📚┃suno-school

[mint owl]

there it is

[vale rampart]

Same

[grizzled cave]

Same problem

[cold zenith]

Is it possible to make a speech with instrumental ?

[final hazel]

hallucinations scat yes

not really words

this is an instrumental... https://app.suno.ai/song/423d3dea-7a36-4621-a3b2-ff92edc066d4

[cold zenith]

i don't know how to make just a dialogue

[final hazel]

you can try with [spoken word] in lyric

[tame lichen]

no matter how many continuatons I make in SUNO, the song always gets cut off at the end, how do I make a song that finishes/concludes naturally?

[final hazel]

I have no problems tbh. There are 2 things I learned though:

1. Make full songs in between and continue from there every time. It has a better memory in melody and is easer to make a full song then later on.
2. Do not remove clips in between.. that does corrupt making a full song.

[cold zenith]

V3 skip some words, a lot compare to V2.

[shadow bough]

I’m trying to get a faster paced song even with mentioning 140 bpm, just DENIES to do so. Not sure what else I can put. It does have lyrics.

[final hazel]

It doesn't respond to 140 BPM

It's often better to leave that out in general

[shadow bough]

Hmm okay

How can I make it be a faster tempo?

Is that possible

[final hazel]

say thing like fast paced.. up tempo, energetic

[shadow bough]

Okay bet.

I put fast paced, energetic. Good call. It definitely helped

I’m trying to get a little bit of inspiration from other songs and I really love the vibe of the future bass / drum and bass.

So it needed to be fast paced lol

Is there a way to get more character space to get into more detail? Is that something that we will be seeing in the future?

[final hazel]

I am not sure about the music style box, but I know they are looking into make the word cap for lyrics higher at least

[shadow bough]

Would be awesome if they do! It could definitely get more detailed results of the vibe we want to perfect or get similar

Especially when it comes to dynamic shifts, tones, majors/minors/ while getting a more detailed atmosphere

[final hazel]

Everything effects the model, and I am sure there is a reason for the cap. I can imagine the model becoming more instable with more triggers. But that's speculation from my side tbh

[shadow bough]

That’s a good point

Just a hope. It’s still the beginning

[shadow bough]

Ugh. I love this 2nd part of my song but it cut off half of a lyric mid word and I want it to end the verse on the 3rd continuation. RIP D:

It sounds awful trying to readjust the lyric

[fallow gull]

Has anyone tried to make a non-standard song structure like a cumulative song?
I removed genre styles and the verse tags because Suno kept getting confused a dropping lines and it would skip verse 5 entirely every time I tried to continue.
https://app.suno.ai/song/933e510e-5216-4801-b00e-041503a7984e

[shadow timber]

Bot seems to be down and there’s a system maintenance banner, anyone know how long it’s going to be?

[north beacon]

Does using quotes in lyrics "quote" have any special meaning to the AI?

[tame light]

is there an API access for suno?

[final hazel]

please don't spam your question in multiple channels...

[cold zenith]

not yet

[thin flint]

What do you mean by in between? So with your first two clips, get whole song and with the next clip, just get whole song instead of continuing from the previous clip?

[final hazel]

Exactly 🙂

[thin flint]

I have no problems with your tip in general about just leaving everything blank and just put end in the style box. I'd love a way to control the ending I get though, such as a fade to end or a big finish.

[final hazel]

We are simply not there yet with AI music. Consistently is hard to achieve, so that is why they are not offering things like that. If they do and it would work it would open another can of worms. 🪱

So that is why I am trying to educate how things can be done.

[thin flint]

Can we colab on playlists? Ever since @final hazel made a style medley, I've made a few and would love to create a playlist if anyone else has done any.

[final hazel]

Feel free to start one and share it in discord. I can give you my songs url if you wish. (But in car now, so have to do that later)

[rose ore]

okay, so, a couple of ways to get around radio edits when the system throws them at you and you don't want them

1. Subversive tag works
2. Just discovered that you can spell your swear words like fck and shit, and the system handles them differently, while still reading out the entire swear word.

Hope that helps.

damnit

what I'm saying is words with stars instead of letters in the swear word, don't know what's up with discord

[broken lotus]

lol asterisks trigger the italics

[rose ore]

lo

urgh

so frustrating

oh well

[broken lotus]

wait so you're just replacing the vowels with a *

?

that would be easier than some of my contorted spellings.

[rose ore]

yep

discovered that by accient when I was collabbing with the rap guy

[teal raft]

hell yeah

Although the context has to matter, I shared a clip with a friend named Derick and there was the lines "eat shit Derick" and it had no problem

[rose ore]

lol

nice

[teal raft]

Processed the lyrics and the voice sang it right off the bat

[rose ore]

I wish one way or the other, the bleeping was a little more predictable

like the one I made today, had not a single issue

and I did nothing

[teal raft]

In v2 months ago I had a similar thing, where the system willingly allowed me to make a clip about 'beating the shit out of a vending machine' and it sang it no problem. Beating the shit out of a person though, no go

[rose ore]

other times, suno gets so prudy on me, it's unreal

[teal raft]

I think it's however it's working with chatGPT

and the censoring is at the text level before the music gets made

Which is why your tricks work

[rose ore]

It's odd. Tentatively, I think it's got two layers of internal moderation going

maybe three

dunno

wish I could see the spec

then I would really bend it

lol

So anyway, yeah, it think this stuff gets caught up in sort of a fuzzy match system that is stricter or looser on any given day

[teal raft]

Right

[neat nexus]

When trying to document anything with asterisks in Discord, always put it in code format:

like this s*** because this is the only way this s*** will show up properly in Discord.

[thin flint]

Hey all, I just made a style medleys playlist featuring such creative people as @untold briar, @final hazel, and me. If anyone feels inspired to make songs that blend styles seamlessley and it'd be good to add, let me know. 🙂 https://app.suno.ai/playlist/df2fe44b-42b7-4fcc-a632-41836eb14e82

I wish we could download songs made by other creators. Just the audio if there's concerns of reuploading or whatever.

[deft totem]

Medley #3 is nice one.

[broken lotus]

you can add my symphonic storytelling thing, if it fits the idea...
https://app.suno.ai/song/f96cb9bd-133f-4f61-ab22-e21dc736eea2

[thin flint]

What do a bunch of those blank lines do? Or is that just what happens when Suno blends together with get whole song?

[broken lotus]

🤔 the song page doesn't record any style prompts after the 1st one...

[broken lotus]

it's just the text formatting on get whole song...

[thin flint]

Added. I continued from it to create a natural end.

[round jetty]

/chrip

[ionic wind]

My kingdom for the ability to cut off the end of a track when it creates a perfect ending, pauses for 10 seconds, then keeps going

[shadow bough]

Anyone have good prompts to have vocal /lyric chops?

[broken lotus]

chaotic stuttering...

vocals...

[shadow bough]

I tried [Vocal Chops] [Lyric Chops] it still tends to say the full word , I’ll try your suggestion

[neat nexus]

Does Audacity not allow you to edit it thus?

[ionic wind]

of course, but I want to stick to Suno's interface instead of going through hoops like that and upload the results to soundcloud/youtube.

[plain agate]

dose anyone have the meta tag for like screaming vocals. like you would hear in metal/hardcore music?

[thin axle]

Tried about 30 gens to get it to replicate a chorus melody, but it couldn't

Not sure if context windows for Sumo are similar to LLMs, except time-based? not sure. But so we know what the context window is for Suno?

Generally I can get a chorus or something to replicate, but this one full song was a bit longer than usual.

[broken lotus]

Yes there is a context window, concept is very similar to text completion.

afaik, Suno has never confirmed the length for v2 or v3 (which seems longer)

[teal raft]

One of the devs (I forget who) said that it's a more-complicated-but-mostly UI thing with why we don't have the ability to prompt as long as v3 can handle

[final hazel]

@teal raft I think you are refering to this convo ^^

[teal raft]

Yeah that one

georg, that ol' yak yelder

[cold zenith]

the biggest problem of the alpha i think is the fact that it skips too much lyrics

too much

[graceful condor]

Currently have this issue. Won't go away

It's for all gen attempts

[iron prairie]

Hey, i am new around here and i have two questions i could not find in the FAQ. Is the quality restricted to 192mp3? and is it normal that some sounds rather distorted?(have good head phones and system) Thank you 🙂

[teal raft]

Yes you are limited to that currently, and yes it is normal to get distortion. It's an AI model so it doesn't really know the difference between music and noise, though you can prompt around that. Suno gives you the entire master all at once, hence if you are from a music production place it's a very different style of making music

[cold zenith]

it works again

[iron prairie]

Thank you. My background is indeed music, although as a roadie. I love what Suno does and had never the patience to make music myself.... this is perfect. I am sure one day the quality will follow.

[teal raft]

If you see what people post in #🐣┃suno-showcase and what shows up in the 'Explore' tab on the website you can see what tags and styles people are using. Suno can be really powerful but there is a slot machine aspect to it. You can think you set up the perfect styles, lyrics, whatever, but not get anything good unless you pull the lever a few times

[iron prairie]

Isn't that the case with any Ai really and i dont mind it to much, as sometimes it comes up with ideas you would never thought off... It is only the start, mad times ahead 🙂

[teal raft]

Very much so yeah

[real plume]

Hi, I'm wondering how can i go into further details when requesting my song, i'd like the AI to understand that i want a verse and a chorus in my song, is there any way to do so ? I'll give the AI my lyrics for the song, but i want it to understand that this part is the verse, and the other is the chorus. Is that possible ? Thank you.

[untold briar]

Yes, in Custom mode you can fully specify the prompt including [verse] and [chorus] and what lyrics are in which

[real plume]

Thank you !! What is custom mode and how to get to it ? I'm completely new to sunno.

[untold briar]

Are you on the web site? It's on top

[real plume]

Oh yes, the custom mode, i get it, but then how to specify which lyrics are meant to be chorus, verse... ?

[untold briar]

[verse]
Lyrics
Lyrics

[chorus]
Lyrics
Lyrics

[real plume]

amazing, and the AI will automatically understand and make the instrumental adapt to the chorus part ?

are there any other specific prompts that i can learn ?

[untold briar]

See https://www.suno.wiki/ for some examples

[real plume]

I just found out about that AI and i already want to subscribe to get the V3 and have a better tool, even though the free version is literally amazing

[untold briar]

Also if you look any any of the songs in chirp-showcase channel that have a video, you can see examples of the verse and chorus and other prompts in the video itself

[real plume]

thanks !

do you think V3 is worth it?

[untold briar]

Yeah

V3 also has custom continue times, which is a powerful feature

[real plume]

what is that ?

also, can you (in the free version) tell the AI to make the song "imitating" the style of a specific artist ?

in the song description

[untold briar]

No, you can look up or ask chatgpt yourself how you might describe the music style and genre of a particular artist, and use those words, though

[real plume]

nice thankk you!

where can i find the exact all new features and improvements of the v3 ?

[untold briar]

The main "feature" is the custom continue time, the rest is overall quality differences

[real plume]

something weird just happened, i asked the AI to continue a song, which has been done, but the "voice" completely changed, even though i left the song description empty, on purpose for it to continue with the same style, the syle is ok, but the voice changed ??

[real plume]

can you describe the custom continue time ?

[teal raft]

If you have a clip that's 1:15, you can say "Continue from 0:58" rather than having to continue from 1:15

[real plume]

nice !

[teal raft]

v2 only allows you to continue from the end of the clip, no matter what. V3 you can pick any time from 00:01 till the end of the clip to continue from

It's very nice yeah

[real plume]

wha does the remix button do ?

[real plume]

?

[teal raft]

Literally just copies all the text from whatever you are remixing from

It's an entirely new song

It copies the lyrics and musical style

and uses that to make an entirely new song

[real plume]

why when continuing a song it changed the original voice ?

[teal raft]

short answer is 'because it's an AI model'

Longer answer is because it's a strange weird thing that nobody really fully understands how it works, really like most LLMs

The way around that is to just keep generating till you get something you like and keep continuing on from there. Any full, complete song I have personally made has hundreds of clips that I made on the way to it

Some songs burn much more credits than others

[real plume]

that's what i have done, i generated something, it was fun, then continued to have a chorus, with same prompt, but voice has changed lmao

[teal raft]

Yeah unfortunately that's just a thing most people shrug and go "welp try again"

Yeah you burn credits but that's just currently where we are

[obsidian tendon]

What's funny is that CEO's letting people go aren't saving money when they're spending so much on credits to get their results just right or learn how to use the tool best. That's is the realty of AI. It's not the sci-fi people hope for. Not yet. AGI will be a good first step for us to shift how to interact with AI. Come to terms with expectations, letting go of ego will ensure we progress.

What kind of text should I use for the prompt to get the voice to sound like this? https://www.instagram.com/reel/C3-zqb_J9RQ/

alittleandroid

Stunning vocal work of @carmenvalentina and @deceptology in the parking garage😂😂😂 This video BLEW UP on TikTok so maybe it will here too haha.
#wbpartner @warnerbrosca #dune @dunemovie #duneparttwo #timothéechalamet #zendaya

Likes

561704

Instagram

Close? Not even? https://app.suno.ai/song/6d425462-ef0d-480a-9bb7-9efea836a10d

[lime coral]

trying to get it to sing in japanese, but it refuses to 💀

[teal raft]

It absolutely can, just put Japanese into the lyrics, it will sing it

[lime coral]

i have, but it wont sing it

just the instrumentals

[teal raft]

That's just a Suno thing. Sometimes it does that, the language doesn't matter. Suno not singing lyrics you want it to happens be it Japanese or English or Spanish or whatever else

[lime coral]

wasted all of my credits trying to get it to sing lmao

[twin halo]

HAPE 30 COINS BEI EUCH GUT ALLES GEMELDET!

[dawn cairn]

Can Suno ai in free version be used without the company's consent in another app application in which their branding is still shown?

[wet silo]

I'm being driven to madness by completely straightforward (and original) lyrics coming back to me as "flagged for moderation", which seems to be a really common problem based on searching this server

[dawn cairn]

Please reply

[warm bolt]

https://youtu.be/9zMgK640cHY

[grizzled slate]

I'm sure this has been asked many times!

I'm looking at the subscription and am unsure of the credits, create, songs, and generations.

