📚┃suno-school

[glacial palm]

now that i think about it, im glad most of my stuff isn't public LMAO

[frosty cypress]

born 84 so yeah, I know, had my ear pierced in 92

[glacial palm]

the left one, right?

[frosty cypress]

lol yeah.

[glacial palm]

and im right there witcha-- 87 here

see tell me thats not an objectively funny topic for a song

[frosty cypress]

goth kid, long black hair down to my ass, wore all black, listened to nirvana, alice in chains, pantera, metallica, acdc, faith no more, etc..

[glacial palm]

like the whole attempt at hitting him was just a cover for him to able to pierce his ear without looking like he likes dudes, but that im on to his rainbow flag flyin ass--- not that there's anything wrong with it, if he is, that's hella cool, but like, own your personality, son-- say it witcha chest! lmao i think that was actually IN the lyrics

[frosty cypress]

yeah, it would be funny

[glacial palm]

bro you and i were very similar

i dig this tune bro good job

i love that high pitched guitar-- standard tuning giving no effs bro i see you. that's probably drop d at the lowest

i prompt all mine in like drop C or A standard, like an 8-string

[frosty cypress]

mines always all over the place, depending on the mood of the song

[glacial palm]

oh i straight tell mine in the lyrics with [] proimpts

[frosty cypress]

i do a lot of times as well

[glacial palm]

like [All Female Vocals, Girl Screamer] then the next line [Post-Melo Technical Brutal Deathcore, Complex Chord Progressions, No Repeats] then [Staccato, Syncopated, Down-Tuned Guitars] and then i'll write the first section, etc

[frosty cypress]

this here is a sound i got through a glitch (originally had female vocals too), did a cover and got a version i could use, but been trying to figure out how to get this style on purpose https://suno.com/song/22a65572-8dac-4757-a9cf-3da7223246cf won't let the video embed because I don't want to get banned

i just loved the raw dirty sound of it.

[glacial palm]

lmao bro your custom pics are A1-- I work in generative AI, so I can tell when I see a model someone's trained on themselves 👍 good job, great continuity between pics. you on auto1111 forge or comfy?

[frosty cypress]

moved from auto1111 to forge for flux

[glacial palm]

feeling myself for cdalling that

almost asked flux

[frosty cypress]

but yet, trained embeddings and all that

didn't use flux for the generations that had me in them because still cant train right on flux yet.

so sdxl for those

but yeah, almost all these I generated myself https://suno.com/@3drcomics

[glacial palm]

bro i figured out my own training script for flux loras by bastardizing that AI toolkit one right when it came out and rewrote it for colab, it works well, if you ever need resources, i have TONS

[frosty cypress]

i do everything local, and have a hard enough time getting flux on the 4090's, let alone trying to train on them

[glacial palm]

flux is cool because you can do stuff like this

oh dude, no, you train with google colab

and you're not using the right optimizations if you cant get it with a 4090

i have a 3090 and i run it all day long

[frosty cypress]

yeah, i did the cover of "the note" with flux, so i could have things written in it.

its not all of them, only a few i struggle with.

[glacial palm]

also now they have AIO flux checkpoints

with the VAE and text encoder and clip state dict baked in

like 4 steps per image. boom.

[frosty cypress]

yeah, the ones that don't have vae, te, and cs are the problem ones, 18gb vram just to load the stupid thing

[glacial palm]

dude your songs are so 95-03 i feel it in my bones my guy

[frosty cypress]

well, yeah, that was the best times of my life, lol

[glacial palm]

nah dude, the flux-dev.1 checkpoint with the t5, clip, and ae.safetensors? you just gotta put your setting right. if you on forge

[frosty cypress]

its all ment to be listened to in a specific order, hints the playlists

[glacial palm]

it should be on CPU and Queue

and one of the f8 t5s

[frosty cypress]

i don't spend enough time doing ai images to really bother with it, my 4 4090's are for a completely different reason

[glacial palm]

and your scheduler and denoising sampler are paramount- if those are wrong you're fked

mining?

[frosty cypress]

3d rendering

[glacial palm]

ahhh

JESUS bro

wtf are you rendering hahahaha

thats so much juice

[frosty cypress]

lol yeah, a lot of juice, I render for visual novels

[glacial palm]

are you like competitive rendering or something, which probably isnt a thing cuz i just made it up, but now that i think about it it probably is, but yeah...

[frosty cypress]

those "ava" songs are for the visual novel too

[glacial palm]

whoa wtf is a visual novel

HAHA

[frosty cypress]

image based video game, each update i put out has around 5000 renders, at 30 minutes per render..

[glacial palm]

i was just listening to one like "yo call carson daily i got a new jam for TRL" 😂

[frosty cypress]

lol

[glacial palm]

its so evanescense it hurts

[frosty cypress]

lol yeah, i wanted a late 90's early 2000 grungeish feel with her songs

[glacial palm]

good shit man thats cool, i dig it-- even ur rap song is good, like, from a rhyme scheme standpoint. like you can tell you understand song structure and shit good job

[frosty cypress]

this is what I do while waiting on renders lol

[glacial palm]

less constructive ways you could spend it, no doubt

[frosty cypress]

also became a great therapy for me, life was hell the past 5 years or so. i feel better now after completing those 2 "albums"

[glacial palm]

bro i heard that

i got divorced, kicked a nasty opiate addiction that landed me in rehab, had two street racing assholes clip me and spin me into a telephone pole going 50, so disability, and then i started dating, boom, pandemic hit, and then my company i worked for folded all in the same 12 month span-- if it wasn't for writing music i'd have gone crazy. if i had this it would have made it so much easier i can only imagine. laid up feeling sorry for yourself will mess you up if you dont got some sort of support group

probably what led to a lot of the suno stuff i have today tbh, hah

[frosty cypress]

yeah, i feel yeah, was cheated on, twice(that I know of), then divorced, mom diagnosed with cancer right after, died a year after, my dad died 9 months after that, my boss at the time made some stupid mistakes that caused me to either take a pay cut or leave, so i left, had some bad shit happen with a few of the woman I dated, and now im just kind of here...

was in therapy for 2 years, no help at all.

[glacial palm]

it make it really easy to get those thoughts out. any time i've started writing feeling sorry for myself or negative, i listen to the song back, and i laugh so hard like "f'n' listen to yourself, bro-- be a friggin man and stop the theatrics, listen to these lyrics you goon brush yourself off and pretend like you have some gatdam answers; do some man s**t" and man im telling you it's helped every single time. just by holding up a mirror to myself like "yeh you just wrote this and you definitely should be ashamed of it you big fat lady" lmaoooo

dude therapy is bullshit for some people-- some people it's fine, but... you know what works for them dont work for everyone. i feel that.

i was in intensive therapy when i did my rehab stint-- 8 hours a day of one on one and group therapy, and i gotta say, i don't buy it, my dude.

sigh

[frosty cypress]

my last really bad depressed moment is when I wrote "the truth" because my brain decided it wanted to uncover some old burried memories, that one hit hard, but after I got that out, everything kind of changed. it felt.. good

[golden venture]

My first Suno contest entry! After 1000+ total song generations over the past few months this challenge was just calling my name!

https://suno.com/song/72b1e438-fedb-4ae0-94d8-fc3d9fa1fa41

& my insta reel is doing well! Super hopeful for the contest! IG: @golden venture

Had a hectic month. little brother totaled my car & Moneys tight, looking to bounce back with a 1st/2nd place to get that official remix & to get my finances back on track. Sending my best energy’s 🌟 let’s goooo!!

[glacial palm]

they're like "it only works if you work it" and im like "yeah, no, that doesn't even make sense what are you talking about" then they kept trying to assign blame to other people in my life, or my past, or parents and im like "alll of those things are just fine, why cant it be just that im maybe a bit self centered, have poor impulse control, crave instant gratification, and just like being high and so that leads to more poor decisions because in that area of my life im weak and need to work on strengthening it? can it just be that?" and the answer to that is apparently nooooo

[frosty cypress]

after that is when I wrote the dark horse, then the sin, and now im doing a rap song, that's not all boo hoo me.

[glacial palm]

exactly

that part dude

thats what made me feel better

once i could see myself through a third perspective and was like "oh boo fkn hoo bro"

and was like "dam is that really what im like?" and it was totally reflected in my song writing. im just glad i didnt have suno in 2017-2019 lol

id hve some real embarrassing jams

and yeah dark horse is kinda like that but also its so heavy that like it fits the genre well and honestly could just be about anyone, and doesn't come off llike that

[frosty cypress]

therapist kept wanting to put me on medication, i refuse, last time I was on medication, i woke up 3 days later in the icu with severe self inflicted wounds

[glacial palm]

at least it didnt to me, anyways

oh bro take the meds

lemme tell you

no shame in the meds game bro

if your brain has trouble producing the chemicals that everyone else does? why tf not?

[frosty cypress]

ive been put on meds 3 times in my life, those 3 times things were 1000 times worse for me

[glacial palm]

i had an addiction specialist do my meds, we tried like 20 things until i found something that worked

[frosty cypress]

i never attempted it, while not on meds.

[glacial palm]

dude me too. one of the things he gave me legit turned me into a crazy homeless person-- i was like whoa that was like taking dextromethorphan or dmt or something but unable to control myself im never taking that again

but i kept trying stuff, and wound up with one that works well.

if you understand how that stuff is working, it may help?

[frosty cypress]

i couldn't survive the trying, my kids are my medicine now

[glacial palm]

like, for me, the issue was apparently when your dopamine, seratonin etc gets sprayed onto the receptors, the parts that release those chemicals suck back into them whatever doesn't get immediately absorbed by your brain-- so a re-uptake inhibitor makes it so your brain doesnt suck those chemicals back up into the glands and lets the receptors just soak in it so it has a chance to suck it all up, and then suddenly you can feel good feelings like normal people.

aint no shame in it, just your brain works different homie, we all different, and not just in looks, theres just as many little tiny differences in how our brains work as there is with how we externally look

some of aren't built to endurance run. some of us are. shrug

[frosty cypress]

i feel good feelings around my kids lol. i was diagnosed bipolar manic depressive, with mild associative personality disorder.

[glacial palm]

yeah if you can feed off ur kids like that thats good. my thing was my poor moods were affecting my relationship with my daughter, because i'd like get annoyed by her and thats not me at ALL dude ive been happy my whole life, and love her to pieces. its actually what sparked me talking to doc about it lowkey

ever since i found what worked, never had that feeling ever again

oh damn dude

thats a rap sheet and a half

i aint gunna even try to pretend i know anything about any of that, so you know you best bro,

[frosty cypress]

i haven't had any irritation around my kids since the divorce, even dealing with 2 deaths, i've been okay around them

[glacial palm]

but what i will say is i feel you on the "screw therapy" thing-- suno has been better therapy for me than anything else ive ever done

how old

[frosty cypress]

yeah, suno was like a god send to me

3 daughters, 9,13,16

[glacial palm]

lemme guess you were quite the ladies man in high school

only thing that explains that level of punishment my friend

😂

[frosty cypress]

at least i think they are all mine, the oldest one I know for sure, she is a mini female version of me

[glacial palm]

i got one 10 year old daughter. met her mom when she was two, so she doesn't remember anything before me. and she mimmics me so hard i cant not think of her as my flesh and blood.

[frosty cypress]

actually no, I was very reserved standoffish in middle school and high school, any punishment i may have is in "the truth"

[glacial palm]

jesus christ dude

that some heavy shit to say rofl

and, not for nothin, but a pretty bitchin line for a song "the only punishment i have is in the truth" that sh hard as f

[frosty cypress]

i had done way to much messed up shit as a kid that I over corrected when I hit puberty

[glacial palm]

ohhhhh dude i thought because you said "at least i think they're all mine" i thought you were talking about the truth of who's they were, and i was like whoaaaaaaa bro-- but you were talking about yourself ahahahaha ok ok ok my bad

[frosty cypress]

oh no. i do question if all 3 of them are mine

[glacial palm]

ok mr. "ive got a mini-me 16 year old"

gtfo with that garbage

[frosty cypress]

she is the only one, the 13 year old looks nothing like me, like at all

[glacial palm]

besides, lemme tell you something, it aint who made the kid who's their dad, it's who raised em. thats their dad.

[frosty cypress]

and the youngest I am pretty sure isn't, she came out jaundice and me and her mother have the same blood type.

oh I know, it won't change anything

[glacial palm]

you ever let any of those girls go live with another guy for extended periods of time ? no? then they're yours, homie.

[frosty cypress]

i will always be their dad

[glacial palm]

ok good

cuz like i said i got mine at 2, and i'd die for that kid a dozen times before i'd let someone fk her childhood up

[frosty cypress]

oh exactly, nothing will ever change how I feel and care for them, how I feel about their mom on the other hand.

[glacial palm]

hey man, good talking to you, i gotta get some work done. what a random chat to have with someone in here, good shit man, i dont talk to people often, you're a good dude, keep doin you. ill see ya around. keep writin bangers 🤘🏻🤘🏻🤘🏻

[frosty cypress]

lol alright, see ya, and yeah, random 🙂

[glacial palm]

and for real, shoot me a message if you're ever fluxtrated and need some resources-- like i said, that's what i do for a living, and if I don't know it off the dome, I can for sure point you to someone/resource that does.

be easy!

