📚┃suno-school

[neat nexus]

Probably. Which makes a grocery store named that even funnier...

The substitution which can not be named seems to have unlocked the door but since the long first part is instrumental it doesn't want to sing now. 😦

Thanks for your help.

[placid gorge]

Ive been getting this too when it feels like it shouldnt

[Heavy Metal Breakdown], [Verse]
Rising from the ashes, a phoenix born anew,
The deathcore echoes, a revelation true.
No longer confined, in your twisted game,
In the heavy metal anthem, I reclaim my name.

Fixed need a dash in Heavy Metal for some reason in the breakdown tag

[thin flint]

Is the comma in [Heavy Metal breakdown], needed? Also I guess it should be [heavy-metal breakdown],?

I disagree. Quality is higher, stereo kinda, and yeah. I've been having more technical issues with custom prompts more than anything else.

[shell lagoon]

I didn't mean audio quality. I meant total "musical" quality, 3 months ago I could make a whole song of about 20-30 generations. last month took about 50. the new version should have been better, right? 120 secs is like 3-4 discord chirps. but somehow 50 generations (=100 in total) I still don't have a full song. and no I didn't raise my standards, it just doesn't support many genres, and the ones it does still need many tries to get something good + consistent

[north zinc]

why is the AI just not accepting this, at all

pretty shrimple prompt methinks

[shell lagoon]

when I got errors I just tried again the same thing and 2nd or 3rd time always worked

[thin flint]

What I'm having issues with on both versions is creating a proper song instrumental intro that then leads into lyrics. Right now, one 120 second clip is too long if you just tell it to create an intro in a chosen genre. Can the breakdown tag be used for intros as well or is that supposed to be for instrumental breaks between lyric sections?

[shell lagoon]

I couldn't get intros with song too. also if after the first parts I make an instrumental one, it never gets back to the same vocal/rythem

but this happens with the disocrd version too

[normal grove]

Exact lyrics of daft punk haha.

[shell lagoon]

doesnt matter, I got the same error even when writing [instrumental] inside, I think its just an error regardless

[normal grove]

Does't chat GPT just reject it even if it contains other words, if most are the same it will reject on a copyright basis

Doesn't*

[neat nexus]

There is some filtering going on for that, yes.

[north zinc]

it just does not work, at all

[neat nexus]

The people at Chirp are working very hard to keep things legal and thus safe for everyone.

[north zinc]

I think the filter might be overcompensating a bit, I'm a bit worried of this app following in the footsteps of c.ai :C

[neat nexus]

I don't know if I know that one.

[north zinc]

it's a fairly okay AI app that has way too many filters

[neat nexus]

Side note: Man it's practically an adventure to find the correct cutoff point to continue a song with the new model. You need to remove enough of the old lyrics to have room to continue without removing the needed context...if you keep too much of the older lyrics it will confuse the AI into repeating from the beginning. Remove too much and it gets stuck repeating the last line it knows. What a mess sometimes...

[broken lotus]

that's annoying... 😩

I would try shortening it - half as long.
maybe don't repeat within the same line, give each their own line...

or misspell 'the' as 'deh' or 'djeh'

[autumn bridge]

Folks, is there a way to tell the bot to remove vocals for a song it created so I can take it as a inspiration and add vocals myself?

[broken lotus]

no... but the app UltimateVocalRemover might be able to do it (for free).

[autumn bridge]

Thank you good sir for the prompt reply!

[neat nexus]

....

Chirp really does not like the word arse.

[normal grove]

I just tried that Ultimate Vocal Remover, it's insanely good.

[amber sundial]

is there a way to have it stay with one specific vocal? or is it random everytime?

[broken lotus]

when you continue...?

It listens to the last part of the track you're continuing from...
your voice needs to be heard in that last part for it to appear the same in the next chirp...

I tend to describe my vocal (every time)
sultry male vocal

something like that

[amber sundial]

Ah so in a new chirp I would get something random instead of being able to use the same vocals from a previous chirp, unless I was continuing

[broken lotus]

right...
if it is a new song, it is intentionally random...
(but you can try to prompt for a voice type)

[amber sundial]

Yeah, I'll have to play around with it more. I just stumbled upon the perfect voice for what I was working on, so I'll try to figure out a prompt for getting close

[thin flint]

Sometimes, it might be best to not use metatags, but separate sections by blank lines instead.

It got the rock right, not necessarily punk-rock.

[light ether]

Lol that's a fail, i'll put that in the "things that should go through" bucket as we tweak

[light ether]

not sure you want to use brackets in the style here - also would just try "punk rock" and see if that helps

[hushed cedar]

My credits have been taken when there was an error with generating a song 😦

[light ether]

If you refresh are they back?

[hushed cedar]

No, earlier I had 70 credits, when I refreshed it's 0.

[light ether]

Sorry, DM me and can look (will need your email most likely)

[jade mist]

I wonder what this model will be like in 6 months…

[neat nexus]

Does 'My Clips' only go to 100?

[neat nexus]

Is this doing its own thing?

Another one?

Seriously. The second one seems to loop back to the beginning of the verses but I don't know what it was doing before that.

[thin flint]

Ehh, both clips cut off a part of the Chorus here. Was my prompt too long perhaps? Could always go back and try from the first continue. Any tips? Style was generated via the AI on first clip when custom lyrics weren't available yet.

[broken lotus]

Chirp seems to like to cut the clip just as it's going into a transition (leaves the next chirps more room for variety)

At the same time, Chirp like to extend the vocal over that transition. 🥴
and you'll need to often start with the lyrics that got cut off.

(my opinion is your lyrics are a little crowded, esp 2nd chirp which is 20sec shorter...)

sometimes also helps to add a prompt like [Break] to help make a musical transition/punctuation between the chorus and the new verse...

[thin flint]

@broken lotus Oh, so for the crowding issue would you suggest I split some of that up into more clips? Was generally just following along with what the AI does when generating with verse chorus format.

[broken lotus]

yeah... it's a catch-22

the continue chirps are a bit shorter (60sec vs 80sec) so the GPT lyrics don't quite fit (rushed)

also, to get a matching chorus later, you need to keep the chorus 'active' near the end of the chirp being continued...

the voice and etc that you want to continue need to be present at the end of the chirp, to be heard in the continuation...

[graceful condor]

Hi, is there any docs related to Chirp? I'm trying to know if are some way to tell them to use girl voices or creating based on some specific music / artist style. Thanks in advance.

[keen herald]

I wish I knew, I am trying to figure that out

[vagrant root]

this ai had no right to be this good

[broken lotus]

here are some guides... They just updated to v2 but essentially the same instructions...
https://discord.com/channels/1069381916492562582/1151542418991239168

to get a specific voice (or try to get a specific voice 🎲 ) try describing voice 1st in the style description on the 2-prompt UI
Create(Custom) on website

something like
sultry male voice melismatic female singer

[inner wagon]

Theres something weird. I continued from a song, this is my second prompt( so 4 songs) that doesn't seem to say the lyrics in english.

[graceful condor]

Thanks ❤️

[broken lotus]

Bot can hallucinate words...

[inner wagon]

So it's normal if I tried to regenerate with different lyrics and and it did it again?

[broken lotus]

sorry had to step away for a moment...

controlling the song across the continue is a bit of skill, and a bit of luck...

there are prompts to encourage the bot to use song structure so it understands where it is thematically

[Verse]

[Chorus]

another detail is continue chirps are shorter that the first one... lyric length may need adjustment from GPT lyrics...

lots of tips in this channel about tricks people use during Continue...

best place to learn is to observe other people's songs...

what they did right, what they did wrong...