From what I see, each song costs 5 Credits, and a Song is the same as a Generation.

Each time I Click, Create.... this is one Generation, or what Suno calls a Song?

Each time I click, Create.... it will cost me 5 Credits?

Can someone please confirm.

[limpid spade]

Each 'Create' will provide 2 generations/songs and cost 10 Credits. @grizzled slate

[digital talon]

One thing I've noticed is adding "vocals" to any style prompt seems to skew things towards pop, which took me a while to figure out when I was making dark industrial tracks.

[cold zenith]

THE INSTRUMENTAL option is very saving

good option

[neat nexus]

It's just the AI being an AI...sometimes it will jsut refuse to do the thing that is asked. I can guarantee you that it is possible to make it happen though: https://app.suno.ai/song/755cb0e3-7e1d-42fb-b044-c0894875c8b7

[steady dock]

vfdvdf

[teal raft]

Yeah same man, same.

[stable sparrow]

how would you prompt the style to get something simililar to jonny cash hurt? I tried chatgpt, but results are not working.

[final hazel]

Soemtimes you've gotta use oldfashion google

[stable sparrow]

voices are always way to jung...

young

I tried J Cash's voice in the song "Hurt" is deep, gravelly, sense of vulnerability and somber introspection. His baritone voice, seasoned by age and life experiences, hauntingly beautiful and unforgettable

to get closer, but voices are always very young persons

I also tried: deep, gravelly, sense of vulnerability and somber introspection. baritone voice, seasoned by age, guitarre only,

[final hazel]

That's a good one.. I have no clue tbh.. you @amber sundial ?

[stable sparrow]

the music style goes into the right direction but voice is always to young

[amber sundial]

id go with more like

weathered, deep, somber, old, wisdom

[final hazel]

raspy?

[amber sundial]

eh but its not really raspy

[final hazel]

https://tenor.com/view/prince-eric-ariel-disney-no-voice-the-little-mermaid-gif-10951277

[nocturne delta]

@versed valley

[broken lotus]

this is the only voice I would consider 'old' that I've gotten. Describing the voice doesn't seem to work on it's own. I think the trick is to figure out a song that would be sung by an old man...

I had a v1 voice that sounded older, and the context was a vegas lounge singer...

https://app.suno.ai/song/282892d8-8a51-404b-914e-14a09787905b

the voice prompt was
low-octave male durge
but none of the others were like this, so very low reliability if any.

[stable sparrow]

maybe the training data doesn´t contain any older people?

I could not find a way yet to create some older rough voices

[dull sandal]

Has any found if the order of prompts in style matters?

[teal raft]

No, but if it is like any other generative thing, the first one is paid attention to more than the second, than the third, etc

or has the most influence

[dull sandal]

Yeah that's my thinking too

[rapid bluff]

type horrorcore with your style probably at the very end, it will bombard your song with a graveled voice

[rapid bluff]

i think it does, depending how long your style prompt is, but it also prefers things more than others ive noticed too. For example put violin, bagpipes, flute, and 7/10 times the flute will be dominant

[fallow gull]

That's a bit of a blessing if the flue dominates the bagpipes.

[rapid bluff]

at least in the song i made which i rerolled heavily, but the flute beat it gave me was solid

[fallow gull]

I found the same with Hurdy-gurdy and Uilleann pipes, if you want to hear those, they need to be the only instrument specified. Adding fiddle or tin whistle will override them.

[rapid bluff]

yeah im intrigued to what those even are lol..

[fallow gull]

https://app.suno.ai/song/91c5a157-313f-436c-a198-99cdefcdf440
https://app.suno.ai/song/77455cf6-fe17-4677-afd2-3c35b21886a6

[rapid bluff]

those are awesome! I make alot of dark comedy hip hop songs and with this V3 i like to use off the wall instruments incorporated into the beat, both of these are perfect

[fallow gull]

Sometimes if you use certain styles the obscure instruments are lost. Like adding bluegrass and you will be more likely to get fiddle and banjo.

[rapid bluff]

yeah ive learned that a handful of times, if you use certain styles you want first and generate the Interlude or Intro with no words, then continue with your lyrics, it stays pretty consistent (most of the time)

[rose ore]

so I figured out that suno thinks "midi" means midi bar. Which means that including the keyword will make the system play instruments in loops on top of each other, especially in hip hop

[fallow gull]

Ooo, I've been meaning to experiment with [vamp] to see if it will repeat a blues/jazz motif.

@rose ore BTW, I'm cautiously optimistic about your chord findings, pending more research. I generated a 4 instrumentals using the sting "Eb - Ab - G - Cm - F - Bb" and they essentially used those chords, not in that order and not always all of them, but it used them.

[rose ore]

Wanna play with a fun one? Try this
instrumental, heavy, fast, midi, sampled, hip hop,
Am - G - D - Am# FM C - B - Am Dm - A - C - G F - B - GF - Em - G

Here's what it sounded like here
there's a middle section I put in for the bridge, but then it goes back to the initial prompt
https://app.suno.ai/song/b417c036-3d64-4ec9-b9c3-e4a1805f34de

That's how I test if it's working.

Broken chord progressions

[rapid bluff]

thats so cool! does putting instrumental in the style bar and checking the button basically ensure it produces without words?

[teal raft]

Not always but it seems to work

[rose ore]

Works like 60ish% of the time

[cold zenith]

the generation is slowing down, is there a problem ? even the pictures are missing

[ionic wind]

So, uh... I just did a couple of rounds of generations where the system told me that my clips had gotten stuck and that no credits were spent, and then those songs immediately disappeared from my Create list. So I tried again a few times until it responded. But now the old "stuck" generations are showing up anyway. What happened here? Did these cost credits or not?

[final hazel]

Seems like the servers are a bit spotty.. I am sure it will be fix it soon

[ionic wind]

I don't doubt it. I'm just curious if these turned into free generations due to the server issues or if I paid credits for them regardless of the error message. I can't actually track credit use, so this isn't obvious to me.

[final hazel]

Normally (and I spend a lot of credits) if they get stuck after 5 minutes you get a brief pop up that your generation failed and no credits were taken

You only pay for generations that are succesful, and not in advance

[ionic wind]

Yeah, that's what's so confusing here. The generations were successful in the end, but I got messages from the backend about the clips being stuck and them not costing me credits after only about 15 seconds or so, and the generations were removed from the list. I've never seen this behavior before, but the eventual completion of the clips regardless makes me wonder about the credit usage involved and whether the error messages I received were trustworthy

[final hazel]

Yeah it would be nice if we have a history tab to show used credits. It was one of my personal suggestions to them. Hopefully they can implement something like that in the future

[thin axle]

Yea, there must be some update, or bug, or server lag. A few times all disappeared gens showed up at once. Just ride it out, they seem to be a pretty responsive team here. Definitely burning through a few Gs of credits, but it'll all be good here

[final hazel]

Yeah trying to ping them but they are in USA timezone so it is still super early there

[thin axle]

Yah, I'm US lol, just had 10 past gens appear lol

All good tho, its nice having a responsive dev team at all 🙂

[willow wren]

Same Problem here. This Morning it worked but now all get stucked, so i will try on later

[red mauve]

a little suggestion for THUMBNAILS :
i noticed that all my finished songs have what i call the OUTRO BIRDIE because suno always picks the last image and i usually use [outro] in the last take. how about picking the 1st one, or even more nifty let the user choose which one of all the ones used ? every suno list would look way more funky and all the songs would have a more personalized look.

[final hazel]

Please add suggestions here: #💡┃suggestions-web

[red mauve]

oh! thx 🙂

[slate epoch]

50 credits seems like nothing lately...For days Suno has given me broken songs under 60 sec., abrupt endings to first generations, and lyrical hallucinations. I've reported bugs on the many generations that were under 20 sec. long that were NOT refunded.

Today, all songs sound great and are exactly 90 sec. long but, zero lyrics. I feel like Goldilocks!

[final hazel]

I understand your frustration with the inconsistencies in song quality. I personally addopted in my workflow to add [instrumental] after my last word in my lyrics, Maybe that is something you could try too? Although this suggestion works best for paying subs because then you get the option to continue from anywhere.

@slate epoch and please don't copy and paste your message in several channels.. it is one or our discord rules

[slate epoch]

Sorry, not trying to spam. Originally put it in #💬┃general-chat but, thought it more appropriate here. Removed the other post.

Hard to understand how a prompt at the end of the song would affect the beginning. I'll try your suggestion but, no offense, a lot of prompting suggestions just seem like superstitions and not real solutions. Wish Suno would "hard code" some prompts.

Normally, I'm fine with the randomness of the a.i. and accept it. It just seems like lately there is something more going on.

...or not going on.

[final hazel]

AI is a different beast than tradional software.. I truly understand your frustration though.. I personally love to puzzle but I understand not everybody is like that

[slate epoch]

I love puzzles, but when you're given broken pieces...

[final hazel]

I have some broken pieces too sometimes.. but I do manage to find the glue to mend them mostly.. If you ever get stuck again, feel free to send me a DM. I don't mind helping you

[slate epoch]

Thanks for the offer of help and I may take you up on it sometime for another issue. My grumpiness is with the a.i. and not you.

Fixing isn't really a thing in Suno right now, espeically for F2P. Once it's generated, there's no going back and adding the lyrics (that were included) to that perfect tune. Best we do can is take a decent start and build on it while altering the blueprints along the way. This week, I've been stuck at even getting something started, let alone building on it.

[slate epoch]

I did get out a single generation song from my very last credits today....you're mentioned in it!
https://app.suno.ai/song/bc1bd5b2-4149-4bb6-b781-dd33e4fe164d

[final hazel]

I am famous! lol @slate epoch 😉

[broken lotus]

I feel like more people need to adopt the 'golf method'. Try to end the song with the fewest wasted gens

• set a maximum, and a 'par' average you're trying to beat
• (if v3, get full song and continue from time count as 1 roll)
• play the ball where it lies
• only a few do-overs, each is a penalty
• 'winning' means using fewer credits
• Bragging rights for the longest song with a perfect score
• "Hole in 1" is when you get a perfect 1st clip that ends itself at 2min.
• less whinging for the rest of us

[stable sparrow]

for the old voice this was the closest I could get: https://app.suno.ai/song/39824e67-542b-4559-947c-e1fc1142f43f

style: legendary very old low-octave male durge raspy voice,emotional, horrorcore country, guitarre

I also tried to change the dynamic a bit by incorporate [Instrumental Electronic Solo, building back to the Chorus]

[pliant cave]

I just caught a maintenance message about a backend upgrade, which then disappeared. Generations were functioning, so I assume I caught the tail end of the process. Hopefully, this means improved capacity? 😄 Gotta be ready for those pesky surges from social media streamers. 😅

Unless it was some routine library upgrade or something, lol.

@final hazel I have to say, I believe the "Mosquito Song," also known as "Buzz Buzz Buzz," is possibly your magnum opus. My family listened to it repeatedly and even shared it with friends. Great work.

[final hazel]

Thank you for the compliment.. You are making my heart melt 🫠 @pliant cave

[pliant cave]

It was refreshing to see a song on the "explore/trending" page that everyone can identify with and that is also catchy. To be candid, most of what I see on that page I tend to ignore as it's either experimental, eccentric, eclectic, or foreign language songs like: "光と音の饗宴 (Kou to Oto no Kyouen) melodic j-pop uplifting."

[stable sparrow]

when there is "Prompt flagged for moderation" how can I figure out why, or which parts need to be altered? same text worked 2 minutes ago

[final hazel]

True.. I am sure there are more gems, but people don't dare to share it because they want to monetize them and are affraid somebody steals them I guess. We all feel our own creations are the best we ever heard. Which is pretty normal. Anything you create is as if it is your own baby. And we love our own babies.

[final hazel]

Feel free to post it here.. I understand German (I saw you also asked in german section)

[stable sparrow]

i just changed the first word from hi leute to EYYY and now it works...

[final hazel]

Ai is really stupid sometimes.. no way it is going to take over the world anytime soon!

[craggy forum]

Hello, I have a question why the /chirp command no longer works

[pliant cave]

I didn't even consider that. What's interesting to consider is where the definition of "your" song begins and ends. Meaning, I can take your song, or anyone's for that matter, and continue from it to generate my own version. I could change the last 10 seconds, or I could change half of it. I could just "steal" the singer, vibe, and structure for a continuation and cut off the part that was the original song.

So, if people try to monetize or claim copyright, I'm not sure how anything could be enforced. To your point, people might just not share them and risk someone getting the song's "ID" in Suno so they can generate from it.

One option is to disallow anyone from continuing from someone else's song if the user who created it makes it public and "opts out", but that could go against the spirit of making it public and also possibly against Suno's overall mission.

[final hazel]

Yes I agree with you. I did talk to staff about it, and they are considering what is best here. Because we do have the advantage to ask a link now if people get stuck. So pros and cons

[broken trench]

hey all, kind of confused here, do I need to pay to access V3, it seems like the chirp beta channels are all still using the older version

[final hazel]

yes.. discord uses V1

Only web app has V2 and V3 alpha

[broken trench]

ahhh makes sense, thank you

[viral atlas]

@pliant caveIf I share this work on YouTube, will it face copyright risks?

[final hazel]

No it shouldn't.. but keep in mind that you are only allowed to monetize the songs if you created it with a paid subcription

If you are a free user, you should add something like "made with Suno"

You can go to the website, and click on "subscribption" to choose a plan. Be aware, the songs you created with free credits, do not retroactively fall under your paid subscription.

[viral atlas]

@final hazelThank you

[thin flint]

Curious what you all think of the track I made intended as a bed when I step away from livestreams: https://app.suno.ai/song/7908bdb2-2dd8-4f9f-803c-a97dd252bad5

Well, the first one anyway.

