, then the first extension of that audio should emulate your voice exactly or nearly, or at least with many of the characters or nuances to sound similar to your voice -- rather than an inbuilt voice bank -- to that audio which you uploaded containing your voice?
[ keep in mind, any recognized vocals in an upload will limit your output to "private" only for your enjoyment and cannot be disbursed publicly. ]
ok thanks a lot for the information But I tried My voice and put the time below 10 seconds but it didn't give the exact results also My vocal lyrics which I sung also mixed with the new generated song in the starting Intro portion and also in some songs some of the parts of lyrics in between the songs also. And I always put My songs private.
As for COVERS. I have gotten my best results, so far, by only using 1 or 2, at most, prompts in my styles area. Using more than that seems to just give back a literal cover of the original. A copy, in other words.
As I'm on pro/prem subscription, I will run a few covers as outputs, with whatever new prompt styles. I then listen and choose one I like. Then "cover the cover" or generate one more cover of this new cover. That seems to get me a different style song, but retain the vocals and nuances of the original. An extra step, and I can say it may simply be a placebo effect. I would need to run that pattern a few times to see if it's a consistent thing.
Again, using one or two prompts is helpful to change up the cover enough to be distinctive.
Ok Thanks for Your Help & Support
Feom your description, It seems to be outputting as it should on an "upload audio" and extend. It is not expected to emulate your voice, but at the most have some of your vocal qualities. But most time, as you say, will apply the inbuilt voice modals. At least that is my understanding.
You can also split the difference and upload closer to 40 seconds to see if it picks up more of your voice qualities.
To point, in some of my uploads with my voice, the output may be more gravely or deeper mixed in with the inbuilt voice, rather than just the usual "inbuilt voice modals" as you describe them.
But I don't think it is Suno intent to copy exactly any inputted voice, so it's designed to avoid that. I would not expect to hear my voice in any of the outputs.
Why do I keep getting 1 and half minutes worth of track. And then I'm unable to extend just stops creating and refunds me the credits?? Any idea guys or help will be greatly appreciated 👍🏻
It does happen. Suno could give you a full 2 minutes or a shorter time. The minimum is 8 seconds before it then credits you back. The good thing is if it is a good 6 seconds, then you are free to use the shorts ones as well.
Yes, it's annoying in the sense of using credits. Thers no way to say give me 2 minutes and end the song exactly at 1:59:59. Further as it's only a 2 minute extension, most times we will have to extend it once more anyway to get back our 3:37 minute song.
So, part of a generative AI persona is to give back a random result that approximates your prompts, we hope. So random can be anything from 8 seconds to not even ending the song and bleeding over at 2 minutes.
But technically, 1) if you have a blank area at the end of a previous generation or original, suno will often just try for something short and give you 4 seconds of a musical notation or just a blip.
- if your total song time is approaching that 4 minute mark, suno may try to accommodate by shortening your extension a bit, and either end it, or fill the rest of the time with hallucinations and variants.
Those may be 2 anticipated reasons for getting a shorter time result. The rest of the short ones are more random in nature.
That's about all I can say, as it's random. Suno themselves say in using suno, the results may vary from person to person, generation to generation. So , at minimum those 10 credits have a valued output of a random 8 second to 120 seconds of either something expected or surprises, hopefully in our favor. I've used lots of the "suprises" [hallucinations/ babbling/sound effects] as just new uploads to spin off another song.
Ok thanks 👍🏻
I tried the cuver beta, but when I click on it nothing happens. I do have the pro subscription so I wouldn't really know what the problem is.
Refresh your browser to see if it helps initiate the generations.
trying to get a song witch vocal and melody and bass and I want to skip the drums is that even possible? a negateive prompt field would be cool.
Add [Fade out] as well.
Not knowing your prompts, anything to do with 1970s style songs will include a fade out at least 3% of the time, is my estimate. Is this just instrumental or included lyrics?
Uhhhh....I thought you were having the issue and was trying to help you. I was also confused because at first it seemed you were looking to slow down the singer and so I made some suggestions in addition to the good ones already posted. Then it looked like perhaps I'd misunderstood you and you wanted to speed it up, which is a different process. So I asked to clarify.
Wasn't trying to confuse you or be confusing. I'm not having any issues at the moment. Thanks for the offer.
Checked off..... and filed. 🫡
✅ Warning logged for puggu_nehal. I couldn't DM them. || For future reference, please do not needlessly ping members of the Suno teams needlessly. The community will be able to help you out in most cases just like the above
My (cover)(cover)(cover)(cover) song worked out just fine. So, it's random luck? Who knows.
https://suno.com/song/f6d441a2-67ff-473e-a497-d64c8816fcd5
This may be the most interesting from a sound perspective and technical I've made
JamesLemonHouse delta Blues, JimiBuilt steel string guitar, #0 yr- poly wat -math,-ina poly slacker song. Listen and make your own with Suno.
It uses the new method, but it also combined it with the old
To get some strange rhythyms
I kept digging for what gives a Jan & Dean, Lettermen, Beach Boys vibe. And got a good running start through successive covers of covers.
It ends up being like an alternate rocknroll
Back when it first split off from blues and the like
I accidentally move this to TRASH and can't find it in trash to put it back in my main library. HELP!
electro funk, disco, male vocals, slap bass song. Listen and make your own with Suno.
