#๐โsuno-school
1 messages ยท Page 38 of 1
Also had some fun yesterday after I found a website that generates a range of unicode characters to see if it would try to "read" any of them ๐
one thing is i dont think a guitar neccisarily means guitar - its like a chorded rhythym -
the instruments definitly arent fixed to like a very specific instrument
I used the "reuse prompt" on one of the above.
https://suno.com/song/6a672203-9e6e-4141-8d69-08a6569d1e3b - genders are reversed to the tags above them
https://suno.com/song/d8dc441a-0b95-4c14-9356-d12264b3cdab - both vocal segments sung by a male, style prompt specifies female icon
have you ever gotten a female?
Yes there's a female singing the male part in the 1st example above
when you use the female icon?
maybe your crosswired
hahaha - man idunno, works like a charm for me
I hate that attitude. AI allows people who don't have the skills to make music themselves to create the music they want. But still some people want to keep their secrets. It should be an open technology and we should encourage others to share their experiences and prompts
"And duh and duh and duh and duh and duh
and "and", and "humma-lumma-mummana"
AND "humma-lummanin'"and "hummalut-sun-chun, summa-chun London"
There emojis are actually kinda awesome to sing along to.
How is that not trending yet? 
I've been repeating it ad nauseum trying to sing along, so give it a couple minutes.
Please tell me you're singing along, too.
Just give the hamster the million already, Summer of Suno is over.
lol
I uh... I don't think I made it public, so let's change ... that...
Done!
the art is perfect
only way i can get a track to stop on 3.5
Just this one a line on its own?
lol
Testing some more with these in the style and in a tag:
Using โ๐I got: 3 songs with male vocals, 1 song with female vocals
Using "female vocal" I got: 0 songs with male vocals, 4 songs with female vocals
what do yuo get with this one โง
As random of a result as you'd get using other emojis.
The hamster's example shows that the full words are understood by Suno ai more coherently
yeah its really hard to control vocals
the ai seems to know a lot of fx names though
https://www.suno.wiki/faq/metatags/voice-tags/
Here's the community's research on the matter, it's all we have until official documentation comes out. โค๏ธ (not an exhaustive list BY ANY MEANS, but examples of what's known to work)
I hope they post a list of all tags used in training
I would be honestly surprised if they presented "their recipe" like that. We'd all love to know the Colonel's herbs & spices as well ๐
In think it's tricky for making a way to a song, but with the same prompt, two songs are always differents, even with the same seed at Stable Diffusion i can't not replicate the same image. I hope to introduce one song to make the same way and style by yourself or by own generated with suno, because i have some ones to replicate and extend functions not works well with them.
also ... is there any way to get Suno to export MIDI directly ?...
...seems a shame no MIDI because at some point in the process there must be clear notes and arrangement info... but it take a while to transcribe...and doesn't do it accurately via Fadr etc..
no there isn't. Neural music generation does not work that way. The final wave is produced directly straight off the input promt
here's a prompt that uses the notes and chords from purple haze https://suno.com/song/80c54368-39f3-43e6-bf9f-9cc5ddaef6a2
70's Psychedelic Rock , ๐ผ๐ถ, Rhythm ๐ธ, Lead ๐ธ, Soft ๐ฅ, โ๐, song. Listen and make your own with Suno.
sounds kinda like purple haze or am i mistaken?
soo seriously, thats my mic drop on this whole debate for anyone trying to argue with me, just save your breathe - you just don't understand it and thats ok
It's not so much that I don't understand it, I'm interested in whether Suno genuinely understands it, and if so, how reliable it is at doing so (which is why I've been trying stuff and discussing the outcomes here.)
Sometimes it's really exasperating ... Suno has actually created a great song ... and then ruins it again by simply repeating verse 2 between the chorus and the bridge. Why? And unfortunately that happens very often. And another good song for the dustbin.
Why throw the baby out with the bathwater? Place an extension point before the content you want to discard, and attempt to continue the song as you want?
if suno didnt geniunely understand it it wouldnt have spit out an almost perfect cover on gen 1
even like kids understand this notation
woiwow!!!! great idea ๐
Based on all of the training data that Suno has been given, as well as the feedback from thumbs down & bug reports, Suno believes that it's creating a great song for you -- subjectively it's likely not a bad song, just not what you were aiming for, but you can still collaborate with Suno to create that ๐
c2,1/2 (note c octave 2 1/2 note) its very basic musical notation
i've never used te rhythmic part before but the rest is very typical
How would I put a [Hook] in a rock song?
There's no need to get confrontational, this is a place for technical discussion. You can show off your work in suno showcase if you don't want any feedback or critique-- a couple other users have noted the same inconsistencies that Furrybeans is pointing out. It's not a matter of whether or not you're able to parse the musical notation, it's whether Suno is capable of processing it as explicit instructions, which I've seen more evidence to the contrary.
yeah. The problem here would just be training data. I am not part of the Suno team, so, I have no idea what annotations is Suno trained on, but basically the model sees pairs of text->wave during training and learns a mapping from text to wave. If the text annotations that Suno is trained on contain musical notation, then Suno is expected to understand it. If they do not, then any hopes of it understanding it are just hopeful thinking. But these experiments are very interesting to see what does it actually understand (which would basically correspond to 'what was it trained on'). Getting notes/chords for training data is hard, as that info is not as readily available as song descriptions or lyrics on the internet.
that being said. When training these large neural models, they often learn things that you would not have expect, since there is always a small amount of exceptions here and there in the training data (e.g., some song lyrics might actually be note names, like in "the sound of music" lyrics), and the model might learn a few associations here and there. We see that all the time, when training models that should only understand English, but understand some Spanish or other language a bit, since there are some times foreign words within english, and models pick up on that
thats a good idea
I just wanna know why i cant get an 808 and a guitar on the same track
Is there a way to force Suno to generate an instrumental after the lyrics have been sang? I'm working on finishing a part three of my song and Suno insists on repeating the previous lyrics. Even though there's no lyrics and instrumental is ticked.
I find that turning off instrumental mode and just putting something like [drop] or [interlude] helps
in the lyrics
@charred merlin .. i suppose the extension point hack would work if you wanted to fix lyrics? like... here one word is wrong ๐ฆ https://suno.com/song/11856117-a887-462e-a0a9-a6bc41a112cc
Time-stretched, tape-fx, tape-loops, experimental psychedelic sound design, complex generative evolving soundscape song. Listen and make your own with Suno.
switching to V3 seems to work also.
The bigger image here is roughly the vocals + bass at 0:05 - 0:09.
Chords:
Output seem like D F C
Specified in "Chords" part of prompt: E F#m G#m A, B, C#m and D#dim
Vocals:
Output uses: A C D E F
Specified in "Notes" part of prompt: A4, B4, C#5, D#5, E#5, F#5, G#5
The smaller image is the guitar at 00:23 - 00:26
Output uses: C D E F G
Specified in "Lead ๐ธ" part of prompt: A4, B4, C#5, D#5, E5, F#5, G#5
It may of course be that it didn't apply the parts of the prompt at the points that I have examined. Can you highlight a particular section where the input had a direct effect on the output?
I couldn't get anything repeatable with the music notation, but this one makes me laugh https://suno.com/song/78d8efbf-c748-4e10-8ef6-2b404dbf06c1
70's Psychedelic Rock, Blues Rock, Hard Rock, Acid Rock, Classic Rock, Funk Rock, Electric Blues, Experimental Rock song. Listen and make your own with Suno.
dude im done -you figure it out
It'll be easier to fix stuff like that in the upcoming models where you can "inpaint" a section of a song to fix EXACTLY something like that. For now, I've been admiring those little quirks and hiccups as "audio artefacts" ๐
if you think any part of that song is in C1 your obviously musically incompetent
and I have no idea what your using to get those notes
such amazing tact
I have brain damage and my social skills are affected by that
not an excuse to be abusive.
and im tired of arguing about this
Thanks for your efforts, I appreciate the work you've put in โค๏ธ
ok ill just block all of them i guess
because i already said many times im done talking
I think it's abusive to persist in questioning someone who is tired of talking to you
take a break.
[Joshuasca Break]
Is there a possibility to play 2 solos of different instruments one after the other?
I would like to have a fiddle solo followed by a steel guitar solo.
Unfortunately, I only ever get one of the two.
I have tried:
[Fiddle Solo] [Steel Guitar Solo]
[Fiddle Solo and Steel Guitar Solo]
[Fiddle Break] [Steel Guitar Solo]
None of them work. Does anyone have an idea or is it just not possible yet?
'@charred merlin looking forward to newer models then and in the meanwhile.. I'll just appreciate the 'quirkiness!! its pretty amazing... but you do have to 'accept what you get' a bit rather than getting what you want... ๐ a bit like working a temperamental producer (#note to self this is how I am sometimes)
I think people often add some unsingable text between the prompts, to fill in time where the one should play, then the other
`[Fiddle Solo]
----__-_--
[Steel Guitar Solo]
.!..!.!!.!!!`
i'll try this
Forgive the (possibly low-key pg-13?) song, but it's a popular example-- https://www.youtube.com/watch?v=JRBBcoANRIE; the first line presumably was supposed to be "forgive me father, for i have sinned", and it came out as "Ahhh, give me father"
I'm sure they could go back and regenerate it, but it's right at the beginning of the song. They could get it "right", but it wouldn't be the same song anymore.
Same when you get to a "hiccup" in your songs, 3.5 is a lot better at preventing some of those suno-sided issues, but even then sometimes you learn to love them because the rest of the song is just TOO good to throw away ๐
Often times when I'm making a song I'll find the point where it stops sounding "like i imagined it in my head", then go back a couple seconds and make a handful of extensions to find the closest fit to "my minds ear". Even then, it's still a roll of the dice as to what you get (though it's neat seeing the tricks that people are able to get to work consistently, because those give you a touch more control over what the song will sound like.
Regarding my analysis above: I transcribed the notes by-ear, which is what I often do with my Suno songs.
As a technical discussion I just wanted to determine the direct effect that part a prompt has on its output. Not trying to start a fight. If a technique works - great, I'll use it and encourage others to do so. If it doesn't - I won't.
Heck, before long we'll be able to just hum along to what we want it to sound like and have suno pick up that tune and carry it!
Can confirm the upload thing will pick up a melody and use it.
and being Suno..., sometimes ignore it ๐คช
Same, I hope the dialogue on the matter wasn't taken personally directed, I just imagine it would be confusing for newcomers looking for technical advice to get unverifiable data presented as concrete fact. Just trying to help the community wiki grow โค๏ธ
I'm looking forward to seeing what it does with some of my guitar noodles. Gonna be fun!
Oh my god, there's a catalogue of garageband projects that I made 8 bars for then gave up on. There's SERIOUS efforts put into those, but I'm not familiar enough with garageband (or comfortable enough solely on the iOS mobile app) to turn them into more fleshed out songs.
Now I can just... work with Suno and hear what they would sound like finished?
I've thrown a bunch of fragments of different random ideas I recorded over the past few years into it and it can give some really good results.
Endless fun! 
Argh, you have access to upload? That's cool, I'm not jealous. I'm not crying at all. You're crying, shut up.
maybe we can actually get acoustic guitar sounds that way that don't sounds like they're from an early 2000's keyboard ๐
hello, i would want to export a sond with the link/video option. On this song, i have apply the function " edit details", to change the picture. When i 'm reading the video of the song : it changes by taking the old ones. Do you know how i can keep the same picture of all the entire song ?
I have the same problem with the text : is it possible to change it for the link exportation ? the text in video dont correspond to the lyrics ( the joy of the "extend" function..)
Since a few days ago, yes. I'm trying to suss out its strengths and weaknesses. Suno really does "get" the tempo, chord sequence etc. of uploaded audio. It generally works well, though I'm picky about the outputs but that goes for my normal generations too.
The way to do this is to use a video editor and pluck the image you want as a higher layer over the old image.
The reason why the old picture(s) appear is because the video generates when the track is initially generated hence why they get stuck with the title & picture it had when it was made.
When extending, it appends the newly generated Part N video to the old part(s) video(s) to help save computational resources.
Updating a video each time someone would change their image would be too costly to do so (esp. when someone is testing out several images).
You can suggest to have an option for a regeneration of the video at #๐กโsuggestions-web
You are much better off making an AI video for it honestly. There are several free ones out there, such as https://noisee.ai (currently limited to ~3x/day free for generating 1 min clips)
Turn your favorite melodies into music videos. Supports Suno, Youtube, Soundcloud url and mp3 files
A few more that I encountered in the past with other YT videos regarding AI Videos can be found here: #๐ฌโgeneral-chat message
thanks for your answer i'll try it ๐
what's the policy for songs started with the free plan, then extended with a paid plan?