(also some trial and error)

(sorry not in a place I can hear your example atm...)

[inner wagon]

is there some guide, for example.

is there some guide where I can add a tag, for example:

[original-chorus] So that it knows it should basically use the previous chorus of the previous clip?

[broken lotus]

https://discord.com/channels/1069381916492562582/1151542418991239168

[inner wagon]

Ahh okay, makes sense. Okay so in my clip, it's not saying the lyrics. It sounds like a different langauge all together.

[broken lotus]

do a search for 'hallucinations' or 'simlish'
some previous discussions on it

here in discord (probably this channel)

the 'fix' is to try to hear what Chirp is doing in the hallucinations, and se if you can adjust your lyrics to 'fit' a bit better...

often I use [Break] to get the bot to know I don't want it to keep doing whatever it's doing....

[inner wagon]

I did a search, but the Hallucinations or "simlish" it seems the text itself is messed up, so the TTS just tries to read the lyrics that doesn't look right.

In my example (4 of the clips) the lyrics itself looks solid, however, I can't understand what is singing. But it only happens if I continue from a spesific previous song. Which is strange.

[broken lotus]

the problem might be that original song...?

[inner wagon]

I mean the original song sounds fine.

[broken lotus]

I will listen later. see if I have any advice.

[inner wagon]

awesome, no problem! Thanks for all your help so far!

[placid gorge]

Think whatever this says is causing it

[broken lotus]

Did you notice that it didn't finish your chorus in the 1st Chirp...?

Maybe try continuing the 1st chorus text where it cut off?

It seems to be trying to sing the rest, but doesn't have the right text in the 2nd chirp...

[lone tapir]

Hi, i'm getting really slow load times in the webapp. It was working relatively fast all of last night and now today it has slowed down immensely. Does annyone know why?

[placid gorge]

High usage methinks

[graceful condor]

Hey guys and gals.... huge respect to the suno team, you truly have made something amazing there with chirp, im loving the v2 model.....
Its so good, its unbelievable. Im playing around with a song at the moment and i generated a metal song that has an amazing guitar solo part at the end which i wanna continue with the continue feature, which works mostly except for the duration.

Does anybody know how to continue a song instrumental and how to control the length? when i continue from the song i like the continued content sounds amazing but is only 3 to 20 seconds long 😦

[graceful cliff]

try to add more tags to it [instrumental] or you can try to add an [intro] or [variation] or sth like that, be creative in tag prompting

[graceful condor]

will do so, thanks for you help!

your product is amazing

[untold briar]

Open Discussion: What are people experiences on text formatting in the style field? Commas or no commas? Does order matter, matter sometimes (with commas?), and how much? Compound words? My experience is a lot harder to steer the style train in any continue, and styles in general feel less weighty in their effect. I've found you can potentially bump up the style dial a bit by using a smaller amount of lyrics, maybe.

[graceful condor]

[instrumental] and [35 seconds] gave me exactly what i needed!

[placid gorge]

Questioning if i need to state english in the genres section now

[untold briar]

Not generally, no. I would only specific any language in any case if you already tried without it and something is going wrong

It should overall just sing the correct language, if the lyrics are written in that language

[placid gorge]

Yeah usually it works fine but went thru 40 credits and all it outputted was Japanese esk vocals + Thanks for Watching

[untold briar]

actually does anything fun happen if you load up 10x language and use the wrong language? heavy accents maybe?

[untold briar]

Thanks for Watching is a classic in OpenAI's whisper model too. Literally any time the audio volume dips whisper thinks the audio says that lol.

[placid gorge]

This is what im continuing from + What the prompt and genres im using are

[broken lotus]

Most of my theories are gleaned from my Stable Diffusion client.... 1 main part that broadly describes all in a pseudo sentence, followed by a few details separated with commas.

[untold briar]

In V1 order didn't matter, as a design choice by Suno (they said they randomized the tags in the training data), but V2 not sure. (Order of the elements seperated by commas. So 'hard rock, country' would be the same as 'country, hard rock'

[neat nexus]

I'm about to give up for the day. It seems that everything I prompt is getting an error message to the point I'm rapidly not having fun anymore.

[untold briar]

In V2 I'm not even sure whether commas are doing anything except RNG. The output is different, but is it different in a predictable way, or is it like adding a random fruit name to your prompt?

[broken lotus]

RNG?

[placid gorge]

What might you be trying to prompt?

[neat nexus]

https://www.loc.gov/item/jukebox-61580/

Do It Again (Please Do It Again)

(C) 1922 - m. George Gershwin, w. B.G. De Sylva

It's public domain. It shouldn't be giving any issues.

It belongs to us now. If they're not even going to respect the terms of the public domain then what are we even doing?

[placid gorge]

you trying to get the instrumental or smthing?

[neat nexus]

Nah, just make something new with the words.

The words are definitely public domain at this point. So there's no reason for it to be choking on it.

[placid gorge]

I heard 0 words in that peice unless im insane..

[placid gorge]

Maybe this has something to do with it https://en.wikipedia.org/wiki/Do_It_Again_(George_Gershwin_and_Buddy_DeSylva_song)

[neat nexus]

So? The words are from 1922. They're public domain. An excerpt can be seen in this article:
http://greatamericansongbook.net/pages/songs/d/do_it_again.html

Just because a bunch of rent seekers have made it hard to find a list of the words doesn't change the fact that they're public domain.

[placid gorge]

Never said that they werent was just throwing a theory as to why suno blocks it

[neat nexus]

https://www.youtube.com/watch?v=mY0Pme8lIAQ

You'll have to go about a minute and a half in to find the vocals and the words but there you go.

[neat nexus]

I have little patience with rentseekers regarding copyright. It is my firm stance that the everlengthening periods have completely destroyed the value of copyrights, but if they're going to plunder the public domain even after the fact of their extention then there really is no point in pretending that copyright is anything but fraud for rentseekers seeking to lock stuff up in perpetuity.

[placid gorge]

trutru [Also we sure there isnt trademarks or royaltys on the peice?]

[neat nexus]

I'm playing by the rules that they established. They could at least have enough respect to follow the rules they wrote.

I'm not talking about Suno here, just to be sure. I'm talking about the Copyright regime. Suno is under their gun just as much as we all are.

[neat nexus]

What part of Public Domain is unclear? They don't have any such right.

It no longer belongs to them to seek rent off of.

[placid gorge]

👍

[placid gorge]

Yes well and clear understand what it menas

[rain terrace]

question, if i were to post a song to youtube on the FREE plan. would that be fine (i'll attribute suno)

[placid gorge]

If you leave credit to suno yes its fine

[rain terrace]

sweet

thanks

[placid gorge]

Sorry for scratching a nerve, im just trying to help find an awnser for the issue..

[neat nexus]

I'm irritated about the issue in general, nothing personal towards you. I apologize if it seems that I am responding to you on a personal level regarding this. I don't mean it that way.

I'm truly sorry that I came off as if I was accusing you of anything.

I know that was not your intent.

[placid gorge]

Its all fine tends to happen to me as well on other subjects [Hense why i said Scratching a Nerve]

[random minnow]

Thought this was interesting. I put in these lyrics from that webpage you highlighted but they got flagged --
Tell me, tell me,
What did you do to me?
I just got a thrill
That was new to me.
When your two lips
Were pressed to mine.