Finally tracked it down. Not fun trying to find a song in the trash when you only have 16 pages worth. Imagine the long term users that might have 100s of pages of discarded generations. See #💡┃suggestions-web for some ideas how to prevent other users having this frustrating experience.
@polar spade a different person was asking for help to slow their song down. Hepcat responded with help, dasvon mistook hepcat as the person who wanted help and thought he was listing things he tried to do to slow the song down (that failed), then Hepcat mistook dasvon as the person who was asking for help. It was quite a convoluted miscomminucation
Haha. No worries. Thanks for clearing that up. Makes for a good movie plot.
No wonder why no one else was helping out. "This cat, Dasvon, is a crazy dude. [Back stepping]"
hmmm... anyone else having an issue with Covers being exactly the same for the first 2 minutes, and only then changing?
I am having in interesting issue when I used an uploaded clip. I took a 59 second clip from a Suno song I generated and uploaded the piece I liked to use and then extended that clip. Am I doing it wrong, because in some cases Suno will make a song for about a minute give or take and then switch into a different song for the second half of the song. It's jarring. I like both versions but none are complete. Lol
Sometimes for me that's an advantage
I'm talking 100% the same, not even any updated lyrics being used
When I want to clean up an older song, Covers make it sound much more crisp. I feel like Cover vocals are approaching Udio quality
Oh I struggled so much with that...
What helped was being 200% certain all syllables matched the original
As in, the count
well, in my case, I'm trying to fill in an awkwardness
And to answer Tiberius, yes, I am having that situation occur too. It doesn't change the song for some reason when I have used covers. I stopped trying because I want to save my credits for when it gets fixed.
as well as changing up some words with same-syllable replacements
okay, so not just me then
Apparently. Hehe
Hi I am gettig really frustrated, it costs me a lot of tokens to find out how to get suno not repeat verses when there is intro, verse, chorus, outro. I tried negative wordings in prompt like No screamo and she does use screamo, I've tried in lyrics different [styles descriptions, breaks, no breaks] still she ignores it all, with no solos she creates long solos etc. are there any tips how to get suno under control a bit?
- See https://discord.com/channels/1069381916492562582/1261146873331318835
- Currently, there is no negative prompting. So when you said "No Screamo", the AI saw "No" (which likely it does not understand what that means) and "Screamo" (which is why you got hit with it harder).
You may want to consider doing multiple extends instead of a one shot generation.
As @charred merlin likes to say. Writing "No screamo" is in fact demanding Suno to put as much Screamo in as possible. Since it doesnt recognize negative prompts.
https://discord.com/channels/1069381916492562582/1253341197917884530 VERY popular request, and I really hope to see it soon.
Covers came out of left field and even the alpha team is trying to regain composure over how cool it is -- they aren't slowing down any time soon, I'm sure. This team is GOOD.
The two gravitational points we find is Sunos somehow pulls from the original lyrics of the original generation , no matter what you do in subsequent extensions. So we often suggest, go back, click edit lyrics, select and copy lyrics first so you have a backup, then delete lyrics. Go back to your next extension and try again.
And/or..... with too many word in lyrics, suno might go off the deep end with non-prompted yells, cries, and mixed lyrics. Try one version with lyrics , of previous full sentence length, cut back a bit. How. You judge. But see if that helps.
And/or.... before an extension begins, put in most any non-lyrical metatag such as [pause], [end], [riff] [ interlude], anything to show a break and it can then sing the following lyrics in the last extension , this rather than suno taking all lyrics from the origin through all extensions just to try to squeeze and structure them all in a v,v, lift, Chorus, bridge, verse tag, chorus, outro form.


➕🇺🇦
(dasvon, I'm sorry you don't have an emoji yet, but you're in there too)
Doh!
Thank you very much that helped me a lot. @hollow hare and @polar spade
@hollow hare <He da man! > @safe hazel <He da man! > @charred merlin
field-tape-recording, post-instrumental cello revival, acoustic viola, ambient chaotic baroque fx, exotica song. Listen and make your own with Suno.
@untold briar this would make a great open to dozens of songs
Or a movie by Spielberg
Hi, how can I make suno not put metal screams on everything?Even to Christmas carols, God
He even adds metal screams to a reggaeton song
What prompts are you using?
The devs have stated a number of times that the biggest influence on Suno's output is what you use as input.
Everything you use influences the style -- even if you use no Style Prompts, no [Metatags], and only put "I'm so angry right now" you can probably guess how it's going to turn out. https://suno.com/song/65a2e520-e54a-4def-be67-d483a08a49eb
A song. Listen and make your own with Suno.
Is there any tool where I can upload a song it gives me description of the style used so I can recreate it?!
You can use a site like this to look up styles of songs https://rateyourmusic.com/
Hyper-2-step, pop, trap beats, atmospheric synths , deep basslines, moody vibes, reverb
What do I remove so that I don't get metal?
I want a falsetto, not a metal scream that leaves you without a throat for a month
I don't know what "Hyper-2-step" is but I would guess "hyper" is making the song a bit, well, hyper.
would you not be better to use uptempo rather than hyper
It's a prompt that gives you the same suno, and it's electronic
but you could add something like "soft vocals" to try to negate the screaming
i'll try
thought it was just me. I've tried doing a cover of a cover, of a cover, etc... can't get anything to change. tried different words. one of my covers came up with all new lyrics - and I was like that is cool, but lot of made up words, so I've been trying to 'force' them into real words with cover 🙂
Is there a way you can use RVC models for custom vocals with suno instead?