[broken lotus]

pleasent. ✨

[graceful condor]

is there a place where i can search for stuff like an ai prompt like how would i make acapella etc

[willow wren]

Funfact: all Attempts were generated later. 40 credits but great Results

[willow wren]

This method works great: start instrumental and continue from it with a new prompt, singing is great too

[rose ore]

Funny you mention that. Experimenting with that now

[rose ore]

one thought. It gets pretty weird when you do that, unless you specify which key the song needs to be in. Like I do a lot of bluesy country, so common ones for me are {A7} and {D4}

[broken lotus]

https://suno.wiki

(it is in progress)

[spark trail]

How do I get on the explore page

With my generated content

[teal raft]

Make your song public and share it with people that have Suno accounts so they can like it in Suno itself

if it gets enough attention it will get on Explore

[spark trail]

Thank you, I thought suno curated this and I had to submit to get on

[final hazel]

See: #📚┃suno-school message

[willow wren]

Yes, they are in the prompt as in the cited Post: e.g.
Am - G - D - Am# FM C - B - Am Dm - A - C - G F - B - GF - Em - G

[oblique stream]

How do I get the AI to make the speedcore more... speedcore

In other words how do I make it go fast and I mean FAST

[slow edge]

I have found that lately, the first 2 parts where ok. I had one where it sounded like someone was mumble complaining about something and blamed someone called Mick! (no 'Mick' in my lyrics) before the tune/style totally changed, as if the person playing the music, didn't like the style Mick wanted them to play, so he was moaning, then snaps Micks name, before then playing their own style of music.. LOL.

Others were great music, but the lyrics were so crackly or faded, you can't understand or hear them or the music becomes distorted, all instrumental, even though there were Lyrics were in the box.

The continued song after 2, the lyrics have been recently nothing much like parts 1 and 2. I end up wasting the day's 50 credits on more Rubbish that's often nothing like the first 2 parts. Yet 2 particular other full songs I've done, were created with very minimal wasted generations. I've got the perfect picture pop up that's relating to the song, but a really bad song/continuation, and then you get a good song, but a totally irrelevant picture.
When it works it's good, but when it doesn't or only part works as requested, especially when the fails rack up.

[rapid bluff]

what style of music are you trying to do? or are you just meaning in general? I did a hip hop song and wanted a verse to do a speed rap and just replaced [Verse] with [Speed Rap] and it worked after the 6th or so try 😂

[oblique stream]

Speedcore

[rapid bluff]

i honestly have no clue what that sounds like, can you share the link to what you have so far?

[oblique stream]

I can grab an example

[oblique stream]

https://www.youtube.com/watch?v=GHHOuW0paBs and skip to 1:40 :)

😂

[rapid bluff]

one sec i may have a way to mimic this

[oblique stream]

I have used Speedcore as a prompt but it doesn't quite hit the same beats

🥁

[rapid bluff]

yeah something like this it needs a more refined prompt

still rendering

[rapid bluff]

https://app.suno.ai/song/47cf2278-8f9d-42b6-9577-b380840566f5

https://app.suno.ai/song/957cd166-9994-4ad6-9f2d-f6340164b820

dont know why the 2nd one wont allow playing

[oblique stream]

150 bpm is quite slow compared to the song I sent

[rapid bluff]

was clocked at 117.45bpm but may have been the beginning portion

[oblique stream]

I think I'm onto something here though

[rapid bluff]

are you using V3 or V2?

[oblique stream]

V3

[rapid bluff]

yeah typically prompting it similar to how i have it, just adding words like energetic, extreme, experimental is typically fast beat and then a high bpm number (wont always get the bpm correct if at all) should lead you to whatever hyper song youre looking to craft (hyper a good word too)

https://app.suno.ai/song/084b5b38-1142-4ee5-a1c1-a0595248dac0

ran another for you, this one quick as hell

[cold zenith]

Those who using V3 alpha, don't forget to go there gives the report ( i've lost the link ) it's a page on discord to select a b c d e f g ( according to the problems )

[thin axle]

Re speedcore: try some related tags. like for example, for some reason it associated 7/8 time signature to prog rock, so putting 7/8 doesn't actually do 7/8, but it does push my genre waaaay off base usually

Blast beats, double bass drum, etc.

Find words that really are specific to that sub-genre

[final hazel]

I have to admit.. not really into speedcore.. but I was trying.. is this fast enough to be considered speedcore?
https://app.suno.ai/song/e58d689f-a6ca-4f4c-bd32-154298ed38b2
https://app.suno.ai/song/c5c3ed54-cfd2-4a64-a20c-6cd42b174352
https://app.suno.ai/song/251d2d5a-1f6f-4e9b-b609-e06538092d37
https://app.suno.ai/song/c38fec75-2634-4e3e-b3a9-dd9ddb1c66db

I guess not lol..

[cold zenith]

I never heard about speedcore

[final hazel]

tbh to me it sounds like electronic music (techno, gabber, hardstyle) but then with double playback speed lol..
I just would throw it in a DAW and speed it up 😂

[final hazel]

Can you provide a screenshot?

That's not a song created in the web app. It says V1

In the web app you can only do that with V2 and V3 alpha

yw 🙂

[limpid badger]

What's the best way to get a continue ending song that's very brief (5-10 sec) and only instrumental?

[versed valley]

Welcome! For help, check out Suno Help

[cold zenith]

Finally made 3 musics i've got 1700 credit left over 10 000.... 4 musics left wish me good luck hahaha

[thin axle]

Delete all lyrics and put (ending) (instrumental) and in genre notes, do similar thing. Like “END THE SONG IMMEDIATELY “ or “End and fade out, no vocals”

You may still need garage band or audacity to cut length down or fade out.

But that’s how I get my songs to end. It’ll still take a few gens, but that type of approach.

It looks at what it’s continuing from in addition to the text inputs, so no need to give full description to keep style consistent for a short gen or ending.

Few times I’ve had success flipping on instrumental mode, but it’s been hit or miss for me, when the above approach has been a hair more consistent for me at least when continuing

[oblique stream]

The speed is correct yes

[limpid badger]

Yeah I did all that. I got one generation that worked. I used style "Ending", "Short ending" etc. Seems hit and miss.

[final hazel]

Mine is pretty simple and works for me 9 out of 10:

• remove all your lyrics
• remove all words in the "style of music"
• put in the word "end" without quotes in the "Style of music"

[oblique stream]

Putting ‘finale’ in the prompt works 9 times out of 10 for me

[limpid badger]

Ok thanks. It's probably my mobile phone workflow that was hampered. I thought that when I clicked "instrumental" all the lyrics disappear. But of course they don't. They just get hidden.

[final hazel]

https://tenor.com/view/homer-simpson-hide-in-shrubs-hiding-in-bushes-disappear-into-hedges-embarrassed-gif-24183287

[limpid badger]

Like Homer!

[storm sierra]

Is there any way to use the same vocalist for different songs? are there commands for getting a certain vocal style?

[thin axle]

Yah, it’s just not at a level where you can truly strong arm it when you want if that’s not where it wants to go, but ya get as close as you can and something starts popping

FWIW there seems to be some cache as well, because sometimes it takes 2-3 gens to clear out where it wanted to go before it’ll start doing what I’ve written in

[neat nexus]

I blame @coral mason for this...

His work inspired me to seek out some hymns and see what I can do...

https://app.suno.ai/song/e583faf6-2bb8-4fce-be53-384cdd753b37

It's always a good song when Suno decides it needs some applause. 😁

[runic dock]

What these (...) are supposed to do?

[neat nexus]

They're spacers that based on my personal testing combined with other factors help space out the music so it doesn't jump right into vocals. I also did a start from time using v3 on the first clip segueing to vocals.

[tribal marlin]

In the Valley of the East: https://app.suno.ai/song/77cc1c2e-ad22-49b9-bbb8-db838aa34acb

Haha cool, finally got something in trending 🔥

https://app.suno.ai/song/32573af6-3f76-419d-a525-986a32f3d7b2/

[pliant cave]

Hey @open magnet , I know you are in ML, but sorry, I couldn't remember which one of you guys on the team was in UI/UX. One possible suggestion/question, now that we are in the era of V3; consistently producing better outputs and fewer "trash" generations, would it be possible to include an input box or even a toggle in the custom mode UI that allows us to specify how many generations we get back in one request?

Meaning, if users want to be more experimental with different styles or lyrics but don't want to use twice the credits on one song, there could be an option to just generate one song and use 5 credits instead of 10. Just a minor quality of life change that could go a long way in giving users more flexibility in how they spend their credits and approach their creation strategy. Right now, users are pretty much locked into sets of two. Not that I've seen many complaints, but it makes sense to me to offer an option here.

[teal raft]

The devs mentioned this last night specically in general, let me try and go find it

[pliant cave]

Oh no kidding, thanks.

[teal raft]

in regards to 1 generation vs. the 2 you currently get

yeah hold on

[teal raft]

#💬┃general-chat message

It was related to the fact they changed the simple mode to give you similar generations like you get in custom mode, vs. the very different from each other generations you'd normally get in simple mode

[pliant cave]

Right, I swear I remember simple mode working that way, so they must have changed it. According to Pat's comment, there actually is a technical compute resource reason to do batches of 2 versus a simple UI/UX change, which I didn't expect to be the case.

Then that begs the question: What is the most efficient number of generations to use on a GPU if one generation isn't efficient, and two are, even three?

[teal raft]

Dunno! They decided on two for some reason to start

Pat said they did change it last night, should be above that comment they said about the compute

Pretty close

[fierce sage]

It seems that when I use the simple mode, the continuing doesn't work very well, even if you get a cut off ending. The continues are where I think V2 is probably still better, but don't try continuing V3 with V2, as one will end up with something that really won't go together.

[cold zenith]

continuing V3 with V2 it's an downgrade and the sound doesn't match at all

[fallow gull]

Even continuing v2 with v3, the v2'isms are continued through. The v3 part definitely sounds better, but not as good as if it was created from scratch as v3.

[cold zenith]

True it's a small upgrade

[fallow gull]

I just tried the Generate Lyrics in Suno and it gave 2 verses and a chorus. That seems new.

[pulsar cairn]

Is Chirp not working?

[storm willow]

My wont generate anything why please

[idle lynx]

this was a cool song , is there a list of commands list in your song structuring e.g. (outro with synchopated groove) or a you tube vid that explores change ups in the song

[cobalt lichen]

Hi, guys rate it, its Melodic dubstephttps://app.suno.ai/song/a42304c5-59c9-4886-a291-c390b67697f1/

[hearty beacon]

Wanna know it too. The continuity is dope.

[regal chasm]

Chirp doesn't work, when I ask to create a melody it says "Accepted terms of service" and nothing else happens

[crude hull]

how accept terms?

[wintry hare]

Me too, is it a bug or something?

There's a new website version available, maybe it's connected with that and accepting the new terms of service?

[dull root]

likewise impossible on the site to generate any sound

[dull root]

how link of the new site i want try this

[formal ibex]

Is it dead?

[final hazel]

https://tenor.com/view/funeral-flowers-grave-gif-5518042

[final hazel]

Try https://www.suno.ai/. It's better anyhow (V2 and V3 for paying subs while discord is just V1)

[formal ibex]

Generations on site don’t work either

[final hazel]

Works for me. What’s going wrong on your end?

More details please

[formal ibex]

“Create” window just doesn’t open, like black screen there

[final hazel]

Have you tried a different browser? Maybe you are using a plugin that’s blocking it?

[formal ibex]

Have tried everything 🥲

[final hazel]

What is everything? Also VPN?

[tribal marlin]

Note: this approach does appear to add a lot of AI noise to the song

Most of the changes are actually through doing continuations. You'll notice I have lyrics and metatags that don't end up being used because I cut the song off before it gets to them.

To change to a totally different style, I will start the continuation with "[instant switch]" and change the "Style Tags" completely.

To contruct the song overall, the approach I have been using is keeping the "start" of the generation (so it might be just a single section, like a verse for example), and then I will start a new continuation 2-3 seconds before this section ends. That way when the new section comes in it gives it a few overlapping seconds to throw a drum fill in etc.

The purpose of this approach is to get more control over the individual parts of the song.

(Note: A downside of this approach is you get more and more AI noise in the song... which becomes especially noticable towards the end of that track)

In most parts I went through a lot of generations until I was happy 😆
That said, sometimes you do get good generations on your first couple of tries

[tribal marlin]

In regards to the metatags they don't always do what you want (or when you want it 😆 ), but I think if you keep them simple and you do multiple generations you can end up getting "close" to what you are looking for.

In regards to a list of tags, I'm just making them up as I go lol. There is some info on the tags here:
https://suno-ai-wiki.notion.site/Metatags-22a69f63b1814aa9afbec9d21f7b55c4

I always forget to use things like [moody Verse] instead of say [verse]. I want to experiment more with that.

[formal ibex]

I didn’t try vpn

[cold zenith]

What do you mean by [instant switch] ?

[tribal marlin]

the AI usually will take it to mean like "there is going to be a change here", it can strengthen the chance of the AI knowing that it needs to swap the style

I saw other people mentioning that here so that is why I used it (and it does appear to often work)

[cold zenith]

i use [BREAK] for that usually

[tribal marlin]

ah maybe that is what I should be using lol

though isnt a break slightly different?

like a break might mean the music mellows or something?? or do I have that wrong

[formal ibex]

I’ve tried vpn now and different browsers. Still COMPLETELY dead. Just see black screen with “uploading” white circle and nothing is happening after

[cold zenith]

break stop the music and bring it back with a different style

[tribal marlin]

i might give it a try then

[cold zenith]

make a space above and below it

[final hazel]

Did it ever work for you? It does seem to be on your end since it works for others. More details are needed to be able to help you.

[north arrow]

Hey everyone! I’ve been playing around with some songs generated by Suno AI (awesome stuff, by the way!) and I had a couple of questions. Is it cool if I cut and rearrange parts of the songs to fit my preferred order? Also, can I extract stems from one song and use them in another song I generated? I am asking since Suno doesn’t allow stem download, and technically it would violate the EULA as "reverse engineering"

[formal ibex]

I tried to submit generations via discord yesterday (for the first time), usually just played with ai with visual, and wanted to try some music. But it just didn’t do for any type of music I tried. Here in discord it’s just “uploading” after creation of lyrics(I use Chatgpt window), but no results. And at web version don’t even show me correct window with “create” options. Just black screen. Tried different browsers, vpn. Nothing

[final hazel]

Have you use discord to login? I think you need to verify an email first. @formal ibex

https://support.discord.com/hc/en-us/articles/213219267-Resending-Verification-Email

[fallow gull]

There are many people using other tools to remix and remaster songs, including using tools to extract stems. Many of the ones seen here with links that point to YouTube or SoundCloud do that.
I'm not a lawyer, but when I looked at the Eula the term "reverse engineer" seemed to apply to Suno's software and algorithms, not user generated songs.