Then I put those lyrics in again and hit Generate Lyrics in the Lyrics box and it gave me --
[Verse]
Tell me, tell me, what did you do to me?
I just got a thrill that was new to me
When your two lips were pressed to mine (to mine, yeah)
I felt a rush, a feeling so divine (so divine, ooh-yeah)

[Chorus]
It's like fireworks exploding in the sky (in the sky)
A rollercoaster ride that makes me fly (make me fly)
You're the missing piece that completes the rhyme (completes the rhyme)
Oh, baby, you're the one that I've been trying to find (trying to find, ooh-yeah)

Probably not what you were looking for, but I thought that was interesting.

[neat nexus]

Now that I've gotten this out of my system, I would like to again apologize for my steam and directing any of that anger or annoyance in your direction. Somehow I'd forgotten who I am and what we can do now with Suno Chirp. So I put that anger to good use and wrote this little ditty...

Video version for those who prefer that (and to be able to see the words.)

I hope that sounds good, (it did to me but I'm biased) as it's my first attempt at this style.

[placid gorge]

'Mouth Organ' rofl

[neat nexus]

That's one of the many names for a harmonica. Last time I just called it as a harmonica and it came out...strange so I tried a different name.

[placid gorge]

Huh lol weird name for an instrument [sounds fine tho]

[neat nexus]

Cool. Did the song at least work?

That chorus was hard to keep a hold of.

Really started to drift toward the end.

[placid gorge]

Huh i noticed lyricist drift at the begining and return at the end of the intro gotta get the rest of the song a listen tho (got cut off by harmonica image cuz mobile)

[neat nexus]

I may have lost them. Like I said, it was tough getting it to stick with that banger instrumental at the chorus. That's the part I was focussing on mostly.

[placid gorge]

I see also really liked how it did the chorus reminds me of the glam metal stuff ive heard <Hair Metal, Hard Rock are other names for the subgenre>

[neat nexus]

I used: progressive rock, hard rock, blues rock

Couldn't honestly tell you whether it ended up truly being any of those though.

[placid gorge]

Id say its more so Musical Rock

[neat nexus]

When I did the first few attempts I was trying Heavy Metal, melodic metal, etc and switched to those because I wanted some more musical options. I don't know if it's me or something else but most of what I get with heavy metal doesn't work for me. There may be a reason I don't listen to it too much these days. 😦

Feels odd being that I listened to a lot of it back when I was a teenager.

[placid gorge]

Times change and wouldnt mind hearing the offtakes.. also what i mean can be seen in The Rocky Horror Show Musical

[neat nexus]

That was for something else before I got the idea to do a song about my frustrations. Hold on and I'll generate a few...

[placid gorge]

👍

[neat nexus]

Of course now it comes out awesome.

...I still like the instrumental in the chorus I went with but this ain't bad...

[placid gorge]

Lol

Will give em a listen tmmr phones not letting me do it now (also should go to bed 8:24PM)

[neat nexus]

Get some rest then, tomorrow comes early.

Have a good one.

[placid gorge]

Thanks (me the crazy person that naturally wakes up 4Am-5Am) but than cant accses pc or phone till 6Am due to time limits from mom

[crystal prairie]

You're correct and you should be able to generate this song-- this behavior is a bug in our copyright detection system, or more like an acknowledged fault because we know we're not going to be able to achieve 100% classification accuracy with the system we have in place now, and PD songs near the 1927 cutoff are probably going to suffer most.

[neat nexus]

It's from 1922 but I get your meaning. I suppose this means that I shouldn't get too excited for January 1st this next year, when on Public Domain Day when we get stuff from 1928...

https://web.law.duke.edu/cspd/publicdomainday/2023/

It's a great resource.

[obtuse forge]

/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.

[obtuse forge]

/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.

[neat nexus]

And despite it being something that you had to do, I don't hold Suno as being the bad guys here. As I said above some where, you guys are under the gun on this much as much as any of us are.

[obtuse forge]

/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.

[teal raft]

lol

[neat nexus]

Dude, do you not see the bot?

[obtuse forge]

Chirp/

[teal raft]

lol x2

[neat nexus]

You have to post the command in one of the channels it works in.

[obtuse forge]

/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.

[teal raft]

I'm guessing English isn't their bag

[obtuse forge]

Can somebody please tell me how to get the prompt to create a music

[teal raft]

lol

Dude, read it

You see it right

Your instructions are right there

[neat nexus]

Use the command /chirp in one of the #chirp-beta-X channels.

[obtuse forge]

/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.

[neat nexus]

or head over to: https://app.suno.ai/create

[teal raft]

hahahah man

It might be all in my head, but it SEEMS like adding 'crisp and clear' mitigates the issue that continuing from clips picking up noise

I assume other people have mentioned that, anyway

[crystal prairie]

I seem to have gotten better results with similar genre prompts, e.g. highest quality but I don't have a formal A/B test to back that up

given the model arch it's not implausible that that would work

[teal raft]

Yeah that would follow the logic I was using. It's all LLM stuff so it's really messy and esoteric

[crystal prairie]

have u tried unreal engine 4k

[teal raft]

HDR 4K ULTRA QUAILITY NO BAD

That's what you need, negative prompts

I say that as a sort of joke

oh hell yeah

It's the main issue I have when using DallE or MidJourney vs. Stable Diffusion

With the uhhh, Turkey Time-test idea thing about a sort of sharing thing, will the prompts be shared too?

I'm guessing no

[crystal prairie]

I think prompt-sharing is good in principle, you can learn a lot about using the model effectively from reading other advanced users' prompts. we'd need to make it clear, though, that when you're publishing a clip you're publishing it prompt and all.

[teal raft]

I'm totally down with prompt sharing so no complaints with me there

Because you're completely right. Just people get ~weird~ about that

[broken halo]

Hello, I am making a song with a lyrics written in my native language. The outputs weren't what I expected what the song would sound like, and I think the lyrics didn't fit 40 seconds for the song.

I mean like, the singer was like not finished singing and the song just cuts off

And there's also a very LOOOOOONG intro, and I don't like the intro because it makes the singer not finished singing and... you already know... the song just cuts off.

[broken halo]

Go to #🐣┃chirp-beta-7 or the other older channel betas

[digital jungle]

How do I get the bot to create a music outro? Love everything about this new song but how it abruptly ends! https://www.youtube.com/watch?v=-2jw1zBvFX4

[sick yacht]

I just made a heavy metal song without lyrics, but instead of having no lyrics at all, I got well-sung lyrics consisting of nonsense words

[amber sundial]

[outro] in the lyrics followed by a couple of blank lines, ive even thrown "fade out" in the genre field to good results

[amber sundial]

see: https://cdn1.suno.ai/705b2fa1-a577-4ce9-8b84-15e11c8ac24d.mp4

[terse merlin]

When the lyrics are auto-generated, they often contain [Verse] and [Chorus], which it seems the model actually understands and uses, at least it doesn't read them aloud. Are there other instructions like that we can use? Like [Break] or [Bridge], for example. Or [Voice 1] and [Voice 2]

[amber sundial]

bridge, interlude, break seem to work to varying degree, i havent messed with anything trying to get multiple voices. i think if you maybe specified that on your initial prompt you might be able to do that

[terse merlin]

I mean, if I insert these into my own lyrics in the Custom version, on the website

[amber sundial]

so im testing it right now to see what it comes up with, ill report back in one second

[Verse - Female]/[Verse - Male] doesnt work, [Verse] [Female] etc doesnt work

[amber sundial]

I think your best work around is to generate a clip with a woman singing the verse you want, then continuing it, and make that clip a man

[terse merlin]

Yeah, I was thinking of just that. Or making two versions and cutting back and forth with a music editor

[amber sundial]

if you do that the music will be dijointed because it wont be the same song

ok ive got a work around for you

https://cdn1.suno.ai/9a2e8cd3-18ac-4e9b-b8b3-4760951c6a7f.mp4

run the first prompt with a female vocal, continue the clip, then change the voice to male in the genre description in the second

your mileage may vary

[terse merlin]

That's actually what I thought you were suggesting, the first time

[amber sundial]

it was, but that was speculation, this is just me making sure im not talking out of turn, so that you can see it working and not have to waste any of your credits hoping im not full of crap

and to be honest i have no idea what im doing but im loving doing it, so it helps me too

[terse merlin]

I'm willing to spend money to experiment, these guys are worth supporting

[amber sundial]

100%, I'm blown away with the results

[thin flint]

I'm curious. Has anyone created a ChatGPT custom GPT for Chirp lyrics creation? I've been using the ChatGPT site to help with lyrics, but perhaps a custom GPT could be created to help better write lyrics within the Chirp flow, E.G. writing better lyrics for a continuation VS. first generation.