Or are suno's voice models just too complex for that to happen with normal RVC
Short answer: No
I have a feedback abou the Cover, I tryed a lot to change some parts of the music using cover but Suno insists in use the same lyrics of the original song. How can I use Cover in a effetive way where I can change only some phrase of the original song without losing the melody? I though cover would solve this problem, but in fact it's getting me a little frustated. I spent a lot of credits and I didn't get the final result I'm aim to.
you have 200 free cover songs for just this reason
ummm @final hazel I have a strange one for you.
https://suno.com/song/8a124cf0-b89f-415d-87eb-8451ad444ba3
This is a song I am making for a friend that loves to hug... I was listening and grining knowing she would love it. then it got to about 2 minutes in and there is a pause.... screams.... then the song caries on as though this never happened.... its not prompted, there are no screams in the lyrics.
It is the oddest bug I have ever seen.
Upbeat comedy folk song with a touch of vaudeville flair (female singer) song. Listen and make your own with Suno.
Sounds like your song had a surprise guest appearance by a horror movie character! 🎃 You can share the spooky details in the #🪲┃bugs-web channel so the team can investigate the scream cameo! 👻
ah thought here was for bugs... will share it there... its the wierdest bug ever
it also had me chucking as its so out of context... yet the music pauses for it lol
guess someone really didnt want a hug
We have players who are not fans of the hug.... it could be one of them... or maybe a player that cheated and is now being dealt with... shrug it apeals to my sense of humour but was unprompted lol
I mean to be fair if someone hugged me till my heart was full and I "started to glow" I would likely be screaming too
** Music Generation Algorithm: Song Ending Improvements **
Hey team! Great work on the recent updates. I've noticed a small issue:
The Problem:
• Song extensions and new tracks often end abruptly
• Lack of proper musical closure in generated endings
What We Need:
Better Ending Detection
- Train an AI model to recognize good endings
- Use pro tracks as examples
Smarter Closure Algorithm
- Tweak the LSTM/Transformer to nail those endings
~~ - Penalize sudden stops in training~~
Genre-Specific Endings
- Build a library of ending patterns for different genres
- Make the AI choose the right ending style
Fade-Out Feature
- Add a smooth fade-out option
- Let users customize fade duration
Beefier Outros
- Extend the outro generation
Penalize sudden stops in training don't do this too much. Sometimes a song needs a sudden end. Though maybe train it so there's a different ending type.
speaking of ending, I'm having that exact issue right now. I use [End] in the lyrics and End as a music style and it's not ending the song
True. I guess it really depends on what genre and what you are going for. Good call out.
switching the extension to v3.0 worked for me
anyone know how to get the singer to sing in a norfolk accent? tried everything I can think off but no luck
It would be very useful to be able to upload audio files longer than 60 seconds, so that we can create instrumentals for complete songs. Do you know if this improvement will be implemented in the near future?
anyone got the covers feature to actually fix some lyrics? cause it's giving me back the same song over and over again, regardless of the changes I make
I can't get it to follow the prompt when I ask it for specific instruments. For instance, all I get is piano music when I type in oboe cello duet and select instrumental.
Also, I tried getting an instrument in a song, but instead the AI singer sang about the instrument and it wasn't playing.
Sometimes I get like a very feint violin (if I do get one) when I've prompted for a much larger instrument. But try generating with v3 instead of v3.5 for instrumentals.
make sure to use this syntax when asking for specific instruments, it helps alot
[bridge oboe, cello duet] or [duet oboe, cello]
but i put the instruments in the section, i havent written for just instrumenta;s fyi
I've been trying to generate an ominous slow "redneck banjo" song similar to the scene from Deliverance, but Suno keeps reverting either to polished country songs or bouncy happy hoedown type music.
Any advice on how to force it away from these styles it seems to prefer?
I made 3 really amazing - top 10-100 hits songs importing some good Udio song ideas and using the Cover functions first to replicate and extend at the end.
Songs Suno would probably never write, cause it writes music differently. And one of the audio clips imported from Udio, Suno manage to replicate 1:1.
And this is lucky case to compare both AIs. Suno has better voice, better sound, better mixing, better everything I would say.
And beside that I never heard a SINGLE Udio song - I made around 200 - sounding better in quality than Suno!
Quite a big difference still. So Suno wins in all categories.
In certain, but not many categories, Udio can compose better or very different, otherwise Suno is better for most cases.
The lyrics from Udio seem more complex somehow the the Suno's.
And I never heard so many miserable songs produced by an AI as from Udio. But sometimes Udio delivers very different and good songs.
I used [end] in a song once and it ended the song abruptly by saying "end' lolol
Wow, duly noted.
guys is the platform down?
No?
What is "just fine" for you? The reason I ask is, my goal is to get higher quality versions of existing songs without changing the Melody or voice. So, in that goal my attempts meandered off course pretty quickly (like a game of telephone).
If your goal is just to produce a banging version different that the original I bet it works great.
Great post!!
Hello, I have two questions to ask.
-
When songs are written, the lyrics in the actual audio sometimes differ slightly from the written ones. Why does that happen?