[north arrow]

Thanks! Good to know

[formal ibex]

Yes, I did all. Doesn’t work 🥲

[final hazel]

Please send a mail to [e-mail]
I personally am a volunteer guide and personally clueless on this point. Sorry I couldn't figure out with you 😦

[formal ibex]

thanks for your help anyway

[wicked matrix]

Hey i just tried the /chirp command on the chirp-beta-1 and i filled it out, then it told me "Accepted terms of service" but how do i find the AI generated song?

[west lark]

I’m unable to generate. My prompts say they are generating but eventually disappear. I have plenty of credits and an active subscription. I have closed and reopened Discord several times. I’ve tried on two systems and still no success. What’s next?

[idle lynx]

it appears my requests are jammed, i get no chirps and it says i have too many in queue .

[final hazel]

Yes there seems to be a problem with discord generations. I suggest using the website which has V2 and up. You might like it better since it has better quality, longer length etc

https://www.suno.ai/

[west lark]

Thanks for the advice

[normal radish]

Moving a discussion from #🪲┃bugs-web to here.
What length of song is considered too long or pushing the limits of Suno? Personally, I have never had any errors or problems Continuing and using "Get Whole Song" on songs up to about 8:30+ in length, with 10+ segments. But apparently others have had issues.

Is duration (time) the problem, or is it the number of continuation Parts? Some of my mid-song Parts have been as short as 3-6 seconds; sometimes a few of those in a row. Which can make the Part count add up fast.

However, I usually do "Get Whole Song" on each Part 2, then Continue from that Full Song. I almost never generate multiple Parts (3, 4, 5, 6, etc.) before combining them. Is that a best practice? Does that matter at all?

The number 1 problem I encounter with long songs is that Suno gets extremely repetitive / overfitting. I speculate that it's because the prior minutes of song data start weighing heavily relative to the prompt and railroading the output, especially if those prior minutes contained repetitive patterns like in EDM. But I also suspect that the "lookback" window is short (~1 minute), so I'm not sure.

[humble swallow]

Please reset my queue too

I hope that they will reset my queue too

[ripe perch]

i'm stuck at generating
what can i do ?

[wet minnow]

use the website instead of discord... it uses a newer version https://app.suno.ai

[ripe perch]

i tried it about an hour ago
it said "too many requests"
but now its working, thx

[grim steeple]

not wokring

omg its stuck for about 4 hours

[wet minnow]

just look at the 6 posts above you 😉 ... use the website instead of discord

[thin axle]

I’ve started doing the same, it retains “memory” better creating full song after each extension.

That being said, I’ve never had issues with number of segments, some songs easily in 15-20. But that’s usually under 5min song, and using lots of short chunks to get what I want.

Only issue with longer time I’ve found is that it doesn’t have more than 1-2 minute look back (experiential conclusion only) so it starts to really lack consistency. Unless you’re doing something that doesn’t require that.

Not saying some haven’t experienced that, but I haven’t. At least never anything programmatically past a certain time or segment duration.

[final hazel]

See my video: https://youtu.be/5uJw5kb05as?si=gGiiztZWEX9rFmV-
😄

[odd ridge]

Is it possible to somehow explicitly indicate stress in words? I tried different options (through using accented characters, with an extra ’ after the letter, and with Uppercase characters that need an accent), but it didn't work

[oblique stream]

Does the title of songs affect the output?

[teal raft]

The song title, that you use the 'rename' option for, does not have an effect on the output

Only the Lyrics box and musical styles

[kindred minnow]

for suno 3.0 is 2 minutes the maximum or is it possible to create a fully fledge 3 minutes song%%

??

[indigo ether]

You can click the 3 dots and then continue from this song and it gives you two choices to extend it a bit, you can keep doing this from the last extension, when you are done, you can click the option to put the pieces together into a full song.

[abstract grotto]

Help me please

Anyone

Hello please help

[barren shoal]

hello, i wrote an issue in the #🪲┃bugs-web channel, if someone can help me because i dont want to waste credits, thx

[fluid pine]

can anyone explain how this tags working?
[e D]
[C H a H]
[e D]
[C H a H e]

[kindred minnow]

I dont know guys, i love using all those tools for the past 2 years in AI but its becoming scary good... almost too good to be true.

[barren shoal]

tips to avoid incorrect spelling of words at the end of part/ beginning of part 2 when generating?

[broken lotus]

Hey, there are a lot of community tips collected here:
https://suno.wiki

Can't anticipate every issue, but we've been sharing what seems to work...

[runic dock]

How?

[barren shoal]

thank you 👍

[gusty saffron]

Is there any list of tags somewhere? I found one earlier, but they don't really seem to work. I'm trying to make a duet where a man and woman sings interchangeably using [man] and [woman] tags, but it just completely ignores it. perhaps it's not possible at all?

[broken lotus]

https://www.suno.wiki/faq/style-and-lyrics/duets/

[buoyant osprey]

Could I get feedback on this song? I wanted to do pop and rap again. https://youtu.be/ZfeXOhjH7HU?si=OCkzecPsi7FTCCeV

[fickle tree]

hello, where can I find a list of chirp commands and sumo prompt examples list?

[normal radish]

As far as I know, it is not possible to force Suno to emphasize certain words or use any specific cadence. The best we can do is increase the probability of getting what we want. I use a couple tricks, with mixed success.
To flow words together, I often hyphenate them. "I-don't-know" Or, I spell them out phonetically (sometimes with hyphens). "I-dunno"
To enunciate words and make them stand apart, I add a period between them. "I. Don't. Know." Or, I put them on separate lines.
"I.
Don't.
Know."
Sometimes, I even break one word into multiple syllables, if I want the AI to sing it that way. "Some. Bah. Dee. Love me-ee."
Nothing works 100%, but these tricks seem to help.

[runic dock]

Please Help getting the intruments/music styles of this song at least for the intro
https://youtu.be/huFra1mnIVE?si=k88kQtorqVlbt9X2
@final hazel
Thanks

[final hazel]

try
worship, Christian music, Christian alternative rock, ccm

[runic dock]

Got it. Will try this thanks
What ccm means?

[final hazel]

Contemporary Christian music

[runic dock]

Thanks for your support. Appreciate.

[kindred minnow]

on the web site, how do i use the remix function with version 3.0 alpha

i know how to continue a song, but the remix dont seems to work

[final hazel]

remix is reusing the prompt from a previous song. It more like "reuse prompt".. it doesn't remix in the literal sense..

But do you know this already?

The continue from "anywhere" feature?

[split vine]

Suggests for what style to use to create a cheer?

Yes, you read that right...like a 6-second teenage girl volleyball team celebration between plays. I assume this type of "music" is outside the training set, but why not try. I'm having trouble getting it to:

• leave instruments out
• keep it short (even with very short lyrics)
• make it sound like a cheer (just non singing voices with claps)

[never in my life thought I'd ever ask a question quite like this 🤦‍♂️]

[fallow gull]

https://app.suno.ai/song/6879a801-1e77-4747-a0fe-1b0ca1c7f502

[fallow gull]

This is not at all what you want, but also kinda awesome:
https://app.suno.ai/song/5598edd2-0662-4571-9cc1-2328f1948bcb

https://app.suno.ai/song/48b0fc93-67d8-42ba-ba54-75658bcf9632

[random abyss]

how well is Suno on the level of understanding Chords on a technical level?

can it understand these chords: amazingly? or okay? or "it's not the best it could be"?

[kindred warren]

Misreadings occur in Arabic texts. How can I solve this?

[wise jasper]

I do [Instrumental] [intro] watching your video and checking it out. Thanks for the video.

[neat nexus]

You could try phonetic spellings of the words that are not being sung correctly.

EX. A while back it was saying lives as in number of lives in a video game and not lives as in lives in this house. So I changed the word to "leaves" and it sounded close enough to work in the song.

[broken lotus]

what happened when you tried it?

[inland kettle]

can we make an api for Suno for external websites like the GPT

[arctic magnet]

Hi. Just registered a free account and looking to use this more of a creative inpiration tool rather than doing everything for me. I’m looking to find how to download the song directly and can I download specific stems like just the drums and vocals so I can add my own guitar?

[neat nexus]

You can download segments or whole songs by clicking the ... menu and selecting the option download -> audio.

There are as of right now no stems but ones you make yourself.

A lot of people have luck with ultimatevocalremover.com which is a free tool. There are others but they cost money.

[arctic magnet]

Thanks it’s a shame as I would be hugely beneficial.

Also looking for help with prompts. On the custom option I can enter my own lyrics but then only have the option for genera? Can I choose the key or bpm? Or even my own progression?

[neat nexus]

You're not alone in that.

[neat nexus]

Not at this time, though some people have had luck with it in their style prompting.

[graceful condor]

歌曲类型有哪些

[neat nexus]

你能想到的任何东西。 （用谷歌翻译写的，如有错误，敬请谅解。）

[arctic magnet]

Where would you put the prompt? In the genra bix?

[broken lotus]

you might search the discord for 'chord progression'
this comes up a lot..., and honestly the results have never been the actual chords/notes...

SOMEthing happens, but it's like the AI has understood the instructions (or is pretending it understood) but has no clue how to correlate that with actual reallife notes and chords...

But people come up with different methods...

So maybe it someday works

[hardy basin]

I see a bunch of tracks in the Explore with keywords in brackets, do those do anything or is it just people hoping the model follows it? Is that documented anywhere?

[final hazel]

Those are songs from the community. Sort of charts.. And you can also learn from them

[hardy basin]

yeah i understand they are from the community i meant in the lyrics they list things like [guitar riff] -- do those do anything?

[final hazel]

You can listen for yourself... And if they use them then there is a chance they do.. Thing is AI is still a new tech in this way with music, and it is a huge model who still has sometimes a mind of it's own. Total control is not yet possible if you want to know that. 😉

Maybe our community website helps you: https://www.suno.wiki/

There you can find things like "metatags" and how to use them in the lyrics

[hardy basin]

ooooh thanks. yeah thats what i needed. metatags are a thing! awesome.

[rose ore]

okay, so technical question: If I wanted to script a complex duet like Wild Ones by Jelly Role.
Can I do that in an organized way in Suno?

[fair moth]

Is it possible to change the lyrics of a generated song? The model generated a song I love musically but the lyrics are terrible, so I'd like to change those lyrics while keeping the song in-tact

[final hazel]

toggle the custom mode

Oh generated no

But you can create your own lyrics in Custom Mode

There you can also click on "random Lyrics"

And you will SEE them before you click on Create

So you tweak them to your licking

But best results is with good old fashion elbow grease and putting in your own lyrics...

Then it has more "soul".. but that is a matter of personal taste

[gray bough]

does any know how to contact Suno? I paid for the year but is asking me to pay again, I have been email for 2 weeks to: [e-mail] [e-mail]; [e-mail] [e-mail]; Suno Billing [e-mail]; Suno Billing but no reply, does anyone know of any contact numbers of emails, please?

[neat nexus]

You have done what you were supposed to do by emailing [e-mail]. The other emails were not designed for that use. I have escalated this by linking this post in the Moderator's section. Please do not send mail to every email address as this is a good way to get mistaken as spam. (Not saying this is what happened but it's never a good idea to do this because that could happen.)

[gray bough]

thank you so much, I trully appreciate it

[arctic magnet]

Is there any prior suggestion to have a duet with male and female singers?

[mossy yew]

so how does the explore section of the suno website work? how do you get songs into the new section? is it just by making them public or do you have to share them elsewhere? cheers

[mossy yew]

how do you get songs listed in the new section? do you just need to make it public? or do you need to share it on discord somewhere? and if it is public does that mean you release the rights to your lyrics etc if someone decides to remix it?

[broken lotus]

Yes, if you make your song public it may end up on the Explore page, and any other Suno subscriber can Remix or Continue From on your song.

[broken lotus]

https://www.suno.wiki/faq/style-and-lyrics/duets/

[rustic mauve]

Is InPainting possible? Sometimes when it misspells a word but the later part of the music is too good, it kinda sucks to let the clip go just because of one little mistake. Is there a way in suno or some other tools to do something like what's possible with image inpaintings?

[teal raft]

Currently is not possible, but boy would I, and lots of others, like it

[neat nexus]

Indeed...just a few word changes here and there on a few songs...nothing major...

Still can't believe I made the silly mistake I did that completely undermined the whole point of the song..

[rustic mauve]

🙂 https://arxiv.org/pdf/2305.15266.pdf found this paper but found nothing useful. Thinking to build this thing myself. Somehow 😃

[teal raft]

Someone earlier made a song that said 'rule the word' instead of world

They didn't notice

But I did

[rustic mauve]

Exactly.

[naive fulcrum]

Is there currently a way to remove vocals and have an instrumental only version of a song I have made?

[teal raft]

Not in Suno, but you can take it OUT of Suno

and use something like https://ultimatevocalremover.com/

Or there are various other systems that can do it

Strip the singing out of a song you've made

[naive fulcrum]

Ahhh I’ve tried this already and the songs always have strange audio artifacts or the mix dips where the vocals should be

[teal raft]

Yeah that tracks

Since you're trying to go backwards from a full master

There are other online solutions, perhaps they use different models

[naive fulcrum]

I see, it’d be cool if Suno adds a feature to export as stems in the future

[teal raft]

I've had good luck with UVR5 in my testing, picking different models and messing with it that way, but maybe I like garbage

[naive fulcrum]

Alright I’ll give it another try, thank you for your help

[teal raft]

They hear that all the time, and the way the devs seem to groan makes it seem like it is somethig they want to do very much, bit is very hard technically with how Suno probably works

no prob

[naive fulcrum]

Also, does anyone have any tips to ‘lock’ a specific type of singer so my songs have a consistent vocalist?

[teal raft]

Nope! Not in Suno

Again, you can do that outside of Suno

let me find the video

I was sure there was a music-centered instruction set, but it uses the exact same software. Something like RVC
https://www.youtube.com/watch?v=nXpBlC6OBw4

You'd run into singing style though. The way Suno works, right now, is each song is a new generation which means totally new singer.
There is no way to re-use a voice.