I'm pretty new to this - was hoping to find a way to prompt using specific chord progressions. Am I understanding correctly that the chord prompts in the linked song do not actually work?
Consider the free song owned by SunoThe free song is owned by Suno. However you'd treat extending someone else's song, treat it the same way.
This is what I was trying to establish. Until recently I have not heard of this working. I may do some further testing later. You can try this out to see what results you get (please do share, if it works for you.) ๐
will do! I've got some songs that I have specific ideas for and I've gotten nowhere so far
Suno toys with me. I've had two v3.5 songs that stuck to the structure & lyrics and ended properly with just an [Outro] - I'd actually left [End] out so that I could tack it on with v3. So I'm like... They fixed it! and then the very next one gets hard chopped at 4 mins. heh
Sorry if this sounds like a stupid question but is it possible to replicate a voice from one Suno song to others, i.e. the vocals are the same vocalist for future songs?
the simple answer is no
What you can do is use other AI programs to replace the vocals in the Suno generated media.
(This method is also an easy way to make a song perform duets by having opposite genders and control who sings when using a DAW).
One tool is kits.ai (see #๐ฃ๏ธโchill-chat message ).
You should look into RVC models as well for this. One popular usage is using the model is to replace the vocals with your own voice (see #๐กโsuggestions-web message ).
I haven't had success with the chord progression prompting unfortunately, but I do have some more questions: 1) Is there any way beyond adding punctuation to make tags work better? I'm having particular issues with [A Capella] - it almost never results in the tagged line being sung a capella. 2) Is there any effective way to make the song end? I used [Fade to End] and [End] in this song and it does fade out... and then comes back with another chorus. https://suno.com/song/0c6923f7-bcd1-4398-8367-5f374e522277
Alt Rock, Rhythm ๐ธ, Virtuoso ๐ธ, Building Layers, Female Vocals, ๐ก, Layered Production, ๐, Ancient Greek Influence song. Listen and make your own with Suno.
Help, Whatโs the best prompt to make the AI stop singing? It just keeps on and on and on. My song is now 7 mins long because I keep extending it to get a fitting end.
Extend where you want to end with the extension set to v3.0 when using ending tags (e.g. [end] in lyric box and "end, short" as the style tag) as 3.0 is much more reliable for ending than 3.5
Thank you, Iโll try that
Hi I was just going to post a solution i have found to similar issue: when i used to put any style descriptions like [Outro - Clean Voice] [Outro][Clean Voice] Outro: [Clean Voice] the chorus starts repeating again for me as well, reaching the 4 minutes during those unwanted loops/extensions. I just found out that when putting no style labels and just an [Outro], the song ends as expected. If you really wanna put any style label, you can just do it and then add an empty [Outro] before the [Instrumental Outro] and [End] labels (this has worked me every time for like 10 songs)
not sure if this is the right section to ask this -- If I generate a song while my pro or premium subscription plan is still active, can I still use that song for commercial purposes (monetization, selling, distribution) after my subscription ends?
From what I remember reading, itโs all about when the song was created. You retain the rights if you did it under a paid plan.
well i mean, do you necessarily have to have copyright over something to use it in a video? as long as you dont violate cr?
As long as itโs on a paid account, you retain the rights to use as you wish.
On the free account, not so much.
What I am particularly interested in confirming is whether I retain all rights to the generated song, even after my subscription ends, provided that I generated it while my paid subscription period was active.
Yes, I believe you retain them, even after it ends.
What if theres a song with no lyrics made on a free account and I use it in a video (hypothetically) isnt it RF because you cant CR the AI part?
I hope someone from official support can confirm this
Read those terms carefully because I believe I read that SUNO can use your track as well
Reading the Terms of Service says the following
Even if you own the copyright.
That's acceptable for me, as long as I can use the song for commercial purposes and retain those rights indefinitely.
I canโt believe how insane of a job suno did on my track โSerenity Prayerโ about a recovering alcoholic https://on.soundcloud.com/4nJyNmtgJRFhqnuMA - I still wish we could split stems
God grant me the serenity to accept the things I can not change, the courage to change the things I can and the wisdom to know the difference
YOU GOT THIS!
As I drowned my sorrows, one by one.
With
The copyright was crucial for me.
Yes awesome job. Thanks for the feedback and share
Anytime
Found in the FAQ (Suno's Help page):
Nice find. Thanks for sharing
Also related (for my future reminder)
Yes, thank you, thatโs what I remembered
what is the text for long-form bark audio?
i tried copy pasting the notebook but it doesn't create an audio file
no file created after it finishes
Hello, i want to ask a question regarding to the use of SUNO, did any of you have luck in creating a prompt that makes the instruments follows what the voice of the song does?
V 3.5 IS NOT RESPONDING TO [End] and so, when it comes to end, starts halucinating! that sucks
@craggy delta this might be helpful
i did it as outro solo and then end
listen what it happens
GOD DAMN HALUCINATIONS!!!
It makes my song unnecesary longer and loger
try making an extension in v3 with just and [End] or [Fade to End] where you want to cut it
Try this structure, this worked for me in v3.5
Chorus:
lyrics lyrics....
lyrics lyrics....
lyrics lyrics....
lyrics lyrics....
Outro:
aaah... aaah... aaah...
[Instrumental outro]
[END]
i'm trying your method of getting back to v3 on the last extention
when doing that, erase all the style of music
and write just 'end'
worked for me, I used the Fade to End and it made me a 11 seconds fade out
You can DM me bark questions, though I can't promise a fast reply, but bark stuff probably gets missed in this general Discord now
I'm getting a lot of bad sound lately, sounds like broken speakers and a constant high treble in over half the songs as well as some kind of volume back and forth on instruments and vocals. It ruins awesome creations.
and the Auto tune effects is verry irregular and they should be possible to avoid by a command [no autotune] or something
That's a feature that all modern commercial music needs. 
classical metal rock, symphony orchestra, Powerful Guitar drop., Deep Male vocal, Male Choire, Guitar Riff intro.
song. Listen and make your own with Suno.
sound broken rions everything
ruins
Yeah there is a prompt in the Guidebook called [Vocal Leading] and i have used it like [Vocal leading Guitar] and it worked about 50/50
Scratch that its " [Voice leading]"
i had to loook it up
same i also use "Dueling vocal melody" and that works maybe 50/50
I used that one time with another instrument like guitar dueling horn and it worked for me
eh dudes !, question, one thing that I have never tried, I don't know if it has worked for you, mixing styles, that is, for example, starting with a solo piano style and then mixing it and starting with power metal, thus having an intro with a solo piano
I do this alot, i either use the tag within the Lyrics [Genre Switch to "genre" ] or I flip it over to instrumental. And My HUNCH is that the genre has to be on both style your extending with and style your extending from, Because the first time you run it, always seems to dud, but that gets the style onto the track, and then when you extend on the dud it will flip. Just my experience, but i literally just did this in a rather dramatic way - let me show you here
Atmospheric Freak Folk, Demented Gypsy Punk, Female Vocal, Tillvocals, Abyssal Jazz Influence, Horn song. Listen and make your own with Suno.
I turned it into a tape loop and sent that miserable gypsy punk into outer space
interesting, but ugh, I already spent a lot of points with the attempts T_T
welcome to suno - i think i bought points like 3 times my first month, after upgrading my sub
most of them made absolute garbage - i dropped like 15k into just experimenting
I have a subscription, but I don't plan to buy points, I would eat them anyway haah+
its worse than any addiction ive ever battled - my friends hate me I will litterally act like im going to the bathroom and be writing suno's at their house
hahhh
new karpathy just dropped https://www.youtube.com/watch?v=l8pRSuU81PU
We reproduce the GPT-2 (124M) from scratch. This video covers the whole process: First we build the GPT-2 network, then we optimize its training to be really fast, then we set up the training run following the GPT-2 and GPT-3 paper and their hyperparameters, then we hit run, and come back the next morning to see our results, and enjoy some amusi...
Does Suno know what I mean when I say " Create a Rock backing track in the key of E. With no guitar solos"?
I never used the word Create - and im not sure if the key is something you can set for individual instruments - but thats a great idea
https://suno.com/song/bc971c04-4581-4668-9c19-94cb32b7fece
Brazilian Funk Carioca
Pagode song. Listen and make your own with Suno.
I expect it won't fully understand this.
You'll get a rock genre at least. There's no negative prompting so you can't exclude things.
I've not experienced it being able to understand finer details like a specific key etc.
It tends to respond well to prompting for a genre, instruments to include, and for musical segments such as verse/chorus/bridge and so on. That's the level of control I find it allows/understands.
I did get it to make one. And it does sound like its in the key of E. But its hard to tell lol
relentless rock gritty song. Listen and make your own with Suno.
Yeah I have been able to instruct it to play certain notes using modern notation like e4 1/2 , d4 1/4, etc...
rhythm gets really hard to tell whats what but i definitly see some correlation
I may have taught suno some extra tricks or something, I can do stuff that only like half the users seem to be able to re-create but I've had alot of success getting into detailed excursions, i have a track that is octatonic and does some gnarly hendrix harmonics
and I have a NEAR cover of purple haze
Thats actually pretty cool!
Yeah and I only fired it once, one was sort of off, but the other one nailed it
This was the notation Suno experimented a bit with in V2, may help: #๐โsuno-school message
thats how it always is for me
At the time I tried quite a bit and mostly found the notation useful for making singing voices do very unusual things, melodically.
You were the first one I saw do that - my first day on discord i went and checked out yourtracks
inspired me to follow along
If you are in a bad mood and want to send your credits into an abyss, style "note" loool https://suno.com/song/2b2c7f98-42ac-4310-ba5d-da769a05cbf0
note song. Listen and make your own with Suno.
I plugged it into a key finder lol Not in E but C minor
Very interesting. I didn't know about this. Did any of this make it into the newer models?
is that the key of the track as well?
I think it worked about as well (very rarely) in V3A, but I can't remember trying it much since then
[Notes: ๐| A4๐ ก, B4๐ , C#5๐ ข, D#5๐ ข, E5๐ , F#5๐ ข, G#5๐ ข | ]
[Chords: ๐| E๐ , F#m๐ ข, G#m๐ ข, A๐ , B๐ , C#m๐ ข, D#dim๐ ข | ]
GPT 4.0 wrote that
As I said, if you want to actually hit the exact notes, it seems like a waste of credits. Not impossible to get correct notes, just not very practical. But I did find it useful as a way to make more unusual style songs from singers, like pitch shifts you would not otherwise get.
thank you!
I corrected that to Voice Leading
Yeah it is
hope you caught that part
yeah go figure ๐
I think theres only one key but you might be able to get the notes or chords
if you want to experiment
Since 3.5 cares more about prompts in the lyrics overall, my guess is maybe it actually works more often in 3.5 than 3
Assuming it worked at all, at least
^^^^^^ THIS!
Im a 3.5 only kind-a-guy
I made 3 hrs of Ambient Instrumental with 3.5
hardly a vocal to be found, works great
they fixed it at some point
I don't have very much musical ear training, so it's sort of annoying for me test. I can figure it out on a keyboard and tell obvious things, but I figured I'd leave the testing to people who can know at a listen how well the prompt works
Yeah I have that ear - i do remixes and time stretch everything into key - lost art of the Junglist.
70's Psychedelic Rock , ๐ผ๐ถ, Rhythm ๐ธ, Lead ๐ธ, Soft ๐ฅ, โ๐, song. Listen and make your own with Suno.
Maybe a decade ago on my commute I used some app for ear training, something you could do passively just by listening, guessing, hearing the right answer, and I was surprised that it only took a few weeks to get way better than I thought possible (relatively, not absolute pitch, but also you can 'remember' a relative pitch for hours, even me). But you lose that ability if you stop doing it every day.
is there any way to get just monotone robot voices? it always wants rap or asian vocaloids.
ive tried using keywords like vocoder, speak & spell etc and it rarely ever works
starting to think it just gets all vocals from the style of music and you cant really control vocals at all
Vocaloid? Try that yet?
yeah its always asian anime sounding when u use vocaloids
Robotic Vocal as well they both sound different
^ Ive only heard that on other peoples tracks i dont normally do robo vocals
not making k-pop, looking for a darker matrixy electro-tech sound
have you tried using bass as an adjective?
ive managed to get voices close to what i need but its not consistant at all
this is close
but could only get that one time
i think they really need to improve vocal control. seems like thats the number one complaint rn
I like to rip the vocals out with stem seperation and feed them to my Timbre Morpher
But not a full retexture - Suno lacks harmonic so just look up a good harmony key and use some stock vocals that fill it in on an enharmonic blend. with a little love from Harmonic overdrive too - sounds so much deeper.
and the mastering sucks - it's always too much treble not enough bass - I use Izotope to master it more like real edm
I'm hoping V4 has harmonies. I lean towards Udio for those if I want them in a song.