-
I made a song last April and tried to download it, but it's only available in mp3 format. Is there a way to get it in another format like wav?
Suno is great for all of us. A whole new begining for everyone
I’m not sure where to put this, but I can’t access covers
If you a subscribed user...
Close your browser and start again
Are you a subscriber? And were you a subscriber when you made the one in April?
i have a issue, and i know it sounds absurd. But some of my songs that i make public just flat out get deleted on their own. And i have to dig in my trash to restore them. I also found some of the songs i liked, being disliked on their own.
any got any clues, maybe my account is hacked, and someone ks
is messing with me
Is there no way to upload a custom image for a playlist?
The search function is just garbage. I’m setting dates for all my tracks and when searching I find nothing or sometimes one or two when it shall be ten. So frustrating…
Hi, How can I contact Suno tech staff to resolve a visibility issue?
hmm. If anyone can assist. Whenever i press create song, nothing happens, and if i look at my browsers console, i get auth error in red. The rest of the site works fine, i can browse the songs ive made, listen to them, etc
Does anyone know how I can make the voice to be only male? in all versions I get a female voice. Or male+female. I want only male voice. I have tried to put in the style things like: Only male voices, or similar things and it doesn't work. Thanks!
As the support email is the only official avenue for support, Support@suno.com is the best, most reliable way of contacting staff.
Guys, I don’t quite get it, is it that tracks based on uploaded audio can’t be made public at all?
Or is it because the song is based on a text written in the 19th century, but the system mistakenly considered this text to be copyrighted?
Try "Male Vocals" In the Style Box. If that still doesnt work, then try putting it in twice.
I try to put it in style. male voice or male vocals or the key a male may sing in I know it’s hit or miss
Heya. Anyone know how I can force the generation to use a specific voice, such as female or male?
Any audio uploads that contain vocals are prevented from being set as "public", but you can still share the links and add them to playlists, they just won't show up in searches on the social media-side of Suno.com
In the genre, i type in stuff like "Female Vocalist" and it seems to work most of the time
Thank you for the explanation and good advice, kudos to you, man.
For some reason Suno seems to love my first verse... it generate now 8 versions of my song repeating the fist verse (sometimes the first and second verse, just for a little change) 2-4 times instead of moving on ... Any Ideas what went wrong here?
Hi ! I have a problem to purchase multiple Premium account with mutiple google account already created. My credit card is block on Suno. Someone of the Suno team can help me ?
billing@suno.com should be able to help you
I don't get it. Somehow, someone made this: https://suno.com/song/7ccd2f3e-e37d-4e75-a6a9-02a1dcea3062 while I tried so many time with my favorite song, it always give me something out of context and doesn't follow at all or cannot generate it.
swing, big band song. Listen and make your own with Suno.
This is a joke we understand: put [verse: female] in the lyrics and you're sure to get "male".
But seriously, as they say try "male vocals, male voice, or even male vocal group, in the styles.
And I don't care if you're genre is hardcore or opera, put "jpop" at the end of your styles and see if it helps. The idea is putting a style in that is predominantly "male" singers. Grunge, may do as well. The more towards the end of the prompts, the more it doesn't make a difference style-wise, but may influence a "male" voice, nonetheless.
I'm having issues trying to "reset" the original, sung lyrics in my Covers. Unlike doing extensions, where we can remove lyrics from previous gens to resolve some issues, I've removed all lyrics from any previous gen in the chain of covers. Let's say I do a cover of a cover 4 links/gens. I delete ALL lyrics but the last cover, changing a an origin typo, emmmm.... "fake " to "face". The new lyrics in the last cover lyrics are correct and so it sings the second half of the song, and 2nd chorus with "face". GREAT. But the first chorus is still "fake". I'm at a loss to know where it's picking it up other than cache memory floating about somewhere.
Anyone doing a chain of covers and find this problem? Any wild suggestions are fine too
And my intent is to use workarounds within Suno than the obvious solution of taking them offside and change in DAW.
I think I broke my suno
Errors on Suno website - when editing simple details - receive repeated error message: "check you upload image size" (I'm not uploading any sort of image) Something is getting in the way of simple text edits.
That too
Errors on Suno website- cannot generate anything
Same. Guess it's server side.
I cant generate anything either
Same
That's a t-shirt.
Though this doesn't answer your question I just want to point out that the person who made the song you linked appears to have uploaded audio that they don't own the rights to. So they should not be doing that as it is against the Suno Terms of Service.
Here to also report same as above. Blank page 1, unable to edit details. Same on mobile as pc
Oh yeah...
the song i generated and look like this some of them are gone....
I don't appear to have a button to edit lyrics anymore?
Just to make sure, are you checking on the full song page?
Yes, but I'm also just being dumb. It's not where I remembered it being. Disregard.
Not dumb -- there's a lot of features that being introduced on a near-weekly basis 🙂
Classic ask for help after being unable to figure it out for 2 hours, then figure it out two seconds later.
You were looking in the right place, you put the work in -- it's like when you wrestle with a jar of pickles & someone pops it open at the end.
Lost my earbuds yesterday. Same problem.
Frustrating as hell, but not a big dill
I actually just worked that out. Of course, everyone else knew. For years the joke was.. "well I loosened the lid for ya" . But what's happening is the first person squeezes hard to open the lid , thus warping or dis-shaping is, causing a loss of vaccuum. Then easier for the next person to open.