[frosty cypress]

Thanks, will do! take care!

[unique plaza]

@all anyone have idea about aisinging.ai model and how its work ?

[round bobcat]

Unfortunatly not, but I saw tutorials on Youtube about singing stuff. Didn't watch it.

[shrewd garden]

I am wondering, if there is a way to delete multiple songsat a time.

And additionally if there is a way to clear up my trash easily.

[round bobcat]

And I got an announcment of Suno 4.0 in my mail and wonder if the generated videos will have that poor content, as shown on X. No song text anymore. Only empty space with a tiny image...

[frosty cypress]

not sure about mobile, but on desktop click a song, hold shift and click another song further down, that allows you select more then one at a time... and for trash.. no idea, I have nearly 100 pages in trash because I would hoping it would empty itself...

[glad valley]

good morning

[frosty cypress]

morning? im still up

[glad valley]

a few hours of sleep after 54 hours didn't help much

but at least it was something

[frosty cypress]

so for anyone who was here when I started this, Here's where I am at on the rap song for a couple of days when I have tokens again, https://suno.com/song/41e2a000-d0ca-4758-9097-dd400cc580fe

[glad valley]

I'm gonna see if there's something to do with your song to get rid of the artificial soundscape surrounding your vocals today @frosty cypress

[frosty cypress]

thanks, if you can, just let me know what tools you used and I'll get what I need to do it.

[glad valley]

Uh that'll cost you a bit of money

[frosty cypress]

not really an issue

[glad valley]

I'm subscribed to Studio One 7 Pro+ for Perpetual License + Extra plugins, samples, Splice integration etc

[round bobcat]

Oh, are you addicted to Suno? I know the problem! It is so much fun to create songs.

[glad valley]

costs about same as buying it and you still get the license

you don't get the extra plugin packs included like in the full version but you get the sub, with samples, splices, extra plugins, tunecore full distribution, soundcloud access, presonus exchange access, cloud service access and whatever else weird stuff

[frosty cypress]

soo $200 for that license, not bad

[glad valley]

yeah or 180 per year for the sub (which grants you a perpetual license which is limited by the version its on when your sub runs out)

but everything you use from pro+ sub during that year of course is also perpetually licensed to you

[round bobcat]

Black Friday is always an option to get a good software bargin.

[glad valley]

I cheated

[frosty cypress]

is that the name of a new torrent site? lmao

[glad valley]

I trial'd Studio One 6, trial ran out. I used upgrade coupon

30% off to upgrade from Studio One 6 to 7

even though I never bought it

not like I hacked anything though, they the ones that activated it like that

[frosty cypress]

reminds me of when microsoft did that and everyone got free windows indefinitely

[glad valley]

https://www.presonus.com/products/studio-one-pro this is the one I have

[frosty cypress]

yeah i looked it up

[glad valley]

I should learn to write tabs

then I can use notion 6 to compose my DAW music

[frosty cypress]

i think fl studio can export tabs

with the guitar rig plugin

[glad valley]

you only need to drag them to the top bar in SO

and it becomes a tab structure

then you can click the tab structure and convert to midi

so audio track to tab to midi

[frosty cypress]

omg suno ! add the ability to build off midi, holy shit that would be awesome!

[glad valley]

a ton of "other" sites have this already

and you can just render the midi into an audio track and upload

[frosty cypress]

really? i only messed with one other site and it sucked compared to suno

[glad valley]

there's a few, I don't think they have vocals but there's a site dedicated to vocals which has super high quality too...but not about to start advertising others openly, I'm dumb but not that dumb

[frosty cypress]

lol

ill google it

didn't even think about that until you mentioned midi

gonna loose my 5 remaining credits

[noble jacinth]

That s interesting. Can you send me a private Note?

[unique stone]

Logically seems easy. Render with soundfont and treat as audio upload with cover would be my guess as to how they could handle it.

You can do those steps manually of course, but it'd be nice if it were automatic.

[frosty cypress]

can you persona off something you upload, and does it work good?

or even cover off an upload?

lol, specifics please? lol, can you persona and its just bad, or can you not persona?

[charred merlin]

You’re not able to make a persona from an upload 🙏

[frosty cypress]

ok thanks, not surprising, persona is new

[old light]

When I try to replace lyrics using the replace section, it doesn’t work well. It works perfectly with English lyrics, but for Japanese lyrics, even when I select the part I want to replace, the lyrics don’t appear. Even if I write new lyrics while that part is selected, the lyrics aren’t reflected.

[outer cosmos]

When V4 will b released

[neat bear]

Ok, about the Edge browser freezing suno.com site, when you wait a VEEEEERY long time, Edge starts to respond again. Very weird crash. All the other sites work so good. 😦

[jovial trail]

Might I just say to the devs - thank you. You guys are pumping out features and updates like the office has bedrooms, and a buffet, and big windows with a garden on the roof... like, it's so awesome. But I hope you guys are okay... haha, like as much as we appreciate it all. I hope they're not chaining you guys up. Again, thanks for the awesome features and improvements. Being a part of the community and being able to use this INSANE service is so special to me. You guys really don fu shi up with this one. Best AI music service since ever and that will ever be, as ya'll are first in the game. Props to you. Biiiig hug! 🖤🖤🖤

[jovial trail]

And that's on Chrome... with the webstore allowing you to make your browser so so pretty. Check out the Bonjour startup screen. Customising it is a bit fiddly... but it's epic... and Dark Reader, damn. And some awesome ChatGPT extensions. Just saying... Edge is wack bro.

[neat bear]

Yes, this suno is so outwordly system, that it like scares me. Ok, I can comprehend todays software instrumens, that sound very authentic. I can comprehend ai composing songs (1960's already). But I can not comprehend how in the hell suno gets everything together, good melody, arrangement, awesome vocals and mix? Whaaaaaat is happening here! 😮

[jovial trail]

Trust dude. They plugged into the future with this one.

[neat bear]

And how in the hell it can create these after it has "learned" just plain single form audio files?

Ok that is the business secret

[jovial trail]

It likely constructs using midi - but yeah the instuments aere MAD! But the vocals are what really fu me up. Hopefully we'll be able to select voices or train our own eventually. It seems like a natural future feature that'll come eventually. But as a producer myself... the way this completely revolutionizes my workflow is INSANE!

[neat bear]

Yes, I too compose and write songs, but the mixing has been weak spot of me. suno mixes well better than me. 😮

[open sleet]

Likely has been trained upon thousands of songs, metric shit ton of hours which is why gens, especially with the new features come out so well

[neat bear]

maybe it has been trained by midi songs?

[open sleet]

I don't think it's midi

[jovial trail]

I knooooow! I mostly download the stems and then just bust it all up in FL using RX11 and Ozone11... changes the game! The only thing that I do not like is how the songs drift. sure it makes it all more natural. But laying down every kick because it doesnt fall on the grid is torture for me! I always match the BPM and key obviously... but adding samples and so forth and especially enhancing the beat is tortuuuure. But I am sure I will figure out a propper workflow soon. But seriously - iZotope is so so useful for this.

[neat bear]

I just tested the stems function, I had artifacts of the singing so I did not like

[jovial trail]

I think it is though - that the bones of all compositions. So it's trained on midi, which allows it to eventually, with enough data construct chord progressions and melodies of its own, which provides the structure of a song, that is why it does it so quickly, as midi is very simple. The sound is what messes me up though. I don't know hoooow they match the instruments so well. Like how the heck does the AI know what sounds good. They out here creating an AI that can actually listen... which is something I've been waiting for in ChatGPT for so long. Being able to upload a track and have it actually be able to listen to the song... not just analyse the waveform you know. Would be so dope in getting feedback, suggestions and so forth. It just helps youknow... instead of mindlessly scrolling through thousands of presets or losing your soul in knob city.

[neat bear]

This is sad too, there isn't many years left for human music creators 😦

[jovial trail]

That's where RX11 comes in dude - literally just run the AI feature on the vocal track and it cleans it up so so beautifully. Like MAGIC! And on the instrument track it does the same. Then you can fiddle with stuff to make it sound how you'd like... boost the bass, add more fluff to the mids and spike those highs. You can even isolate sounds to an extent - so you can enhance certain aspects or instruments in the song so you can fool around like adding effects to the focals. From reverb to delay... all that. Just know... it is VERY heavy on your computer. Like I have a beast and I can't listen to playback with a full quee in RX11. Ozone11 is fine though.

[jovial trail]

Isn't it SO funne dude how they said 'artists are safe from AI, it will never be able to do art or music...' we are the first to get fu-ed... Hahaha

[neat bear]

Actually, I have seen it already, making art or music is much easier for AI than making those dirt, low paid jobs that are automated at very last. 😦

AI takes first those rewarding jobs leaving humans those bad jobs

[jovial trail]

Yeah it really does. But it all still needs a human touch you know. That touch will become less and less however. But not everyone will use AI. And there is still a lot of value to learn traditional art. It's fun you know. Good for the soul. So it will stay forever. Digital art though. Wrecked

[neat bear]

Maybe wh have to be optimist, human art is needed in the future too

😮

[jovial trail]

Yeah - I mean this helps with making music... Someone still needs to get up on stage and play it... or fake it... or use holograms... Haha

[glad valley]

RX11 is expensive as all hell though and if you're a commercial producer you can't use pirated

[neat bear]

So you just have to save many to buy it

😄

[glad valley]

the plugin is slightly cheaper than the stand-alone software

yeah I'll save for it sure...yeah

[jovial trail]

There is a cheap-ish monthly 'cancel anytime' subscription. Paying $19 for such amazing plugins, and a whole lot of the good ones... is completely worth it.

[unique plaza]

anyone give me suggestion like which model i can use for text to music ? need to be open source model

[glad valley]

No I'd rather learn how to repair everything without it before I use it

[neat bear]

Sry for offtopic, but I judge monthly subsciption based services, still I also have them! 😉

[jovial trail]

Just imagine Sune gives us a 'Download Midi' one day... bruuuuuh I would lose my mind

[glad valley]

Suno doesn't have midi tracks though that's not how it makes stuff

it doesn't even have instrumental models

[jovial trail]

Again though - literally all music has 'bones' its the rule of all composition - you can play anything on a piano right? ....so.... midi

[glad valley]

no...

that's not how generative ai works

[jovial trail]

My guy - that is how music works

[neat bear]

Have noticed, that suno songs have "human" time drifting, "swing", so they are not machine like super in-time presentations?

[glad valley]

You're applying music theory to programming and noise

[jovial trail]

It's an added feature... there are knobs in plugins to add that no? So it adds a persantage of swing to make it natural

[neat bear]

I think there must be midi behind suno

[glad valley]

there isn't....

[jovial trail]

For an AI that makes music bro???

[neat bear]

yes but I think, suno songs already have the swing

as and added sweetie

[glad valley]

you know what, if you want to act like an arrogant ass about it, I'm not gonna waste time on you

[neat bear]

If there was no midi behind suno, it must be some alien computer science, I think

😮

[jovial trail]

Oh look - a troll. You know I am a producer right? With a music theory background... if an AI makes music... then the rules apply. So maybe... educate yourself?

[neat bear]

Please try to speak politely

[unique stone]

Suno might have some kind of annotations in the training data indicating which notes are playing in the audio, or it could be frequency analysis of the source material, or it could just be that it's trained on so much audio that the algorithm sways towards producing "good" output.

Either way, it's very unlikely to use MIDI. Otherwise MIDI export would've been an incredibly easy option to implement.

[neat bear]

Very interesting. Then, how could it produce these sounds without using software instruments, thich are based on midi?

*that are

[jovial trail]

Certainly. Yet the speed at wich it constructs compositions implies that it writes midi then applies the instuments over all that... it's literally just the most probable 'souce'... Look at Unison - the AI Midi plugins for Bass, Melody, Drums... it's all there

[unique stone]

It gets trained on millions of songs and works out the patterns of sound, just like image generation does (there's no "midi" equivalent of images?) Then it's tagged by genre, instruments, other info, so when you prompt, it knows which patterns to generate from.

[jovial trail]

Probably using the frequency stuff. Based on the training... but damn... when you make a guitar track... those slides... the slaps... it blows my mind

[glad valley]

See this is why I don't want to deal with people that has your attitude. Oh you're a producer? Who cares? You could be an astrophysicist and still you wouldn't know what generative AI is unless you learn about it

[neat bear]

It has to midi

*to be

[jovial trail]

Go away

[glad valley]

......no cause midi wouldn't sound so natural

[neat bear]

today soft instruments are super real sounding

[jovial trail]

BRO? Midi is the BONES of all music... you do not understand dude...

[unique stone]

A question then... How would MIDI represent this?
https://suno.com/song/e67010d8-9887-4c50-bb65-e2b9af85669e

[glad valley]

oh my god....how old are you? 15? The way you're talking to me, good god you are full of yourself. I'm trying to explain how a model like Suno works but if you don't want to know. Then stop reading my text. Cause maybe others do

[dim pond]

This. As most of the other generative ai models of any kind, they get trained on loads of data which were prompted thoroughly, and the ai will basically put together stuff from its database based on what you prompt them..