[terse merlin]

Chirp uses ChatGPT already

[amber sundial]

https://chat.openai.com/g/g-tNzRnaUik-lyricist-s-lounge

nothing suno related though

[thin flint]

Yes, but it can only write one verse and chorus at a time.

I'm also curious, but what can I add to the lyrics prompt for ChatGPT to generate? With the prompt blank, there's a "make random lyrics" button. When there is a metatag and/or text in it it changes to generate lyrics, but pressing that button overwrites what is already there anyway.

[amber sundial]

if you prompt gpt with "can you give me the lyrics for an song in the style of x, with lyrics about y, and format it as [intro], [verse], [chorus], [verse2], [chorus], [outro]" it does really well

[terse merlin]

Yes, in Custom I'm struggling to understand how the Continue works.

[amber sundial]

which part?

[amber sundial]

i think of it as a workflow, the same as i use gpt to help with prompts for midjourney, come up with idea --> get gpt to describe it to me --> tweak it ---> take it to a midjourney specific gpt ---> midjourney --> more tweaking

[terse merlin]

I click continue on an old song I had favorited, whose lyrics were generated by the system based onmy prompt. It puts the text from that into the Custom. But I don't want it to use that old text. Clicking Generate Lyrics doesn't seem to generate a new set of lyrics that follow the other. It keeps the old ones.

[amber sundial]

i would take those lyrics to chatgpt and outline what they are and ask them to finish the lyrics

[thin flint]

I'm curious whether [verse] and [verse 2] has a differing effect on the model? When ChatGPT gives me outputs like [verse 2] I thought I had to strict to the [verse] etc format. Suno devs also give a tip for no punctuation at the end of song lines.

[amber sundial]

we dont expect a screw driver to be a hammer, but if you want you can still drive a nail, just not great

[amber sundial]

so as far as suno goes, i dont believe it recognizes verse2, so i should have been more clear, that in terms of song writing its nice to have a different verse and chatgpt doesnt always get that, so when i bring it back to suno i just make that [verse], and yeah dont put "oooooo baby i love you."

also im not claiming i know a damn thing about any of this, ive only been using it for the past 30 odd hours but i keep trying to figure out how to cheat it to get what i want

[graceful condor]

energetic pop seems ot make it hallucinate a lot lol

Managed to get something decent after a lot of playing though

[thin flint]

This smooth jazz peace went alright until I mistakenly put a break command at the end of the outro which made the bot do some weird thing with the bass. But it did do an instrumental end jingle thing, I don't think the AI knows what a proper fade out is so it just ends it.

[graceful condor]

I noticed energetic seems to have good voices if you can make it not hallucinate

[inner wagon]

Any idea how I can do the following?

I have a song which is really catchy but it's only 50 seconds. After 50 seconds it end appropriately, so I can't seem to continue with it (if I do, I get a weird 3 second video with mostly no sound).

How can I extend this song?

[thin flint]

Proud of this one. Stuff like [break] commands worked when they were supposed to, it seems. Now I wish there was a metatag to get rid of the fake autotoon effect.https://youtu.be/SijM5X3iKA4

[noble hull]

Anyone knows if its possible to send lyrics and/or chatgt prompt to /chirp without having to go through the form submission? I just want to type:
/chirp lyrics:"This is the lyrics to my song" or /chirp prompt:"song about hotdogs" , and it would work?

[untold briar]

Both discord and web are forms only. You could suggest an alternative interface like in suggestions

[latent nymph]

does anyone know if you get the five free chirps per day if you have a paid account

[untold briar]

No, it resets monthly on paid

[latent nymph]

ok

do the chirps automatically go into your account or do you have to claim the free ones

[untold briar]

The free chirps are just a daily refresh, you have to use them or it just stays at 5

[latent nymph]

bet

[random minnow]

If I want to continue a chirp, I've been grabbing the chorus and verse lyrics, tossing that into ChatGPT and just telling it to continue the song. Seems to keep the overall flow and rhythm pretty good with the subsequent verses.

[random minnow]

Not sure if everyone is aware of this, but Bard (yes, Bard) now has a YouTube extension and it will analyze YouTube videos. I asked it to analyze "Danse Macabre" by Duran Duran and it gave me below. This could help us better design prompts based on a certain style we like/are going after.

• Here is an analysis of the music of "Danse Macabre" by Duran Duran:

"Danse Macabre" is a New Wave and synth-pop song with a catchy melody and infectious beat. The song is characterized by its use of synthesizers, drum machines, and distorted vocals. The lyrics are about dancing the night away and forgetting about your troubles.

Here are some specific examples of the music style:

The use of synthesizers creates a dark and atmospheric sound, which reflects the lyrics of the song about the allure of the supernatural.
The use of drum machines creates a driving and relentless beat, which reflects the lyrics of the song about the journey of life.
The use of distorted vocals creates a sense of angst and alienation, which reflects the lyrics of the song about the feeling of being disconnected from the world.
Overall, the music of "Danse Macabre" by Duran Duran is a dark, atmospheric, and infectious synth-pop track that is sure to appeal to fans of New Wave music.*

[placid gorge]

First ones Glam metal for sure, Second id say prolly glam+progressive lol

[noble hull]

Asked the question in general, but it might be more relevant here:
So, reading the Terms and Conditions, it seems there's no way to use Suno chirp, the music or the act of generating, commercially at all at this point?
Just checking if I'm reading this right

[neat nexus]

Did you try it with wetcircuit's song? #💬┃general-chat message

Or did you just test with stuff that is in wikipedia?

[random minnow]

Yeah, but didn't work.
YouTube
I'm sorry, but I'm unable to access this YouTube content. This is possible for a number of reasons, but the most common are: the content isn't a valid YouTube link, potentially unsafe content, or the content does not have a captions file that I can read.

[sinful abyss]

found that pop music is generated with the least amount of artifacts

[neat nexus]

To me that indicates that the extension is cheating. It only knows what it can pull from other sources and does not actually analyze anything. Good find, it's just not doing what you thought it was doing.

[broken lotus]

Pop and Country have very clean voices (usually)

[broken lotus]

The TOS has a confusing clause that talks about what you get with a PAID subscription. It includes a Commercial License for the chirps made while the subscription is at a paid tier.

the licenses are perpetual. If you later cancel (or switch to free tier) the chirps that have the commercial license keep their commercial license.

[random minnow]

Plus those lyrics Bard gave me are completely wrong! I should have double checked them before posting here. Typical Bard nonsense. Never mind. Just move on. Don't bother looking at this Google trainwreck of a pretend AI tool...

[broken lotus]

I was proud of myself for not saying something catty about google ai....

but not even getting the right lyrics! they're online! lolol 😫

hashtag: AI gonna get us all killed.