One I worked that out, I now squeeze a jar lid from two sides first, then give it a go. Pop!
This has been a public service announcement from the pickle jar industry of America.
Ergo, also why the smart people "tap" the side of a lid first.
It was intended as a colloquialism, but a genuinely interesting follow-up!
How can you tell?
Am I a dum dum, I don't see that anywhere?
Does this help?
Hi, I wanted to have 2 verses and 1 chorus, but Suno decided to repeat Verse 2 twice. I really like verses and especially the chorus. Is there a way to recover from this without using external tools?
This occasionally happens for me too. One thing to note is having the same amount of lyrics (syllables) per verse to avoid it. For example, if one verse is shorter than the other (4 lines vs 8 lines) it may repeat the shorter verse. Ive had pairs of generations where this didnt matter though. kind of luck of the draw sometimes
I have many "Covers" that end up sounding the same as the original. Pretty annoying
otherwise it's an amazing feature!
yeah, but its beta, so its not all bad. at least youre not wasting credits
I had 4 per verse and even in the chorus. The amount of syllables - that could vary a little
I've thumbs down/bug reported them
how long are your choruses?
i just don't want to run out of covers!
it's so much fun
4 verses and I actually have the same amount of syllables in each of the verse in the chorus. It's a shame I can't just cut it.
I did cut the double verse out and imported it back to Suno, but it get confused after that 😕
ah yeah... that sucks
Had a great rendition of something that I was working on, you know, you just know it sounds great, and then Suno decided to take a break and didn't 'save' that version. No matter where or what I tried, gone...AAAARRRRGGGHHHH
and stem downloads sux...there...I said it...sounds like the music has been processed through a phaser...
its because its doing the same thing any stem separation tool does when it comes to a finished wav/mp3, its pretty lossy. unless it generated the stems separately by default, theres no way to get a clean stem separation, short of recording the music yourself.
they get close, but its never perfect
sometimes its not bad, but usually either stem will have a bit of the other stems embeded into it.
I get that...But it is almost unusable. I just hate the one and done approach that you get though.
FL Studio has a stem separation tool which can separate bass, drums, "instruments" and vocals, but its not perfect. i dont think we are there yet with stem separation tech in general
vocals are easy enough to fix in post production though. but youre right, kinda unuseable
guess it depends on the genre youre going for
edm, dubstep, or anything electronic, you can use it 😄
With me, I find I am getting about 75% good stuff, but then I am blocked. What I mean is I like the lyrics generated and the music, the vocals sound good...but then theres the 25% where you're like, really? And there's no way of going back in an editing the 'liked' file and 'fix it'. If you don't get it right the first time, your kinda done.
i personally dont use the generated lyrics for this reason. and since playing with the way i spell or punctuate things, it really gives me more control. you might wanna do that. Like for example "read" vs "read" (past-tense) or Live or Live... you gotta spell them differently to get the pronounciation like Live vs Lyve or Red vs Read or even Reed. things like this make a difference
then again, im a lyricist too. i know not all of us are
Interesting
Well, I am too, but sometimes what it generates is pretty good. It's the 'almost there' syndrome that is driving me crazy.
yeah, lyrics will always be "almost there" if generated imo. That human input really matters imo. Not a knock on suno or ai in general.
the more you "feel" or believe your own lyrics, the more that feeling will come out in the song
and that goes for human music too lol
Dumb question. Is it possible to have the lyrics a held for a song like at the end of a verse?
You can use Instrumental to get a 2 to 3 minute beginning. Then extend that and choose custom, and put in your lyrics.
it doesnt always work, but it has worked for me
thanks!
yup 🙂
Oh. You want to stretch or prolong the voicing of a word? Yes... do what Song-arrhea suggests.
The more letters you add, it seems the more it wants to shout the word. I have had to test different amounts of letters to get a nice held note.
That's not a bad idea. I'll try that
something to note, you can also type in something like this to extend the instrumental in the beginning:
[Instrumental Intro - 4 Measures]
I had one song with just two "i"s in FIIND.... it went for about 4 seconds lol.
you can also tell it [30 second duration]
it has worked for me, but not exactly 30 seconds
more or less accurate. sometimes it will put two versions into one longer track too whcih is nice
its like getting 4 generations for the price of 2
Or go all out..
[Intro]
[Instrumental]
[Interlude]
[Solo]
. . . .
[Verse 1]
i do that every time. i label everything to make sure
Hopefully they come up with a way of editing what it produced in a future release. Happens to me to often and it stinks. LOL I do label to make sure.
but even then sometimes weird things will happen. sometimes the weird stuff is preferred
Can covers have the same length as the origional song(Up to 4 minutes)?
Having the 3:12 cap sucks
and once and for all, if I want a male vocalist, [Male Vocalist] ?