THis new song radio feature...is it really infinite? Seems it just makes a playlist. Will it find more songs once this playlist is over?
Good call on that
suno is great but its still a lot of hard work to make good music with it
stuff like this is going to change music forever. wouldnt be surprised if we get entirely new genres of music
ive spent like 30k credits so far just learning to prompt ๐ ๐
I love it to death - even with it's quirks
sheez i even had a whole mp3 upscaling autonomus agent designed to run rx11 in reaper while i was asleep lol
This sounds like nothing ive really heard of - Tape-loops is an OLD genre - but it sings a different sort of song
i made like 20 different flavors of tape loops to show it off
alot of old genres can be ressurected quite easily with suno - especially now it only takes 3/4 gens to get a track if your not too picky
i love the vocals but its very efficient to make instrumentals - less weird picky variables to catch you up
what! amazing
I made 12 of those in the last 12 hrs - chugging away before i go off grid until next month. Im gonna play all that stuff at the end of the fest
Do you have a youtube channel or something where you put your remastered stuff?
I had a good flow with videos before noisee went paid and cut everyone down
theres still like 2 ir 3 in the pipeline to be edited down
Im gonna have to find a different flow though i might just use Vizio to get some basic visualizers for them but it would kinda be less ideal cause they sound like a great soundtrack
I really like all the development in Freakiest Creature, looks like you have a good relationship with a GPT4 instance too
But my favorite thing to do is run stem seperation and timbre morph duplicated stem layers - so i take 4 tracks and make 8, the doubles all recieve timbre morphs using morph3 in fl21, then I add some synths and restore harmonics. Iput this stuff on Soundcloud. its like 60% human 40% AI
https://soundcloud.com/joshuasca
My all time favorite so far is an unreleased track - weird genre bending stuff
Original SUNO file
theres a lot of sounds I make with uHe stuff, suno has nothing like it so I usually toss a few Bazilles and Hives and maybe zebra 3 if im feeling spicy and I want to push my computer
Looks like soundcloud is your real home. Youtube is a pain unless you really care about videos.
yeah for sure, I was one of the first 25,000 sc users, that thing took me all over the world - they gave early users free pro accounts for life, awesome
Took a couple years off to assist the NAC - I had to drive a Sundance chief around for like 3 years.
I don't know anything about mastering but I'm beginning to hear how Suno tracks are "the start" of a song, I love to see this kind of development and see what can be done
I use Izotope its an AI assisted mastering tool
i suck at mastering - but im getting better
yeah ive heard a lot of stuff ive never heard before just making music with suno
sounds like DnB to me
but a bit aggro
pretty cool still
Can't login via Microsoft account
is there a way i can add english lyrics next to my japanese lyrics without them triggering in the song
Not now, but editing lyrics after the song is generated is a popular feature suggestion. (You can [use words in brackets] and maybe if you lucky they will not be sung, but if you try this with all the lyrics, the music will become quite strange. Because the brackets are for stuff like [guitar solo] and so on.)
alrighty thanks so much
After typing that it won't work, I am actually a little curious at exactly what the song sounds when it doesn't work though...
Both samples sing the third translated line, interestingly:
https://suno.com/song/281646df-fb1e-4d31-b9f6-a1a2cccdaf11
j-pop, anime theme song, energetic jazz song. Listen and make your own with Suno.
yeah, and parenthesis for sure wouldnt get ignored, hmm
rock, male, Deep voice with vibrato
song. Listen and make your own with Suno.
Please don't post songs in multiple channels at the same time. If you just want to share a song, it should go into #๐ฃโsuno-showcase
ok
sorry
Hi I have a question
I'm making a song with a male and female Duet
No matter how many times I try
Wherever I restrict the use of male voices in commands, they will all become female voices.
https://suno.com/song/4ca51655-2392-4bba-a614-950ac207755d
Here are my instructions:
Style of Music:
Duet, male vocals, female vocals, C-Pop, Gu Zheng, love, romantic
[Intro]
[intro music solo]
[Redfox: male vocals]
[Verse]
Leaving my hometown with a heart so heavy
Something seems missing, the sunshine so cheery
A red fox's journey, painted with regret
Missed my golden sister, a bond I won't forget
[Redfox: male vocals]
[Chorus]
Under the canvas of a colorful moon
Dreaming of the day she'll be back soon
In shades of red and gold, our story's told
Waiting for the moment our reunion to unfold
[intro music solo]
[Goldenfox: female vocals]
[Verse]
Across the fields of memories, where echoes play
A golden fox whispers in twilight's soft sway
In the silence of the night, I hear your call
A bond unbroken, through the distance, after all
[Goldenfox: female vocals]
[Chorus]
Colorful Leaves fluttering in the wind
Dreaming of the day, together we'll renew
In shades of red and gold, our story's pure
Waiting for the moment, our hearts will reassure
[Duet]
[Bridge]
Stars above will guide us, lighting up the sky
Paths entwined forever, no more goodbyes
In every gentle breeze, in every song of old
Our spirits intertwine, in tales yet untold
[intro music solo]
[Chorus]
[Redfox: male vocals]
Under the canvas of a colorful night sky(hoo~oooo)
Dreaming of the day she'll be back soon
[Goldenfox: female vocals]
In shades of red and gold, our story's told(hoo~oooo)
Waiting for the moment our reunion to unfold
[Duet]
[Outro]
Through the seasons' changes, through the ebb and flow
Our love remains constant, in the warmth it shows
No matter where we wander, or where we may roam
Together we'll find solace, in the love of our home
[intro music solo]
[Fade Out]
[end]
Duet, male vocals, female vocals, C-Pop, Gu Zheng, love, romantic song. Listen and make your own with Suno.
Hi everyone, I have a question. I just want the singing without any instruments and I also want to control the speed. Is there a trick for this?
Why is my Microsoft account not working today?
I also wanted to create a Duett. I've given up.
Duets are one of the lowest percent things to get Suno to create. And even if you are successful, my experience is that every voice change tends to push the quality down when they work. I had a trio singing a 6 minute song and by then end they had all merged into one weird voice. My best duet was actually two women with similar voices and I actually stripped the voice changes out and just let human conversational ear switch you between them.
It can be done but so far it's rarely worth the heartache. Maybe v4 will make it easier.
lack of control over the final efect disturbs me a lot... this is great tool but to be honest, just a shiny random toy ๐ฆ
for instance, sometimes it does not want to end and keeps repeating random verses ๐
is there any howto/documentation describing [commands] that work? ๐
https://www.suno.wiki/faq/metatags/
Don't think of the metatags as explicit instructions, but strong suggestions.
Instead of trying to generate a 3-4 minute clip and being upset that it doesn't end specifically where you want, go back earlier in the song & choose an Extension point from which you can focus on creating "the end of the song", where you throw everything at the wall to see what works best (and get hands-on experience with what prompts work best). It's easiest to create exactly what you're looking for if you break the song up into multiple pieces and use "get whole song" to place it altogether later.
[end] as the only lyrics, fadeout as the only style instruction, for example. (not a 100% success rate, but have fun playing & learning ๐ )
Please tell me how to pay for a subscription from Russia?
You might need to contact billing@suno.com
Suno is very opinionated ๐ "No no... we don't actually need a hook here!"
Thank
"Go away, User, I'm in the middle of making something really cool." - Suno
haha
Sometimes Suno is right. I've adopted its hallucinations occasionally as they improved my lyrics. Not always, by any means. But sometimes it's good to be open to happy accidents.
Hello Everyone
I have canceled my subscription, but it keeps attempting to charge my card, resulting in failed transaction charges. I have been trying to reach support and billing through email, but I am not getting any response from them. The website doesn't allow me to remove my card details unless I update another payment method. There have been three failed transaction charges; two more and my card will be blocked.
Kindly help.
Is there a rough ballpark estimate to how much material is being generated by Suno Daily/weekly or across any tmeframe? Has there been any public statement that gives insight into this?
sorry didn't see
So I got a REALLY good piece of instrumental that I want to extend for as long as possible, but no matter what I do I only get screams from hell and the continuation from the singing from before it.
This is the song with the amazing instrumentals at 0:35~0:49: https://suno.com/song/9d376f68-96f2-4249-8c31-c632155b6d72
these are the results, this one did exactly what I wanted (for 11 seconds) but is straight up is barking at me on top of what I want: https://suno.com/song/5069948f-37cc-4ee2-963c-f40dbc7341a9 (barking one)
another example: https://suno.com/song/af9a43df-467c-47bb-b6a5-6d9e9c319cf2
Nice to bump into you in chat ๐ 100% agree, sometimes a "hallucination" makes me entirely rethink the direction I was taking a song.
It might not be what I imagined or had been trying to coax out of it at the time, but arguably, it's almost always right about it being a pleasing and musically satisfying composition. That's why I find it funny when people call it "bad" without any further explanation.
How quickly they've come to be bored with what would have been considered literal science fiction 5 years ago, if that.
This would be SO COOL to know-- it would be neat if there was somewhere that kind of stuff was tracked so eventually a rogue data scientist (or anyone with ChatGPT) could make a chart to see how quickly this has all grown!
It definitely tries to smooth out places with bad lyrics with hallucinations. Some people definitely need to look at their lyrics if they have a lot of trouble with them.
I tend to have a good feel for line lengths/emphasis structure, but I get a fair amount of hallucinations due to a changing emotional tone in the song. Sometimes a little repetition by Suno smooths that out naturally. Sometimes it sends me back to the drawing board on my lyrics.
Suno is amazing, and it is interesting to watch it work. Yeah, if you told me a couple years ago this'd exist, I'd have said you were insane. Actually, I'd have said that last December before I discovered it LOL
I'd been watching Google's MusicLM closely, but they really like to keep Canadians away from the fun stuff. Even then, it seemed more like a proof of concept than a generally accessible tool. However, v3 really seemed to "understand the homework assignment".
Looking at the growth between v2 > 3.5, the lyrical adhesion & knowledge of when to contract, expand, or delay words on a per-syllable basis is absolutely jaw dropping. Of course there's still some instances where it'll nail the first verse and start to lose the thread by the second verse, but that's where you use extension points.
What did you use to master this?
Izotope
Hello, for me I can no longer connect to my suno with my account, thank you for your help
Kelvin is driving harmonics
I tried but failed to make it instrumental only within my single try
Though, in my biased opinion, the first came out pretty good:
https://suno.com/song/4f456022-29ad-4108-99aa-3307b25d30da
The second one... God, the beginning was horrible: https://suno.com/song/7ef44519-2061-47b6-a7e6-44973db1d442
I should have tried it with an instrumental break at the start
and all the stems were put through morph3 then recombined with the original unaltered
They usually all end up like that, maybe it's actually pretty good, but it's totally not what I wanted
With Instrumentals, its all about getting the style prompt right you know, make it sound how you want it want then switch suno to instrumental and let it run
Joshuasca, เธฟโฑ โณโตโญ สวษขษจฦ, ึสษฎสษจสษจีผวส๐, ฦISโฉW หฅIฮฦแก หฅโNIWIหฅแ โฉS, odd atmospheric tape-loops, tape-fx song. Listen and make your own with Suno.
I asked at the last meeting...
the question was sidestepped. ๐
The voices of hell were really strong on that second one
I figured when chat gpt told me to ask you
Do you guys plan on doing any sort of brand abassador workwith musicians who use the service?
I am part if one of the most ANTI-AI music scenes and I love rubbing it in their face "I dont even know what im doing I just said chat gpt - make it trippy" and I got all this!!!" lol
hasn't worked for me since yesterday
I tried another attempt with break but no good luck, if we only had start trimming ๐ข
https://suno.com/song/187ee23d-c4d8-4f85-be8d-d9fb7eaadbcd
https://suno.com/song/a384467f-a2dc-4e8a-b3ef-6134298017e3
I was thinking about doing it in v3 or v2 instead of v3.5 but don't have spare credits to try.