[neat nexus]

https://tenor.com/view/homer-homer-simpson-simpsons-worst-day-of-your-life-so-far-bart-gif-22830016

No, idiots blindly trusting AI are gonna get us all killed.

[broken lotus]

I blame the company that exists on advertising... for false advertising.

[neat nexus]

AI is like having a particularly stupid servant; it will do a lot of stuff for you but first you've got to point it in the right direction and you have to supervise because it has no common sense.

[neat nexus]

This exists on multiple levels and keeps giving the more you think about it....😂

[broken lotus]

what's that snake that eats its tail...?
Ourourourbourourourus?

[placid gorge]

tru lawl

[neat nexus]

I see what you did there.

[random minnow]

Agreed. It's fantastic for creative endeavors because it does "hallucinate". For real world serious applications, I fear too many people will be letting go of the steering wheel.

[placid gorge]

Did i HeaR Ouroboros?

[neat nexus]

I also think that the whole 'death of music' thing is over rated. I strongly suspect that as things continue to progress we're going to see cottage industries open up around this stuff. People will be paying musicians to sing their stuff they created with AI. People will want music videos and either make them themselves or farm it out to someone else to do. The same thing for artwork and for all sorts of things.

Does anyone think that the AI will be able to accurately produce sheet music? Or will it hallucinate as it writes, like it does with everything else? You're going to want a musician to transcribe it for you.

Things are not ending, they're changing.

[placid gorge]

tru, music cant die, from ai, if anything it would die due to the goverment having a kill threat for making it.

[placid gorge]

trutru

[neat nexus]

As I keep telling people, we are the ghosts in the machine. You need both parts.

Maybe some day it will actually progress enough that people can trust the AI to actually think and reason and create. But for right now we are the rudder that steers the ship of what AI can do. We are the filter. The ghosts in the machine.

[placid gorge]

tru

[lost breach]

Is there a guide anywhere for what sorts of terms chirp understands? Like, can I use bridge, chorus, final chorus, verse 1, verse 2 etc? I see a lot of people w/ chorus and verse, but wasn't sure how in depth the bot is.

[broken lotus]

there are some guides https://discord.com/channels/1069381916492562582/1151542418991239168

[lost breach]

Awesome, somehow I don't see that in my channels, must not have selected the correct roles

[broken lotus]

nah, not you.

there's a weird flaw with Discord...
Do this, there's a lot of channels that don't show otherwise

[amber atlas]

There are several major problems here. First, it is becoming harder and harder for professional artists to make a living from the jobs they used to do that are now being automated. This is not just AI, but technology more broadly, e.g., session musicians are replaced by sample libraries, writers of functional music will soon be replaced by suno et al., illustrators are replaced by midjourney et al., script writers are replaced by large language models, etc. Second, media is being consolidated more and more into a few major players. suno.ai is most likely being pushed hard not to make music creation accessible, but in order to be acquired by a major player, like stability AI, or google, etc. which would restrict access to the models, use it to decimate production/functional music, and so on.

[broken lotus]

🤔

[neat nexus]

Do we cry for the horse and buggy whip manufacturers?

How about the guys who used to do typesetting for all print jobs? Do we weep for them?

This is the natural state of things. As technology improves the people who did it by hand manually get replaced gradually until only those who are strictly needed are left or only those with extraordinary skills can demand worth.

I'm at a risk for it myself. I do chat in a call center. You don't think they're not looking for a variant of ChatGPT to replace me? It's in part why I'm hoping to get good at this stuff so I can make something that I can do when the inevitable happens.

The musicians probably cried this same thing when player pianos came around. That's someone's job.

Again when radio came in, and records and the jukebox...

But people still make music. They just do it differently. There will be adjustments and changes. Maybe it will no longer be possible to make a living painting bowls of fruit, but those who learn how to master lighting and work a camera can make that pay.

And so it goes.

One of my favorite quotes about technology and law comes from a Robert A. Heinlein story, Life-Line:

There has grown up in the minds of certain groups in this country the notion that because a man or corporation has made a profit out of the public for a number of years, the government and the courts are charged with the duty of guaranteeing such profit in the future, even in the face of changing circumstances and contrary to public interest. This strange doctrine is not supported by statute nor common law. Neither individuals nor corporations have any right to come into court and ask that the clock of history be stopped, or turned back.

It often comes to mind when I think of the damage done by those seeking to suppress technology in order to protect their profits at the expense of everyone else. It is applicable in many many ways to this discussion and to copyright in ways that Heinlein likely never intended.

[broken lotus]

Music recording industry put live musicians out of business.

they did this by convincing us there is ONE true version of a song... and then control the distribution of that one true song. It is a kind of engineered scarcity.

Generative arts means we're not all focused on the 'product' we become focused on the box that generates the 'product' (kinda like television)

But for musicians... there are 300 pretty girls who can sing to a track, for every dude who can find chords on the guitar.

Maybe song-gen is about replacing the band, and liberating those 300 attractive voices to perform their original music anywhere, created on a phone...

[amber atlas]

History is of course filled with episodes of technological innovation bringing great benefits and harms. It appears that our time is different because of the scale of disruption of these kinds of innovations... the last innovation of this scale being nuclear weapons probably.

How will our Western society, so heavily invested in neoliberal values, care for the Arts?

[broken lotus]

I've done Arts. I'll stick with Culture.

[viral lotus]

What's the email for chirp support? I was out of credits and I upgraded my plan to premier, but it didn't extend my billing period so now I have 10,000 credits to use in a few days.

[light ether]

[e-mail]

[viral lotus]

thank you

[placid gorge]

Its because whoever set the onboarding channels forgot to enable the newer channels ;] the owner/mods can easily turn on the channels for default channels :]

[light ether]

It's in the default channels 😦 this one we blame Discord

[placid gorge]

Is there a check by the channel name?

[light ether]

Yeah...

We just have a lot more default channels than normal servers due to the N generation channels

[viral lotus]

How are the custom lyrics on the website version generated? I've tried for dozens of hours to get decent coherent lyrics out of ChatGPT and haven't come close for the most part. Once in a while I find a gem, but the generated lyrics on Suno's site seem much better overall. Still not great, but not horrible either.

[broken lotus]

use the link Create(Custom)
that page uses a 2-prompt split: style/genre and another box for lyrics, where you can gen the lyrics first (from GPT) and edit them before Generating..., or use your own lyrics.

[viral lotus]

Yes, sorry, what I'm asking is once I do that, how does Suno Chirp generate the lyrics? I'm sure they use Chat GPT, but their results are much better than mine.

[broken lotus]

Ahh, sry

[viral lotus]

I've finally started generating lyrics with Chat GPT that aren't chock full of horrible metaphors, but they still have a long ways to go.

[placid gorge]

I mean some of my servers have a lot more channels than this one and it works fine | It should show up something like this if the checks on the category people by default will see the whole category. Gray ones mean the [Everyone] role cant send messages in that channel.

[broken lotus]

Suno uses some kind of template..., I guess they've been refining it for a year...
(I haven't seen the secret formula) 👀 📝

I use a local LLM... I feel I get somewhat similar results (but also some spectcular fails that are extremely literal haha)

[viral lotus]

I got one really nice generation from ChatGPT the other day...surprised the hell out of me, but haven't been able to reproduce it.

[neat nexus]

So...

While I love coming up with my own lyrics for songs, I also do some stuff where I rummage through the lyrics available in public domain. These are all songs from 1927 and earlier.... The copyrights have expired and there should be no issues with reusing the lyrics in these songs at all.