Because I just did a generate and female LOL
in the "style of music" section (where youd throw in your genre), add "Female Vocalist" or "Male Vocalist" (could also type in singer instead of vocalist)
thats where i throw it in and it helps. Putting it into lyric space doesnt always work. Ive done one where the first verse was female and the 2nd is male when i typed it out, but suno reversed them when i generated 😄
these things will be ironed out over time though
for me, its just fun to be able to generate songs/ideas when i dont want to spend the time to make it manually
It sometimes depends on your style of music too
and your lyrics
the AI thinks some styles are predominantly female sung, like an anthemic metal or symphonic metal gets a female choir sound almost every time regardless of using male singer or anything
but i had one hard rock that the AI mustve thought sounded like Evanescance style of music cause it always gave me female vocals
im gonna have to disagree here. cause the song i was talking about was written from an obvious male and obvious female perspective
but i could be wrong. im just stating my results
Anyone else have an issue when making songs, where sometimes the song will just stop? It'll get to 4 minutes ish, then just cut off mid song
I've just seen a couple people say symphonic or anthemic metal would give female vocals most of the time
i have been able to consistently get a female/male voice when intended for the most part by using the method i typed above
throwing it into the lyric section not so much though
yeah most of the time i've had that work, like i say the one i had with "aggressive hard rock" kept doing female then i added male vocals and it stopped
yeah i always do it in the style
it will default to male if you dont specify for some genres, i can see that
if i put in "metal core" it will almost always be a male voice
unless i specify in the genre list
yeah rock tends to get male most of the time
cause yeah there's not as many female rock singers as it is male for it to train from
there are plenty of female singers in metal though, especially nowadays, so maybe that will change
a song generated in v3.5 can only be 4 minutes long. You need to make extensions to complete it. Extensions can be 2 minutes long
v3.5 = 4 minutes
v3.5 Cover = 3.30'ish minutes
v3.5 Extension = 2 minutes per extension
another trick @keen lintel when you get to where you want to end the song, swap it to v3 to extend and do
[Outro]
{END}
or just
{END} to just get an ending without extending for another minute and a half to two minutes
Legends!
anyone getting any good results with the new Covers? I have created 20 covers of some of my previous Suno songs giving them completely different genres to the original and apart from one generation it just generates the same song as the original.. Just wondering if anyone else is experiencing this also?
first generation usually is very similar to the first, the 2nd one usually changes enough. It is in beta though, im sure it will get better
can we add our own voice and let suno generate the song?
yes, use the uploads feature
As a moderator at Suno, I'm thrilled to give you an exclusive sneak peek at an upcoming feature from Suno.com!
Here I am, singing my own chorus. Watch how Suno picks that up in the song! THIS IS MAGIC!
I'll be posting more examples soon, so if you want to stay updated, hit that subscribe button and get all the latest news from Suno. Also, don...
It's very hit and miss today, mostly just generating the same song. It was very different yesterday, every generation was a different song. This was with quite old pieces.
I'm now wondering if it depends on when the original was produced. I'm now having success with every cover I'm producing with new pieces that I upload. Every cover is different even when using the same prompt.
I keep repeating my verse 2 for some reason
Know how to stop that?
It always plays twice
This is a known "bug" sadly, the best you can do is to include 4 bars in each part, and hope that it won't do it anymore.
Sorry, but I'm no expert on Covers 😔
I can answer my own by just testing. So I'll check. Thanks for acknowledging my question.
This is almost always a problem with your song structure or tags. Suno is expecting something from the structure and filling it with the pieces you've supplied.
Suno tries to build a song based on what you give it, plus the style prompt, plus the random seed and instructions behind the scenes. Think of your lyrics as train tracks. The verses are straightaways, the chorus is a tunnel, the bridge is a bridge, etc (for example).
Suno tries to build a railroad across the landscape generated by all your input, using the tracks you've given it. But if it needs a straightaways and all it has coming up is a bridge, it'll reuse something it already has to fill that spot.
But also, unexpected characters or overly instructive tags can give Suno permission to noodle. I've noticed ~ and ... especially are susceptible to this. And sometimes weird characters get inserted as we cut and paste between apps, that we can't see but Suno can. If you can't figure out why Suno keeps doing the same thing, try deleting the problem section and re-typing it.
Hi, I'm currently having a problem with the extend function. When I try to change the music style with the extend function, I always get the same result, the music style stays the same or has only almost inaudible changes. This worked perfectly 2 weeks ago. It seems to me as if when you extend, it also tries to cover the rest of the song and therefore no matter what you write in the style of music, the same result always comes up. Can someone confirm this for me?
But most of the time in my experience it's a matter of working with Suno rather than trying to work against it. If it just insists on repeating verse 2, try giving it a new verse 3 there instead. Or move a later verse up. See if that changes things. A little troubleshooting like that can help you figure out what's wrong.
Will extends ever be 4 minutes too? The time limits kill me
Hi, I had subscribed to Suno, and unsubscribed, but even a few months now I am still getting charged. i even went back into the accound to try again and its not letting me but I am still getting charged.
The only reason I chose to unsubscribe because I was getting charged and not getting the credits I was paying for.
email billing@suno.com
It's not a super accurate comparison, but for simplicity's sake, Extends are half the length of the first section because it's spending half of it's attention listening to the section it's extending from to make sure that it's following along.
After you extend, you might find that each extension after that sounds more and more "off" -- which is why it's good to "get whole song" to lock those extensions into a Full Song -- this way you can extend from this new Full song and keep adding 2m sections each time 💪
The extensions sometimes cant replicate the voice for a song that I really like, if they where atleast 4 minutes I could rebuild the song into something close-ish
covers REALLY really need the 4 minute cap too
The 3:12 makes them pretty much useless
For me atleast
@charred merlin @safe hazel Hey folks. I worked on a Christmas song to throw in the SSC5 contest and it got restricted. Any way to find out why. I'm stumped.