[jovial trail]

Didn't I say MUSIC? And even with that... where does it place the sounds? It's literally the code for music if you want to apply it to computer stuff dude. If you can play a song on the piano... then you can write the midi for it. Not make monkey sounds? Are you okay?

[jovial trail]

Your opinion is invalid

[jovial trail]

Exactly... how it applies the sounds so well is just amazing.

[neat bear]

Shit, almost freaked out. Just the freaky "song" stopped, bird hit my window. Huh!

😮

And that was not a joke

[unique stone]

It's trained on audio, the models are algorithms made from that audio, and when you generate, it produces audio. If there was MIDI I see no reason why that's not offered to us. There's no stems either. None of this works like traditional music composition.

[neat bear]

Then it is freaky computer science. Like Nobel prize

[stone tundra]

Suno's terribad at separating stems. Use an external one for it. Ultimate Vocal Remover is really good at it, and is free, just as one option.

[glad valley]

Suno scrapes the web for millions of songs (including copyrighted material) and trains the model on it. The data becomes part of a large "noise" of no particular sounds at all and when somebody prompts for something it uses the mathematical algorithms it learned from the existing music to generate the songs. Unfortunately, Suno is a single model generative ai. Which means it has no separate instrument models, otherwise it would have stem tracks without having to do a poor stem separation. It doesn't have midi tracks and since it doesn't have stem tracks either, it will have a hard time converting them to midi tracks cleanly. This is why it's unlikely to have midi in the near future

[jovial trail]

There are stems... and again bro... IT IS MUSIC... midi applies. It's just not a feature in Suno. Please... like really... understand what I am saying. If it writes a melody... there is likely midi behind it. Look at Unison... seriously. 1 sec and you have the midi to apply ANY instrument to it. It's BASIC music theory... everything else comes after the midi. Don't the compositions need to be written first guys? Doesn't the notes need to exist before it becomes a piano or a syth? Uh... hyeah

[dim pond]

I've been using Fadr's stem separator, for me that works well enough, just to input another option if someone needs it

[jovial trail]

RX11 and Ozone11 like I said erlier

[glad valley]

There are AI sites that use different models for different instruments in a node based system but Suno doesn't

so Suno has no stem tracks cause it generates it all at once

this is also why there are often artifacts present

and why it has to run a stem separation (poorly) to give you vocals + instruments

[unique stone]

It is music but Suno makes music completely backwards to how you would expect. There are no stems. Suno would need to obtain tens of millions of stems to have the equivalent of their current data set which as far as I'm aware is fully mixed and mastered songs. And there would be no need for the kind of stem splitter they currently offer. Again why wouldn't they just provide the stems?

[jovial trail]

We have a suno developer in chat....

[glad valley]

if it had stem tracks it wouldn't need to do that

[neat bear]

I'm starting to get convinced, that suno is not based on midi

[glad valley]

you're telling a guide he's wrong dude

[dim pond]

Suno's AI model doesn't know midi itself or what chord progression it generates, sure, the music itself has melodies in them, which were taken from the training data and combined with other stuff that the prompt requires.

[glad valley]

Music is math and that is why the model can generate it, to simplify it to an extraordinary degree

[neat bear]

And then this, AI is trained by copyrighted music. But how about human? All musicians are also trained by copyrighted material. Why AI not? All music is based on earlier music,

[glad valley]

I'm not gonna get into a legal debate in a technical channel

[jovial trail]

Because the noise knows where to apply the notes and write the melodies and chord progressions? It makes all that appear from noise

[jovial trail]

Sure brodude

[neat bear]

Basically, EVERYTHING is noise. Human speech is also noise, all languages are also noise.

[jovial trail]

There's only so much that you can do with sound dude... thus ABCDEFG... it's a structure. How it's played, is random

[unique stone]

That's why I felt the animal noises clip I linked was relevant. It's made by a music generating AI but it's all sounds that are mostly without a particular pitch. There might be rhythm in there somewhere I guess? But no way that could be represented by MIDI.

[neat bear]

Like every popular music songs have some kind of structure, represented in the training data also, so those progressions tend to present in the final songs also.

[jovial trail]

Charlie Puth can tell you the key of a fart so lets calm down

[round bobcat]

No, is not random! Randomly would be disharmonic. There are harmonic rules based on the frequencies of tones.

[stone tundra]

I realize ozone and rx11 exist. they're not for everyone, and they are also VERY much not free, like the option I suggested.

I use other things for stem separation, myself, if and when i need it. However, I rarely use it, because I have the ability to make the stuff Suno puts out sound good without having to go through that effort.

[jovial trail]

Let's not confuse them more

[round bobcat]

If you know the rules, you can generate music. There was a pretty good music generator for the C64 in old times.

[jovial trail]

I get it. This is my job though as an artist. So I don't mind paying for the stuff I use

[glad valley]

there was also mario paint

[neat bear]

I think they have just invented a way to recognise harmonies and so on. Then the AI produces pleasent outcomes after training it with human traines for a time.

[unique stone]

This makes sense. All music has clusters of frequencies corresponding to the notes played. The AI would've learned what frequency combinations and patterns of sound fit together well, basically. Or more accurately, which ones are commonly used together.

[jovial trail]

Yup yup - Alicia Keys... her exact piano trained as a VST...

[glad valley]

frequencies, harmonics, subharmonics, cents and whatever else is in the end mathematics. A computer model isn't going to think like a human

[neat bear]

If there have been an army of ai trainers, just like text generative ais represented in the media

[glad valley]

it's going to think like a computer

[stone tundra]

Mine as well, but it's not for everyone.

[neat bear]

Some abroad guys there clicking thumbs up and downs so the results are getting better and better

[glad valley]

everyone does that though

[jovial trail]

You are totally missing the point fam. It's is music. You're stuck on the process of creation. No one cares how the AI muches it all togeter. point is... midi applies regardless

[jovial trail]

AKA midi

[glad valley]

every answer on chat ai's have thumbs up or down

Suno does too and Suno uses that as well

[neat bear]

I mean before it has been published, there can be human ai trainers to make it steady enough in results

[jovial trail]

By the way - it makes you get closer to a generation you like. If you use the thumbs... it kind of learns from the things you like and don't then gives better results.

[glad valley]

there are some that work on them, outsourced to third world countries. Then the mass results come from the users that rate it

[unique stone]

The thumbs up/down only tends to matter for beta features where devs want feedback (even then it doesn't have a direct effect afaik.) If you thumbs up/down things that aren't beta features it doesn't do anything other than just indicate for your own reference what you liked/disliked. It doesn't influence your future generations.

[neat bear]

Before publishing ai it could be too bad, and of course, after it has been published, the training continues by users clicking those thumbs

[jovial trail]

Think about Kontakt - and the INSANE library of instuments they have... training the model on that alone would make it sound incredible... much like it does

[jovial trail]

I'm on private. The cover song feature is also amazing. The second generation is always the best

[round bobcat]

Ihmo training soneone to become a musician and training an AI is the same - but the AI much much much faster

[glad valley]

So it's a useless popularity feature then? 😛

[neat bear]

BTW, what's the best way to make Cover songs melody match the uploaded song? Should it be like piano + singer

(uploaded)

[glad valley]

get the right tempo

that's a first step

[unique stone]

Look at trending on the home page 😆
I mean I used to use it for my own reference so I know what sounded good, because I can't be bothered to trash stuff.
But now I use it more for feedback to the devs when doing testing.

[round bobcat]

My grandfather - who was a musicain - always said: Most songs are only music scale up and down...

[jovial trail]

Yup - Google Unison... one second and you have a melody and a chord progression completely created by AI. Add a VST on that... MUSIC

[glad valley]

I'm well aware that trending is a popularity contest that just like high school popularity. Means nothing in the rest of the world

its funny though

[jovial trail]

Prompt it in the lyric section and at the front of the styles section like this [120bpm]

[glad valley]

so many invest so much time in it, it's a dopamine trap

[unique stone]

Suno does not understand tempo expressed this way.

It'll tend to give you something in the range of the genre you specify

and you can tell it fast or slow or other descriptive words, but numbers don't really mean anything to it.

[round bobcat]

First I used the remix feature to change single words in the song text, but recently Suno ignores my text changes. 😦

[glad valley]

did you go into song details and change it there too first?

[jovial trail]

Or it's an incredible way to create legit art. Get a song... re-record everything... and uh, yeah... you have music... where you own the masters

[jovial trail]

Works for me honey

[cloud shore]

👀 V4 soon? 🤯

[glad valley]

if song details match the new prompt on remix, it tends to work for me

[jovial trail]

It applies swing though

[unique stone]

I'd be curious to see some examples as it never has when I've tried it.

[round bobcat]

no. I thought that is only for the displayed text. I changed it only in the generation section on the left side. In one song I changed it on 2 places and it did one change, but not the other one

[jovial trail]

Set the tempo using [] then after... download the track... google Find BPM, and run it through that tool to check it. It's always super close for me. Or exactly right

But again... it applies swing. So it won't alighn on your grid in a DAW

[stone tundra]

Are you also using genre tags? Because certain genres tend within certain tempos and Suno follows those "rules of thumb"

[unique stone]

I usually find it is fairly close but ... yeah as above, very likely depends on the genre

[stone tundra]

But if I tell Suno to do Swamp Blues at 150BPM it will not do that lol 😉

[jovial trail]

Yes - I use specific genre tags. And sometimes stuff like specific instrument instructions ie. acoustic quitar, adlibs, slow male vocal, acoustic guitar, drum rolls...

[stone tundra]

Then it's very well possible it's just following the rule of thumb

[unique stone]

It usually takes a bit of work to align stuff from Suno in my DAW but I can usually get it eventually. Assuming it's a clip that doesn't decide to do a random tempo change and start accelerating.

[stone tundra]

Or at least just as likely /shrug

[stone tundra]

Those can be fun sometimes lol

[fallen plume]

Is therre a thread fo mixing and mastering

[unique stone]

not to edit! lol

[jovial trail]

Bruh - this kills me... aligning the drums... nah fam, it takes me so so long. That's where iZotope comes in... I isolate the drums, make them sound SPICY and then boom... happy camper

[stone tundra]

what kinda help are you looking for? Dunno if there's something specific

[round bobcat]

@unique stone I deleted it

[fallen plume]

Just a way to clean up my stems, I have audacity and audition and would love to learn

[jovial trail]

My suggestion would be iZotope... it has awesome ai features that you can then tweak to your heart's content

[stone tundra]

Ahh gotcha. In my experience, both of those are really difficult to get proper mastering done in

[round bobcat]

Can it be, that Suno song links have a time out after 24h?

[stone tundra]

Startin to sound like a salesbot rofl

[neat bear]

Maybe he is just izotope sale ai...

[jovial trail]

I'll stop suggesting payed services to the freeloaders. Arrrr

[glad valley]

lol

[fallen plume]

It would be but cant be impossible and I would prefer to have these skills than not

[fallen plume]

I'm gonna see if I can get a copy

[cloud shore]

The time and effort is not worth it. Download a free trial or free DAW and practice mixing/mastering technique's.

[glad valley]

it's not impossible, audition has the tools to get rid of artifacts, noise etc, it has a parametric EQ (dunno if it has dynamic or active)

[jovial trail]

Just get the $19 subscription.

[jovial trail]

Or let AI do it all for you

[stone tundra]

Not a freeloader myself. Shit, i've spent almost a grand this month alone on plugins and subs lol but there's also a thing of the toolbox having more than one tool, and a single tool isn't always the best option

[round bobcat]

I could improve the audio quality a little bit in Audacity with using Amplify and Loudness Normalisation and Normalisation and Noise Reduction in differend combinations depending on the music style. But there is much room for improvement.

[cloud shore]

You can do that also. I like to do it myself as a personal prefernce.

[jovial trail]

Maaaaaybe look at the features of RX11 and Ozone11......

[stone tundra]

Who said I don't use them already?

[jovial trail]

Same - I run the AI then go in DEEP

[stone tundra]

I'm just not a procyletite for them

[round bobcat]

Suno noise seems to be around 2,5k and 5,0k, when I use the equalizer in Audacity.

[glad valley]

There's no point in arguing with him, he seems dead set on trying to piss people off is all

[jovial trail]

Cool. Then watch some tutorials. They are nearly all you need. Will save you some cash. I've been down the PluginBoutique hole myself. not worth it

[glad valley]

there's also some above 16khz and around 150-200hz

[jovial trail]

Googles procyletite....

[glad valley]

if you're talking vocal artifacts

[jovial trail]

??? Sure bro

[stone tundra]

I never said I was inexperienced with it, either. I know what I'm doing... been at this for years, as a sound engineer for radio stations as well as producing music. Stop making assumptions and learn some freakin' humility or something ffs

[jovial trail]

Or am I providing professional advice with flair?

[fallen plume]

thank you guys for your help

[glad valley]

just ignore him

[jovial trail]

Chill loca

[stone tundra]

Suno isn't my sound engineer job. It's a fun hobby that I can use the tools I use for my REAL work on too.

[jovial trail]

He own this channel by the way

[round bobcat]

@fallingforever Yeah, I saw the one around 16k, too, but it destroyed too much of the voice decreasing that in Audacity.

[jovial trail]

Same - rerecord and own the masters....