Not right now.

[sudden lantern]

can we select male or female sound in suno ai

[limpid spade]

Generally a single metatag male vocal or female vocal will work for an entire song. Duets or switching from Male to Female are more luck than good prompting though.

[undone cypress]

uhh i just had a really weird experience. first suno started generating what sounded like old cd skipping artifacts from a bad disc, and then the generated track went on for 4 minutes instead of the standard 2. not sure if these two are correlated lol. i thought it was genuinely mimicking cd glitches at first lol

[broken lotus]

This might have been the audio clip ‘streaming’ as it loads.
It is probably not more than 2min, now it has finished streaming.

[undone cypress]

oh, you're right now that i'm looking at it retroactively. it must have been a ui bug then because the timer genuinely went up to 3:55

[broken lotus]

Yeah…, the web ui doesn’t always keep up with what the AI is doing…

[undone cypress]

but i distinctly heard those "hiccups" that you would get from playing cds and not just silent gaps

[broken lotus]

It probably started playing again from the beginning… maybe more than once…
(That’s how it played for more than 2min…)

I like the streaming as it loads…, but it gives some confusing info)

[north arrow]

Anyone had success with speech to speech/speech conversion tech to get a consistent singer? I tried Eleven Labs, but it jumbles up words and the results are super weird. Saw RVC mentioned here earlier

[mild basin]

Hello! Newbie I am, and no matter I entered my lyrics, the result was instrumental only. Though the lyrics was Hungarian, I don't know how to specify language. I know Suno can sing in Hungarian.

[neat nexus]

Bad luck happens.

But formatting your lyrics differently can help. Are you structuring the lyrics? [Verse] [Chorus]

Do you have some kind of mention of vocals or vocal styles in the style prompt? This can help sometimes.

[mild basin]

🤘🏼 It was structured but probably too long (3x4 lines). Then I learned I can add one verse then add another by clicking to Continue. Now with a 4-line lyrics, the singer arrived but it tries to read my lyrics as it was English. How to specify the language? I heard Suno singing perfectly in my langauge. Thanks for any help! I am so excited that Suno may songify my childhood poems. 😄

[neat nexus]

I've only experimented with one or two languages. Those times I just put the words in the language and as far as I know it sang them in that language. So it should just work. You can try priming it though by putting Hungarian vocals in the style prompt to see if it helps?

[untold briar]

Here's a fun tag. I'm trying to use **[incomprehensible] ** for *targeted * insertion of hallucinated lines of lyrics. So you prompt one real line, hallucinate the next, and so on. I think 3 tags per line may work a bit better.
https://app.suno.ai/song/1776f434-0562-448d-9437-2775e87d5615/
https://app.suno.ai/song/30578980-8373-4a85-81e1-976f9ba10072/

[shut tree]

Devs, did you fine tune the musicgen models with this new system?

specifically the MIT meta/audiocraft music model

[lament bison]

Hello everyone!

I'm trying to achieve a unique sound using Suno.ai that resembles the ancient, Nordic folk atmosphere of bands like Wardruna and Heilung. However, I'm facing challenges due to the limited selection of instruments available (for example, the tagelharpa or the shaman drum). My goal is to at least come close to this sound in my compositions and use the available instruments in Suno to mimic the vibe of the tagelharpa and other authentic instruments as closely as possible. Does anyone have any tips or tricks on how to create a deeper, ancient atmosphere with the tools at hand? Perhaps some of you have experience with specific commands that could help me approach this unique sound more closely?

I appreciate any help and advice in advance. Thank you!

[sturdy flax]

Hey frens, best way to repeat a unique measure in a song? For example , it gave me a full measure of a very uniqe beat that it basically wants to use as an intro , but I wanted to repeat it as the verse.

However every time I choose to continue it from that point it just goes into some random next part of the song instead of repeating.

[final hazel]

Repeating a specific measure in Suno can be challenging if it hasn't been explicitly labeled or called for with a metatag (or lyrics). If the AI isn't recognizing the measure for repetition, you might consider editing the track in a Digital Audio Workstation (DAW) to manually repeat that section.

[broken lotus]

are you adding lyrics...?
I think Lyrics are the strongest part of the 'prompt' sometimes... overrides style and metatags...

Instrumental might be easier, or starting with an instrumental and then switching to lyrics once you have the sound...?

[static wigeon]

Hey Suno fam' ! i hope you're good ! i've a little question, can you tell me what you would use to have the same style of voice on chirp? the prompt ? i really dont know what i can write please 😭
i try glitched voice, autotuned glitch etc... idk :/
https://www.youtube.com/watch?v=DNpidNsSWzA&ab_channel=Jima

[sturdy flax]

Ty for the reply, I've definitely considered this but was just checking to make sure there wasn't some sort of trick I'm missing.

Right now v3 seems to have a tendency towards a certain genre sound when you're asking for metal even when you're prompting out for a very specific genre

[hallow sinew]

Is the Suno V3 model open / does it have a nAPI to play with on ReplicatE?

[final hazel]

No.. api is on the roadmap. There is no ETA known for us users.

[lament bison]

How to add the prompt with Lyrics?😮

[broken lotus]

use Custom Mode...
that gives 2 prompts: style/genre and lyrics

[lament bison]

Yes, I understand that. But when I bend the genre section to use tagelharpa, it doesn't. Can it be instructed directly in the Lyrics section? Can you give me a specific example of this?

[final hazel]

tagelharpa made be too rare and niche

Bagpipes are for instance too

[broken lotus]

here's some links to walk through how the metatags work...
https://www.suno.wiki/faq/metatags/

[slow shore]

is there a way to continue the same voice of the person im using for my first song i made i want to make albums lol

[final hazel]

Sadly no...

[slow shore]

like my post i know they have the smarts to implement this all

[final hazel]

They are smart.. and they will if it is possible 🙂

[slow shore]

ohh it is anything is possible with this tech haha

[buoyant cargo]

What models and settings are you using for UVR? I haven't managed to get anything usable out of it

Or any other stem splitter for that matter. With Suno tracks at least

[neat nexus]

This is the one I generally use. It's slow but seems to do what I need it for.

[buoyant cargo]

I've used that in conjunction with the VocFT model with mediocre results. I'll have to try that one by itself!

Thanks

[lament bison]

Metatags it's not working...🫤

[broken lotus]

do you want help?

advice? Suggestions?

[dense owl]

hi guys, my suno songs always ended abruptly. any tips to have a nice fade out?

[lament bison]

Yes. I've described my problem before. I want to add extra instruments to my song, like the tagelharpa and the shaman drum. It won't even add the simple violin and I use the [ ] sign. Why does it not work?

[broken lotus]

Have you added these instruments to your style prompt?

[eager topaz]

Try: [Fade Out 0.5] [Song End] it mostly works for me

[eager topaz]

I use RipX DAW Pro, which does a great job, but it's not cheap, although there is a free trial:
https://hitnmix.com/downloads/

[buoyant cargo]

Awesome thanks! Just curious, have you compared it to any other specific splitters?

[eager topaz]

Add Violin in your Style and then wherever you want it in your song in [] It usually works for me. Try and find an Indian drum, perhaps Banga and try that. With Tagelharpa, try Harp or Lyre. Add it in your style, then in your tags [] . Experiment!

[eager topaz]

Yes, none compare. The editing in RipX is like nothing else, pinpoint accuracy, I even copy and paste the smallest piece of a word, even a single letter. I've used it to copy a chorus when Suno hasn't made all choruses the same, as it does sometimes.

[kindred minnow]

When using the V3 on the website, how do i use the Remix function?? An another question. V3 is greally great and consistant to create a full song, the only problem i see is sometime, the more parts you generate, the quality becomes lower in the later parts.. not really something that anoying because its still sound good, but i was wondering why this was happening, i mean how if the ai s able to keep the complexity of a melody adn create thing so amazing but are a bit weak on the sound quality in the later parts? I notices its often in a song where there is power full signer

[buoyant cargo]

Sweet thanks! I'll give it a shot then. For choruses I usually just copy and paste the audio in it's entirety with the instrumental

[neat nexus]

All "remix" is as of this time is a re-prompting command.

If you are already in custom mode, it copies over the text from the lyrics and the style so you can try another version of it. It does not retain any of the original flavor of sound or vocals.

Some of us think it should be renamed to be less confusing.

[eager topaz]

Yes, you can within your DAW or audio editor, but sometimes hard to do it with pinpoint accuracy.

[kindred minnow]

@neat nexus ok thank you

[drifting parcel]

Anyone has any prompting tips for Suno skipping words in lyrics? like no matter how I format them it just skips them sometimes
Is it something I'm doing wrong or just limitation of the model, does the word skipping count as a hallucination?

[barren spear]

I have part 2 and part 3, but when I get the whole song, i can get only part 1 + part 2 or part 1 + part 3, how can I mix just part 2 and 3 togheter picking them from library?

[limpid spade]

I hope V3 final release has "Remix" changed to "Reuse prompt" or something similar.

[undone cypress]

exploring the boundaries of this as a sound design tool. lmao at the results
https://app.suno.ai/song/cb51d3e5-3c52-439e-b793-a4fc2265f7e0

[lofty merlin]

Hey, everyone, quick question if anyone has any tips for this (using Chirp v3): I had a clip with a chorus in it. Then I continued it, and the intended continuation begins with the same chorus before moving to the next verse. Except Chirp refused to sing the second chorus (first chorus at the start of the second continuation). It just kept skipping the entire chorus and jumping straight to the new verse. I regenerated like 10 times, wasting credits, and every time it skipped the whole chorus. I even tried changing some things to no avail -- changing the [Chorus] tag to [Chorus x2] or [Chorus -- Important!]; or adding (Ooh) (Yeah) before the chorus, which it did sing, but then skipped the chorus again anyway.

Is there a trick to forcing it to pay attention to lyrics that it's stubbornly skipping, without changing the lyrics themselves much?

[lofty merlin]

It's a type of hallucination. I've noticed v3 seems far more prone to this than v2. If anyone figures out tips to avoiding it, I'd like to know, too 😄

[kindred minnow]

I have a question on what is the music Suno Is trained on??? When i listen to what it can do it feels it had absobed every single music in the world to a high level

[eager topaz]

I agree about the music of the world! The more I use Suno, the more impressed I become. I just can't imagine how a machine can do it!

[eager topaz]

I've had some success when that happens with putting in [Pause] or [Break] , but not every time.

[knotty kite]

Lol yeah its pretty cool. Enjoy it while it lasts… the napster of a generation

And by pretty cool i mean the coolest thing ever

[lofty merlin]

Nah. Napster shut down due to lawsuits for copyright infringement, since they directly copied and shared copyrighted works. AI training is not copying, nor does it violate any copyrights; the few major lawsuits on the topic that have been settled so far ended up with judgments that agreed on that front. And I can't imagine any judge or jury, given an accurate and clear explanation of how AI training works during a trial, would overturn that precedent; it would be inaccurate, for one thing.

(Point being: I don't see Chirp getting shut down.)

[onyx swan]

Give this gpt I made a try, type any song and artist and see what happens, I like these results so far https://chat.openai.com/g/g-utsMc0LBR-essence-explorer

[static wigeon]

What can I write to have a voice tone similar to Fase Yoda please? in [...] idk what i can write 😦 like the tear in the voice of fase yoda

[fathom steppe]

Chirp bot command not working for me

/chirp

[flint haven]

The /chirp command is not available in this channel

Go to the #🐣┃chirp-beta-1 channel to use the /chirp command, and check out the #🐣┃how-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.

[fathom steppe]

Still no dialogue box on beta-1

[flint haven]

The /chirp command is not available in this channel

Go to the #🐣┃chirp-beta-1 channel to use the /chirp command, and check out the #🐣┃how-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.

The /chirp command is not available in this channel

Go to the #🐣┃chirp-beta-1 channel to use the /chirp command, and check out the #🐣┃how-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.

[limber gull]

My Gosh... I wasted 500+ points on that. And even here it fades away xD
Suno is unsteerable at all: "You asked for an... intense blast beat, yeah? Okay! Got you covered, mate! Chill synth-pop it is! By the way, 'intense' is a musician's name, so I will not do that. Good luck!"
https://app.suno.ai/song/298e5b6b-cde0-4d10-b8c8-7fb6c54f020c

(And no, this is not just with this genre - but a problem in general.)
You need to add weight, special commands (example: [backvocal: #something]) and so much more. For now, it kinda follows the first part and then - "Bleh, I will do whatever I want here! But with your points."

[zealous raven]

Is there an outline of instructions that the AI is trained on that we could put within brackets in the lyrics and it would convey style, chord progression, key, description of the vocals etc

[radiant hazel]

is the server down? i am unable to use it

[limber gull]

Ha... kinda. But AI follows them very not very.
[start]
[Intro]
[Verse]
[Chorus]
[Bridge]
[Pre-chorus]
(record scratching) *if you put (la-la-la-la) here it will be performed by backing singer(s) or (bird song) you'll get just that.
[Drum Solo]
[Guitar Solo]
[Intense Guitar Outro]
[solo vocal, outro]
[Outro]
[Fade Out]
[End]

But again, it may finally follow that in 100+ tries if this is a continuation of the song.
But the first time... it may follow more-less fine.

Other stuff like instruments, vocals, rhythm, etc. better to put into the Style of music input. Or AI may and will sing them XD

[cold zenith]

what would be great is to be able to upgrade our V2 music to V3. Because there are some I'd really like to upgrade. But even when I do continue, it's not up to the level of V3.