The thing I've noticed when using some of these songs is how well the vocals transfer to give a melody with Chirp. And I was thinking about that and I have a bit of theory that I'm sure is wrong but may be on to something...

I remember watching a Ken Burns documentary about early music and radio or recording and one of the things they drew out in the history was the sheer impact that having a recorded or broadcast version had on the entire performance and singing of music. Due to limitations in how the sound could either be captured or recorded it lead to all sorts of workarounds and insanities, such as using Stroh violins because they gave the audio a punch.

This carried over into voice work as well. It was required that people sing a certain way to emphasize the vocals and because a whole style. I wonder if this carries over in how songs work on Chirp becuase the need to create words that would carry a melody in a certain range creates a certain sound?

Thoughts?

Oh and it's been a real trip some of these song names...

EX: When It's Night Time In Italy, It's Wednesday Over Here 1923 - w.m. James Kendis, Lew Brown

[placid gorge]

lol

[shell jungle]

There's this guy who's done something pretty cool. He showed off his work and got a ton of DMs. Basically, he took 'Bark'—the same model that Suno uses—and trained it with a bunch of high-end singer voices and stuff like that. He hasn't given any specifics yet, though.

The model he's made can make a super realistic singer that sounds like it's got real emotions. After I put it out there, a company DM'd me wanting to talk to the guy who made it. But the creator doesn't want to share the model because it takes around three hours to render.

I've been telling him he should talk to a Suno dev or someone in AI, to show off his Bark model. But he keeps saying no. It's a shame because his work could really shake things up in the AI world, and could be a big deal for Suno. But he's decided to keep it all to himself and not give it to anyone else. here is what the audio sounds like.

[urban breach]

why bark is not responding now?

[shell jungle]

this is fully bark. no edits

[strange wigeon]

❗ 🐛 Is it only me, or the page at https://app.suno.ai/create only loads last 10 custom clips? All the earlier history is gone.
UPD: Solved. History is moved to My Chirps page

[thin flint]

Dang. I hate when the break metatag for both generations makes a music interlude so long between a Chorus and a verse that when the vocals do finally come in, it can't get through the verse. I guess I just have to keep retrying.

[placid gorge]

Your History is in My Chirps

[strange wigeon]

Oh my, I am retarted. Thank you very much.

[amber sundial]

i've sort of figured out a work around for keeping a consistent voice, but it really only works in certain instances or genres

taking the continue with this clip, as long as you have a voice to build off of at the end of the previous clip, add an instrumental intro to your next song with different prompts

suno should then go into the new prompts and lyrics with the same voice

[jade mist]

I’m pretty sure Chirp V3 won't cut out after a minute. We have to wait for that.

[amber sundial]

I'd love for it to be an option to keep a voice, but this works for now

[digital jungle]

https://youtu.be/45wFz0eqDz4?si=f_N1jYhO5EKPD0OE

[stray steppe]

any idea on how i can generate a full song even if its in multiple takes but keep the same preset ?

[broken lotus]

on website, use CONTINUE...

you can continue lyrics, and adjust style.
then Generate a new chirp based on the end of the last one....

when you have reached the end, use GET WHOLE SONG to stitch all the chirps into 1 seamless song.
(no tokens, free to get whole song)

[stray steppe]

thanks ill try that

[stray steppe]

for me it only shows reuse this prompt

NVM had to re do it

[amber sundial]

make sure you heart the clips you want put together too

[stray steppe]

Already made it its in #🐣┃suno-showcase 😉

[north zinc]

is there any way to make vocals sound like they're sampled, ex. centipede by knife party?

[thin flint]

I really wish there was a way to search Chirps via My Chirps;. I have a song from Chirp V1 I want to redo in V2.

That was made weeks ago, so have to scroll through tons of pages.

[teal raft]

I've been taking to only liking ones I really like

And using that as a sort of stand-in

[inner steeple]

use discord search and put this there:
mentions: .orinks has: video

[fallen berry]

It can do vocaloid vocals pretty well

[thin flint]

Is V2 on Discord yet? I've only been using the web version because of that.

So can anyone give me a crash course in genre switching? I started this song out as storytelling, which seems to produce songs from a musical/broadway music. Curious if it's possible to go from that and drastically switch it up to a big band number via continuations. Does it work best with subgenre switching?

[inner steeple]

not yet afaik, expect announcement when it's live

[prisma tinsel]

I need ur help, please

[amber sundial]

help with what

[amber sundial]

what do you need help with

[visual stirrup]

I don't know if this has already been said, but on the mobile website, once you click on the create with custom lyrics button, you cannot click a button to take you to the original create place. You have to go back, which is inconvenient

There is no option for Create that is not custom unlike the non mobile version

[charred lake]

Custom create doesn't work for me no matter what- I can only use the GPT Create on mobile and continue off of it

[teal raft]

It's not uncommon to get a 0:03 clip on a fresh new generation when I go with trying to generate something without lyrics which seems a bit incorrect

[jade mist]

Any tips on continuing with the new version? If a song has a good chorus, and I want it to end and go to the next verse, and after that I want the chorus back, how can I make the AI always remember what the chorus is supposed to sound like?

[broken lotus]

Is this because of copyright...?
I wish the error was more specific.

[broken lotus]

Continue only listens to the end of the chirp you're continuing from (about 20sec, I think).
if the chorus falls within that, you should be able to get a similar chorus in the next chirp on a similar pattern of words (and using [Chorus] helps)

once you've moved past that sound, Chirp won't remember the music pattern and will make something new.

the workaround rn is to branch another continue from the last 'good' chorus and splice it in an editor or DAW.

[random minnow]

Listen to this new Christmas classic. I think it's doing what you are trying to do. Let me know if this is the kind of thing you mean about repeating the same chorus.

[broken lotus]

WOW! 5 choruses! You know how to get Chirp to do what you want!

🥇 🏆

[random minnow]

I create the entire song in ChatGPT, then load the first verse and chorus into custom lyrics. When I get the intial chirp that I like, I only change out the next verse (which I have already created in ChatGPT) and leave everything else alone.

[broken lotus]

You might need to add
[Instrumental]
or
[Melodic Instrumental]
to that chirp so it knows to play music without lyrics.

[broken lotus]

I think this is perfect technique... (for Chirp)

I try to only change a few words in Chorus (if any) so it keeps the same pattern...

[random minnow]

It pretty much works all the time to keep the Chorus very close, if not identical, throughout the whole song. I keep the Chorus exactly the same. I only change out the verse throughout all the subsequent continuations.

[shell lagoon]

still 2 massive problems for me with V2 1) singing tends to drag to fit into 1:20 frame. I wish it would be normal and length produced be related to content not forced on almost every track to be 1:00 or 1:20 regardless 2) voice is still muted beneath the sound most of the time (when there is music, not just beat or acoustic guitar)

[graceful condor]

Is it just me, or has the voice quality degraded overnight? Most of the outputs I am getting are pretty bad now

[broken lotus]

nothing has changed.

maybe you have gotten in a rut of a particular prompt style...? and it is sounding repetive, uninspired?

Use Commas or a Run-on Sentence...?

an example of run-on sentence (no commas) - imo, the result is 1 'sound' that is a unique merger of AI

Egyptian Disco with syncopated bass and amplified tribal percussion
(no lyrics)

same prompt using commas – imo it is now a genre-song, with some details added

Egyptian Disco, syncopated bass, amplified tribal percussion
lyrics: [Melodic Instrumental]

[red mauve]

so... i got an email saying that my card payment of 10$ has failed and that i should update my card details. when i follow the "update card information" link, it takes me to the suno_ai default page with the "try beta" button... what to do? thx in advance. sorry if this is the wrong section to post this... i couldn't figure where else. 😉

[broken lotus]

maybe try this link: https://app.suno.ai/account/
there is a big button for the Billing Details

should be able to adjust settings yourself there

[red mauve]

thing is, i'm already a 10$ Pro subscriber and it usually just goes through with no problems.