[round bobcat]

I really was not able to get something usuable using stems seperatly - sigh. But atm. I do only have Audacity. Have to save some money for better software first.

[stone tundra]

Unfortunately sometimes those compression artifacts can really kill a track and sometimes it's just easier to make a cover of it

[glad valley]

yeah

[fallen plume]

one more other thing, does anyone use Cakewalk

[stone tundra]

I've tried it in the past but haven't given it a look for a long while

[cloud shore]

Studio 1 on my end. Sorry.

[glad valley]

SO7 ftw

[round bobcat]

Talking of compression - if you use the music in a video, you should use the wave file in your video editing software. Using mp3 and then let YT recode the video into their formats, does not make the audio quality better.

[cloud shore]

It's up to 7 now? 🤣 I'm Still on 3.

[glad valley]

cheap now too

and a good midi editor in 7, good new vsti's etc

[round bobcat]

If version 3 does all you need... Then I would not upgrade. I always buy updates at Blackfriday as often software costs only half the price.

[glad valley]

well you didnt know 7 existed so you dont know what it does ^^

[round bobcat]

Got a cheap PS subscription on BF. Saved me a lot of money since then.

[glad valley]

can always check if it has anything you need, if not, then yeah

[round bobcat]

of course

[neat bear]

You think there is noticiably noise in suno songs?

*audible

[jade mist]

Excited to try V4. 3.5 sounds very metallic.

[neat bear]

In my ear, these are ready for publish quality songs

[marsh adder]

Has anyone tried to troll people like that? 🤪

[neat bear]

And I'm not too old, to say ;P

(older makes higher hearing low)

[unique stone]

Yes, many times 😆
One day someone will go to do that and find it's actually there.

[stone tundra]

It's kinda dependent on the genre and the type of voice you're doing. Metal songs still have a lot of issues lol 😉

[cloud shore]

It's very hit or miss. With metal.

[stone tundra]

Yeah sometimes I can get one that's great out of the box and then others that i'm like... is this really gonna be worth the time or is just spending another 10 creds going to be better lol 😉

[cloud shore]

You mean.. '100's, or 1000's' going to be better xD

[stone tundra]

bwahahah

[cloud shore]

Do you find it's the lyrics that cause the issues, more often than not? A few changes here and there or switch positions of the verse's?

[stone tundra]

It does sometimes yeah. Other times, it's when all the instrumentations kick in and the vocals are trying to go, too. Just seems to overwhelm poor Becky (that's what I call Suno lol)

My Becky's a stress-bucket

[neat bear]

What is you opinion? Should you go through (audio) mastering process with the output or straight to publish? :

[stone tundra]

up to you, but i always master everything

Suno's output is always a bit on the low side, so at least doing normalization would be a good idea

that's at the very least

[cloud shore]

Low side?

[stone tundra]

volume

[cloud shore]

Really? I find it's loud.

[stone tundra]

And i always try to be compliant with things like Spotify for their needs. -14 LUFS for a start

[neat bear]

If your doing some kill'em all metal, then maybe, but when you are doing loving schlager, maybe not more volume

[stone tundra]

sometimes, but most of the time fo rme it's down by about 5db from what it really should be

[stone tundra]

i do those too 😉

I genre hop 40+ times a day rofl

[neat bear]

check, top is about -6 db, way too low volume 🙂

[cloud shore]

A few days ago we were experimenting with some prompts and I wonder with metal tending to be excessively loud, it drop's the db a little so vocals can better sit in.

[stone tundra]

That's possible, though i've not really noticed that with the waveforms most of the time.

Then again, I am legally blind sooooo... lol

I just do everything by ear

[cloud shore]

I think suno has a compression addiction. maybe that's why.

[stone tundra]

That, yeah

[charred merlin]

Did you try Stable Audio Open? You asked this a couple of times, so I’m wondering if you’ve tried it and it didn’t meet your needs.

[charred merlin]

Reliably? I’d love to see it.

[glad valley]

is that harmonai?

ora re they two different things

ah yeah it is harmonai, I remember when that could barely string together PS era midi notes

[sharp shard]

So... What's EXACTLY the queue length for free users?

I thought it was four songs, but I'm thinking it's actually five.

[charred merlin]

Harmonai is made by the same team, but this is their open source project -- https://stability.ai/stable-audio

[glad valley]

yeah

[charred merlin]

"a stability AI lab focused on open-source generative audio models" according to their github, but their website is BARREN compared to the stable-audio page, which lists it right beside stable audio 2

Seems like it's still delving into open-source projects?

[toxic ridge]

Does writing songs in a language other than English change the style of the song?

[charred merlin]

It absolutely can!

https://suno.com/song/65a2e520-e54a-4def-be67-d483a08a49eb
Only using the lyrics "I'm so angry right now", without any other style prompts, I bet you can guess what this song sounded like.

[warped furnace]

Good question. Mostly, I write English lyrics when I use Spanish it's for Latin genres like Merengue or Bolero. So I change the genre myself. Maybe I'll try a round with a blank Style box to find out. Have a song in two versions for that.

[toxic ridge]

I'm just wondering if my song I used Claude to adapt into French actually was helped by the French lyrics. I was trying to get more like Justice or something but I'm not sure if it did much

[charred merlin]

Incroyable! 🥖 🇫🇷

[warped furnace]

For sure. I think, the lyric mood has a big impact on the genre the AI picks.

[glad valley]

what about on huggingface?

[charred merlin]

https://www.harmonai.org/ I dunno -- the website is bare as hell, essentially pointing to a discord channel. I gleamed what little info I could from the github link in the top corner.

Huggingface isn't listed, but I wouldn't be surprised if HF hosts any models they've created -- that's sort of HF's jam -- recently passed 1,000,000 AI models being hosted 👀

[toxic ridge]

Also, the AI sucks at understanding stuff like French House and Jersey Club

[sharp shard]

Actually, nope! If the four songs are generating, and one finishes, it allows for two more! Meaning you CAN have a total of 5 songs generating at once! (I just found this out while generating)

[raven turtle]

Most commercial label tracks are mastered around -8 to -9, with clipping, but Spotify will "turn down the track" to -14LUfS. It's why clipping is important in ensuring that the perceived loudness is there on the waveform.
The -14 thing is a guideline but you don't have to strictly follow it.

[sour thistle]

Hi there (Hope this is the correct place), I am learning about Copyright for AI and I am wondering, since Suno can generate the song from a prompt, does that mean that any lyrics generated are Public Domain? Can anyone explain how my music is protected. Does this mean anyone can take my generated lyrics or music and use it. Do I have any legal recourse? Also, does Suno train their models using copyrighted content or do we know that that is not the case?

[charred merlin]

are lyrics generated by Suno entered into the public domain?
Resounding no.

Can anyone explain how my music is protected?
Your music arguably has no protection unless you take steps to protect it. Copyright law is evolving wildly fast from country-to-country, as are your options for recourse & what you need to prove you have the right to do so.

Please contact a copyright lawyer for your area if you want to know specifics, this is a community server & isn't the best place for serious legal discussions (saying "well, the orange-shirted man with dreadlocks from Discord said such & such" won't hold up in court 🫂 )

does Suno train their models using copyrighted content
As far as we can tell, copyrighted content is likely included in the training data, but they've taken sufficient efforts (both publicly and under-the-hood) to prevent unauthorized use of copyright.

For related reading, here's a comment from Oct 18th from AudibleMagic on how they're working with Suno to prevent this kind of infringement, keeping the platform safe for users like you and I who want to use it as an instrument 🙂 https://www.audiblemagic.com/2024/10/18/ensuring-content-integrity-suno-partners-with-audible-magic-for-user-uploads/

[polar spade]

As a member, I can read the specifics to Sunos policies and what my agreements are within the Terms of Service and FAQs, as well which help shed light regarding ownership, if any, of the content created within the site itself. And it's just a good idea to know what we've agreed to by reading the TOS.

[sour thistle]

Thank you for your response, I agree with you regarding the Copyright lawyer. I appreciate the effort you've gone through for me.

[warped furnace]

If you live in Germany it is not possible for an AI to own Urheberrecht (proprietary right), this is only possible for humans.

[sour thistle]

I have gone through what Suno has in their documentation, however my questions weren't answered. They say you have ownership and not copyright, but that doesnt mean I can stop someone from using the lyrics or music as far as I can tell.

[sour thistle]

That's what I have been reading about, apparently its the same in the U.S. but Suno is a U.S. company, at least I think it is.

[warped furnace]

I don't know how this works in the US but here I am only the proprietary rights owner of my own lyrics. As I didn't create the music, I have no right whatsoever to it. The AI can't own rights to anything.

[sour thistle]

This is what I've read, which as far as I understand means that anyone could use the music but not the lyrics. I'm not sure though, that's why I'm checking with the community.

[charred merlin]

Good to check with the community for discussion, but please don't use it as justification for anything 🙏

Suno is based in the US, but the relationship in terms of copyright, ownership, & authorship are dictated by your local laws where you used the service. (now I'm wondering if we agree to US law by using Suno. This is why I tell people none of this advice is legally binding 😛 )

[warped furnace]

Depends on how the laws of the country have decided on that matter. I only know the German Copyright Act, as I live there. For now, I'd say, it is a grey area. One thing is sure though, if you wrote lyrics, they are yours. Nobody has decided yet, what to do about AI training and using copyrighted material for that.

[sour thistle]

Agreed, otherwise I would need you to come to testify..👨‍⚖️

[polar spade]

A copyright is just that, a right to your copy or creation. Since the 18th century, there have been copyright laws. Since the 70s, they have been made to be more enforceable regarding song, writings, scripts and so on. But unlike human rights, we really don't have a public advocate or police to stop someone or defend you and me from someone taking our content and claiming it theirs or using it as theirs. Or maybe there are and I don't know about them. Always good to check. We have to enforce on our own or with the help of an attorney has been the prevailing direction.

So we file a copyright registration on those things that are copyrightable which helps us at least be rewarded from our attempts to enforce our copyrights.

[charred merlin]

Ugh -- even "copyright" is a word with shades of grey -- some places grant weak copyright just for having created something (the amount of technical skill required varies by area, and can be interesting to see what passes as minimal expression of creative skill in different countries -- making a jackson pollack is copyrightable, while spilling paint may not be, regardless of how good it looks if it wasn't intended to be art)

But because all of these words are seen as interchangeable to the layman, the assumption is that creation of a song (authorship) through a service that gives you exclusive rights to a song (ownership) is the same as a HARDCOPY paper-copyright

[sour thistle]

Exactly, If I press generate and I get the Mona Lisa, its not copyright-able.

[charred merlin]

😬 Depending on your jurisdiction, it may be? Some places would absolutely say no because it was made by a computer. Some places would say maybe because YOU CHOSE the prompts to use

There's absolutely ZERO concrete answers, except the fact that contacting a lawyer is the only way to have certainty for your specific cases.

[sour thistle]

Essentially I own the output but not the content unless I wrote it?

[sour thistle]

This makes sense, which is why it's best to speak to a copyright lawyer.

[warped furnace]

Again, we're talking as laymen. 😅 But here you don't own anything you did not make yourself. Clicking on "enter" after a style prompt of 120 characters is not a level of creativity that grants you proprietary rights. At least for now.

[polar spade]

As any app or site, to simply say I have a copyright and this or that app has my timestamp and documentation, is a slippery slope. Document as much possible ,on your own, in as many forms as you feel comfortable, and according to what your entertainment attorney says would be enforceable. Once a site goes down, you will be much out of luck enforcing a copyright infringement to just rely on the site or app for safekeeping that information.

[sour thistle]

Yes, the general response has been, "You need proof." so document your steps when creating anything in AI.

[toxic ridge]

Why does the AI suck at understanding stuff like French House and Jersey Club?

[polar spade]

Simply put, the more you can prove something is yours, the easier to prove something is NOT theirs. And they may certainly try hard-- even file a copyright in front of you. That's about what it comes down to.

[sour thistle]

Thanks everyone! I know what the next steps are now. Have a great day!

[polar spade]

I remember in 2007, I moderated a "lyrics" website. Only lyrics. It was nice that it timestamps your creation, had some good filing system, the works. Nice site. Went down two years later. No notice. So that's helped no one. (I printed all my lyrics, which 17 years ago were junk, but 1 I got cut, anyway... lol). Document. Document. Document!

[charred merlin]

Those genres might be underrepresented in the training data, it might have been listed as something different in terms of genre tags -- no idea.

You could use ... > report > bug > doesn't follow style on songs that notably miss the mark, and it'll help train future models on what outputs didn't meet your expectations. Isn't going to give immediate feedback or improvement, but this is a way to be proactive about the matter 😄

[toxic ridge]

Is it also because the AI assumes that Jersey and French are just regional descriptions and not actually a part of the genre name?

[charred merlin]

There's so many factors it's hard to say what would & wouldn't be likely to impact the output.

When I'm working with Suno & it doesn't understand what I mean by something, it's the same as speaking with someone through a language-barrier -- if they don't understand what I mean by "ska", instead of thinking they're stupid, or saying "SKA" louder & more slowly at them, I might try to think of different ways to describe the same sound -- "you know -- horns, happy upbeat guitar, dancing music?"