(The more likely thing of getting what you want is to wait for that audio sampler to come out in the next version as that definitely will get you what you wanted)
there are some more employees operating on Twitter who seem to be in charge of promotional campaigns... they seem less visible here on Discord (probably discord is not their target)
I will disassemble and reassemble it in another program anyways, so long I get a longer version at that peak instrumental, or just more of it I am happy
Thanks for trying anyways
Are you trying to do the Microsoft login? The team is aware but they need to get in touch with the MS team according to #๐ชฒโbugs-web message
Edit: just saw 3Daizy posting the same reference, sorry for missing that
Well I was thinking more like live performers with ties to the music industry (LIKE ME!) not more online media (saturation imminent!)
yes for me it's Microsoft
I was just suggesting how to find them.
Im gonna do 2 hours of suno live on July 1st
epic chillout end the big festival on the top of mt shasta dealie
I'm a core member of Starseed Productions, we bring about 800 folks to our camp ground on top of the mountain from around the world every year
I used to work for Symbiosis and Burning man LLC for 7 years
Awesomesauce but not really the place for this here as this is the tech. discussion channel
Thank you for informing me @hollow hare
Hopefully it gets resolved soon hasone
Me too @hollow hare
If anyone uses Diktatorial Suite or other AI mastering tools, the guy who runs that tool is asking for Suno specific feedback on the reddit sub: https://www.reddit.com/r/SunoAI/comments/1dck8y3/diktatorial_suite_needs_suno_users_feedbacks/
Thank you very much to @hollow hare for giving me information. I was afraid of having lost my account. Good luck everyone and happy creation ๐
Hey can anyone help me in generating the audio in my preferred choice of singers say male or female @amber maple as i checked there is no option to do so in the playground
try using "male vocals" or "female vocals" in the style prompt
didnt worked. tried all the prompts as well
are we talkin in the custom mode where in the genre i can select genre
Yes hence z mentioning the style prompt section.
If it still doesn't work, you can repeat the prompt for more weight, especially when it is a style that does not normally use that gendered voice (e.g. "clear female vocals, female voice, female, woman, ...").
Alternatively, you can use other AIs to swap the vocals in the final content: #๐โsuno-school message
And if you were not aware about the wiki: https://suno.wiki
i was not Thanks for this !!
Any information for V4? Will it be available soon?
Nobody knows. We have to be patient.
i need inpainting because the model accidentally produced the most awful "oh oh oh oh oh" screaming in one song that is otherwise pretty ok
should probably just delete it and remake....
Split to stems, clear vocals as best as you can. Re-create the instruments using a DAW, instruments, find the chords using melodyne...
Or something similar...
I know inpainting would make it a lot easier, but you could place an extension point before the undesirable part, then try to continue the song as is?
(Sometimes the resulting lyrics on the final song might seem jumbled, but the important part is whether the song sounds good!)
Unfortunately it's the first 3 seconds of the original
any future plans to train Suno on data from Souncloud songs? it would highly improve the quality of generated subgenre songs
๐
Can someone help me with voice cloning application of bark?
So unless you get lucky almost every suno song needs a de-esser out the gate. Like a warbled tape. It cool that you offer 4 minute tracks but after 2 minutes it can to completely off the rails. Instruments, vocals, they all get a bit wonky. Its a real roll of the dice what quality you will get. Even with specific prompts. I know we're on the bleeding edge of this technology and it's pretty amazing that it works at all but please don't let yourself get carried away simply trying to compete with the other guys. Suno has the advantage in instrumental composers and decent balancing with vocals but udio has the corner on clear vocals. Between the two of you there is something very special but individually each track needs a lot of work. And frankly if everything worked together in concert with something like ripax this would be an amazing sweet and then I'd love to get all of this into VR at some point or ar so that it can be implemented in a truly 3D space like a composer. I have a few ideas on how this could all work but I won't get into all that right now
I've made a song, but now I want just the instrumentals for that song. Is there a way?
I'll extend the song and click instrumental only for the extension
hmm not the same acoustics as the original when I did that
Sometimes you can get lucky and have the AI generate the same thing without words but there's no guaranteed way to do that.
If the vocals don't have a lot of effects/reverb on them you might be able to download the MP3 and use something like https://vocalremover.org/ to separate the vocal + instrumental but it's not a "clean" separation a lot of the time. Worth a try though.
I have to play around with time of the extension for instrumental only, trying different sample points to generate from
agree with meowmeowbeans, https://vocalremover.org/, or try ripx free trial. it separates everything.....mostly. https://hitnmix.com/ripx-daw-pro/
Hi. I suggest stopping the deduction of credits if only the [END] tag is present in the song's text and the styles are cleared. Sometimes, the bot can't finish the song and keeps trying to continue it endlessly despite the [END] tag and cleared styles, which consumes too many credits.
When I do [End] in the lyrics section and end shortly in the style, I usually get a refund for the few second clips
hey guys do you think suno will add sound uploading
so we can extend songs like udio
Yes 100%. It's available to early-access testers right now and seems like it's coming pretty soon.
Not sure if anyone else has discussed, but I find that v3.5 is really bad at instrumentals, v3 is really superior.
can someone how i make this style? Is Manele style, gipsy romanian or balcanic , but i cannot get it ๐ฆ https://www.youtube.com/watch?v=zvZMhBoPOdU
FLORIN SALAM - Tiganco sau romanco
ยฉ & (P) BIG MAN Romania http://www.bigman.ro
Big Man Online:
- facebook: http://www.facebook.com/bigman.ro
- Twitter: https://twitter.com/BigMan_Ro
Licensing/Contact/Booking/Marketing:
- audio@big-man.ro
- (+40) 0764.80.80.90
Despre acest canal:
Aici puteti asculta intreg portofoliul casei de discuri Big Man...
Does anyone know how to get SUNO to make the singer a young person?
I'm trying to write songs as if they are performed by a character in a game, but the vocals are almost always too old sounding
From my observations, they've highlighted the different model's strengths in the selection box;
v3 produces a broad range of genres and styles, and can be incredibly flexible when trying to create specific sounds. Also, the "hiccups" or "quirks" tend to be rather unpredictable.
Subjectively, v3.5 is phenomenal at writing instrumentals, but does have a tendency to include vocals. However it's a night and day comparison for v3.5's ability to pick up on the structure of a song & continue/refine.
The trick is knowing when to use which. Like PB & Jam, they're great on their own, but even better together. ๐
try Miku Voice kids ๐
idk if this is bug or not, but i'm not receiving my daily credits, can someone explain me why?
Have you waited a full 24 hours without it updating?
when server is full you cannot use suno, just for subscribers
almost a month actually, the last time i did a song was in may 13
you payed ?
no
but we used to recieved a daily 50 free credits dont we?
ah
under heavy server load, free users aren't given daily credits
@mods/team, is that true? First I've heard about this.
no, when server is full, is just for subscribers, for who payed . Not all days are free.
Do you have any citation for that? Like a website, or a message link to a member of the team saying that?
Not saying you're wrong, this is literally the first I've heard of that, and it would be helpful to know what to reference when sharing this in the future (if that's the case)
@dim glacier @neat nexus Sory for pinging, you're just the only mods with active statuses ๐
so the server is full every day since last month?
i agree with that
i dont know, just that i see. i stoped to pay because im out from city, after that i try to get free credits, but every day was full server, so i pay again ๐
how do you know when the server is full?
now, im not see when server is full because im subscriber now, but when i not pay, on pc i see when i try to make song , i see one text down to the screen write SERVER IS AVAILABLE JUST FOR PRO CLIENTS, like that.
If the service is extremely busy and starts becoming unusable for everyone, free users may be unable to create songs temporarily. The team will put a banner up on the site to tell you when this is happening.
It is rare that this happens.
(there were issues in early April I think? Since then it's mostly been ok)
This does not affect whether you get credits or not
It just means the Create button won't be enabled.
so why i'm not receiving them?
That, I don't know. I've seen other people mention the same issue. Try emailing support (support@suno.com) with your account username/email as there might be something affecting your account. It's probably a bug.
oh ok
i'll try, thx for the help guys
I subscribed day one that I found out about Suno, so I'd never seen that screen ๐
Thanks for your help!
You'll know when it happens because the general chat channel explodes ๐
Does anyone know if its possible to prompt a violin solo with no background instruments?
Luck of draw..
Although I donโt know an annotation to invoke a violin solo, you might find some inspiration in this song: https://suno.com/song/4ad86fc1-fee9-4cfe-8b13-ac4355ec91f7
Iโm amazed at what @unborn quiver was able to pull off with detailed bracket instructions.
High Octane, Electronic, beat drops, bass reach, lead metal guitar song. Listen and make your own with Suno.
not working lol the hell
Ill try again once it finishes loading. I just thought it was pretty cool lol
Metal song. Listen and make your own with Suno.
โค๏ธ
I wanted to ask you, how did you discover all those instructional tags to use?
I just wing it with the less conventional tags and see what works
Lol success! Iโm experimenting with this approach now
https://suno.com/song/1b6f3a18-0fba-4572-bdd5-d4b3c164ea91
Also it is a lot of cutting off the song where I don't like it, for example this song is 13 extensions total
emotional metal dynamic range solo rich low bpm progressive metal slow intro dynamic male vocalist bright and dark song. Listen and make your own with Suno.
The audio quality I think drops a bit at the end which is a shame
But flow is more or less what I wanted
There is a solo on 5:35 that is a bunch of few second extensions because it kept making it short
So you can see that thumbnail changes like a bunchload of times lol
It sounds good! Yeah Iโm really fascinated by the verbose instructions. I saw another example of this except with vocal instructions from @mrwholesome https://suno.com/song/2f3f5c61-e090-4e09-882a-a786c3fb410d
KEY: G Minor, BPM: 80, Choir, Choral, Orchestra, Symphony, Heroic, Atmospheric, Epic, Heavy Metal, Thunderous song. Listen and make your own with Suno.
Yeah but also its a little RNG that instructions sometimes will and will not work
I also like putting it wacky things like [Explosion] or [Speeeeeed] lol or whatever just to see what happens
whatever you put inside the brackets is processed by the AI, but usually not sung (unless you stuff a tag right in the middle of other text). They don't always have an effect. Sometimes it seems like they rarely have an effect. Occasionally, tho... you'll get a gem.
I wanted to see if I could get You Tubas to be more farty, so I extended from 00:21 with [Sound Effects - Fart, Fart, Fart] . It's definitely more farty. Because of the tags, or just randomly? Who knows. ๐คท https://suno.com/song/4cf79d93-efef-4bcb-8543-c40f81ce581a
black metal anime electroswing, tuba intro, orchestral, tuba solo, dubstep tuba, glitch tuba, trap, song. Listen and make your own with Suno.
Oh nice!! Super helpful! Thanks so much ๐
You're very welcome, and ps. there can never be too many Nates.
๐
Is there a website to upload and compare multiple song tracks?
@unborn quiver hereโs my first attempt with tweaking verbose tags: https://suno.com/song/a6e95d22-224f-432c-bb05-1da192e1a45a
heartfelt, experimental, synth, trap, techno song. Listen and make your own with Suno.
Whats in your heart if this is heartfelt lol
My friend asked me to work on something synth hybrid eske thingy so I am doing something like this, still in the works tho https://suno.com/song/ea4ed7e7-2426-410f-a48f-af7799c43dd5
Industrial Metal, Synthwave, Breakbeat, Acceleration, Drum and Bass song. Listen and make your own with Suno.
I missed this because I was away at the time. Sorry.
Not a worry in the slightest ๐ ๐ซ
would be cool if there was a feature to use a generated vocalist haha
I created a chatgpt that can be used to create a song using the tags available. https://chatgpt.com/g/g-7m01NDd1w-suno-ai-generator
How does one apply or receive alpha testing access?
Keep an eye out on their social media (e.g. https://x.com/suno_ai_)
to get early access?
Yes, they had a contest there before which is how a few got access to it iirc
Let me double check
Want to make a song from any sound? ๐ง Help us make a video for VOL-5, and weโll pick 3 winners to receive early access! 1 winnerโs video will be made into our next VOL-5 to be shared on social media ๐
Comment on this post with a video of audio that youโd like to be turned into
I don't use social media like at all so i would miss this kind of stuff.
I don't know Suno's plan to expand testing, just a guess, might be a good idea for people interested to take the Suno Survey. #๐ณโmini-polls message
Perfect, doing that now, thanks!