[kindred minnow]

can we do narration on the website

[tribal marlin]

As far as I can tell, the more your lyrics are balanced out the less this occurs. It can still happen regardless, but certain lines seem to flow better (perhaps because they match more common lyric structures). I sometimes change lines because of this until I get it to work. Having said that, it can be annoying when everything else is perfect but the missing word 😆

When Suno adds that "paint in" function I imagine it could be useful in this situation

In regards to it being a hallucination, it seems to skip words to create a phrase that more suits the timing that its using in this generation - at least that's how it sounds (which I guess is a type of a hallucination)

Certain phrases will almost never skip a word in a generation, while other phrases might sometimes skip a word and other phrases again almost always skip a word. Though when this specifically occurs for a given phrase depends on the context of the phrase (the rest of the lyrics, metatags, styles etc)

So you just gotta keep playing around until you get it

I imagine it is something they can improve, but you can work around it even already

[tribal marlin]

Something I've noticed when I've got stuck like this is changing the point of the continuation either a second back or forward (or more if possible). I think I recall having a similar situation and I was able to get around it (though it did take a number of attempts of adjusting the prompt text and the continuation point)

Like it might not fix that exact situation, however there have been multiple times where altering the continuation point made Suno do closer to what I was expecting

I kinda wish you could specify the continuation point with more precision (kind of like selecting the exact point on the wave)

[devout parcel]

🗿

[limber gull]

Kinda... Just say in the Style of music field that this is a narration, male(female) voice, etc. Just do not put a lot of text)
But in my experience, the most interesting TTS was in novelai, because it can produce some sort of emotions and sound more-less natural.

[gray valley]

whats difference between chirp here and suno website

[limber gull]

I guess that chirp is just v2's naming.

[final hazel]

Discord has V1.. Website had V2 and V3

@gray valley @limber gull ^^

[deft void]

Do we know what the background prompt was for the V-Day demo? It had a distinct style that i want to recreate.

[rose ore]

okay, more on the verse calculator. I figured out something interesting and wanted to share it.
When the system is calculating how your input is going to be read, it runs a calculation that figures out what gets played in what way and how. And it's hashed out into a value, that everything else is based on. Which is fine, we all knew that.

But...

What the new Continue from does, is it changes the math.
Might be good, might be bad for your song, but it's there.

And knowing that... you can exploit by initially calculating the verse one way, and then clipping in something that different (mathemetically), that'll force the math to rejigger.

Then, you get to enjoy watching the system try to balance both sets of conflicting math, and it's fun.

[kindred minnow]

I noticed something and i want to know if there is something to do in the prompt to prevent that. in the version 3 the result are really amazing but as soon as a signer push his voice there is something whong in the mastering

the voice become with noise in it

and also another question, how do we lower the volume on the website when listening to the music

[rose ore]

I hate it when suny gets screamy. Okay, so try tagging it with, cirsp, hdaudio, studio album, mastered

sometimes remastered works better than mastered

[rose yacht]

I'm not familiar with what you're talking about, is the math the rhythm or something? What changes when it has to struggle? I'm always in to making AI do weird things, sounds up my alley

[rose ore]

Okay, so

[rose yacht]

btw I'm digging your klezmer genre mashups

[rose ore]

Yeah! so much fun. I'm also experimenting with jazz. I know I know, but it's so much fun

[rose yacht]

I was a jazz band kid in highschool, I've got a soft spot even if I don't listen to it anymore

[rose ore]

So the way it works is by figuring out stuff like how many pieces are in a line, syllables, notes, backups, whatever. And it does some math. And it remembers that math as a hash, which is an abstract representation of whatever you've sent it, relative to the larger composition like if it's an extended piece

[rose yacht]

So what're you doing to make it conflicting? Like changing the chorus, or it's just automatic when you continue? Not sure I understand the context where you introduce something new

[rose ore]

Right

You can do it when you continue too, but it would be continuing from rjiggered hashes that wouldn't or couldn't exist by default

which makes more interesting music. You can also splice and splice out

like putting a section in that doesn't match. Then, cutting over it before the lyrics kick in

[rose yacht]

Interesting, half the time I don't even use real music terms for my style prompts so a new way to get weird with it sounds fun

[inner drum]

Hi, I have a little question if someone can help me

Im using suno.ai V3, how can I make my song last more than 2:00 minutes?

[spring sail]

suno v 3 is bad about cutting off verses, etc, even when i just have 4 lines for the entire clip. i was wondering if there was a hack that users could use like continuing from the middle of a verse or chorus just to get the rest of a great sounding clip. i dont know, in the new clip, how to indicate that it is just starting from the middle.

[subtle adder]

Hi, I have an opposite question.

Im using suno.ai V3 and the "Song Description" prompt method, how can I make my song that ONLY lasts 30 seconds?

Would specifying "1 verse, 1 chorus" in the prompt help?

hmmm seems like 1 verse 1 chorus is pretty long 😄 will experiment and report back in case anyone is interested in how.

[obsidian tendon]

How can I mimic or create something similar to the popular Dune singing theme song

[subtle adder]

so adding this to the prompt helped a little "... 1 verse, 1 outro, no chorus..." managed to consistently generate songs with 1 verse and 1 outro. not sure why we had to include a "no chorus" as it would intermittently add in a chorus if we didn't specify "no chorus".

[lament zenith]

Ive created a song and I like the voice on the song how do I get the same vocals consistently on different tracks

[teal raft]

You currently aren't able to do that. Everyone wants to be able to though

You are not alone

[lament zenith]

Hmm has anyone tried including prompts that describe the voice

[proven rapids]

tried it. "evocative vocals, kitty vocals"

[tribal marlin]

In the new clip, start with the whole verse or chorus again, including the [Verse] or [Chorus] tag. That's worked for me before at least. Sometimes adjusting the continuation point too can help (if you can change it by a second or so - if this is possible).

There is a certain randomess that you kind of have to work with, but sometimes these things help.

[lament bison]

What is the problem with this lyric structure?
I've already spent several credits on it, but it carries out the instructions. Does anyone know the solution?

[Bridge With Cello]

At the river's edge, where the waters meet the shore
The Morrigan's magic, through the mists of lore
In the cycle of life, death, and rebirth
Her wisdom prevails, across the Earth

[Chorus]
Morrigan's veil, woven in the night
In the dance of war, in the flicker of twilight
Goddess of battle, of fertility and fate
Morrigan's call, through the ages, we await

[Outro]
So heed the call, when the crow takes flight
In the depth of your soul, in the dead of night
For Morrigan watches, her power unseen
In the heart of the wild, in the spaces between

[Violin Melody Ending]

[broken lotus]

the lyrics look like 3 verses, despite how the sections are labeled. There is no change in meter or phrasing.... they are all verses.

I'm not saying it's 'wrong' but this will leave your song a bit 'flat' and sing-song. Some ancient songs follow this style (all verses, no chorus) but I think modern ears prefer the change, and the satisfaction of change and repetition...

generally a chorus is something to repeat (the part that gets stuck in your head), something more melodic/uplifting with a different meter. It could be as simple as 2 lines, or longer but probably fewer words per line, more emotion and less storytelling.

[lament bison]

This is not 3 Verse!!! This is just an excerpt from the full song. This is the last section. The Chorus there is the Chorus that follows, they perform usually in one version and are good too. That's not the problem.

[broken lotus]

Ok, no need for extra !!!!!!!

If you'd like to TELL US what problem you are hearing, rather than making us guess from a wall of lyrics like a game show, you might get an answer you're looking for.

[lament bison]

Ok. Here is the full Lyrics and everything:

[Violin Intro]

[Verse 1]
Beneath the crescent moon's soft glow, through the misty veil she treads,Morrighan, Phantom Queen, in the battlefield she weds.Shape-shifter, crow and eel, wolf and heifer's guise,In the shadow of her wings, the fate of warriors lies.

[Chorus]
Morrighan's veil, woven in the night,
In the dance of war, in the flicker of twilight.
Goddess of battle, of fertility and fate,
Morrighan's call, through the ages, we await.

[Verse 2]
Her voice, a call to arms, a whisper in the heart,
Guiding the fallen, from this world they depart.
In the raven's cry, her presence is near,
A guardian of the realm, she casts away fear.

[Chorus]
Morrighan's veil, woven in the night,
In the dance of war, in the flicker of twilight.
Goddess of battle, of fertility and fate,
Morrighan's call, through the ages, we await.

[Bridge With Cello]
At the river's edge, where the waters meet the shore,
The Morrighan's magic, through the mists of lore.
In the cycle of life, death, and rebirth,
Her wisdom prevails, across the Earth.

[Chorus]
Morrighan's veil, woven in the night,
In the dance of war, in the flicker of twilight.
Goddess of battle, of fertility and fate,
Morrighan's call, through the ages, we await.

[Outro]
So heed the call, when the crow takes flight,
In the depth of your soul, in the dead of night.
For Morrighan watches, her power unseen,
In the heart of the wild, in the spaces between.
[Violin Melody Ending]

Style of Music:
Cello, violin, flute, hurdy gurdy, Aggressive, Dark, Ambient, Black Metal, Folk-Celtic Metal.

Problem: The singer doesn't use the instruments, doesn't sing thunderously like in Black Metal bands, doesn't follow the metatag instructions!

[final hazel]

Your prompt is intriguing and well-crafted, but here are some thoughts on why Suno might not be responding as expected:

1. Complexity of Prompt: The blend of instruments and genres in your prompt is quite sophisticated. Suno’s AI might be struggling to synthesize all these elements cohesively, especially when combining genres like Black Metal and Folk-Celtic Metal with a specific set of instruments.

2. AI Interpretation Limitations: AI, including Suno, sometimes interprets complex prompts in unexpected ways. This could lead to a deviation from your intended outcome, particularly with nuanced metatags and a mix of aggressive and ambient styles.

3. Adjusting the Prompt: Consider simplifying the prompt or focusing on fewer elements at a time. For example, you could start with a more basic genre definition and gradually introduce additional instruments or style elements in subsequent generations.

4. Experiment with Metatags: Metatags are powerful but can be tricky. Try adjusting or simplifying these tags, or experiment with their placement within the lyrics.

[lament bison]

These styles exist, and the instruments are also given here. Unfortunately, it has cost me thousands of credits to experiment. Maybe you can give me some concrete examples?

[final hazel]

Examples are useless really. What worked for others are not the same for your style, genre, lyrics etc.I understand your frustration with Suno not following your detailed prompt and metatags as expected. It seem you are expecting total control which is not possible yet. The AI’s unpredictable nature means it might not always interpret or execute instructions as intended, particularly with intricate stylesl. Each genre, lyric set, and instrument combination can produce varied results. Unfortunately, this unpredictability is part of the current state of AI music generation, and there isn’t a surefire way to achieve total control over the output. The best approach is to continue experimenting with different styles and prompts.

[lament bison]

I want Black Metal snoring, because so far the vocals are clear. How can I solve this?

[broken lotus]

Find other black metal songs by searching the discord (soon possible to search the website, but not yet)
Copy the prompts from the songs that sound like you expect them to.

[hybrid heart]

try linking a youtube video in your prompt. IDK if it works, they say it doesn't, but I have had some luck by creating a clip of a song at a good vocal point

[static wigeon]

impossible the amout of caracter... suno take 100 carctere in style of music x) the link is really useless i think

[versed valley]

Welcome! For help, check out Suno Help

[supple frost]

Prob a stupid questions, but, where is the archive on the website? I left off some songs I wanted to work withtoday and as I logged in.. Well, half of the stuff I generated yesterday disappeared xD

[static wigeon]

but yes... i really need to brain the security of Chirp... becaus i CAN'T explain the style of the song i want to make because is new... experimental and i can't explain 😢 is not copyright is inspiration 😫 I'm going into depression x)

[versed valley]

Welcome! For help, check out Suno Help

[full kettle]

Did anything change from v3alpha to v3?

[round jetty]

/chirp

[flint haven]

The /chirp command is not available in this channel

Go to the #🐣┃chirp-beta-1 channel to use the /chirp command, and check out the #🐣┃how-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.

[sacred willow]

does anyone know where i can find some documantation about how the program works, such as used external models? I am not expecting to find all of the source code or something but still...

[tight frost]

"don't" ?

[neat nexus]

Yes. They don't recognize references. If you try to put an artist's name in you will likely find it unable to continue.

[alpine venture]

Does this mean v3 has had an update?

Like, the announcement mentions more graceful endings

[teal raft]

It's different than alpha

Subtly

But it is different

[alpine venture]

fuuuuuuun

[teal raft]

[alpine venture]

oh boy

[blissful raven]

When they said watermarking how does that work

[blissful oak]

/chirp ukrainian song in country stile

[tough sandal]

Does Suno have an API? Would love to integrate songs into my games. Imagine a bard singing a song about a player's adventures as they play.

[final hazel]

Yes, music created with Suno includes a proprietary, inaudible watermark. This watermark is a security feature embedded within each song, which doesn't affect the listening experience. It serves the purpose of identifying the track as an AI-generated creation by Suno. This technology is particularly important for ensuring the responsible use of Suno’s capabilities and can be crucial in legal scenarios, such as proving the synthetic origin of a song in case of disputes or copyright issues. While the watermark is designed to be undetectable to listeners, it plays a vital role in maintaining the integrity and proper use of Suno's music generation technology.

[lime grove]

Hi guys, I need help. I have used my own lyrics on the free plan but now I have subscribed to the pro plan. How can I use the music I’ve made with my own lyric for commercial use

[jaunty apex]

Hello, if I like a music created with V3 but just the percussions are not the right instrument, can I change just those?

[final hazel]

Your rights to use music created with your own lyrics for commercial purposes are effective from the start of your Pro subscription and do not apply retroactively to music created before subscribing.

[jaunty apex]

(is Suno V3 chirp or bark? I'm not sure)

[lime grove]

What about the song I’ve made with my own lyric when I was on FreePlan can I use it for commercial purposes?

[final hazel]

Currently, Suno V3 doesn't offer the option to specifically change only the percussion instruments in a song.

[jaunty apex]

Thanks, is it possible to restart from an existing music file or only from the prompt?

[final hazel]

When you created music with your own lyrics under the Free Plan, those songs are only for non-commercial use. Suno retains the rights to music generated while you were using the Free Plan, which means these songs cannot be used for commercial purposes. You only have the copyright on your lyrics.

[lime grove]

Meaning that I can make new songs using the same lyric without problems for commercial use?

[final hazel]

Suno allows you to continue from a specific part of a song you've created within the platform using the 'Continue From This Song' feature. However, it doesn't support importing and continuing from external music files.

[final hazel]

You've got it! 😄

[jaunty apex]

Nice, thank you!

[lime grove]

Thanks a lot!!!!! 😎👌👌👌👌

[vivid cedar]

Why the step backwards with saving the video? The lyrics are no longer included...

[fierce star]

Is Discord Bot using v2 or v3?

[broken lotus]

Lyric videos will be back (maybe by tomorrow). They've changed how they're being made to a new system.