[toxic ridge]

French House isn't so hard to prompt out. Jersey Club is the hard one because of the kick pattern

[charred merlin]

I'm not wildly familiar with Jersey Club -- as it could help another project I'm working on, would you say any of these are relevant in regards to that genre?

Or is it a bit generic?

[toxic ridge]

Party vibe, up-tempo, drum machine, looping, energetic, maybe vocal chops. It's an urban EDM style

It uses 808s as the bass

The kick pattern is 5 kicks in the same span of four on the floor

That's the main issue

[charred merlin]

🫶

[stiff sable]

i also tried generating jersey club a little while back as a test and noticed it struggled with it too - your best workaround for now is probably create a kick drum loop yourself in a couple different tempos, upload them, then start new tracks with that and describe the sound you want overall, and it should do it while retaining that groove @toxic ridge 🙃 *hopefully*!

[charred merlin]

Heck I missed seeing you around here, dude -- always a red letter day when you hang out 🙂

[stiff sable]

always keeping an eye on here when you least expect 😉

[charred merlin]

hurriedly tidies his desk up

[stiff sable]

a little v4 announcement recently may have just reignited a spark in me so you may see my activity spike in here again soon lol

[normal quartz]

Hey team, does Suno perform better vocal separation when the vocal frequency doesn't conflict with the rest? I separated one where most of the frequency was in the middle, and I got a lot of artifacts. So I'm wondering if I should try a different voice.

[normal quartz]

Should I ask Suno for something like just keys and vocal, so it's easier to split the vocals from the rest? I'm afraid that if I do this, then the singing won't be what I expect. E.g. how can I make a grunge vocal with all its nuances while having only the keys 😂

[charred merlin]

There's definitely a lot of bleed between the stems if you're using the in-house Stem Separation -- it's handy for quick demonstrations, and getting rough instrumental/vocal isolation for use in other generations (extending, covering, personas, etc), but I'd recommend using external tools for vocal isolation

Even then, it's not 100% perfect all the time, as we're attempting to create stems from something that was generated in whole without the use of layered tracks, like we're attempting to recreate

[toxic ridge]

Didn't think to ask earlier but what's this project of yours?

[normal quartz]

I guess using something like UVR is the recommended approach then? If it's not too much to ask, do you recommend any particular model?

[charred merlin]

Furrybeans speaks very highly of UVR -- I alternate between the OpenVino plugin for Audacity (optimized demucs, afaict) and https://www.bandlab.com/splitter

[charred merlin]

Not fancy, but it can help turn ideas into style prompts https://websim.ai/@theharber/song-prompter -- enter the genre (or anything, really) and it will present a list of prompts.

The prompts you select get added to a pool that you can rearrange, then when you're ready, you click "copy keywords" and paste into Suno's Style section

[normal quartz]

Nice I'll have a look, thanks

[rough gust]

Hey all, I am new. Probably its asked before but is there a way to add the songs to each other or I need to use an editor

[sage grove]

cmon yo help my ass over here, its been like that for over a week

@charred merlin maybe u'll know?

[charred merlin]

You can stitch the extensions together by using "Get Whole Song", but you'd need external tools to stitch together unrelated songs

[nimble sierra]

Hello - a question. I wanted to change the chorus to be my own lyrics, and I did that but forgot to change the second chorus. So I changed it again, but found that it then creates you a new tune. Is there a way of avoiding that, and just being able to change the same underlying version?

[rough gust]

Thanks. I have pro version, but I don't see that option. If its a 2 min song, I just click on extend and add something new.

[charred merlin]

I've seen error more recently in place of the "your song has been flagged for moderation" when something about the lyrics or style prompts doesn't pass muster (https://help.suno.com/en/articles/3198209) -- if you still get the same message when generating an instrumental song with a simple genre as a style prompt, you might need to contact [e-mail] to resolve the matter for you.

[charred merlin]

You only see "Get Whole Song" on extensions (clips labelled part 2 or something), ... > Create > Get Whole Song -- that adds a song to your library containing the original section & the extensions all as one. Doesn't cost any credits, either 🙂

[rough gust]

I don't have that option just realized

I think it fails and refunds the credit

[random brook]

Songs are ending while incomplete well before 4 minutes more like 3 and a half minutes. How can I prompt to get the song to use the whole 4 minutes before cutting off or ending. Apologies in advance if this question has already been answered somewhere, couldn't find it in search.

[charred merlin]

Can you take a screenshot? If you're not seeing a feature as basic as "get whole song" specifically on an extension, there might be something being misunderstood, or something for the team to investigate.

[rough gust]

It fails to generate part 2, maybe I do something wrong with the prompt

[charred merlin]

if you click ... > create > ... does it show "get whole song?" -- the vertical 3 dot interaction menu at the FAR right hand side

[rough gust]

It showed hecking now

So the extension is only 4 seconds?

[charred merlin]

This happens more frequently if you're extending from dead silence at the end of a track -- if that's the case, move the extension timestamp earlier to when the song is still going strong

[rough gust]

I understood the logic now. Thank you. I thought I can compile different songs, each from 2-3 min

[rough gust]

Can I make it to half an hour by keep adding something like this?

[charred merlin]

You could do that externally in Audacity pretty easy -- fade out one song while you fade in another

[charred merlin]

Feasibly yeah -- tbh, I start to get really annoyed with my own tracks if they start approaching 4m, so I might not be the best person to ask in regards to longform content 😅

[rough gust]

You cannot dictate the length as I understand, when i first tried it gave me one 2 min song, one 4 min song. Can we add this to prompt?

[leaden lark]

yes, you can, but eventually the "noise" will get more and more noticeable

[charred merlin]

v3.5 starts off with 4m, the extensions are 2m. v3 is half the length

[leaden lark]

i'm assuming v4 will also be 4 mins?

[rough gust]

I had only one song which was 4 min, with lyrics. Without lyrics instrumental its around 2 min. Even on v3.v5

[charred merlin]

So excited to read the official documentation to see what all will change 😄 Just realized that I've thought about GTA6 considerably less than I thought I was going to throughout 2024. Thanks Suno

[rough gust]

btw @charred merlin thank you so much for the help. Its more fun to play now hah

[charred merlin]

📸 🖥️✍️

Suno is a life-changing amount of fun as it is, so that comment means a lot to me

[rough gust]

Practice question: What if you keep extending the song with same prompt? Doesn't make sense or make sense because it keeps generating new song

[charred merlin]

great question -- arguably, you don't need to enter the same style prompt when extending, because it uses the clip you're extending from as context to continue from -- it can help reinforce a style, or you can shift the stylistic direction by adding something different

#1256301881966198805 message -- a related tip that I find helpful when playing around with extensions is that Suno can only use the clip you're extending from as context, and arguably can't "hear" any segments before that.

This means if you extend from an extension, the new part 3 you're making can only continue from part 2, and doesn't actually know what the original segment sounds like. This is why "get whole song" can give different results to extending from extensions -- but once you get up to 10+ minute songs, you might be far past the point where Suno can keep everything in context -- diminishing returns for sure

(mind the clutter in my old workflow thread, I don't have the ability to tidy it 😅)

[leaden lark]

i have been testing out a tagging thing, and i have gotten fairly consistent results

[glad valley]

there's always gonna be bleed especially on anything under 32bit float with 192khz

you can clean it up afterwards though, but the lower bitrate makes it harder to get a pristine stem split

It's a shame Suno doesn't run on multiple models cause the stems we could get would be a godsend, truly. Maybe v4 will surprise us, aye @leaden lark ?

[leaden lark]

Hopefully

[worn flicker]

Woah, lots. So, Replace Section is acting a little loopy still. Over extending. While I don't care if it gets fixed since you can just do the timing by ear, but we need to be able to crop the replacement. Is that a common thought? I didn't even notice crop wasn't available for sections. I was confusing it with extend

[worn flicker]

What do you mean by this?

[glad valley]

it's hard to explain but more data = more info = easier to stem separate

a file with that though will be massive

a raw file with at least 192khz and 32bit ? maybe 500mb for 4 minutes? If not more

[worn flicker]

Sorry, what I meant was what you meant by bleed, as neither of those have anything to do with splitting a track, but since you meant it by simply having more data and less Artifacts bleeding into other frequencies, I get what you meant by that.

[glad valley]

yeah

[cerulean relic]

hey everybody! new user here, just installed from the github repo and am trying to install a webui, but I'm having trouble (finding) where my Suno installation actually is

[charred merlin]

Out of curiosity, can you link to the github you're using?

[cerulean relic]

installed from here

whoops lol, ran into a bot

[glad valley]

....

[charred merlin]

Gotcha -- it's not exactly Suno -- it's the text-to-speech model that the same company open-sourced

[cerulean relic]

ok

[charred merlin]

I don't think the github gives a webui, pretty sure it was a commandline interface, but there's a couple playground sites linked on their github that show where you can demo it -- huggingface, replicate, google collab

[fallow pollen]

Then who should I ask if I want to know the trigger and conditions of a certain style?

[cerulean relic]

right, I got that far, got the github open-source version installed and working

wanted to try and install the "bradsec/barkwebui"

(trying not to linkspam)

[charred merlin]

It's all good, I think it's just 10 seconds between links -- can you link to barkwebui?

[cerulean relic]

yeah, here

https://github.com/bradsec/barkwebui

[charred merlin]

Ahh -- this is a user-developed webui -- the resident bark expert/historian @untold briar made similar tools, but troubleshooting with them would likely be up to the owners of the respective repos

https://github.com/bradsec/barkwebui/issues is the discussions/issues page for that tool

[cerulean relic]

tbh I'm not even trying to troubleshoot the install itself since I haven't even gotten that far yet lol

I haven't been able to determine where bark actually got installed to

[charred merlin]

I'm considering installation tips as troubleshooting -- while technically the server has shifted toward Suno instead of Bark, it becomes harder to give advice for bark-related projects not developed by suno.ai, like user-built extensions

[untold briar]

Maybe look at https://github.com/rsxdalv/tts-generation-webui/ it's frequently updated

[untold briar]

If you use an installer, it usually gets installed all in a subdirectory, using anaconda, so it's also got tons of python files. But the big models may be in a user folder.

[cerulean relic]

thanks for the answers, guys, I think I can puzzle it out from here, just needed a point in the right direction

I really need to brush up on the latest tools and methodologies lol

stuff's moving so fast

[charred merlin]

ugh -- I know -- I stopped paying attention to image-gen stuff for a couple months and they've managed to get mochi running on 12gb GPUs.

Pretty sure that was a 320gb model 👀

[worn flicker]

Agreed. I don't want to chase down GPUs that have an insane amount of VRAM just for AI generation.

SD 3 and all the new models are freaking enormous too. It ends up being cheaper to subscribe to cloud based...Suno. Even running LLMs locally is just too slow on my current gear. It was hard enough to keep up 10 years ago, but now it is like a year before you are dated.

[charred merlin]

What GPU do you have? Just asking because I've recently gotten 16gb flux running on my 12gb 4070ti using webui-forge, and I was running local llms a while back with kobold (?) -- provided you have the vram, I'd be happy to grab the links

[lost scaffold]

Anyone have tips on how to create separation between the intro of a song and the actual song? Like, I want a short monologue at the beginning (ideally with no instrumental) and then for the beat/lyrics to begin. I've tried:

[Monologue]
"Whatever I want to say"
[Instrumental] - to indicate a separation
[Lyrics]

Unfortunately, the results are very inconsistent. Maybe 1 in 16 songs comes out ok. Any feeback appreciated.

[charred merlin]

The more specific of a song structure you're trying to create, the less likely you are to achieve it as a one-shot generation through the use of complex prompting.

My advice would be to create the song in segments using extensions and use "Get Whole Song" to piece everything together afterward.

• original section, monologue, narration, dialogue, stage play, interview, stuff like that to suggest just vocals
• part 2 / extension, whatever genre you want -- this way you get to pick specifically when you want the music to come in
• "get whole song" potentially if you want the narration to continue in the song, otherwise move onto the next step
• part 3, lyrics -- this way you get to choose exactly when the lyrics come in after the instrumentals start in part 2

[verbal mortar]

sigh I am so frustrated with suno lately. something changed for sure.
after 20 redo's I finally got the thing to give me a drum and bass track using my usual drum and bass prompts. but...the vocals are so hot that they almost sound distorted like they peaked out.

I'm pissed, there's no way to fix this. if i do a cover its going to change the song dramatically, if I do a cover and add vocals -3 to -6 db, music -6 to -8 db it isn't going to follow it. What did the devs do to this platform because it really feels like it took a step down the last few weeks.

and i'm not halucinating. something is off, very off from what it was doing not even a month ago.

[lost scaffold]

Big thanks!

[verbal mortar]

There's more "drifts" in songs now, audio qualiy is worse it is not clean it is very dirty.

@charred merlin I just sent you a DM

[lost scaffold]

I'd be curious to see your prompt/lyrics. I do dnb songs often. Might be able to help.

[rose jasper]

@charred merlin Can you create songs atm? It's not working for me some god awful reason..

[verbal mortar]

i just sent it to you

[leaden lark]

What are you using Suno on?

[rose jasper]

I figured it out, it had one word in Portugese (Brazilian) that got flagged, fixed it!

[leaden lark]

So Suno had no idea what the word was I’m assuming?