Worked it up yesterday. So initially no issue, but it's been flagged. I don't know where to look for a reason. A "Do they Know It's Christmas" theme song.
The SSC5 team would likely be the best people to contact -- it makes me happy that you thought to reach out, but technically the moderators (lighter blue names) and the guides (darker blue names) run the show here 🙏
Wait. I should check way back a month ago, see if it was initially restricted since back then it had vocals. I did covers yesterday and finalized it, but didn't think about checking.
Yes. I'll send a message to support, then.
Important distinction --
The Flosstradamus "Suno Remix Context" is official, and questions should be directed to support@suno.com.
SSC5 is a community-run unofficial contest, and questions should be directed to their community context organizers. https://discord.com/channels/1069381916492562582/1272505779554881590 (they were just given green names in discord today)
No. Not originally restricted in April. Same music. Same lyrics. Strange. Did Scrooge become CEO?
Well, I can just send any old song. I'd get only 2 votes no matter. Ok. Thanks for the insights.
did scrooge become CEO?
I might be confused -- is this about SSC5 (unofficial, not affiliated with Suno) or the Suno Remix Contest?
Its my Library thats an issue. I assume I can't post in Discord contest because of it
Can I post it here... let's see...https://suno.com/song/31f3a4d3-9d95-47cd-8185-4b0b4293e0a9
brill building, piano song. Listen and make your own with Suno.
Yes.
It's got a content restricted notice... though it's been up since April. And the older versions are fine. Other than using covers feature, the lyrics are the same.
Does it sound like a celebrity artist? Does that do it. ?
Beyond any issues with that particular song -- I just want to highlight that SSC5 is not an official Suno contest.
(apologies if I misunderstood -- I thought the misinterpretation was Suno had denied the song submission from SSC5)
That's different. Different vocals
Sound like maybe, Amy Grant-ish and ummmm... Michael Bublé-ish Cool but hardly close.
Yep. Ok. I'll see it can be submitted while I wait for an answer from email.
What's the issue? Can't be covered/extended or can't be made public?
.... and everybody enjoy a new [restricted] Christmas song to get you in a holiday spirit. Lol.
Can't be made Public.
with this?
Yes. I'm no sure Which content as it's same content that I know , since April. The older copies are OK.
I though. Wait, maybe because vocals in the upload, but no, again, the older copies that I made covers for, are fine.
Is it from an upload originally?
But the ones I did yesterday, I just noticed today, restricted.
Yes. An upload in about April.
Audio uploads were added in June 🤔 unless you had early-access (which would've likely had bugs)
And an upload with vocals at that time, would not be able to be Public. Ironically all the old ones are fine.
I just looked at the creation date of April. Let me. It's the same melody. I assume from my sing I uploaded.
https://suno.com/song/8f12183c-95d8-4574-a1f5-9fc0e19e8e52. I did it with ver 3.0. ( this link is an old version and variant, not my melody) But I don't know about the original upload. It's my melody in the scrolled up final version. . And somewhere is my singing the original
not created randomly.
It looks like the song was created from an audio uploaded 2 days ago
Covered, then extended, I think
Oh. I may have pulled I tover from soundcloud again. I did in April... obviously. OK. Cool. I uploaded it . Did an extension and cover.
Did you have early-access to the upload feature back then?
I remember the content restrictions were added a bit later on
So maybe I DID, just type in text back then. And got dang close to my melody that it was indistinguishable. Well, cool for me.
So the cover tool, recording a voice or uploading anything with vocals still restricts public access? Maybe that's the issue.
Hold on ill check a jpop song I did yesterday from just scatting in the phone mic and making a cover.
Anything that originates from an uploaded audio that has a vocal or voice-like sound in it makes the entire "chain" of extends/covers restricted
False positives (content that shouldn't have been flagged) & false negatives (content that should have been flagged but wasn't) permitting 🙏
Yep. I assumed that to begin but then said, wait, what's the bit about creating melodies with the cover feature, by just humming a tune. So that threw me. No worries. I'll get a confirmation from email. Apparently I can still submit it to the contest, oddly enough. But Scouts honor, it's my lyrics, my melody, my livelihood on the line. So all mine. I don't put thing public much, but wondered if it affect the contest. It might still, lol, as one might not be able to listen to it. Beats me.
Thanks for weedling and talking it out. I guess I originally just typed the lyrics in. Most of my songs lyrics are sung by suno nearly exactly as I had recorded the original cuts. That's the secret to good lyric writing. Thanks again.
Yeah you're allowed to upload audio containing vocals, it just limits making it public afterward is all. Not entirely sure why. I assume it's to try to avoid people uploading vocal stems of famous singers and then trying to clone them and hit the home page/trending with fake songs by them. That kind of thing. You can still share the links, download, do everything else though.
Yep. I'm having fun with it all and finding every hack possible. Posting Public is not of much interest for similar reasons as everyone else.
guys
I tried to make a cover changing the lyrics of an already created song
it repeated the same original exact lyrics... how possible?
it didn't care at all o.ò
Piano tends to be what Suno falls back to if it doesn't understand something
When was the already created song created?
I have the same problem if I try and make a cover of songs created months ago.