[broken lotus]

still on v1 afaik

[fierce star]

So even worse than v2

[inner steeple]

are the servers down / hammered? 😄

I mean, I don't wonder why

[exotic fiber]

a looooooot of people testing it... its also happening to me

[inner steeple]

wait

[final hazel]

bot is using V1

[inner steeple]

I need to accept terms

I tried remixing so it didn't pop up that

[exotic fiber]

btw... im really glad with v3.. testing it right now... great improvement...

[inner steeple]

well.. it's still the same, lol

[inner steeple]

ah

funny how it finally started gen for me now

but I redid this

[spice bloom]

How to share my song to Explore?

[inner steeple]

I imagine this

[final hazel]

To share your song to the Explore feature in Suno, simply go to the song in your Library, click on the song settings, and change the sharing status to "Sharing Public." This makes your song visible to other users in the Explore section.

[inner steeple]

finally testing v3 and it's really cool, you still need happy accidents for cool things tho

I haven't checked yet how people got chorus done more reliably

it seems to be pretty hit and miss still on v3

most of my generations got some individual cool pieces though, even when the whole thing as one doesn't work

good thing about the v3 is that continue from timestamp is there, so I guess I can just brute force before chorus till I get it done like I want

[delicate fractal]

Why no more lyrics in videos? 😦

[inner steeple]

I get it

[final hazel]

It's temp.. update coming

[delicate fractal]

Ah great, that was one of my favourite features 😄

[inner steeple]

I can imagine it breaking when you continue from timestamp,, or if gen cut off before it was done etc

would be nice if one could manually adjust the lyrics to show what is actually there but I get it's probably a lot of work for minimal gain

[delicate fractal]

Am I right in remembering that it used to generate different lyrics per song when generating two songs at once on the website? Now it's the same lyrics for both gens.

[inner steeple]

do you use custom gen?

custom gen has always given same lyrics

I mean, there you specifically ask it to

[delicate fractal]

Nope, that's just regular mod

[inner steeple]

but yeah, v2 did give two versions if you let it gen the lyrics

[delicate fractal]

Yeah, now both songs have the same lyrics :/

[inner steeple]

but

well yeah

but you can now pregen the lyrics on custom mode

so you'll see if the lyrics are what you like, before you gen

[delicate fractal]

I'm not using Custom though

[inner steeple]

just saying 😄

[delicate fractal]

Just think it was nice to get two different songs and pick my favorite, rather than twice the same lyrics with a different soundtrack 😄

[inner steeple]

but yeah I'd really like if the gen would just do one song per gen tbh

or have some batch setting

so you could like do 5 variants of the same thing if you wanted to

just forcing it to two gens is sometimes unnecessary and wastes resources

especially if you continue, because often the first version is fine for what you do

[delicate fractal]

Yeah I wish that was an option, I wouldn't be against being able to pick for it to gen only one song, it would also be easier on the servers

[delicate fractal]

I read that three times and still don't get it

[quartz robin]

Thanks for V3. Are you going to add it on Discord?

[broken lotus]

he is hoping for a music-to-music feature, where he can upload his original audio to generate (continue) with Suno....

OOP it was removed

[final hazel]

atm it's V1, and they are still thinking about upgrading.. So not sure

[quartz robin]

Got it, thanks!

[lime grove]

Why can’t I download my creations?

[final hazel]

it works for me?

Go to 3 dots menu.. and there you will find the download option for mp3 and mp4

[terse musk]

Question - ever since the coverage around Suno has exploded, it seems to me that the generation quality has drastically dropped (with V3). Has anyone else noticed it? Possible they're reducing per-generation compute time to cope with the larger number of users?

[lime grove]

Like…where?

[final hazel]

yes

[lime grove]

Guess what happens when I press download:

[final hazel]

I am sorry.. I am a volunteer and try to help to my best ability.. Not to play guessing games.. Please be clear

[lime grove]

This happens…

[final hazel]

Maybe try a different browser. As I said, it is not global it seems. I can do it just fine. So could be on Suno side, but since I don't have it it also could be something on your side like a plugin or browser. Please try a different browser.

[lime grove]

Ok

[zinc coral]

Does anyone know why sometimes in suno the video has lyrics and sometimes it doesn't? Did that feature change?

[final hazel]

in past it had lyrics, now it doesn't, they are working on new display

[zinc coral]

Do you know if I can get lyrics again?

Or I just have to wait

Also thank you for responding

[broken lotus]

i doubt current videos will get re-rendered with lyrics....
but the new videos should be working by tomorrow I think.

(according to a dev)

[zinc coral]

Thanks

[buoyant osprey]

I just released a new song on youtube. what do you guys think of it?https://youtu.be/XSXP3VBJBdI?si=OZHpsyGzCLY9QvWL

[unborn turtle]

I also noticed this. The voices in getting aren’t very distinct. They are slightly muffled and the songs sound tiny and have no flow, as if the system can’t determine where the lyrics should fall in relation to the music. It sounds more like a bad version of V2 from my first impression.

V3 alpha was amazing. There must be a hiccup somewhere that they will fix

[unborn turtle]

@final hazel

[final hazel]

Feel free to post it in #🪲┃bugs-web so the Devs can see it. I myself feel that it is more genre related. The genres I play with sound pretty good. I mean it is not prefect, but better than V3A for me.

[neat nexus]

Alright, @static wigeon you were warned. This is a violation of rule #3️⃣ and you've earned yourself a day in timeout.

[graceful cliff]

nope this is not a hiccup and won't fix

like @final hazel said, this is heavily genre/lyrics dependent. If your lyrics doesn't "match", or you choose vocals in certain genres, the voices will sound slightly worse, to keep the flow of the music/genre consistency.

Of course a world where everything is perfect would be nice but at current capacity there probably will have to be trade-offs.

[raven condor]

Does anyone know why they chose to abstract/water down prompts for V3? This version ignores large parts of the prompts 90% of the time and doesn't get close to feeling of the descriptions, ect. comparing it to V2.
I can understand if they wanted to make it only generate along more strict list of genres as they upgrade its capabilities , but I wanted to know their reasoning for it, if mentioned somewhere.

[vivid cedar]

Wait so with more server load the generations get worse?

[teal raft]

That's what you took from that?

[vivid cedar]

"At current capacity there will have to be tradeoffs"

[vivid cedar]

Why are you confused?

[teal raft]

I read that as model capacity

Not server

The model, Suno, is trying to make good music out of prompts people feed it

[vivid cedar]

Ah, well I guess we'll have to wait for the official answer to know for sure

[teal raft]

Before they've talked about the GPU as being an either/or thing

Not a sliding scale. It costs them as much to set aside a GPU to generate as it does to do it fast or slow or only 1 gen at a time

Sometimes people ask for 1 generation at a time rather than two to 'save' credits

[graceful cliff]

Kaker is right, I mean model capacity

[unborn turtle]

Thanks for answering, but why does V3A sound better with the same prompts ?

[lament zenith]

Hey so i created a song with lyrics. Is there a way to go back and edit the lyrics or at least take just the beat so i can rewrite the song

[neat nexus]

Resume from an earlier time but from before the vocals begin? Or just before where you wish to change the lyrics?

[graceful cliff]

This is a great research question and I have been thinking about it for a month lol

unfortunately I don't have a great answer

but we understand your feedback

and I agree to a certain extent

so we will address it in the future

[lament zenith]

How do I do resume from an earlier time?

I can only see remix, continue from this song

[graceful cliff]

continue from you can choose a time

[lament zenith]

Thanks

[limpid spade]

It really needs to flash with neon lights #moreneon

[unborn turtle]

That’s okay! I appreciate the response.

You guys Will figure it out 😊

[raven condor]

--- Came back in to say there's a night n' day difference in the quality I was getting from earlier today and now.
Changed my mind, V3 is awesome

Though I would like to see some attention to the breakcore genre, ill put in suggestions

[pliant cave]

Have you really had time to get a large enough sample size here, or is the general theme that V3 is "worse" than V3 alpha? I'm not seeing that at all so far on my side. V3 seems great, but Tony's comment seems to imply that somehow V3 is "worse" in some way.

[tepid wedge]

Is there a reliable way to set male and female voices? Especially to switch between them?

Like male for the verse and female for the chorus

[limpid spade]

Reliable. No. Search in Discord for duets or male female vocals. It's a common question

[tepid wedge]

Oic thanks

[limpid spade]

My best duet was all Suno. Not prompted.

[lament zenith]

How do I stop my songs from stoping abruptly I want it to slowly fade out is there a prompt for this

[supple frost]

Try to continue with choosing instrumental for the end

[limpid spade]

[fade out]

[lament zenith]

In the song description ?

[limpid spade]

You could also try getting the song to continue longer then fade out manually in a DAW or other editor

[limpid spade]

At the end of the lyrics section. As with all prompts, there's no guarantees but it has worked before.

[graceful cliff]

how should I clarify this...you shouldn't be biased by my comments.

v3 is on average better than v3 alpha, doesn't say it is 100% better than v3 alpha. If you compare head to head, you may still find you like v3 alpha in some cases. But overall v3 has a higher success rate.

By the end of the day what matters is on average you get better generations.

[upbeat depot]

The audio quality is distinctly grainier than the alpha

Feels like going from 320 bit rate to half that

Honestly sounds kinda terrible so far in comparison

[graceful cliff]

go put in the feedback: #🗳┃mini-polls message

[upbeat depot]

Can or have any of the devs confirm if a setting related to bitrate, steps or general quality has been nerfed for the official release?

[lament zenith]

Also a lot of the time I do this it completely changes the beat

[terse musk]

Yep exactly the same experience. Thanks for confirming

[cold zenith]

Is there a document where we can see the differences between alpha and Final ?

what have been done

[cold zenith]

Sorry to bother, I have a question. Are you planning to allow music generated in V2 to be upgraded to V3 quality with a single click? Because there's some music we've generated that we'd like to see in better quality.

[willow wren]

To me V3 sounds different to Alpha. Not worse the wording is more correct now. But i miss the Natural Voices the sudden screams i had before. Suggestion: make a Choice in the Box, so we Could use v3 Alpha and V3

[unborn turtle]

I am just saying this is from my first impression compared to the V3 alpha. My sample size could be larger, but I was really just reusing prompts that worked really well and V3 alpha, that didn’t work well with V3. I also tried to generate some non-custom songs that didn’t really sound that great

@graceful cliff is it possible to keep V3 Alpha in the options for us to choose from when using Suno? At the moment, V3 Alpha isn’t there anymore.

[cold zenith]

Maybe you have to do it, in other way

[unborn turtle]

I tried custom lyrics, auto generated lyrics, I’ve tried more descriptive music genres, simple music genres, re-running old prompt, making new prompts.

I hear a mix of these things, partially grainy or white noise / muffled songs and voice, less distinct voice, more choir like or a group singing together, and the flow of how the lyrics are sang doing really fine with the flow of the instrumental.

So far, I didn’t have that wow factor as I did with V3 Alpha. In theory, it should at least feel the same, if not only slightly better. @graceful cliff

If you can roll back or just include V3 alpha as an option for us, that would be greatly appreciated.

[echo creek]

i would also like to request adding an option to choose V3 Alpha
prompts that worked great in the Alpha result in absolutely horrid generations in V3 - super noisy, inconsistent and just overall musically unappealing

this is the case with the new generations, while continuations seem to still work decently well

[misty dust]

What’s the recommended verse amount for songs, so I don’t get hallucinations?

[silent minnow]

why does it generate only up to 2 minutes in subscription on v3?
needs 4 to 5 minutes, not 2 minutes.

[cold zenith]

but you know that you can continue the generation....right ?

[neat nexus]

Please take ten minutes to reread the #📜┃rules, particularly #3️⃣

[tribal marlin]

Is it just me or is v3 producing fuzzier output on "heavier" tracks, take these for example:
https://app.suno.ai/song/8e12e356-95b6-43ba-b474-0e9b1bdb6f79
https://app.suno.ai/song/0deb70d7-4431-4a55-b07a-1ef10c9f6971

especially towards the end of the tracks. These are just single tracks first generation for the sake of testing

[neat nexus]

Enjoy an hour in timeout This time kindly read the #📜┃rules please? Thank you.

[unborn turtle]

I noticed the same thing, the voices are less distinct. it sometimes sounds fuzzy or as if It’s a group of people singing. The instrumentals are good, but the vocals and V3 alpha were much much better sounding, and the vocals were placed on the instrumentals better.

V3 Alpha:

https://app.suno.ai/song/be760285-8dec-462e-a991-f7b2b9a5d0e5

V3 official release:

https://app.suno.ai/song/ef692428-2051-4e3c-962f-14409d06079b

This is a good example of how V3 Alpha sounds better.

[final hazel]

It's like this.. sometimes you have to break a few eggs to make an omelet.. While you guys observer problems with voice, others find in their genre better results.. It's extremly hard for the Devs to make sure everything is perfect for all people who all have different ideas, different prompts, different languages and different genres they want to use.

[tribal marlin]

yea that does make sense. Though I have noticed this quite a lot across the board, at first I thought it was just heavier tracks though it does seem more prominent than prior. Regardless, what Suno generates is amazing so i'm always happy lol

[unborn turtle]

Definitely! I just wanted to point out my observations to hopefully help the devs get perspectives of what users desire, and what changes I’ve noticed coming from V3 Alpha.

However, I would love access to V3 Alpha again in addition to V3 in the drop down menu.

[tribal marlin]

Nothing like Suno exists so I know it's tracking new ground

[final hazel]

Try to do things differently as you did before, don't fall back to old patterns.. it helps

[tribal marlin]

yea i am trying to do that! as much as I can, i guess that's part of the fun

[unborn turtle]

Anyone discover any tips on what works better in V3?

Such as using brackets [Intro], [Chorus], etc.?

Or if you have to start with an instrumental and continue from there and add lyrics, orrrr, idk, something that can get me to a starting place.

I have also seen the guide on Notion.

[final hazel]

I find it helps to get better result with more minimalistic music. So I often add "minimalistic" in my "style of music", and I try to stay away from electronic music since AI has a hard time with the SFX in that kind of music.