[rose jasper]

Yeah, it was a mix betwen ChatGPT also flagging it, and just Suno flat out refusing to make a generation due to this, with no alert. I guess it's normal when it comes to Brazilian Portugese, since most of their lyrics are so borderline sexy related

[worn flicker]

Based on the norm of drift, I find that anything that has more Reverb or any type of high frequency is going to have more drift. The drift is worse with female vocals and any electronic music along with any type of music that uses a natural flanger/reverb to it. Like everything else in audio, streaming, recording, encoding, the first thing to suffer is the high frequency due to their instability. Like how higher frequencies attenuate way faster than low frequency. So the encoding process can end up garbled on the highs as the higher the frequency the more "effort" (resources) it takes to keep them stable. Same reason as violet is the first color to go unseen on a rainbow in the sky. I am sorry for using you as a rambling thought point...lol

In short, depending on how Suno utilizes its resources, like development servers vs test servers. If they are doing testing and resources are being used to test v4. If those resources are coming from the same server then they would have to lower the resources offered to the community. This would be done dynamically and automatically of course. At the same time I have no idea about anything that Suno has or uses for its backend services.

[leaden lark]

For all we know, there could be separate servers for testing and live deployment, but I obviously dont work for Suno so I can’t say. But if servers are shared, that could be interesting theory as to why people sometimes say the service degraded

[runic oracle]

Covers NOT WORKING. Everytime i generate a cover, it kicks it out as "taking too long". Pro User. 2 days this has been going on.

[stable isle]

Anyone figured out if its possible to get the AI to "reset" itself so that it doesnt end up slowly degrading a v3.5 generation - i have generations that have been extended so many times that the generation has become tinny like its being played in a large metal can

[marsh adder]

It's a wild guess, but thumbs up and down are probably for you to tell AI which direction it should go with next try using similar prompt. "The similar prompt" would be the most important part here. My guess would be to change prompt enough for it to start over.

[stable isle]

this tinny ness is a problem with v3.5 a multiude of extension generations end up with this tinny ness - i hope v4 fixes it

[verbal mortar]

they could have separate servers, we do on our game for testing different versions then merging once testing is done.,

they could also not be using the other servers, something we did for a year for some dumb reason.

[worn flicker]

They should for sure. I'm just throwing out possibilities. Since we all see the same degradation without reason

[marsh adder]

Each time you use those buttons you are narrowing the path it can go through. I'd say it's a feature not a bug, so I doubt it's gonna be changed in v4. Unless I don't understand how it works 🤷‍♂️

[worn flicker]

The drift is definitely due to the high frequency vs resources required for the higher sampling rate to grab those high notes. The tradeoff. Then something like HDR comes around and changed the game in video. A genius software algorithm or process that is able to be downloaded and backwards compatible to make old things better...that is just insane.

[verbal mortar]

they need to either address this, fix it, both or release V4 already cause i'm pissing money away for a downgraded experiance.

also...anyone know how to do a mass trash removal? or do I have to do one track at a time? like i'm in windows 3.1

[worn flicker]

This is probably accurate. Same with drift. The new version would have to be better at resource allocation. I have no clue what they are working on or what they encompass in what they call versioning. If they don't either figure out how to use less resources with the same data or increase their hardware, then it won't fix the drift.

[verbal mortar]

HAHAHA why facepalm theharber LOL

[charred merlin]

#1270928841614823546 message You can select multiple songs at once -- make sure to click the 3-dot interaction button to move to trash

[marsh adder]

It works so good that if you don't select any song and you select delete from the dropdown menu, nothing happens 😂

[charred merlin]

You're allowed to have that opinion -- it just deflated me to hear such a defeatist POV considering I spent 30m talking about ways that I approach song generation in an attempt to help, and how I haven't noticed any sort of degradation over the last 10k songs I've made -- more of a statement on the lack of efficacy in my own efforts than a critique on your opinions 🙏

[verbal mortar]

I need to follow the tips you gave me, it may give me the results I want. Since I got these new monitors though I have noticed a lot more in the songs I missed before. I'm just frustrated. I'll get over it. Always do.

It's like I mentioned to you, to be able to just make 4 versions of a song and have 2 of those be SOOO good that I could just master from there to now...27 versions of a song just to get it to be right and not have that insane gain on the vocals. Just seems odd to me. Maybe I dodged a bullet for a while that others were just getting?
I do appreciate your help greatly. I think i'm being a bit of a jerk cause I also am sick so my tolerance is about zero today.

[charred merlin]

😅 please don't feel like you're obligated to follow any suggestions -- suno is all very subjective, from what I've learned. Very few things that apply concretely or ubiquitously, and enough things vary depending on what you're combining as style prompts -- I understand getting frustrated with the vocals getting louder and clipping, definitely would be annoying.

I apologize for the facepalm react, I'm definitely tense as well 🫂 -- I wasn't trying to shill,or make people feel like they can't express dissatisfaction

[verbal mortar]

@charred merlin no need to apologize man I have thick skin and i wasn't offended at all

you are a great help and a fantastic part of this community

[marsh adder]

You have to treat Suno like a living organism. It can take directions but it has a mind of its own. Another way of thinking is that the more professional artist you are, the less Suno can be useful for you, because it tries to be very creative and if you know how to make music you would like to have more control over it - and that's not how it's been built. That's my take on it.

[verbal mortar]

sorta. I made music for 10yrs before I walked away. when I started using this 2months ago. i fed it 30 songs i never finished. it finished all of them and either as good or better than I would have. I'm really impressed with Suno which is why I guess I get bitchy when it isn't working like i'm used to.

[marsh adder]

Real artists also burn out and come back ... keep trying 😜

[fossil osprey]

In my opinion, these are very interesting observations and this information may be useful

[verbal mortar]

i burned out hard man. wouldn't touch anything regarding music from 2015 to about...August this year LOL

[marsh adder]

I heard stories like that. You can still find "the thing". It can be some inspiration, it can be a collaboration with someone, it can be suno 🙂

[worn flicker]

Thanks, I think too much...lol.

[worn flicker]

Same...lol

[visual anvil]

i have do the lyrics about what i have live and what i was feeling, and when i do the song and listen it, i have cry because i have reconigzed my self on this song, i feel very stupid so please, that happen to everyone right ?

[frosty cypress]

Yeah, this happens to me, i write about my life, a few of my songs i cant listen to without crying

[visual anvil]

if you wanna listen it, this is the song https://suno.com/song/521691e0-f6e5-4414-840d-fcd3e22b66c4
there is no reason to cry, right ?

[frosty cypress]

Ill listen when i get back home

[visual anvil]

ok

[frosty cypress]

That was good, suno can some times capture the emotion. @visual anvil

[graceful condor]

Works for me on windows, select song by song you want to put into the trash by holding shift (works only when no song is playing) and then click put in trash. I do this constantly. I trash more songs then I retain. Hope it works for you

[thin falcon]

Question: Personas from upload not possible anymore? Why? Anybody got Infos?

[charred merlin]

If I had to guess, I think it's as simple as the devs wanting to prevent infringement -- "look, I uploaded the beatles and made a new album by them" or whatever

[frosty cypress]

yeah, thanks to copy write crap we can't do legitimate things

you would think though if youtube/twitch can auto detect and block copywrited music, suno could do the same

[glad valley]

same as you do with files in windows

if nobody answered yet

ctrl and click several or ctrl-shift (not sure if just shift works) to select everything between first and last pick

then whatever action you take will be done to all of em

[frosty cypress]

yes, just holding shift works

[glad valley]

oh by the way who was it that wanted that song which had two lyrics on top of each other?

[frosty cypress]

i saved his song, hold on

@worn flicker I believe

[glad valley]

was it you @worn flicker ?

[frosty cypress]

and I think he was trying to get like 5 or 6 overlapping too

[glad valley]

yeah but he didn't say anything about 5-6, he said it couldn't even be done with two

[frosty cypress]

at one point he did, lol

[glad valley]

yes but he didn't ask me to prove 5-6, cause all I said was that I knew it was possible to get 2 sung on top of each other, sometimes accidentally

especially with personas

[frosty cypress]

lol okay

[worn flicker]

Yes, I never negated the accident...lol. I said I wanted to create a Broadway style and I said it can not be done controllable. You can't run two lyrics simultaneously is specifically what I said, not nonsense that has nothing to do with each other. If you did that then show me and I said i would love it and I would want to know how you did it and that I would say I am stupid. All of this is true if you show me a true controlled lyric stacked on another lyric for an entire section, then I shall be and say all these things. I love being wrong.

[glad valley]

well I dunno if it was by accident or not in this one. Cause it's not mine...but personas tend to make it possible I think. If you name the persona and maybe do a prompt like
[Jane Verse 1] and [John Verse 1] with a previous style and mix prompt that describe both being sung together. Or something. If I had to guess at a way to do it. But I have no sub, so no personas

[worn flicker]

Make the song boss, I said show me. Send me some kind of link please. Otherwise you are telling me things that I already know about. I can't go holding up my end if you don't hold up yours...lol. I would love to see it.🤣❤️

[glad valley]

I can't make it

I don't have a sub

and if I want info on if they made it on purpose, that will have to wait until they answer and tell me if I can share it

[frosty cypress]

create a free one, if it works you should be able to do it in 5 free songs right?

[worn flicker]

Yup

[worn flicker]

Ok. Awesome when that happens please please show me and I will appreciate it greatly. We all want to learn.

[glad valley]

I can't make personas

so I can't experiment or try anything with them

I'll let you know what they say

[worn flicker]

Ok. Then I appreciate your comments. I don't know what to tell you all I asked was very specific and you said you could when I said you couldn’t. I didn't ask you to do anything I said if you could then I would say all that stuff it was literally just for fun. Thanks. For real. I appreciate your time. If you want to continue please do, I will appreciate that as well.

[frosty cypress]

test some idea's I had for this, While I couldn't do it, I've nailed the male to female in a verse or chorus, lol

[unique stone]

Just so you're aware... Using more than one free account is against Suno's ToS.

[frosty cypress]

:/ i have a day until my account renews..

burnt through like 24000 tokens last month, lol

[unique stone]

You can have 1 free and 1 paid and your paid one should also get daily free credits when you run out.

[frosty cypress]

do i have to run it to 0.. because it has 5 credits...

[unique stone]

Ah... Yeah I think that might not work. Not sure. It was noted as a problem a little while back... If you get stuck after your sub renews then I'd suggest asking support if they can reset it to 0 so you can get your free dailies

[frosty cypress]

thanks, because I didn't want to spend another 30 1 day before my account renews

[untold briar]

There used to be a bug if you got stuck at 5, but I think it's fixed so you should back to 50. If you don't, email support and I'm sure they'll make it good.

[frosty cypress]

ran to 5 yesterday, still at 5 now

[untold briar]

Was it yesterday before 8PM EST?

[frosty cypress]

probably not, lol, i work crazy hours.

[untold briar]

Oh it should have already reset at this 8 pm though, I think

[frosty cypress]

i don't remember what time it was

@worn flicker this was an odd result of my tests... liked it though.. its close at times. https://suno.com/song/4cdaea57-f7f0-483c-831d-7a18ca061328

[worn flicker]

I'll check it out now

[frosty cypress]

it seemed like if I had the lines longer they would of overlapped

what you seem to be looking for is round/quodlibet

ill play with it more tomorrow when I have credits again. lol

[worn flicker]

Sorry got hung up...now I'll listen

[worn flicker]

Hey that is awesome. I think we can steer that well. I never used those descriptors in concert. Awesome combo, maybe try adding Broadway, well swapping it out or capitalizing Duet or did you do that already, I am gonna steal that if you don't mind I want to play with that. I have been searching for a good epic choir. Awesome!

I doubt we can get two lyrics to run but nice descriptors, you packed em in there...lol

[frosty cypress]

Oh yeah, your welcome to play with, i was just curious if it could be done, and i have a feeling doing weird brackets will help.. maybe even try something like ((this is a line) in the song) and it may double the first part, then single the second

[worn flicker]

I tried that before, I have tried everything I can think of, and anything ai can think of because I test with generated lyrics. Any song of mine you see without copyright at the top is one that is all AI and I just thought it came out well. That is why I said I like your descriptors, but I think that is the right direction if it can be done, a combo of descriptors and parentheses mixed with curly braces and/or both or whatever.

[frosty cypress]

And some of my tags i used are just remainders from what i started with, anc cinematic. Exposive, nuclear is in all my songs, it seems to help chorus hit harder on slower songs

[worn flicker]

I use epic and soundtrack in most of mine

Caps

Epic Orchestral Dubstep Alternative Rock Soundtrack, Soprano Female Vocals, .....

I like the fusion but I rarely like the outcome of the descriptor fusion

[frosty cypress]

I like using crazy unrelated tags because it gives me unexpected outcomes. I did notice in most my attempts at that. The (this is)a chorus() would almost always swap between male and female, which is a cool thing to know.. edited to fix bad typing, dont discord and drive people. Lol

One of my songs, when it first generated came out as synthwave because of my crazy tags, but it had the perfect melody and chorus for the song, so i just covered it made it rock .