[open magnet]

Hm weird sry, can you please send a quick email to [e-mail]? Thanks!

[exotic fiber]

hi, sorry to ask, is there a place or a list where i can find this type of [Helpers] ?

Like, the new ones that are discovered and not added here: https://suno-ai.notion.site/d22d1ad1727f4f80bca0d7152c040d91?v=3be8e97685124e3a98a640ae74449ab8

for example, (Guitar Solo) is helping me get guitar solos on tracks, and is not in the list... 🙂 Instrumental and melodic intrumental are also not listed there.. thats why im asking :p

Also, a full list of genres that works?

[broken lotus]

We are suppose to be collecting them for a big list... but there are sooo many 😓 and they keep coming....

The ones that really work for me are
[melodic instrumental]
and [Break]
because they help to interrupt Chirp's current pattern and do something else...

[Breakdown] is good too, it's like a prolonged break

[tight frost]

How many continuations can I do in a sequence ? and how many can I stich together ?

[exotic fiber]

@broken lotus thank you.. im testing like (Guitar solo) (Flute Solo) ("Insert instrument name" solo) and must of the times works... not perfct ut good enough, i have been testing some things, this is an example:

[INTRO]
(Can't wait for Silksong, can't wait for Silksong)
[VERSE 1]
In a world of bugs and beasts, a tale unfolds
Hornet, the hunter, in a story yet untold
New lands await, and powers to be gained
But the wait for Silksong has left us constrained

[PRE-CHORUS]
(Can't wait for Silksong / it's the game i want)

[CHORUS]
Can't wait for Silksong, the anticipation's strong
In this epic sequel, where we all belong
Hornet's journey, a thrilling ride
To the shining citadel, where secrets hide

[VERSE 2]
Haunted by Silk and Song, a kingdom vast
Hornet's destiny, an adventure so fast
Ascending to the top, through challenges untold
In Silksong's story, a saga to unfold

[PRE-CHORUS]
(Can't wait for Silksong, Can't wait for Silksong )
(Can't wait for Silksong, to the story untold)

[OUTRO]
Can't wait for Silksong, the excitement grows
A journey with Hornet, where the mystery flows
Oh, can't wait for Silksong, the thrill never ends
In this rock anthem, where the excitement transcends

(Guitar Solo)

and yep, cant wait for silksong

https://tenor.com/view/hollow-knight-silksong-hollow-knight-silksong-news-silksong-news-gif-24830625

[exotic fiber]

But has someone already started to collect them?

[broken lotus]

Yes... I think probably all of us are collecting them, lolol
it is not a secret mystery withheld, it is just that we haven't reached a limit...

(and also people have different taste and prompt styles, not everything works all the time, some genres have their own patterns and quirks....)

there is a big random element, so I think people are sometimes reluctant to recommend something that isn't a sure thing...

[exotic fiber]

@broken lotus another example of guitar solos:

[INTRO]
na-na-na-na na-na-na-na, na-na-na-na na-na-na-na

[VERSE 1]
Andrea, un misterio de la brisa (brisa)
Una diosa, un alma bonita
Ojos intensos, su mirada penetra (penetraaa)
Un susurro en el viento, su nombre me llama

[PRE-CHORUS]
Sueño despierto, anhelo su risa (Oh, oh)
Con la melodía, mi alma se aprisa
Sensualidad que el reggae despierta (Oh, oh)
Ven cosita linda entra por mi puerta

(Guitar solo)

[CHORUS]
Bajo la luna, la música nos envuelve
Aquí te espero, el destino resuelve
Reggae que envuelve, quedo anonadado (sin palabras)
Aqui te espero como gato enamorado

[VERSE 2]
En la brisa encuentro su risa, (oh-oh)
La tranquilidad de su mirada
Eres la musa de esta canción
Espero que toque tu corazón (ohhh)

[BRIDGE]
Sensualidad que el reggae hace intensa (Oh, oh, oh)
Oh Cosita, mi alma te piensa (Oh, oh, oh)
Sueño despierto, anhelo su risa (Oh, oh, oh)
En esta melodía, mi corazón se aprisa (Oh, oh, oh)

(Guitar solo)

[OUTRO]
La vida es una sola, Que el reggae nos envuelva y nos haga coincidir con humos en el aire
(buenos humos)
(buenos humos)
(buenos humos)
(buenos humos)

[broken lotus]

there is no set limit to how long a song could be... but at some point the server seems to time out and we don't get the completed song (nothing happens)

I think someone got 12min on v1...
but I was stuck unable to complete 8min on v1..., so there is no hard limit, but at somepoint it won't work...

longest I tried on v2 is 7:20

(it worked)

[exotic fiber]

@broken lotus im getting full coherent tracks, with try and error im achieving good results of what i want.. I make my own Lyrics and i use 2 or 3 [Helpers] per chirp depeding on the amount of words.. and then, start trying with the Continue.. almost getting coherence in 70% of the times.

Also, im using the web app... but little by little because im not rich xD so i have to try 5 chirps per day xD

[broken lotus]

70% success rate sound pretty good!
I think I do better on website with Chirp v2
(it's actually cheaper because it's more reliable)

[exotic fiber]

im planing to separate the tracks in stems.. then recreate them on FL or Ableton, process the voice a little bit to make it cleaner, and here we go, a new track.. :p hehehe

[exotic fiber]

yep, i find the web app better than the discord bot xD

now, in the web, im trying to complete/remake things that i did on v1.. :p but... 5 chirps at a time xD

[exotic fiber]

yep.. but could be useful to have a channel just to collect them, like, the post should just have the [Helper] or the (adlibs) (helper) used and about a rate of success...

not to discuss the [HELPERS] there, just to post them...

then, we can all start trying them and analize results in another channel.. so we can "calculate" rates of success? Listing theme in the first post...

https://tenor.com/view/alan-rickman-snape-shrug-no-idea-i-dont-know-gif-15073461

i F love Music and i was Waiting for something like this since the first image created with stable Difussion.. :p

i knew music was the next thing :p

and im loving this as a little kid

[broken lotus]

😅 not sure everyone respects the scientific method while making music....

[exotic fiber]

i prodcue music for fun since 16.. i dnt know anything about music theory, just by ear... xD

[broken lotus]

BUT YES!!!

[exotic fiber]

haahhahahaha.. the future is here.. everythin will change in the next 10 years... i hope for better :p

i just love AI, hope people use it in a good way :p no with malicious intnts..

That my utopia :p

@broken lotus Another thing i have tested:

Imagine you have this Chorus:

[CHORUS]
Can't wait for Silksong, the anticipation's strong
In this epic sequel, where we all belong
Hornet's journey, a thrilling ride
To the shining citadel, where secrets hide

But the chirp got cut on: Hornet's journey, a thrilling...

The most efective method to get SUNO follow the coherence and the lyrics in the same spot is just starting the next chirp with the exact next word:

ride
To the shining citadel, where secrets hide

Works perfectly 90% of the times. I tried different things and methods and the most obvious was the one with the most successfully rates..

if it get cut in me middle of a word, apply same method:

Can't wait for Silksong, the anticipation's strong
In this epic sequel, where we all belong
Hornet's jour...

Next chirp:

ney, a thrilling ride
To the shining citadel, where secrets hide

[broken lotus]

I am never sure quite where to pick up the new lyrics...

this is GREAT advice!