Experimenting with allowable tags and trying to integrate with generated AI lyrics: https://suno.com/song/2182aa3f-5518-48b7-86ec-dfde6eaef559
heartfelt, futuristic, dark experimental dubstep swamp blues song. Listen and make your own with Suno.
Feel free to copy and re-integrate with the tags you like best ๐
@glacial garnetand how does it work? can u make a short tutorial somewhere?
@glacial garnethow does it work? step 1 , step 2? i get a chat line and on backslash i see nothing related to suno tags?
Thanks, i'll try some tags. It's a shame about the ending that it's skipped, but it's already a matter of version 3.5. Let's see if it doesn't take long with version 4 and in addition to adding, they improve.
Servers having issues right now? Noticing songs get to about ~1:50 and restart from the beginning again. Maybe itโs just me but itโs happened twice
Hi, I have been waiting for a response from billing@suno.com for over a month. Does anyone ever respond to this e-mail from support?
Email support can be behind, but it shouldn't be that far behind. Double check the obvious stuff like spam filters just in case and re-send a new email, if you haven't.
This is the first time I've sent an e-mail again after such a long time and I check the filters every day. No answer.
if i acedentlly deleted a song wile purging some other songs can i get it back? i have the link to the song but i want to get it back in my library, not just put it on a playlist
If you trashed it, it's in the trash menu in the library. Top right icon on desktop.
anyone know, how to make screams vocal?? what command front..??
Anyone has an idea how to generate a suppressed, light effect on the singer's voice? I'm looking for a result similar to the start of this song:
https://www.youtube.com/watch?v=gfJyKq7zhhA
Stream/Buy: https://idol-io.ffm.to/Arise
Pre-order the 'Arise' 12" vinyl: https://hy-records.com/collections/store/products/clann-arise-7-vinyl-lp
Composer/Producer: Sebastian McKinnon
Lyrics: Sebastian McKinnon
Vocals: Charlotte Oleena
Violin: Chloรฉ Picard
Arrangement: Francis Choiniรจre
Conductor: Francis Choiniรจre
Orchestra: Orchestre...
This is nice to create lyrics:
This is very good, it also works in Hungarian
Yes I tried it is nice. Also this is good, both.
Your GPT understands Hungarian, too ๐
Yes, it uses user's language
Even I created Turkish, French, Swahili languages songs
It is more difficult to make rhymes in Hungarian, my language is so hard. But the GPT works fine.
You may try. Each language has a different sentence structure of course
how to make a song using /chirp
Chirp bot is retired, use the website https://suno.com
Some helpful starting links: https://suno.wiki
https://youtu.be/hsecBccWETM (edited to newer link)
https://youtu.be/hsecBccWETM is a newer video @arctic crow
Hello, I'm 3Daizy, a moderator and guide at Suno. Welcome to my channel, where I share my journey and experiences with AI-powered music creation. Whether you're a budding musician curious about AI or an experienced artist looking to innovate, you'll find useful insights and practical advice right here.
Join me as we navigate the fascinating int...
This is some interim community leader stuff right here.
Immediately bookmarking this video to share in tandem with the wiki-- great quickstart guide for new users!
Thanks ๐ฅน ๐ผ ๐
It's also in 14 languages (links in the video description)
after like 6 songs i let suno create i can't click the create button anymore?
Do you still have remaining credits?
So why is the Lyrics section empty when you Extend a generation that have Lyrics on its track page?
Unless you "reuse prompt" it only extends afterward-- you can use whatever lyrics or style prompts you want to continue from.
Sorry, I was in the wrong channel ๐
It's very easy to get the lyrics back; 'Extend' > 'Reuse Prompt' > Remove any lyrics that have been sung already.
edit- not a worry! Becoming familiar with the default settings & adapting a workflow around those is totally a technical discussion in my mind ๐ Unless you want to make #๐กโsuggestions-web or report #๐ชฒโbugs-web
https://plugins.audacityteam.org/realtime-effects/plugin-suites
I'd recommend the MFreeFX bundle that comes recommended through Audacity's "get more effects..." option in the effects section. Also the Muse tools are wonderful (both beginner and pro options included, a lot of basic settings as well)
And the shining gem is the OpenVino AI plugin that can separate stems of any song (provided you have Intel architecture)
Oof.
(on the bottom of the same page linked above)
To be fair, I haven't experimented with a lot of different effects suites, I just recommended the ones I'd tried. It looks like OpenVino is windows Intel only, but hopefully there's something here that can help! ๐ค!
::cough::
Is there a reason they pulled all the play stats off the songs in your profile?
They did?
(Or was this to someone else?)
right! I have about 6 songs out of 30 ish that show stats, and it appears random
Is anyone else having the issue or is it just me?
how can i know what tone are my song of suno ?
In custom mode you can motivate the direction of tone with your input into "Style of Music" for exmaple: "upbeat irish rock, ancient dwarves, electric new wave pop, Celtic monks, wings of the fae"
Custom "Style of Music" for exmaple: "upbeat irish rock, ancient dwarves, electric new wave pop, Celtic monks, wings of the fae" https://suno.com/song/d15b4e9b-fd54-4fe0-b6a7-e9fd1b7296ec
upbeat irish rock, ancient dwarves, electric new wave pop, Celtic monks, wings of the fae song. Listen and make your own with Suno.
I know there's other ways to separate stems, but OpenVino is the easiest way I'm familiar with; absolutely life changing to be able to treat any song like it's the original master track-- removing vocals for karaoke tracks, or isolating them to appreciate the timbre and minute vocal fluctuations that are otherwise lost under the other instruments, being able to compress & master individual tracks (drum, bass, vocal, and all other instruments grouped together unfortunately, but still!)
Nice but how u can know the tone of your creation
Do you mean which musical key the song is in?
I use Audacity but i dont know what this extension do
yes
Wait...is this better than UVR5?
Zero comparison, I'm just a scrappy enthusiast-- I suggest this because it's absolutely free to get started, provided you have an intel CPU (which I THINK is the leader still? AMD might have overtaken recently. I dunno)
There is a lot of randomness in your songs creation, using tags (https://suno-ai.notion.site/Using-Metatags-34944efe09ec41d693e314f13af44695) help and setting your background echos or repeats can help with the flow and tone. However I will say it usually costs 100's of credets or more for me to get a song the way I like it. Because in the end there is a lot of randomness.
Metatags guide prompts but arenโt verbalized themselves. Like stage directions or parentheticals in screenplays, you can use metatags to describe a specific part of your prompt (e.g. [laughter] and piano interlude) or to influence your entire prompt by describing speaker characteristics like [whispering] or music genres like rock.
I'm not trying to fanboy on UVR5, I'm trying to find out if this could be useful for me to use instead?
UVR5 for those following along and lost:
https://github.com/Anjok07/ultimatevocalremovergui/
I can make some comparison tests to show you if you wanted some more references before installing everything?
I'm happy to try out new tools anyway, so just let me know ๐
Just a surface level response from someone who uses it. If I wanted marketing speak I'd do a Google search. You're a real person who can tell me their thoughts, so I'm interested from that perspective.
HAHA, NOT SURE WHAT THAT TYPO WAS
What we do with Suno is kind of its own thing, so getting perspectives from fellow users is much more useful.
Okay cool, I'll shoot you a DM later today!
Reallly impressed with UVR. Even the default models it ships with work well but then someone showed me that there's models to do other stem splits besides just vocal/instrument and I'm now going to just obsessively stem-split with different models to judge which output stems are cleanest for different clips. So, that's going to keep me busy ๐
And how fast it is too
It doesn't work well on my Thinkpad but there is a fork if you look around that is supposed to work better on AMD Ryzen if you have that hardware and are lucky enough to be supported.
I don't really use a DAW much so mostly what I have used UVR for in the past is to split out the vocal for use in music videos as the track for the lipsync. It helps to remove as much distraction as possible so you get a better sync.
Nice.
I'm envious.
This Thinkpad is my highest end hardware at the moment and I got it used, before I got into any of this stuff so it's just good enough to run the one game I planned on needing the Ryzen for. And I'm so distracted by music and music videos these days I don't even play that game.
I was into VR which, before this, was THE thing to push GPUs. Then LLMs came around and, welp
That's the biggest appeal to OpenVino & Audacity for me-- incredibly low educational bar of entry.
As I'm getting UVR5 installed, it's just vocals, right?
I see no reason to upgrade now until we get more than 24g of vram
It can split more than vocals
I think it's name was when it only did that
I gotcha, no point in rebranding once you've reached a certain level of success. Like half of "Prison Break"
One thing that could be improved with UVR is if it gave you a brief explanation of what each downloaded model does. I had to search to figure out what the purpose of each model is for and it turned out a lot of it was down to the range of music they were trained on.
I have the gui open; any models to recommend starting off, or just jump in and play around?
Ahh, I gotcha
These are the ones that interested me most (ignore the highlight, that's just what I have selected)
Oh damn. I forgot to download them again after the last time I did a clean install of my system. I need to test them all out again to find the one that worked best on my potato...
hi folks - since Monday I no longer see my remaining credits on my profile, has anyone else had this issue? I'm only about 10 days into my subscription.
No, the credits section doesn't show up at all anymore.
Have you tried clearing your cached data for the website?
I think it has moved to the top right of the website.
Yeah mine is still in the lower left
Who keeps yelling 'Log Off' at my house.
I will never log off 
if I squish my browser width, it moves to top right...
(mobile, smaller browser window)

sadly it doesn't make a difference. Even in incognito mode I log in and there is no indication that I have a subscription or that I have any credits.
ah yeah there it goes
try this link:
https://suno.com/account
aha.. I squished my browser and it did pop up on the top right!
Sounds like a UI thing then
Does it show up on the left again after you "unsquish"?
no it doesn't.
Why squish when you can
W I D E
And it looks like I've lost my credits. I only have the default 50. Damn. Something went kaka.
Go wide or go home, imho. If I don't have to pivot my head then I ain't browsin
has anyone heard whether Suno plans to publish an API?
"maybe later" but nothing concrete
Unless there has been something else more concrete since then
They aren't against it, lot of other work they want to do before then
^ It is in the plans from what I heard but that's pretty much it atm.
Likely we will hear about it in an official manner when the officials apps are ready for some testing, whenever that would be
check on the subscription page, find the link button for EDIT BILLING DETAILS
this should have a list of your billing invoices... it may have been a glitch when your bill came due, and the transaction didn't complete...?
there should be a link to retry payment (once you confirm you weren't actually billed through your bank/cc)
Sounds like you signed into a different account. Not suggesting thatโs exactly what happened, but itโs presenting as if you only have the 50 daily credits.
When I go to the subscription page I get an endless spinner.
I think Suno is mad at me for using an unofficial API. That's my bet.
Yeahโฆ that MIGHT be against ToS? Iโd be curious.
oof, I am sorry.
If you have another browser...?
(fwiw, there were recent steps taken to reduce bot-generating... it may be related to unauthorized API... I don't have more detailed info, unfortunately)
otherwise I guess contact
billing@suno.com
Iโm looking through the ToS to see if thatโs mentioned anywhere, so far I just see they have the right to cancel your service if they think youโre abusing the free credits with multiple accounts.
https://suno.com/song/812c5903-393c-4bee-bdb7-59005348812c
thought some people might find this interesting
Borat soloist, comedy show solo, comedian virtuoso, laugh track, stand-up impressionist, Laugh House song. Listen and make your own with Suno.
not a song
just... weird, and may be useful
1- Did you try with another browser? Ideally with incognito or no add-on installed on the other browser in case they may be interfering with something.
2- If that did not help, Did you try checking by creating/logging into another account and seeing if everything was fine in that one?
3- If that appears to not change anything, try accessing the website on VPN in case your ISP was throttling something.
I combined some of these with musical prompts with interesting results
good suggestions. I tried 1) but I will try 2).
like, adding cello to the end basicslly creates thay Nashville radio show:
https://suno.com/song/e1756e59-e64a-4b8e-b596-79130816640b
Borat soloist, comedy show solo, comedian virtuoso, laugh track, stand-up impressionist, Laugh House, cello song. Listen and make your own with Suno.
Not necessarily โtechnicalโ, but just wanted to share other snippets of the ToS that might be relevant, at a glance
First image is things you agree not to do
Yup - it's my account. f-me-in-particular. Works fine if I log in with a different account.
Out of curiosity, and with zero judgment, do you ever make songs with one account and extend them with another?
these seem reasonable
if you're asking me - no I never played any account shenanigans. I am (was, I guess) a fully paid subcriber doing legit music generation for my own research project. But I needed to do it programatically which is why I was using a wrapper as an API.