This song, you can hear the synthwave when it starts, but man, the melody and chorus were perfect. https://suno.com/song/a6f6a2c1-8f9c-46a6-a5d7-d5b31ec23c62

[worn flicker]

I'll check it out shortly. Thanks

[stone tundra]

One of the weird ones I tried - throwing tags at the wall lol - resulted in a really cool feel for a bunch of my lyrics...

[Dark Waltz, Dark Ambient, Gothic Doom, Glam Rock, Theatrical, Cinematic, Horrorcore, Funhouse, Female Vocals]

The results were... surprisingly good lol

[tiny jetty]

/help

[flint haven]

How to use Suno

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

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

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

[frosty cypress]

That is fun set of tags, i might try those

[stone tundra]

This is the song. Bear in mind I did upload audio I'd made in a synth that took up the first 20 seconds or so, just to give Suno the feel I was looking for

https://suno.com/song/3566c612-f2f5-40a9-980c-78e2a49dd781

but I think it ran with it really well lol

[frosty cypress]

Yeah, it finds it places as it goes and sounds pretty cool

[merry creek]

So
.. will Suno allow everyone to re do their creations for free? Hopefully we can keep the same music whilst upgrading quality.

Also. Looking forward to SUNO on android.

[stone tundra]

lol doubtful. they need to make money somewhere lol 😉

[frosty cypress]

Considering i use 2k to 3k credits to make a song, 50 or so credits to cover not a big deal

[merry creek]

That I can agree to, but a free user can hope

[worn flicker]

Good sound except for the freaking drift, did you break it up or no you are a free user aren't you? If that is the case have you tried using Tempo Change to help bust that down.

Reminds me if James Bond went to Tron at the beginning. What is at the end! That is funny.

[stone tundra]

Not free, no, but I took it into Studio One and worked the heck out of it. Got it sounding great in the end

[stone tundra]

hahah yeah i left that so i could extract it and play it for my wife haha we get a kick out of those hallucinations sometimes

[worn flicker]

Hey, I used your setup and swapped out some styles, I didn't get multiple vocal layers but your layout managed to pull off 3 separate vocals while still swapping. Two different male vocals and a female vocal all play distinct roles. Not sure if the other song did that, but this one is real slow. Not a good song. Still it shows stuff
https://suno.com/song/327b383c-a4d4-451f-855a-545c98bca965

[polar spade]

That's just the style of songs needed for syncs. Maybe there will be more acceptance to AI work soon.

[untold briar]

Layers, Layers, Layers Your Boat https://suno.com/song/2e2866ad-2405-4e46-b53c-04505ae00ec7

[worn flicker]

Did that use pure suno or external audio first

[untold briar]

The actual layering is mostly from messing around with Replace, not that prompt

Replace with spammy/confusing/repetetive lyrics, Suno goes off the rails and way too many voices, then I think I used that sample as a Persona, then Covered, and just happened to generate that song from it (or Replace, I forget) and it kept it up there

[worn flicker]

I was getting there. Still a nightmare to control but it is happening that is the first time that I heard two distinct lyrics go for a period of time.

Using the layout that @frosty cypress used, got me close to over lapping I was about to persona that and then cover it.

To be clear I used that layout using the extend over silence trick, that got me closer

[untold briar]

Actually, I "Covered" A glitchy/skippy singer, that sounded like a CD skipping, and one of the Covers instead of copying that got a bunch of layers. This is a song I think I branched from https://suno.com/song/23bed5df-0380-432f-b162-bdf396cd2637 and did various Covers and Replaces

TLDR: if you want unusual voices, you probably want to use iteration, lots of steps, rather than try to get it going in one single prompt

But "confusing" lyrics is good too in the prompts

[worn flicker]

Too bad it won't accept a for loop

All kinds of voices in this extended over silence https://suno.com/song/ebaeb739-ffc8-47f4-b555-cc6dcccc9ff4

[untold briar]

Oh I see you mean Uploading silence and Extending from a silent upload. Extending over silence (like overwriting a silent part of a clip) shouldn't be special, but if you use Upload, that is a good way to start a song that is differnet. Also can try: Using silence as a Persona, Covering a silent clip. But don't forget about using your own close-to-layered clips as starting points for new songs instead of silence, as personas, crop the best piece of them too

Weird brackets like Happy Birthday (Happy Birthday) Happy [Happy Birthday] [(Happy Birthday)] ([Happy Birthday]) maybe, so Suno mixes up when it's singing and not, and adds voices at weird time

Random old "confusing" prompt example, kind of layer like, but best to also use Replace and other tools too, now we have them https://suno.com/song/a1ca4ca4-e5c7-448c-be80-438ff24ea82c

I forget the workflow, so not sure this prompt was actually that useful, but overlap example again https://suno.com/song/bdc1f817-bb5c-43ca-8062-853fc9726de5

[worn flicker]

My goal on the AI side is always going to be to get whatever I am using to do what I want first. I test boundaries. I do love music so it does meet together.

[worn flicker]

That one is it...are you able to achieve that consistently?

@stone tundra what @untold briar did up there in that last song is what will make me say, I am stupid.
I am stupid...lol

So it uses the normal instructions tags incorrectly in a way, it swaps them over to a third marker because of the volume of them inside lyrics

[untold briar]

I think you could be pretty consistent NOW if you also use the new tools like Persona, Replace, can even use Upload and edit two clips together and then generate. But text prompt alone is way harder. Here are a couple earlier bits, kind of just Extended with the same prompt at points, and got closer over time. Also if you use Suno normally and "keep a lookout" for opportunities, rather than prompt specifically for one thing and one thing. I think I might have testing this kind of prompt in this specific case, but I know for others, I was just like "oh that sounds a little layered, branch of that one later..."

Original gen: https://suno.com/song/466fa297-e300-4a8c-bb2e-7be43e1eddfa

1. https://suno.com/song/34afeeae-9867-4704-938b-fd2ab9a322ee
2. https://suno.com/song/9a9452f5-7071-42e9-aeea-a2d8e87af178
3. https://suno.com/song/66ca0a99-2e4f-409a-b7c3-07edae1c84cd
4. https://suno.com/song/ec9f00e6-e1ae-4951-a087-7a60018d027d

[Using lyrics in brackets] are not supposed be sung. But if [all the words in the brackets] look a lot like lyrics, Suno will often sing them. Kind of a bug, but can also be used as a sort of extra (version of a backup prompt) and push for more voices.

[worn flicker]

It will always be getting AI to do it if it can. I know I can overlap stuff in a DAW. The stuff you have shown me has already broken my boundaries of testing. I like this...lol

[worn flicker]

Yea I know structure is stronger than much in Suno, more lyrics on a single line push rap and so on

[untold briar]

There is another way to use the DAW though. You can use the DAW as raw audio input. Like you aren't making a song, you are just making an INPUT, a thing to give to Suno, and then letting SUNO sort it out and make sense of it.

To give a different example of that idea, you can use a DAW and take two different songs and overlay them, or interleave them. This sounds terrible, garbage, right? But then use that as Suno Upload and generate from it. (Or Replace between left and right different songs, also good.) I have to link twitter can't find the raw video file example https://x.com/jonathanfly/status/1797779432163160276

For Upload, it was actually interleaved at first 10 or 15 seconds, then overlayed for 10 or 15 seconds, using both worked best.

[worn flicker]

The inference that [...] became a tag was because the similar words prior to each one had lyrics in them and not instruction. Good idea, I wouldn't say a bug. Just an inference push

[untold briar]

But basic idea, you can make total musical nonsense in the DAW and give it to Suno. Suno will sometimes make more garbage, but also sometimes make it make sense somehow

[worn flicker]

If I have to use external sources it breaks my grounds...lol. I talk a lot so it can get confusing. I am into both, I love music and making great stuff that gives me chills but I also am a computer nerd and that side is what gets me down right arguing with AI knowing it doesn't care.

[untold briar]

I think using Persona, and trying lots of crops of your clips with unusual voices as Personas, you will find some that you can reliably do it with. If a txt prompt sort of works, then you take the best sample from the sort of working prompt and Persona it, then use the text prompt again with that Persona, then it can work better. etc. and Replace is real weird with voices, and of the weird ways it can go off the rails is lots of layers

[worn flicker]

Yea. My goal would always to be getting it to a controlled hallucination if possible, using mechanics that may get patched up reminds me of forcing FPS in little big planet...lol

[untold briar]

Crop is powerful too, easy to forget about, but lets you narrowly focus on just the part of the song where the thing are you looking for is strongest, and then start from that again

[worn flicker]

Yea, I could crop. If I can control it enough to get two Vocals singing 2 sets of lyrics Broadway style and have them come together, using all AI. That will be a feat. In the current atmosphere without my own models...haha. I assume if I could have all the resources of Suno dedicated on me generating stuff it could do all kinds of stuff

[frosty cypress]

this is pretty cool, nice to see that this is how I can switch vocals almost on command now.

Dear suno, please remove the ad from the side of the webpage for video thing I can't even do, it has really messed up the performance of the dashboard.

[frosty cypress]

apparently it wasn't all this, I had logged in using my microsoft account which triggered the passport thing in windows, which proceeded to use all 120gb of ram...
no idea... but.. okay

[worn flicker]

Why in the world are you using 120 gigs of ram. You running numerous virtual instruments or something? I mean like 40

[leaden lark]

not me with only 16gb ram

[frosty cypress]

well, like right now.. I am here. rendering 3d animations across 3 4090's, playing with stable diffusion on another 4090, playing in suno, on discord, usually have youtube running something. (the rendering along uses around 80gb on average)

[untold briar]

are you really going to have 4 4090s and 64GB of ram? c'mon

[frosty cypress]

3d rendering requires system ram to be higher then vram because it keeps a copy of vram in the ram, so 96gb of vram...

[worn flicker]

I mean so you got cash to spend...lol.

[untold briar]

96 is actually what I have, it's just two 48 sticks, actually not that crazy compared to even a single video card it's a lot cheaper

[frosty cypress]

3d rendering is my job, business expense, tax writeoff

[worn flicker]

I know about all of that, if it is computer stuff I have at least dabbled. Blender, Maya, so on.

So you get paid to render

Or to create

[frosty cypress]

yeah, almost 10k in gpu, the 4 4090s and 1 4080 that I use to play games and run my displays on

I create visual novels

which is why i got into suno, wanted to do custom music

[worn flicker]

Oh. I see what you got going on, we talked about this

Ok

You know that doing all that is overkill right.,,write off or not you could have gotten at least a Matrox

[frosty cypress]

the actual cost of gpu's is almost nothing now compared to power bill. lmao

[worn flicker]

All top end 1k+ gpus and the cooling alone...man.,,that is a lot

[frosty cypress]

I had looked into using the professional cards, and rendering in iray im stuck with nvidia, so while the cooling and power requirement is lower on the pro cards, to get the performance of a 4090 in a pro card is 8k per card

[worn flicker]

Matrox is or was, I haven't kept up but was pretty much a machine made to render things in real time a computer just for that pushing through the interface of a computer. They are expensive but I don't think it gets that high. Things like real time after effects and that was over 8 years ago

I learned about them from commercial editors

[frosty cypress]

yeah, matrox cards have been a bit out over the past few years due to physical based rendering being the way to go now.

[worn flicker]

Gotcha

I figured they would have adapted to the high need of high speed components. Top end NVMe with PCIe 4.0, chock full of NVIDIAs...damn that would be a machine.

I still think the VRAM is an issue, that much VRAM as it stands is hard to sustain at least not without causing major climate change...lol

[frosty cypress]

yeah, 3d rendering requirments and ai requirments are pretty much identical

[worn flicker]

When I used blender and after effects, I watched it go from sitting over night on a commercial environment to an hour on a laptop. Won't be long.

[frosty cypress]

in most scenes i spit out an image every 5 minutes, in some others where I crank the detail, around 30, each scene containing between 2 and 20 million polygons

[worn flicker]

Well I'm happy for you...lol.

[frosty cypress]

lol

[worn flicker]

And a little jealous

[frosty cypress]

its just where my life went

if only i could run suno locally to get around restrictions lol

[worn flicker]

I remember the conversation. Bark is a start. No one said it had to be suno directly, are you the one with less computer knowledge? I forget because we all have white names...lol

[frosty cypress]

lol no, ive been a computer nerd since the early 90's, my entire life has revolved around computers, even at one point wanted a break, went and got a job at a pizza place, in 2 years I was the head of their IT department..

[worn flicker]

Ok so it's another. I was born in 81 and picked em up and had a knack and never put em down

[frosty cypress]

yeah same here, i was born in 84, so pretty much as soon as I could I was on one

[worn flicker]

Where I am from, people still look at me like I am speaking some weird language as they nod their head then ask, "Does it Work?"

[frosty cypress]

yeah, same here, eventually I just got tired of it, stopped working for people, started picking up freelance work, odd jobs, and started making visual novels, now self employed

[worn flicker]

Gotcha well I'm off. Holla soon

[frosty cypress]

later!

[halcyon coral]

ok, just a question. if i try and extend a sound byte i uploaded , why does the song it makes last only a set amount of minutes? (for example, i uploaded a 1:59 song i wanted to extend and every iteration lasts only for 2 minutes or the other one, i upoloaded a 48 seconds sample and it gives me only 3:11 for every song it's extends)

[untold briar]

4:00 is the maximum song length per generation. But Extending an Upload is up to 4:00 - length_of_your_upload. But you can Extend the Extension, of course, to make it even longer. Or you can Crop the front of your Upload to make more 'room'.