Try some instrumental music.. just for fun.. like piano.. it's amazing good quality. You can get the midi notes almost in every DAW and can play around by using your own VST's

[candid fox]

Have I written something wrong? I told suno to use a female voice also for the chorus but he used a male (autotune?) voice. Can't I choose if female or male ?

[broken lotus]

some genres are very female or male heavy...
and you need to work a little harder (make your style prompt more specific) to get around the bias.

and sometimes a style prompt can conflict or override...

for instance, if I use 'bubblegum' anywhere in my style, I almost always get a female voice. (genre overrides my ask for 'male vocal')

[candid fox]

well when i write female voice... how can you write it more directly ? So you can bypass the autotune ? Why not in a prompt directly or is that possible ?

bubblebum instead of female voice... ok thanks for the tip

sorry if the questions are strange... in chatgpt etc. the promts are more direct and more likely to be implemented

[exotic pewter]

I need help. I forgot my login details, I don't know which email address I was logged in with. I subscribed, I have the transfer details, so I can prove that I subscribed. I need help. Thank you

[final hazel]

please send an email to [e-mail]

[brittle gulch]

Dumb question.. what if you get a song that is perfect but you missed a word when typing from Google and you want to rerun that song, not a new mix.. would you do "continue from' to keep it the same evening?

Same *everything

[teal raft]

Your only chance is to cut from before that word was pronounced and hope you get something similar

But Suno might decide to shift tone in the continuation

[exotic pewter]

I wrote, when will I get a reply? are there still customer service hours at this time?

[daring dew]

https://app.suno.ai/song/a3d04c9f-2158-4351-a6ad-c0354159a82e

Brazilian music guys.

[spice bloom]

Thanks!

[candid yarrow]

does anyone have any general prompting for making UK Bassline similar to BurgaBoy, feel like ive treid everything ?

[willow wren]

I use the Open Source MT3 model for that. What do you use for MIDI?

[maiden monolith]

What excatcly are these “new styles” and genre?

[pliant cave]

Good explanation, Tony. My take on this is that the "better" model is the one that yields more consistent, high-quality generations on average, given the primary use case for Suno. The average Joe or Jane, who just wants to hear a fun song about their lives or things they care about, is simply looking for clear vocals, good-sounding instruments, and the general style they request on the first try, without having to create ten versions. Regular, non "power users," (for better or worse) expect something "right" on the first attempt and aren't making 50 runs and piecing them together to get it just right.

I mean, looking back at v1 and prior, people here who used those heavily know you could get some really wildly creative outputs but more often than not you got a lot of "trash" outputs or something incoherent. Most of that is gone now.

I give two thumbs up to V3 after about 100 test runs, aiming for variety. 👍👍

[pliant cave]

Agreed in general, but I've had some really fun outputs with "electronic" and "robotic" vocals. See this one, for example.

[pliant cave]

Interesting comment, I was thinking about this today and what the future of these models could be. I'm not an ML guy, but I do know that transformer-based models are all about scale and data. Just thinking ahead to what a massive-sized model akin to a GPT-4 level parameter count could be like with audio like Chirp, if one could ever be trained with enough data.

We know almost for a fact that GPT-4 (probably), Gemini 1.5 Pro, and Mistral use the MoE (Mixture of Experts) architecture to scale, get better, faster outputs with less compute usage. Just thinking how a Chirp-style model could perform with a similar design where you had 8 submodels all trained to be "experts" in highly specific musical styles, genres, or on certain types of data.

Maybe you could get that "one model to rule them all" effect for Chirp. I could only imagine something like that is in the realm of possibility.

[final hazel]

and why can't you use minimalistic then? It still helps adding that

[untold briar]

Any musically literate folks around who can evaluate the accuracy of this style prompt? https://app.suno.ai/song/7f91c2e0-1328-454a-8e3a-5525ab2ff68e/ Very interesting

[broken lotus]

👂 : it's pretty.

I am not hearing any of these extra meters...
(just 4/4)

OOH! I like that bass coming in there at the end... it was just getting warmed up!

[inner crypt]

hello, I wanted to know if it is possible to run suno ai locally?

[final hazel]

Hey Pixel, Suno AI is a cloud-based thing and can’t be run locally. It needs those server resources to make its music magic happen, so you gotta use it online. 😊

[inner crypt]

Ok, thank you

[unborn turtle]

I think a power user doesn’t mind working through bad generations using the continuation feature.

[pliant cave]

You are absolutely right will try, thanks!

[fresh hemlock]

how to adjust the volume of the voice in v3 ?

[static wigeon]

Hey all ! I really need help
sorry to come back to bother you
Is there a "strong skill prompt" person available for like 10 minutes ?
I would like her to listen to a YouTube song and for her to help me find the best prompts to make similar sounds please ? (No Copyright ! is just a fresh new style of rap...)
https://www.youtube.com/watch?v=DNpidNsSWzA&ab_channel=Jima
a fairly eclectic person, because these sounds are not easy to explain

Please and thanks 💜

[steep peak]

im stuck on an infinite black loading screen. any tips?

[untold briar]

Please just ask one time. But also accept that you may not get any answer to your question. There is no easy way to find a Suno prompt for a unique style of music without a lot of trying and failing so maybe nobody can help you. Also, Suno V3 is brand new and everyone is re-learning how prompts work so this is extra difficult right now.

However, I would recommend linking the YouTube video you are referring to in your question instead of expecting people to DM with prompt advice. I think you will have a much higher chance of getting an answer this way, because then anybody who reads your question can just listen to the YouTube song and then they might have an idea for a good prompt. But like I mentioned, V3 is brand new, everyone is learning now, there may not be a way to capture a unique song like you want without more time and experience.

[untold briar]

Try Control-R refresh.

[steep peak]

no luck

[static wigeon]

Where i ask a second time ? 😮 show me please because i lose my brain if is real....

i delete to general to have more chance here... but yes, see me where i ask many time this ??

[untold briar]

Yeah it's fine, I just mean, don't keep asking if nobody answers, because I think it may be a difficult question to answer! (But also you shoud link the YouTube song so there is some chance of getting an answer.)

[static wigeon]

linked 🙂
but where i ask many time ? for my mental health please tag me where i ask a second time

I'm mentally unstable and I don't like people picking on me for nothing... I'm not angry, no worries, I just want to see where I asked that recently?

[untold briar]

I'm trying to get ahead of things by reminding people to only ask once. Everything is OK.

[static wigeon]

k.

[gilded hound]

no idea if "earrape" would be accepted or work but that's one thing that immediately stands out

[static wigeon]

i need at the time for the instrumental and the effect voice 😦 but yes i gonna try this ! thx bro ❤️

[untold briar]

I don't know French music, do you have any ideas for what genre this is? That might help. Or is there a word for that style of voice?

[fallow gull]

I'm still figuring out v3 compared to v3a, but does anyone else think that less is more in v3?
I was struggling to complete a song that was doing well in v3a using stuff like "romantic 40s jazz, violin, piano, old soft female vocals, swing" and every generation had problems.
Then I stripped everything out and just used two terms "baroque violin, swing" and got what I was looking for.

[gilded hound]

just so you know it basically means that the sound quality of the vocals is reduced while being made louder for comedic effect

and those vocals are definitely lacking in sound quality

[static wigeon]

No, I asked lots of French servers to ask them but it's a very recent style and it's extremely hard for me to find the right vocabulary for Chirp (and for other french guys in this style)
I even asked a rapper lol but even he couldn't really describe the right vocabulary to me

[gilded hound]

less is more if you want to make something cliche, so like idk about jazz for example if most romantic 40s jazz uses piano you dont need to say piano - if most doesnt then keep it

[fallow gull]

I was using terms like romantic heartfelt, nostalgic and the vocals in particular were terrible.

[gilded hound]

what does something nostalgic or heartfelt even sound like

[untold briar]

I haven't specifically noticed that trend. I know for sure I preferred some long V3 style prompts over short ones. I usually just try both extremes.

[broken solar]

omg hii, so just joined, was curious if anyone had any like disco music they;ve generated theyd like to share

[fallow gull]

This is what soloteller thinks of that french song:
https://sonoteller.ai/DNpidNsSWzA
Its analysis looks about right to me, or at least a place to start.

[static wigeon]

DAMN thanks a looooot for this link bro !!! i think you just save my Chirp life !

[gilded hound]

til sonoteller exists

[untold briar]

This looks like a good start, maybe add in 'glitch' or 'chiptune' or something as well.

[fallow gull]

Sonoteller is ok for a lot of popular music. It's not so great with folk, world, or anything niche.

[static wigeon]

The problem when you're French is that you have a lot more difficulty looking for this kind of useful link. Or even the tutorials are much less in France than in English

[static wigeon]

this type of vocabulary i need x)

glitch i always know it but chiptune never hear x)

[fallow gull]

Lol. Well I don't speak french, but my wife and daughter do.

[untold briar]

Do you have the lyrics for that song by chance? Or just a verse? I want to get the approximate line length.

[gilded hound]

id actually advise against chiptune here because at least in my experience it skews it towards itself pretty heavily whereas its not that big of an element

[untold briar]

Maybe 'glitchcore' but I haven't tested that in V3

[static wigeon]

If you ever have useful links like this I'm interested! Really ❤️

[static wigeon]

for my song ? yes i can found this

[untold briar]

'melodic' should be in there I think

[gilded hound]

glitch works well

[static wigeon]

sorry for the long txt x)

the problem in the new style of French rap is that the flow is practically never the same, sometimes the end of the word is extended, other times cut... it's extremely hard but already with all your help here, I think I can have a lot of fun!

when is good for you, you can delete the french txt x)

[untold briar]

Sure, delete it, it's super big in chat. I copied

[static wigeon]

OH i have an other question ! this time is about the order between the [ ...]

it is better to keep it as short as possible for example [acoustic guitar melody] [distortion] or it will work just as well if we do [soft and catchy acoustic guitar melody with distortion] ?

[fallow gull]

I don't know how to get that distorted excessive autotune:
https://app.suno.ai/song/25702d73-935a-48a0-99f8-00bd2ba907cc

[static wigeon]

You think i can add a thing like [sidechain effect on voice] ? for the effect on Jima song ?

is really cool 😮

[untold briar]

Maybe, but in V3 only use small amount of stuff like that I think. And keep very simple not a whole sentence in the bracket.

[static wigeon]

is better this [acoustic guitar melody] [distortion] in v3
rather than [soft and catchy acoustic guitar melody with distortion]

Sorry sometime i missunderstood x)

[fallow gull]

v3 has only been out for a day, so I'm still experimenting, but so far I think less is better, at least for what I was working on.

[static wigeon]

ok I see I'll try that then, thanks a lot

[untold briar]

What is that bassline, it's something specific that music people probably know the name

[static wigeon]

is a mix between bass line and some guitar i think

is really full electronic song x) Fl studio etc

i gonna try 'Slowy repetitive bassline' or background bass line, something liek this

but yes BIG thank for the website

[untold briar]

@static wigeon I kept getting EDM but here's what I tried, maybe something to learn in there https://app.suno.ai/playlist/124dd085-329f-4dc5-879c-10a0e0db4e2a/

I know that bassline is one very specific name of a thing I've heard a billion times before I can't think of it

[static wigeon]

oh dmn thanks i go listening all this 😮 i come back after :3

just 2, is really perfect ! 😮 Now my job is to find the prompt to alternate the voice/flow effect.. idk how i can explain this x) but all you send me is really perfect for me ! 😮 BIG thank too

[untold briar]

Remember you can continue from a specific song to start your song, because prompting exactly the same may be different

When you use your own lyrics try to use approximatley same line lengths

[static wigeon]

Yes for the lyrics I noticed that... I asked 3 AI to generate texts and the result on Chirp was bofbof...
I tried with my texts (5 years of writing) and it was a direct banger x)

[static wigeon]

yes i see, i can restart to the second i want... that is really good feature !

If I didn't have to go pick up my puppy in a week I would have already paid credits to do lots of tests lol

[fallow gull]

I was using https://audiomaster.ai/ to quickly clean up suno generated songs. Are there other quick and easy audio remastering sites?
For the song I currently have it is a bit questionable whether audiomaster is "better", the vocals are crisper and brighter, but not sure if that is what I want.

[static wigeon]

I know of software that allows you to isolate each part.. it even allows you to remove the delay and reverb etc.. but I think the best thing is to isolate the voice and go to sound software like FL studio and improve it with tools

my biggest problem is the lack of vocabulary to better describe what I want... I really have no musical terms, damn im stupid... i just need to find a good beatmaker or sound engeiner in french x)

[fallow gull]

Do you have access to Google Gemini or Chat GPT? You can the AI to describe a song, particular if it is popular. I prefer the way Gemini describes them right now.
I ask stuff like "How would you describe anarchy in the uk by the sex pistols?"

I'll also ask other questions for musical terms like "What is it called in music where 2 people sing different parts overlapping on top of each other?"

[static wigeon]

i try something like this for Juice Wrld on GPT and yes the result is really cool but he give me many many word x) not real prompt :/ and i try to make it in prompt and is not really Juice wrld vibe :/

[static wigeon]

That's the kind of thing that doesn't cross my mind 😦 😅

like a "balade" ?

in french is balade but if i translate is Ride XD

[untold briar]

I think the fastest way to learn Suno prompting is keep browsing Explore NEW songs page. Just keep looking at example prompts and listening to what they sound like.

[static wigeon]

Can we search with a specific word? like Hyperpop for example?

[untold briar]

No. Probably in the future, but not now. But there is still plenty to learn from.

[static wigeon]

This is just the beginning 😄

[fallow gull]

Do you mean 'ballad? like an often slower love song?

[static wigeon]

ah yes XD my bad x)

[fallow gull]

I haven't tried "ballad" in v3 yet. In v3a, "heavy metal ballad" often resulted in a song that was too slow for what I wanted.

[static wigeon]

write the bpm ? or up the bpm

[fallow gull]

I'm not sure if bpm makes a difference in v3.

[candid frost]

Does.

[stone bluff]

Sometimes I can't finish a song. Usually if I press continue, it completes in about part 3, but when it doesn't, it becomes a blockbuster, like part 12. I wasted 80 credits today lol.

[night dirge]

Like using Moises.ai, isolating all the stems, mix them, and mastering again? I could try that haha

[night dirge]

https://sonoteller.ai