I'm by no means an expert, and was just trying to shed some light on potential snags-- I don't want to insinuate anything untoward might have gone on, and like wetcircuit mentioned earlier it's probably best to contact the team directly by email if there's any confusion.
I hope everything gets sorted out, the world deserves to hear your music ๐
yeah no worries at all. I really appreciate the help from you all.
I emailed them yesterday - let's hope they don't take too long!
guy, wonder you all encounter when extend the song, it still on encounter error and refunded. I have face this issue since yesterday till now. Generate 12 times. all show refunded
If you're extending from a quiet part, sometimes it has a hard time "picking up" from that point. Try moving the extension point back a bit-- although I had one song a month or so ago on v3 that just refused to accept any extensions longer than ~10 seconds, even v3.5 gave up after a minute ๐
๐ this
the issue is probably at the end of the clip you're extending from..., it has a bit of silence at the end after the music has stopped.
the extend clip hears silence, continues silence for a few seconds and gets refunded
theharber has the solution: set the Extend from time timestamp to before the false ending...
Lovely guy, it solved my issue. New thing learnt! Thank you so much ๐
@neat nexus @fierce bloom
From some quick tests comparing UVR and Audacity+OpenVino, I don't hear a heck of a lot of difference. There's a few times when OpenVino captured a bit of the bass into the guitar track, which means the bass track would sound a bit empty at that part, but beyond that the audio quality was largely comparable.
OpenVino;
- pros -- splits song into 2 or 4 stems, and places them directly into the audio timeline for immediate playback, editing, etc.
- cons -- based on Intel architecture (works with AMD), and doesn't accept nvidia graphics cards (afaict), ~30-40 seconds to separate all 4 stems into separate audacity tracks on my AMD 3700x
UVR;
- pros -- multiple models to choose from which presumably handle their respective styles with better accuracy, works with nvidia, ~7 seconds to create a single audio file with the selected separation on my 4070ti, able to "negatively select" instruments for separation (ie. save file with only vocals, or save file with vocals removed)
- cons -- files are saved to a folder, requiring extra (if minimal) setup to compare audio, edit, etc.
~~ ~~
Conclusion -- Both are great tools situationally, I feel like I'll more commonly use Audacity+OpenVino when experimenting with a song or exploring it for the first time, and look to UVR when I have a specific modification that I'm looking to make, given the finetuning settings.
I don't know if I expected the resulting audio quality to be noticeably different, but the biggest difference just seems to be use cases.
Thank you for doing this. This is the type of technical discussion I enjoy seeing and I appreciate that you took the time to test it out for us.
i wonder if the CPU openvino is RAM Speed constrained or compute constrained on the 3700x... i have 13600K and overclocked RAM maybe i try to reproduce your result (i too have openvino)
are you willing to provide the same audio file? or maybe there is a internet archive audio public domain we can do another quick comparison?
Does anyone know how you format chord prompts? -- is it like this
[Chords;|| Ebm / B / Abm / Bb ||:]
S/O to @charred merlin and @hollow hare for helping so much! Truly appreciated โค๏ธ
Thatโs pretty much the gist of it, but very hit or miss often enough
Personally havenโt tested it enough to have a confidence in it
ta mucho.. im not sure it really does it tho...
it would be gret if you could just type in TAB ... land promtp with -->. [arpegiate through chords:|| Gmin / Ab / Bbmin / Aaug7 ||: ]
i imagine that "acoustic Tabs" in the prompt and Ebm / B / Abm / Bb lyrical stuff under them on a different line singing might work acoustic, i haven't tested either method however
@neat field thanks ill give it ago...
next time i'm on free tier i will work to remember you for my experiments (i spend half of some months on free teir testing and half on $10 a month 2k credit creating)
Don't forget with UVR you can pick "Demucs" as the process method, in which case you can pick a stem, and then either isolate or remove it, e.g. You did mention negatively selecting instruments, but as you mentioned specifically keeping/removing vocals, just wanted to make sure you'd noticed this usage.
From what I had seen in the past, several people tried but cannot seem to make it work consistent in v3/.5
I would not be surprised if the AI is able to do such a thing but it is a matter of finding the right terms that it may have encountered during its training... easier said than done ;_;
JFly mentioned this link earlier that this type of chord notation worked in v2 but does not appear to work much in v3/.5:
#๐โsuno-school message
#๐โsuno-school message
Someone said they got a consistent rhythm here but maybe it was a coincidence? Maybe not? #๐ฌโgeneral-chat message
I remember someone once talking about getting a trichord rhythm working but others said they could not replicate it.
Hey guys, any tips (I guess out of the Suno) how to make the vocals sound less "artificial"? ๐ Cant really describe it, but I really like the melody and tempo, but it just does not feel polished and I have no experience on how to do it...
https://suno.com/song/0315b45d-af97-44dd-a689-2958d65622bd
sci-fi hopefull rock ballad song. Listen and make your own with Suno.
Try using "clear vocals" as one of the style prompts in custom mode. For example, if you wanted a female voice, I would do something like:
Clear female vocals, rock ballad fusion, sci-fi hopeful
Why the commas? To distribute equal weight chances among the different sets (which I split into voice, genres, emotions). For a bit more info, see #๐ฌโgeneral-chat message
Er, well sci-fi is not an emotion so probably shift that somewhere else, maybe replace the fusion keyword...
By the way, fusions can work with something like this was well: Fusion of [style 1] rhythm and [style 2] melody
I guess it's too late for that. I was unable to recreate this style and it was combined out of several generations - now I am happy with the song, just want to polish it...
Ah, I see. Well, for cleaning up vocals in the final audio without redoing from scratch, that is going to be quite hard to do.
One easy-ish method is to use other AI tools to replace the vocals with one of your choosing (see #๐โsuno-school message)
There are other tools that do something similar but in a more manual way, it can be quite tedious though... though those were the ones I know of, mind you.
I'd use a stem splitter to separate the vocals from the instrument, and then use a service to replace the vocals (Audimee, Kits, Lalal.ai) on the vocal part, and then mix that in with the instrumental.
I don't generally do this though as I end up liking the original vocals more and can't find a natural fit elsewhere ๐
thats what I was affraid of when replacing vocals. Well I got what I expected ๐ Thanks @hollow hare for the tips, I will look into it and probably try'em all.
Hello, since yesterday, after generating any style of songs, I always have a strange robotic voice and an echo (as if it was recorded close to the microphone, or I would compare it to the sound of an accordion). Don't know what it could be? https://suno.com/song/3ff627ec-4c08-4040-9838-51963c2fa3bf
pop song. Listen and make your own with Suno.
kids song song. Listen and make your own with Suno.
I saw mentions of stem splitting here and it actually blows my mind that noone ever mentions RIPX
it's literally the BEST stem splitter available that i've found
by default does a great job and also lets you manually edit every single sound on each track to clean up to perfection before processing with another daw
I have a hunch ...... now hear me out coz I have looked into this.... i think.. that suno thinks a kazoo is a glockenspiel and a harmonica is a blues electric guitar. Any one agree? (note: this is a tongue in cheek request for support feedback ๐ )
Does anyone know the trick to getting suno to play the conga or bongo drums?
try african drums
thanks
next question... does amyone know te best routine for creating MIDI out of Suno ?... FADR or Rip DAW... also when you write chords you would do them in box bracets -> [ ] wouldn't you ? like [ Bmin / A7 / Dmin / Caug7 ] ?
this doesnt seem to work... or maybe is does,..[Extended instrumental intro]
[Chords: Ebm / B / Abm / Bb ]
[Record scratching]
not gonna lie it would be great to have a Download only instrumental button. on songs wich have a singing voice
cuz the ones with voice have always a WAY better structure to my ears
I was just using one of my instrumental Suno tracks, but I'm sure there's a better example of a song with dynamic range & variations within the potential stem that we could agree on as a "test track". I didn't think the differences were noteworthy enough to justify sharing the audio tracks of my findings, but let me know if you'd be interested in me sending them as a DM o something.
Search the above conversation for "stem separation" "openvino" or "uvr" and you can learn to do it yourself! ๐
yeah it's all good, no worries, thanks for your awesome contribution either way
ive found that when I write the lyrics as 'oooooh' and a'wwwwww' and then it act like a song with song structure but doesnt have loads of off putting lyrics...
yeah i use already a voice deleting software
but its not as good and clean compared to a clean pure instrumental part from scratch withougt cutting manual something out
im using FADR but looking at Rip now... as well as im not convinced that FADR doesnt MIDI notes well enough...
maybe it comes in update 4.0
Oh the amazing that will come! Still very early days! Lots of room to grow!
stem separation built into free version of virtualDJ 2023 too as an FYI : https://www.virtualdj.com
With over 100,000,000 downloads, VirtualDJ packs the most advanced DJ technology. Both perfect to start DJing, and perfect for advanced pro DJs.
let's say i have a 2 minutes song almost good just need a 30 secondes extension to end ......not another 2 minutes of pain that i cannot ask to end when i want.........any suggestion for extension less than 2 min
tried simply using [Abrupt End]?
Mmm never test ...will try right now
because [end] does not work lolon my side
maybe [Short Outro] ?
I've had some success prompting [outro] and then [end]
so its a technical issue it respond better to [end] or [abrupt end] with v3 than v3.5
please, why is it that songs don't play at times on the platform?
well its black on white ... v3.5 ignores End prompts
Although v3.5 can end the song, it often doesnโt want to. It will sometimes create a false ending then start the song again.
another thing.... does [verse] [groovy] [intense] [melodic] in the same line influence something or i should just leave [verse] alone
Yeah, it's not an explicit instruction, it's a suggestion. Finding the right extension point & stripping all lyrics and other style prompts is also important.
Most important is to experiment, and realize that it's not a matter that Suno isn't listening, it's a matter of needing to figure out how it processes the instructions and try to work alongside it.
I'm not sure if this is the right place to ask this, but here I go! I was wondering if the subscription package that you buy replaces the 50 free credits you get daily or if the purchased credits were in addition to the free ones. Like, if I have the subscription do the first 50 credits I use in a day still take away from the amount I purchased or is it only after using 50 credits that it starts using up the purchased ones?
You get the entire month's credits at once, and don't accumulate the temporary daily 50. (afaik?)
Thank you! If you use up your monthly credits do you start getting the 50 free daily ones or are you just done until the next month?
i have premier plan 10k credit per months reload on the date of invoice also at the time you subscribe for me was yesterday at 22:45 was long time to wait haha and premier plan dont have free credit i run out of my 10k credit and dont have the 50 free credit perday that the free one have .... we should have it too so the wait is more bearable ๐
Thank you for letting me know! And yes I agree I think it would make sense if the 50 free credits daily began when you run out of the purchased ones
premier plan is 328.76 credit per days if you manage yourself .... i cant ahahHA
That's not possible. I need at least 1000 daily... on a slow day.
I feel that my refill was yesterday already down to 8470 lol just trying to catchup
You guys can't even assist and anwer a simple question
I dont understaind the question ๐ฆ
songs won't play when you click on the play icon
I have not had this issue
maybe its a browser issue
thank you
sometime just refresh the page helps
Done all that, still the same thing
windows or mac
windows
yesterday i think people complain about microsoft OS and SUNO
Are you clicking on it from the library where all the songs are listed, or from the song's own page (with the lyrics and prompts etc)?
That issue was to do with users not being able to log in to the website with a Microsoft account, not due to using a Microsoft OS.
from all areas at the moment and nothing is playing
Can you still download the song?
not just my song, but all songs on the home page as well
@languid dagger Do images load?
Some users in the #๐ฌโgeneral-chat channel are mentioning that they cannot access the site at all at the moment.
haven't tried that, wanted to listen to the song before deciding if to download or not
okay
There was a problem between Cloudflare and Suno, the requests were timing out.
Feels ok now.
But it's just streamin audio.
One thing, you might need to give permission for audio maybe?
Or volume might be zero for the browser?
So sorry for my attitude earlier, I am creating a birthday song for someone, and still need to create a video for it this night, the birthday is tomorrow
I guess i will just forget about it
You can also download the generated song, and listen like that.
hmmm... let me see if i can download it
You might need to do this:
System -> Sound -> Volume Mixer -> choose your playback device then hit "Reset"
Probably there is a problem with the driver.
Question: does [verse] [groovy] [intense] [melodic] on the same line change something or [verse] alone is the corecte way...
For 3.0 it might not work. 3.5 most probably it will override itself.
Try [verse: groovy, intense, melodic] instead.