[ebon hinge]

folks, how can I get rid of noise - some of my songs generate background noise that should be easy to remove, but I have no idea how to clean it up? If anybody knows of some tools to do it, or if more info is needed, please let me know.

[trail drum]

iam can not use get whole song. it not working, how to fix?

[frosty cypress]

doesn't extend only do a max of 2:00

[stark epoch]

I am new to Suno, and when I create sounds, it is sometimes displayed as 16 hours. I would like to know, when I purchase a subscription, will I be able to access 16 hours of music or not?

[halcyon harness]

Every time I try out the personas thing it uses 2 credits per song? Is anyone else having this happen?

It generates the two songs as normal per a single generation and removed 4 credits from my number of free persona credits.

[pale mica]

can somebody gimme a feedback on this 😀 https://hf.co/chat/assistant/66835a83511a0c887e2ea979

[wet epoch]

Hello, I'm having trouble replace section, it gave me the same part as the original one.. can anyone help

[patent coyote]

uploaded an audio file of someone farting how do i add lyrics to it so you can hear the fart in the background while someone is singing?

[pale mica]

sigh

[pale mica]

https://www.audacityteam.org/

[cyan spire]

Why don't I download? Songs that I couldn't create

[trim glade]

dont know whats wrong here.. i send friends an invite they make accounts and create songs , but we don t become 250 credits... i think Suno have problems... it woud be help wen we try it together: https://suno.com/invite/@tasfariel thats my invite link..

[marsh adder]

nice bait

[trim glade]

thats not a bait.. i dont become anything from friends v.v

i have ask the support but i dont become anny massages

[polar spade]

Generally, it's likely most in this channel are already members of Suno, so the link would not be of much help to test.

As with any email correspondence, its helpful to have documentation of the invite, the friend's registered profile and maybe even the 10 songs they created. I assume a "song" is a generated output, but I don't know if that means 5 clicks of create, or 10. So all these questions would need to deal with Suno directly.

This channel is more about technique of prompting and some general technical discussions as well. Most direct support and billing questions would need to be addressed to Suno directly.

If you have waited a couple of days for a response, then follow up with another email. But certainly let us know what the resolution is. I am curious.

[trim glade]

thank you! I try to wait more for an answer of suno by self 🙂

@polar spade I've had this problem for a long time. I haven't reached 2500 points yet either. Not for a long time. The funny thing is, for 2 friends it worked without any problems, but for 3 other friends it didn't work. They have already tried it several times. But I'll let you know when I know more and if Suno has contacted you directly by email, I've had the email open for days

Me*

[polar spade]

I have looked at the simple instructions on the senders end. It just says send someone a link. Invite them. They join. You each get credits.

It goes no further to say when the credits will show, any specifics as to what a "song" is, and who to contact. So I can understand there can be confusion when the expected does not happen.

I'm happy it worked a couple of times. So, we know it works. If working out bugs 🐛 on one problem means other things get squashed, it can take longer to resolve. It more becomes a one-on-one assistance and may involve just manually giving your friends credit.

Yea. Just ring them back soon and see what they can do.

[trim glade]

thank you for your detailed help! I'll wait now and see how it goes. 🙂

[charred merlin]

Not 2500 points-- 250, two hundred and fifty

-edit. I see you received credits for referring other friends so I'm sure you already knew, I just wanted to ensure 😅

[trim glade]

show here.. Max Share bonus is 2500.. you can send you email 10 friends to become 2500 credits when they make an account with the invite and create 10 songs.. now you understand?

[charred merlin]

It means that you are limited to inviting 10 friends

After you earn 2500 credits through this method, you are not eligble to earn any more.

You can still invite more people, but you will not earn credits for inviting people #11 and onward

[trim glade]

yes and i have send this invite to 5 friends.. i have become from 2 friends the 250 credits , the other 3 i dont become anything when they make 10 songs

[violet mesa]

Hi there, any news about 4.0 launch date

[charred merlin]

I misread at first -- we are saying the same thing

[trim glade]

all good 🙂

[charred merlin]

😅 🫢

[trim glade]

my english is not the best. xD thats the point here.

[charred merlin]

Your English is flawless, truly. I'd like to blame my severe lack of coffee, but I simply made a mistake. 🫂

[trim glade]

were all make mistakes. I gues. Where are humans. 🙂

[charred merlin]

The email support team is off for the weekend, but gets back in tomorrow presumably around ~<t:1731247200:t> -- they work fast, so I'm sure they'll respond soon 🙂

[trim glade]

oooh thx for the information 😄 now i understand why i must wait so long

i mean its litery normal its weekend and no one like to work everyday xD

[stark epoch]

Hello, please, I am new to the group and I have a question. Sometimes a sound appears with '16h00' marked on it. I am using the free version, and when I download it, it only lasts 4 minutes. I would like to know if, with the paid version, I would get a sound with a duration of 16 hours?

[charred merlin]

I swear -- every time I need to contact a company, it's always minutes before they close for a long weekend 😂 "harber's luck"
Sorry -- we'll be out of the country for the next month on a team bonding retreat. or something 😛

[trim glade]

very nice 🙂 thank you for the information

[charred merlin]

Nothing yet -- make sure you follow the https://discord.com/channels/1069381916492562582/1083140371980234852 and https://discord.com/channels/1069381916492562582/1159295652531019777 channels so you're the first to get notified!

[unique stone]

Do you mean it shows "4:00"? That's minutes:seconds, not hours:minutes
(and even reading it this way I am not sure how it'd be 16 hours)

[violet mesa]

I am, for a long time.. also the Android app

The demo song sounds awesome, loud CLEAR voice.. good music track..

They should implement the same things what FADR AI does

Up to 14 stems and midi tracks

And DAW implementation

[placid hemlock]

iv just subscribed. so suno... is it not possible to have the instrumental stem of a track, completerly without the vocal's? surely to create the track the ai created the instrumental and then merged the ai vocals into the track?

[unique stone]

Huh weird. That must be something your browser is doing. The first part of a song can be up to 4:00 (4 minutes 0 seconds) long - that goes for free use and also paid subscriptions. No difference in length.

The only way I can think that this makes any sense is if your browser is converting "4:00" into a 24-hour time (16:00) but that wouldn't explain why it is showing it as "16h00" like a length...

Try a different browser and see if that does the same?

[glad valley]

hai tharr

I really hope they give out some info on what to expect with v4 soon

then I can resub earlier and play around a bit with personas

if I want it

[worn flicker]

You saw what Jonathan did?

[glad valley]

nope?

[worn flicker]

That was what I was talking about. Accident or no it was continuously Overlapping

[broken tundra]

idk what happened yesterday, but my whole libary was suddenly deleted

[worn flicker]

Do you have a song that you shared before?

[glad valley]

I still don't know what he did or what was happening?

[broken tundra]

yea, 2 links i shared still show up

[worn flicker]

I thought I mentioned you, it is above

[glad valley]

cause the song I had, was continuous overlapping for like 30 seconds

[worn flicker]

Share one here

[broken tundra]

but for me it doesnt show, libary is clear

[worn flicker]

Remove the embed

[broken tundra]

all gone

[glad valley]

but I don''t know if I have permission to share that one

[worn flicker]

No I mean a link

[broken tundra]

what link?

[glad valley]

I have no link

to what he did

[worn flicker]

Sorry

[worn flicker]

You the link that you shared in the past

[broken tundra]

i shared it only for myself

to whatsapp

[worn flicker]

That is fine if you can copy paste it here

[broken tundra]

how can i delete it if it doesnt show up 😄

yea wait

[worn flicker]

The embed that shows in here, so it doesn't flood the channel

[broken tundra]

idk what you mean by that but

searching it

[worn flicker]

Layers, Layers, Layers Your Boat https://suno.com/song/2e2866ad-2405-4e46-b53c-04505ae00ec7

That link

[broken tundra]

aa ok

[worn flicker]

Once you post it in here it will make the embedded picture, and like the link I just did above I removed the embed so it doesn't make a huge square in chat and makes it harder to follow the conversation

[broken tundra]

song link

but i can't share it

[worn flicker]

Is it not in whatsapp

Shut up app

[broken tundra]

https://suno.com/@mxxl

these tho i shared for myself, but even there are gone

these*

[worn flicker]

You can edit the message and put < > around the link

[broken tundra]

i have no access to song links anymore

[worn flicker]

That removes the embed

https://suno.com/song/b0b0f651-bd8e-4e58-8651-ee9b199cac70

[broken tundra]

that's what im trying to figure out, wtf happened

what part you don't understand, i can't share the song link, since libary is empty

[worn flicker]

That seems fine. Did you login the wrong account

[broken tundra]

😄

[marsh adder]

He doesn't mean for you to click share on Suno, just find a link you already shared on your whatsapp and copy it here

[worn flicker]

I see two songs in your profile

[broken tundra]

yep, and these aren't in my libary,

• others what i lost

[worn flicker]

Ok those are the ones I am asking if you ever shared before on whatsapp or anywhere you can access

I just shared one of your songs in that link above

[heavy salmon]

hi everyone, do you get stuck while create and get error "a clip get stuck ..." ? thanks

[broken tundra]

so what?

my libary is still empty

[worn flicker]

Since we don't work for Suno, I can help you up to the point of needing support. Which making sure the singe still exist is the first step. It is not normal for them to disappear, I have not seen any disappear completely

[worn flicker]

If you want help then this is the route or you can email suno support. Your choice

[broken tundra]

oh i thought thats tech/support chan

[worn flicker]

I could tell by the way you were talking to me, as if I was a tech support something giving copy paste solutions...lol

[broken tundra]

well managed to save 2 of them but shit, i had to start polishing them,.. now all gone 😦

[worn flicker]

So you never shared any links to the however many other songs?

[broken tundra]

nah

[distant carbon]

Yes the stems feature is pretty poor. Best to use something like audacity with the AI module added (all free) & you can split to vocal & instruments & even individual instruments & vocal. OK eve nthat isn't perfect but far better than the stems feature on Suno, they really need to either improve that or get rid.

[broken tundra]

straight to youtube + visualizer

[worn flicker]

Oh. How about the profile name, or internet history

[broken tundra]

zer0

[heady condor]

Does anyone have any understanding on what to do with this virtual ticket?

[worn flicker]

Dang, well I can tell you I have only ever heard of it happening but not since I have been around. The majority of times it is a login issue. Which support would need to direct you to.

[worn flicker]

Virtual ticket? I haven't ever contacted support is that what they give you?

[broken tundra]

funniest thing i was logged in working on another one, when i saw how it got wiped

thought it was some glitch in my browser... but nope, today logged in and same emptyness 😄

[heady condor]

The early access ticket, you know?

[worn flicker]

No but whatever it is, I want one

Oooooh

The iOS app

[heady condor]

Right

I don’t know how to us it

[worn flicker]

No idea I thought it was open to everyone now

Did you check the store

[heady condor]

The store?

[worn flicker]

App Store, Apple

[warped furnace]

I got one. Cant use it. No iOS phone. 🤣 But it's nice to look at. ^^

[heady condor]

Understood, alright. I already have the IOS though.

[worn flicker]

In other words if the app is publicly available then there will be no point to the ticket

So you searched for suno

[heady condor]

That is what I was trying to understand.

[worn flicker]

I don't have ios

[glad valley]

I see

[worn flicker]

It is international access

[glad valley]

well I knew I had done it in the past at some point, finding where or what has been the problem since I've done so much random crap on Suno that I deleted. I didn't use this method though

[heady condor]

International?

[glad valley]

that I know

[worn flicker]

If you can see it then you are supposed to have access to the app

[glad valley]

it also wasn't consistent

[worn flicker]

If you are local then it went to you in error

[glad valley]

and it was more simplistic like one singer and then the background might utter a word

like a choir or something

but yeah, I know that's not what you had in mind

[worn flicker]

It was a continuous Overlapping lyric, it isn't what I am looking for but it is further than any others I have seen.

[glad valley]

yeah

I know that

[worn flicker]

It required nonsense, but at least I heard multiple lyrics without using persona

[glad valley]

I'm limited in what I can do since I also can't cover on free sub

or inpaint, or crop

[worn flicker]

Both

The whole conversation is above it isn't that far up

[glad valley]

So if it's possible to make layered lyrics easily now, then it's not something I can

yeah I'm in an adhd brain today, can't focus on reading much

[worn flicker]

All I was ever saying is that it was an AI limitation, and that I would like to be able to do it. And we are all ADHD these days...lol

[heady condor]

Understood, thank you for your time.

[worn flicker]

🙂 np

[broken tundra]

for US users only

[glad valley]

I'm very ADHD 😛 and a lexicon of other stuff

not "american here buy some pills ADHD"

[worn flicker]

Yea I read the small print that said international.

[worn flicker]

Congratulations??

[glad valley]

for not being American? thank you

[worn flicker]

Agreed

[glad valley]

Seems like an annoying place to live

[broken tundra]

@worn flicker aa btw this one i shared to my whatsapp https://suno.com/song/0bbdcc27-46a0-4e9e-b96d-55d7e550dd47