I don't have the problem at all changing the lyrics of songs that I've created in the last couple of hours.
5 mins ago
Hmm, well I have no answer for you. I'm changing songs right now, changing lyrics, changing genres with covers. Silly question, I know, but are you deleting the old lyrics and replacing them with new?
I created a song for my niece... but I really like the music... how can I tell SUNO to take this sone and make an instrumental-only version of it?
The song: https://suno.com/song/3efc79bd-557f-47d6-ad48-2a5e56d4005d
Jersey Rock song. Listen and make your own with Suno.
Hey everyone, I have a question. It seems that the majority of the music I'm generated has that "tin can" type of sound to it. Sometimes its worse than others. Is there any fix for this? Is there something I can do to avoid that or fix it? Today, I have generated 10 or so versions of one song and all of them have that sound component making them useless to me. I have been search for more info, and find lots of people complaining and identify the same issue, but no solutions. Any help/info would be appreciated.
yeah of course
I've been getting it a lot recently too. or the vocals sound like they have gone through auto tune seven or eight times. Even the instruments sound like 1980 "electric" syntheized instruments. Which is great when I'm doing synthwave. Is there a new patch or model coming out any time soon? Just checking and not wanting to be rude.
Hello. I'm pretty new to this and I was extending the Soundclash Synth from around the ~02:40 mark not realizing that it would keep the ~02:40 of the original song. Now the song I created has a very odd intro, is there any way to cut it off or say just extend from 02:20 - 02:40 ?
not perfect... but there you go.
Now I can do AC/DC It's a Long Way to the Top.
Yeah... I just threw all that together because I was fairly certain it would muscle something akin to bagpipes. But ... fine-tuning that -- well, it's all you.
I'm still working on Troll Management vs Ronnie James Dio.
Thanks for the info. I am a subscriber but I'll just download the song and cut it myself. That'll probaly disqualify as well but 🤷♂️
- Covering the song after generating it will nearly eliminate the tin sound. I have near crystal clear renditions with covers. 2. See 1.
So... we are having issues with covers... they will go WAYYY off script..... turning songs into Simlish songs. Incoherent words all the way through...
Me: I wish I had a song that nobody can understand but also slaps.
AI: aight... bet... hold my microprocrssing intergalactic gargle blaster...
I'm not sure it's either possible or necessary to listen to all the entries... The winners will be based on likes, shares, and plays -- which can be sorted with an ascending or descending sort button. Then just grab the top 500.
Yeah. it's not perfect.
I'm only pointing our the issues in case it becomes a problem
But if you work in the "what's most likely" zone instead of the "what's strikingly difficult and might not have an audience" zone, I predict less frustration and fewer credits used.
I wish there was a better way than "chance rendition" at achieving greater variety in sounds / styles.
😄
That's funny.
I'm having a fair amount of luck with mine. The trick, it seems, to coming out ahead is limiting the costs to just Suno. I'll keep the $6 Moises subscription for stems, but I'm dropping the mastering ... I learned how to master with Audacity for free. Saves me $30.
Or "The Sims" music lol
SaaS anonymous. If you have a problem suno-ing... call 1-800-music-wrecked-my-life.
I literally have the best one.... genuinely thinking of pitching to The sims lol
I edited to take away the strange boise
Noise*
After 10 seconds it's a hit
Tell me 1 sentence 🤣
But.... notice.... I take away the grainy static voices.
Hi, i had a suno subscription with 10k credits... i ended the subscription and all my credits vanished. is it normal ?
Oh, ok. I'll give that a try. Thanks so much for the reply.
Ok, I tried your suggestion. Seems to have help some. However, about 1/3 to 1/2 way through the song that "tin speaker" sound comes back in the vocals.
Does anyone know why there isn't a text to voice option? Suno can clearly create good speech. It would knock Eleven Labs out of the park.
Suno stopped working on that project (Bark) in favor of the Music AI.
The source code was made available but only less than a handful of people will be able to help you with it here.
Ah, thanks. Yeah, I tested Bark, but still, the king is Tortoise since it can read moods and you can prompt engineered results.
Am I just terrible @ using this? I mean there has to be a better way to download music from Suno. I'm using it in a bit of my work and I need to download large quantities of songs. Is there a way to for example download an entire playlist that I just don't see?
No. AFAIK, there is no option to mass download songs
Damn, that's a shame
you guys ever have that issue where u add something to a playlist and it goes "0 songs were added to the playlist". I would love a feature that blacked out the playlist youve already added the song to, and when adding songs into playlists on mobile, after a few additions the playlist drops to the bottom of the list for me. Not complaining, just admiring
is that an "l" and an "i" youre using the "1" for? or is it caiamity, or calamily? either way im here for it, its unique
Haha! Well it's hard to spell. But I NEEDED the name Calamity when I started my youtube. But that was spoken for so I had to practice leetspeak and it ended up with one's.
So ca1am1ty
ok I'm extremely confused regarding with the cover feature right. It said "Transform Audio Uploads: Try uploading a clip of your singing, humming, or playing an instrument and try covering.". Am I wrong to assume that if I would upload myself singing, it would keep it?
like the singing
cause right now it just makes a new song entirely from what I understood
or am I supposed to manually add the two?
None of the Covers I try to do work and I don't end up using a credit for it either. Everyone else having the same problem or am I missing something?