Too many tags bleed into music as lyrics.
Great idea will try that out
Hello, folks! Do You know if I am allowed to sublicense music that I create with my paid subscription to others for their use in their own projects?
For paid subs the copyright is yours, so you can do whatever you want with your generations.
It's kind of weird the auto moderation generates the song, but then won't let me generate a second copy. lol
https://suno.com/song/a3ab9b1c-6c45-417f-8c27-f694022c702f
After hundreds and hundreds of points spent, SUNO finally recognized the promts, albeit half-heartedly, but I was able to make a gender change with the initial melody.
melancholic, soft-piano, piano fill, dream space, ambient, space song. Listen and make your own with Suno.
Not really suited for technical-discussions, maybe share in #๐ฃโsuno-showcase?
Not sure what's being discussed aside from you generating a song that you're excited to share?
The other days we discussed how it is best to make a genre change within an already created song, I think it's something technical, right? Or in this section the promts are not discussed or how to achieve such a style of music?
Are you saying you've found a reliable way to change singer styles? Besides 'luck of the draw', I'm not sure how you're saying you achieved that with the lyrics/style prompts used, and the only technical workflow you've shared is that you spent hundreds of credits. Not trying to gatekeep, I'm just not sure how to recreate your success beyond "rolling" a bunch of different generations.
Exactly, of let's say, 40 attempts, 2 of them I managed to add a style or change it, but it is something poor, of course, note that I spent a lot of points to achieve it, so I don't know whether to call it luck or technique XD
[guitar solo] seems to "work" but i cannot get [bass solo] or [drum solo] never make any difference
Hi everyone - I've noticed today that all of my generations sounds really overdriven and distorted. Some of it thought I could chalk up to the genre [I'm trying an experimental/metal/electronic ... Thing] but then I had it so a sweet acoustic country thing just to see, and it was still all fuzzy and clippy, except for the quietest parts.
Is anyone else experiencing this?
Im trying to make a song sound more oriental, or for a better example, get brackgound instruments similar to master kogahs theme.
But no matter what i do it just sounds like basic rock/pop, even when I list chinese/japanese instruments
Any advice?
I've had some success getting regional sounds by specifying the region in the prompt, ie: whatever style 'with strong middle-Eastern elements' or something like that .. so you could say 'with traditional Japanese instruments' or something like that in the style box .. might work?
I am. I think it started after Suno 3.5 came out.
Is there any more word on the API? My developer hands are itching
Hopefully in the future we get an ability to tell the ai what we dont want in the song
If i want a song to be fast paced, 9 times out of 10 it becomes rock ๐คฃ
a "negative prompt" is something i've suggested before. it would be super useful.
Or even something as simple as "instruments i dont want"
I'm trying to make a grogarian chant. But the ai is always adding instruments. I'm adding no instruments in the prompt and the instructions in the songs. But nothing seem to work
I think thats like, the only real complaint i see against suno ai
Which is saying A LOT
Is the original output audio format of Suno the MP3 format? I.e. when I click "Download WAV", does it just convert the MP3 to WAV prior to sending, i.e. it will be the same quality?
No, the wav is slightly higher quality than the mp3. But the vast majority of audio problems are not caused be compression, so the difference is fairly small.
Is there something wrong here? I've went through about 1000 credits and every sample it just repeats the verse.
I've had the same issue, i dont know how many clips in one of my songs it would keep repeating verse 1
Usually the best thing to do is keep backing up the Extend From time, further and further, until you find a spot where the problem doesn't happen. If you keep retrying without changing the time, it could happen almost 100% of the time.
ahh ok ty. How do you end a song now? I used [end] then [mute] while (description of a picture) but that doesnt seem to work now
Well if you have paid plan, you can change picture after. If you just want to end quick then "end" in style, [end] in lyrics, and also try using V2 or V3, V3.5 is the one that really doesn't want to end.
Oh i didnt know i could change the picture like that... thank goodness... im sick of lute playing birds! xD
Subscribers don't get the 50 free credits per day... which does seem a little odd, it almost like we are penalized for being a subscriber lol ๐
I have been so frustratingly experiencing the exact same thing more so since v3.5 about 95% of my generations are so distorted they sound like trash. I've had some absolute massive mammoth monster jams that I have just deleted because of this distortion issue.
I've had earmurdering screeks happening in the middle of growls in my metal songs xD
usually happens neear the end of a song
I do a lot of metal songs too and experiencing this too but I have also changed genres to to test and its happening even with ballads or just a piano piece. Some songs are great till it gets to a build up or half way through the song then it becomes trash.
is there a cut feture on suno? to cut a song off sooner?
Not at the moment. You can try to extend a song with [outro] and [end] in the lyrics, and "end" in the style, and hope that the AI will listen. I've heard selecting the "v3" model when doing this can help.
Easier is to download the song and use a free audio editor like Audacity to fade-out the song, but that does mean you won't have the edited version on the website for others to hear etc.
damn ๐ฆ i use [End] but when i just rando generate and the first 1 is a banger XD i dont always have the kinks worked out
I agree, it's a disincentive to subscribing and it's why I've decided not to.
damn no matter what i do the bloody ai keeps playing hahaha, even after a while when its gotten silent it restarts, starting to shout blubber.
Someone did mention there might a possibility of this being reversed in that when you have used all your subscriber credits you would then start getting the free 50 daily credits which would be much better as people do run out and have to wait till there renewal date. This is not guaranteed to happen just something one of the admins here told me as we were talking about it the other night. I look at it like this, right now this is all experimental some good changes happen along with some bad ones. I guess Suno need to see what works and what doesn't based on feedback and adjust accordingly. I wouldn't let it put you off subscribing my friend even at if just at the pro level as you could be missing out on some absolute monster jams.
I just know I'll use the credits up very fast and then have a long time to wait before getting anymore. If they update the policy then I will probably subscribe.
I do blow through a lot of credits trying to get things to work. I am having a lot of issues with v3.5 in the quality and seem to be throwing away song after song because of this insane distortion added ruining the creation, BUT I know I am not the only one and am sure it will get sorted. Patience is key even though I don't have any lol. As a subscriber you can also purchase top ups or you can revert back to a FREE user again.
how i get suno bot
I know it doesn't feel like it, but when Suno does things like increase song length from 2:00/1:00 -> 4:00/2:00 when they went from v3 -> v3.5, that is indirectly giving every plan more credits, because the same credits can go further than before the update.
that is so very true...
I suppose then that each month I could subscribe then revert to free when i run out and keep repeating that, but that seems like a hassle
You have to work with any kinks my friend.. its literally a click of a few buttons. If it gets you creating again it has to be worth it. ๐
nasil sarki yapicam yardim edin
thx
i dont understand this
nerde sarki yazabilirim
chrip komutunu hangi kanala yazmaliyim
Chirp is at Suno now
https://www.suno.com/create
calismiyor
https://discord.com/channels/1069381916492562582/1154780629079162963
Do you have credits?
Thanks for letting me know, man. Yeah I've gotten 2 straight up bangers yesterday, but I'm gonna have to trash them and start over because of the clipping. It feels like it just started doing that when 3.5 was released to the masses, tho. It wasn't doing it for me, for example, last week.
I spent all day yesterday and it was producing banger after banger BUT every one sounded trash and I had to delete them. Such a shame as there was some really good stuff. I wasted over 1500 credits doing this over and over again trying anything just to get a good clean outcome. I gave up in the end. For me too its certainly been more so since v3.5, like you I like Metal and Shredding but its very difficult to get some clean creations without sounding like every musical element is at the max volume running super hot. I really hope this does get sorted,
we are not the only ones. Quality seems to be one of the biggest issues right now, BUT I am a realist at the end of the day, what we have now is still super amazing and in time all the creases should be ironed out, at least I hope so. No point having a 4 minute song if for those 4 minutes it sounds distorted and terrible.
Can I share the songs we made on platforms such as Spotify and Instagram?
@wet minnow
If you use a free SUNO account to make songs the ownership belongs to suno. In this case you can share the song (Youtube, Socials, ...), but not monetise it and you have to mention suno โMade with Suno.comโ.
If you have a paid subscription all songs made DURING this paid period fully belong to you. You can monetise them in any application you want. The songs made with a paid account stay your songs even after the paid period ends. โ๏ธ
If I download spotify now, where should I write suno?
hi guys: small technical question.... the trashbin doesn't automatically empty itself after a period of time, does it?
arent they automatically monetized on spotify? i dont know how this exactly works on all external platforms
no, everything stays there until you delete it again from there
idk
but
blablabla ft,suno
perfect... that's for letting me know
thanks for ..
How can something like this happen?
@wet minnow
if there is a chance for you to earn money from it... its a monetized application...
Did you read my last post?
yes. but i dont really understand it and still think about it ๐
Are you asking how to get your content onto Spotify?
yeah
You don't do that directly through Spotify, you need a distributor.
How
Can't I upload this myself?
Which you can't do if you don't own the song. Suno owns it if you made it from a free account.
do you mean in the audio itself? or as the title? i got the paid sub to be sure i can release my stuff just everywhere i want without problems... thats 8(10) bucks well invested...
Did you read my previous post? ๐ (joking because you used this exact line on Mr Tree)
You can't upload directly to Spotify or Apple Music. You can upload to Youtube
ohh somebody that actually knows stuff joined the conversation ๐ thanks for chiming in
1 dollar is 30 TL, 10 dollars is 300 TL and my father gives me 50 TL per day. heheheh
Do you mind if I upload it?
ahahahahh
You don't need to ask permission from me or anyone. You need to give credit to Suno because they own the free songs you generate.
Can you give an example for the title of the song that will be on YouTube?
@wet minnow come to valorant
on youtube it is enough to put "made with suno.com" in the description of the video.
I think there might be some misunderstanding.
You choose the name and content you upload to upload ^^^ what Mr Tree said
oh my god this is genius
@wet minnow are you coming
i have no time to play games at the moment. i have to help people using suno ๐ but thanks for asking
ahahhahahh
You can't imagine our need for a fifth player
im crying
But thank you for your time.
let me know if this makes sense or not. completely understand if it sounds like im talking baloney
is it possible to copy an mp3 link from one of the suno tracks?
instead of actually downloading it
one sec. looking it up
ty! reason im doing this is because of a lua script project im working on that only accepts mp3 links. cant use actual files
I guess downloading them and putting them on a server somewhere is cheating...?
do this but with mp3 at the end instead of mp4 #๐ฌโgeneral-chat message
Can we use v3.5 but I want it to be 2-3 mins
bet, ty so much
has Suno just gone down? All of a sudden i now have a cloudfare timeout error as I was typing in lyrics?
?
if emptying the browsercache in the settings doesnt fix the problem it might be the issue some users are experiencing right now...
This used to be the default behaviour on the mobile web app, but it's been "fixed" now.
oh word? thats interesting
Anyone?
for me it depends on the amount of lyrics in the textbox. and if i put a [end] in the ylrics box. 3.5 is known to make long songs. 3 might work better inthis case
Ok
ahh ok thanks buddy its been fine all day so far it literally just went.... I am just trying my work machine in the USA and that's the same. Two completely different GEO locations.
It's just a maximum of 4 minutes, but you can still attempt to end the song with extension points & metatags. https://www.suno.wiki/faq/making-music/how-do-i-end-the-song/
hate when you get what seems like a nice ending... and then it picks back up. Tricky v3.5 ๐
Trying to get the download link from an mp3 I downloaded, I got this 404 page -- I'm not super familiar with browser workarounds to know if there's another way to approach it
https://suno.com/4840425b-b760-47bf-bd72-6dbba56cd56b
I understand what you're saying. If that's the case, I don't want to do it. Ideally, I'd want it to be separate from the play count system on Suno to ensure it doesn't 'cheat' the system. This is just for a radio script in my FiveM server that people would actually have to manually adjust themselves. No bots or anything
the tip above is also super good... use a extend with the style "end, short" and [end] in the lyrics box while switching to v3... that should make like 9 second parts in the best case if you extend on silence
Yup. I've force ended sooooo many with v3
I wonder if v2 has any application in trying to end a song, or if it just makes it sound like you're listening to a euro-dance version of your song from outside, muffled through the back doors of a crowded nightclub.
i found a working solution some time ago for a sharable mp3 / mp4 link #๐ฌโgeneral-chat message
I got that same issue at first-- take the song id from the "Share Song" link instead of from the downloaded mp3
I love that friggen tree.
@wet minnow My eyes might be deceiving me, is your profile picture the first boss from Kirby's Dreamland?
its a ai interpretation of the pokemon sudowoodo ๐
ik this is a lazy take, but I would love for that to be a built in click feature somehow. But for now, its a simple 3 step process
fair my bad. didnt think about it
๐
Not a lazy take, I just think it's a matter of trying to balance available options with a clean UI. I'm sure if they added a 4th download option for .mp3 link or .mp3, it would just be another thing to confuse the less savvy users.
ah btw @wet minnow your [Ending] tip gave me the first "credits refunded" in V3.5
Needs more testing though if it actually works or was just luck. But, from the few tests it did it did actually end the song in most cases.
i think the mobile version did that instead of making a proper sharable embed... and we had a ton of people asking how to make a embed link. yes. they get confused ๐
https://www.suno.wiki/faq/making-music/how-do-i-end-the-song/ Pretty well documented, but still requires luck/understanding ๐
#๐โsuno-school message Anyone sufficiently bothered enough by it has us ๐
I know... this was 3.5 specific as it tends to be a bit stubborn and Mr Tree here tried [Ending] instead of [End] which seems to work better in V3.5
i experimented with the specific metatag [ending] and it gave good results even in 3.5 that is hard to end. is the wiki updated yet to include this? if not... then i might have to add this by myself ๐ .. i looked and i will add it ๐
Oooh, my misunderstanding, Bites! I've never gotten specific enough to test [end] vs [ending], but that's on my radar now... ๐
All good. I would test it more myself but i am running low on credits ๐
wiki page on how to end songs updated... thanks for the reminder.
dumb question, but architecturally, is Suno an end-to-end multi-modal model, or is it just traditional multi-modal?
I guess for clarity, I'm wondering if text to vocal synthesis and instrumentation happens in the same pipeline
I'm trying to explain end-to-end multi-modality and based off of the quality of the output, this may be a good example to point to
You could look at the source code of bark on GitHub it is basically the previous interation. But there is no guarantee both work in the same way. And the only ones who can answer without a doubt are the devs and I am not sure they will go that in-depth.
I had taken a quick skim and then shelved out of laziness. I'll go back and give it a closer look, thanks for taking the time to reply, @glacial scarab
Hello Guys, Im kinda new in this ai genre if i may ask is there any way that i can train the pre-trained bark TTS model by using my own dataset? Can i get a reference link? Ty
Totally possible, if you know how to use transformers and/or pytorch and have your own dataset. Idk if there is a tutorial out there or not, though.
Looks like you would need a big GPU with 12gb+ memory
I have my dataset but i've been trying to find how to train the model but i seem to can't find it.
I have been trying to create song in Suno where I can change the song tempo or style or genre within the song. Like with the next Verse or Chorus. I've had no real success at that. Should I give up on trying to make changes mid-song because its not possible?
how do i delete my acc? i wanna go on a different one instead of my discord????
!
!!!!!!!!
I personally haven't tried switching styles and whatnot yet with metatags personally but did you try to see if Extend helps you have the style?
(E.g. extend at that midpoint, change the style prompt, and see if it induces a change)
For tempo, avoid using # BPM as it usually does not work. Instead, use up-/mid-/down-tempo and/or the Italian tempo keywords like Presto for 168~200 BPM (see #๐โsuno-school message )
If you log into another account, does that other account work? (Then log back in and confirm your main account is still not working)
If that is the case, you will want to contact support@suno.com as a few others reported a similar issue with them about that.
Other checks you can perform for reporting:
- Clear cache (Shift + F5 does a cache-bypass refresh in most browsers) or go into incognito mode
- Trying with another browser
- Ensure you are logged in an account
Probably a few more Other checks I am forgetting to list
thank you, simple ctrl+f5 worked
If you want to permanently delete your account, contact support@suno.com since I did not see a delete option on the site yet.
If you want to log out of the current account, click your small avatar at the top right > sign out (or delete your cache/site cookie)
thanks
Does anyone know what that whirring is that keeps appearing in music?
Is it like ai images with fingers?
Going to enjoy checking out the new upload feature. I've just tried clicking on Upload Audio and got no responce. I supose it could be getting plenty of clicks at the moment. But it got me thinking about the Browser I use. At the moment I'm on Chrome, but that has a few problems lately. So, is there a Browser that you would recommend?
wait i just saw what i kind of asked for is litterally already a feature now lmfao
guess its not that impossible to categorize input that isnt words afterall
PLEASE!! let us choose the duration, sometimes I don't want songs that are so long that they take time to generate. ๐ฅ
Oh man oh man oh man oh man. I'm going to upload my soul directly into Suno.
Hi! I just started using Suno today and I was wondering if there's anyway to add your own voice or tracks of you singing for the AI to learn it and create the songs with your own voice is that possible?
Hi! I sometimes find the ideal voice, tempo, melody, and other elements for a song. I want to use that form in other song lyrics as well. Is there a way to do that?
You joined at the best possible time. #๐ขโannouncements message
Hey there! Welcome to Suno ๐ Youโre right in time! Suno just released this option today. Please check out #๐ขโannouncements . If you have any questions feel free to ask
Hi all! I'm having an issue with my Suno songs, the last 1/3 of the audio seems to be distorted or clipped, is this happening to anybody else?
HOLY TIMING BATMAN
Certainly, have you seen the wiki page dedicated to Suno? Lots of great things there that could help you better learn how to use Suno https://www.suno.wiki/
Lots of room to grow and things to come. If youโd like to, feel free to share what youโd like to see come next in #๐กโsuggestions-web
This can happen for sure. Lots of room to grow. Try to make sure that you use the extend function before it gets to a part of the song that might be displaying any distortion.
im still a bit confused how vocals land in my tunes when i select instrumental only
and those vocals usually have way better toplines than when you let it generate vocals
topline meaning melody
If youโd like to share the prompt you are using, maybe we can figure it out
[HELP] absolutely stupid questions - I just upload an audio of my own handpan playing, suno just plays is back to me! How do I use the new audio input feature to actually generate a piece of music?
Extend from it
Thanks for reaching out! As @charred merlin Brought up, you will need to extend from it and will need to include the tags that support what you are trying to make in the style of music field.
Yeah the uploaded audio is essentially "part 1" of the song.
hmm, I'm not seeing the extend option for those clips
Only available after they've finished processing
I was also wondering this, and checked the wiki, but couldn't find anything specific about reusing something from the song. Is like for example a seed we can see and re-apply to get the exact same voice, or same melody, etc?
I mean my point exactly is that I would like to do a small adjustment in the lyric, but keep everything else the same
Got no extend option in dropdowns for uploaded files
The result is SICK. Love it โค๏ธ . Thank you!!!
Technically you โcouldโ use the new feature to accomplish that kind of, at least to be able to build a similar foundation. At this time there are no seeds though
Top song hasn't finished processing yet
how long does it take to process? I have one uploaded ~8 mins ago that doesn't have the Extend button
Refresh your browser? I know the server might be a bit slower because of the heavy load, but it also roughly gets processed in order, so if you have a lot of songs processing you might need to wait,
that UI looks unfamiliar - I've only seen "extend" as a menu option
Yeah, there is menu options for smaller screens (I use Macbook Air), but not for uploaded clips. Anyway, button does work
oh, got it - need to make the window wider
honestly are the admins actually sure that this app holds ALL rights to its training data? Cause im kinda feeling like that might not be the case here.
I think this is a mild bug, then
The picture is when the screen is large enough, e.g. desktop mode. The Extend in the menu overflow is when your screen is squished like on mobile
it SHOULD presumably also be there in the menu, like for regular (generated) songs
Yeah this looks like a bug
to me some outputs sound like copyrighted material is slipping through the cracks
Itโs a menu option on mobile browsers, at least
Obviously, and theyโll be copyright struck, theyโll get Suno in trouble, and it might ruin things for everyone, but while itโs here, make responsible use of it and have fun! I misunderstood. What existing copywritten songs do you feel are being spat out?
@daring lava
hey, these were the tags I had
mariachi, male singer. accordion, trumpet, backing vocals, mexican music, upbeat
Is the chirp bot working properly? I'm trying to use it on my own server and it isn't responding at all.
Chirp bot is retired, use the website https://suno.com
Some helpful starting links: https://suno.wiki
https://youtu.be/hsecBccWETM
(Check the video description for the links to other languages if needed)
Retired like 3+ months ago
nvm you corrected yourself
This looks like a bug to do with the size of the browser window. After reading the above messages, I've reported it in the bug channel here: #๐ชฒโbugs-web message
Try male vocal instead ๐
I misinterpreted. Once I got a rendition of Pachelbels Canon in D, but that's not exactly a Suno-specific issue if you listen to the radio ๐
Idk when it spits out a random vocal thats associated with a certain style and i didnt even ask for a vocal and the topline is wayyyyy too good i get the feeling it might be VERY inspired
to be fair pachelbels canon isnt copyrighted anymore right?
I will do, thanks
Let me know how it goes
i have no specific example yet, but i am still curious what data this is even trained on to replicate modern music so well
Release a beat-for-beat instrumental cover of Vitamin C's "Graduation" and try to get away with saying it's Pachelbel's canon.
Suno states on the ToS that due to the nature of generative music, and the fact that it's based on commonly available tags and random seeds, users might generate content that sounds similar, but that's not the same as spitting out copyrighted material.
There's open-sourced generative text-to-music programs if you wanted to look into different ways models can be created, but Suno hasn't shared details regarding their training methods.
Is it just me right now, I am trying to make a song public, but if I refresh the page it is still on link only? I tried doing it from both my phone and PC and same thing is happening. Can someone check it or is it just some issue because they just dropped an update
the extend option isn't appearing for the uploaded clips, I have pro plan for that matter (It does appear in generated songs)
still getting female vocals here and there
see the discussion from about 10 mins ago above!
Looks like there's a slight bug, where the option isn't showing up in the menu, but it DOES show up as a button if you make your window wider
ooo Im dumb lol
yeah I made the screen wider but was still looking at the drop menu and didn't see the new button 
thanks โค๏ธ
Does it show this if you put your mouse over it?
It lets me actually press the switch
And it appears as if public
Its an instrumental
What content
xD
Ah, sorry, assumed it was to do with audio upload ๐
I wonder if just the phrase "mariachi" is doing it. good grief, another 2 both had female vocals haha
I just tried with one of mine and it worked so I'm not sure what's up with that
If you have an audio which has problems with copyrights, you can not make your song public
The funny part is that one of the previous versions of it does publish fine, but this ones isnt' could it be the title? Is "DEFCON 1" illegal or something
It's if it detects vocals. You simply agree to not upload material you don't have complete ownership over (such as copyrighted material, someone else's public work, someone else's voice, etc), and you face the consequences if you do (including, but not limited to, discontinuation of the Suno service, and legal ramifications)
That sounds like a bug with that particular song tbh. Maybe Suno thinks it's not finished generating or something.
after removing it and just leaving "mexican music" that gave 2 male vocal songs. lol goodness
Not sure whatโs happening, itโs working for me pretty consistently ๐ค
Glad you got it worked out!
What is it?
content id i guess
Are you trying to upload someone else's Suno song?
no its a vivaldi song
i dont think vivaldi are on suno
Ah, they DO have content detection
Even if the copyright is expired, you need to OWN the content -- from above, " You agree to not upload material you don't have complete ownership over (such as copyrighted material,*** someone else's public work***, someone else's voice, etc), and you face the consequences if you do (including, but not limited to, discontinuation of the Suno service, and legal ramifications)"
Vivaldi's compositions have expired copyright but most recordings do not. they cant really know whether the recording itself is copyrighted
Not entirely a matter of copyright, it's whether you own the content.
unlucky
copyright also applies to a recording
the content ID matches recordings not sheet music
there's also no way for Suno to know if you've properly licensed it, so they just block everything
I'd imagine for LEGITIMATE purposes, you can use the special tier to sign up for an account rather than the basic free/pro/premiere; there might be some kind of wiggle room or considerations made
i assume they'll partner with Beatport and other similar licensing sites eventually
Does the audio upload feature work only with v3.5?
No it works independently and can be extended with either
So all my work that is uploaded to soundcloud can not be uploaded to suno
Can I prove my soundcloud page to suno devs to get an exception?
I have lots of WIP on my soundcloud that I want to see what suno can do
Since the feature is so new, I'd say this qualifies as a bug, or at least it'll have more visibility in regards to the team. They might be able to clarify if this is expected behaviour or an unintended side effect
Sure thing, very doable
3 old unfinished projects not on soundcloud, they work