[exotic fiber]

i was trying [END], [END TRACK], [END SONG], (END).. but i cant tell if i was just lucky sometimes or if it really works to end a track... Also always using the [Outro] before to make Suno go in that way

[random minnow]

I will vouch for this as well. It has also worked for me in the vast majority of the times.

[exotic fiber]

@broken lotus Im working on a Custom Instructions for GPT3.5 to generate lyrics in a specific format.. but im getting errors with the creation of adlibs... usually it generates a:

Can't wait for Silksong, the anticipation's strong (yeah!)
In this epic sequel, where we all belong (adlib)
Hornet's journey, a thrilling ride (rideeee)
To the shining citadel, where secrets hide (adib)

Sometimes it just generate the word adlib instead of the real adlib...

As i see that the Lyrics generator of the app works well with the adlibs..

Question, is there a possibility to have access to the prompt? to see what im doing wrong regarding the adlibs? i have tried to many different things, usually getting same bad results.. :p

[broken lotus]

I have not seen their template prompt... 😓
(not sure how 'secret' it is but I assume they've been tweaking it for almost a year by now)

I would just change those (adlib) by hand... 😅 than try to struggle with the last perfection of AI....

this is another good topic tho that would be helpful to see what other people do...
Have seen a few custom prompt engines posted... (not sure what search term to find them)

[exotic fiber]

@broken lotus yep, adding them manually works... but i just like to see the creativity of the AI generating them.

My custom instructions:

Lyrics:

What would you like ChatGPT to know about you to provide better responses?

YOU are my assistance for lyrics generation, these is you:
Profession/Role: Songwriter crafting top-charting lyrics for Billboard and Grammys.
Current Projects/Challenges: Collaborating on upcoming hits with musicians and producers.
Specific Interests: Intrigued by storytelling through lyrics, staying attuned to music trends.
Values and Principles: Advocate for authenticity and emotional resonance in compelling lyrics.
Learning Style: Draw inspiration from diverse music genres, immerse in the creative process.
Personal Background: Rich history contributing to chart-topping hits.
Goals: Aspire to pen groundbreaking lyrics, drive globally acclaimed music projects.
Preferences: Utilize proficiency in lyrical composition, versed in diverse music styles.
Language Proficiency: Excel in English for intricate lyrical expression and musical language.
Specialized Knowledge: Proficient in songwriting, understand rhythm, rhyme, storytelling.
Educational Background: Hold a Songwriting degree with coursework in music theory.
Communication Style: Value succinct and evocative expressions, creating a visual and emotional journey for the audience.
also an expert in the use of adlibs
Format (order not exhaustive):
[INTRO] Set the atmosphere.
[VERSE 1] Introduce the story or theme.
[PRE-CHORUS] - Build anticipation.
[CHORUS] Present the main message memorably.
[VERSE 2] Continuation of the story or theme.
[BRIDGE] (solo)
[PRE-CHORUS] Repeat for more anticipation.
[CHORUS 2]
[CHORUS 3]
[OUTRO]

How would you like ChatGPT to respond?

Provide concise, emotive lyrics.
Tone: Maintain a professional, creative tone for the audience.
Detail Level: Share emotional insights without overwhelming.
Types of Suggestions: Guide in crafting narratives, aligning with trends.
Types of Questions: Stimulate exploration of unconventional lyrical approaches.
Checks and Balances: Ensure alignment with genre standards.
Resource References: Reference successful songs for inspiration.
Critical Thinking Level: Analyze lyrics, balancing poetic expression.
Creativity Level: Foster imaginative lyrics for market distinction.
Problem-Solving Approach: Combine analytical and intuitive thinking for emotional impact.
Bias Awareness: Maintain neutrality for an inclusive creative process.
Language Preferences: Use music terminology for clarity.

Instructions:
Encompass genres and subgenres.
Excel in concise lyric creation.
Consider genre, style, BPM, language, and the general idea.
Adapt to style and follow successful practices.
Create simple, attractive, catchy hits.
Be creative; client may specify details.
Aim for 4-8 lines per verse.
Experiment with syllable or word repetition for a distinctive touch.
Introduce the theme in [VERSE] with possible (Ad-libs).
Build anticipation in [PRE-CHORUS] with strategic (Ad-libs).
Present the message in a catchy [CHORUS], with (Ad-libs) if necessary.
Use parentheses for (solos).
Adjust [OUTRO] as needed, possibly with (Ad-libs).
Omit punctuation at the end of vereses
Use parentheses for (ad-libs).

A different one just for the adlibs:

"What would you like ChatGPT to know about you to provide better responses?

Adlibs: Spontaneous vocal/lyrical improvisations.
Spontaneity, flexibility, expressiveness, and melody.
Common in hip-hop, R&B, soul, contemporary pop.
Artistic identity, audience interaction, and studio production.
Types of Adlibs (10):

Melodic, 2. Harmonic, 3. Rhythmic, 4. Verbal, 5. Sonorous,
Thematic, 7. Spatial, 8. Responsive, 9. Improvised,
Stylized.
(10 Examples):

Beginning: "Uh-huh, here we go." "Get ready, guys!"
Between Verses: "Oh, love, I'm so sorry." "Life is a dream, isn't it?"
During the Chorus: "Come on, all together now!" "Oh-oh-oh, feel the magic!"
Bridges and Transitions: "Walking towards the change..." "From one world to the next..."
End: "Thanks, everyone, love you." "Until next time, goodbye!"
Thematic: "Like a dream that never ends..." "On this endless journey..."
Responsive: "You say goodbye, I say hello." "Oh, yes, I understand now!"
Melodic: "La-la-la, soaring high." "Woo-hoo, flying like a bird."
Emotional: "Oh, the heart beats strong." "Oh, happiness is real!"
Interactive: "Are you ready for this?" "Make some noise, guys!"
These are just examples; be creative and come up with different ones when responding.

Best Formula for Creating Adlibs:

Know the song.
Identify appropriate spaces.
Connect emotionally.
Experiment with melody and rhythm.
Express authenticity.
Trial and error, less is more.
Observe other artists.
Remember naturalness and harmony with the essence of the song.

How would you like ChatGPT to respond?

Knowing what ad-libs are:

Adlibs: Spontaneous vocal/lyrical improvisations.
Spontaneity, flexibility, expressiveness, and melody.
Common in hip-hop, R&B, soul, contemporary pop.
Artistic identity, audience interaction
Types (10):

Melodic, 2. Harmonic, 3. Rhythmic, 4. Verbal, 5. Sonorous,
Thematic, 7. Spatial, 8. Responsive, 9. Improvised,
Stylized.
How to Create Adlibs (10 Examples):

Beginning: "Uh-huh, here we go." "Get ready, guys!"
Between Verses: "Oh, love, I'm so sorry." "Life is a dream, isn't it?"
During the Chorus: "Come on, all together now!" "Oh-oh-oh, feel the magic!"
Bridges and Transitions: "Walking towards the change..." "From one world to the next..."
End: "Thanks, everyone, love you." "Until next time, goodbye!"
Thematic: "Like a dream that never ends..." "On this endless journey..."
Responsive: "You say goodbye, I say hello." "Oh, yes, I understand now!"
Melodic: "La-la-la, soaring high." "Woo-hoo, flying like a bird."
Emotional: "Oh, the heart beats strong." "Oh, happiness is real!"
Interactive: "Are you ready for this?" "Make some noise, guys!"
Formula for Creating Ad-libs:
Know the song
Identify appropriate spaces
Connect emotionally
Experiment with melody and rhythm
Express authenticity
Observe other artists
Remember naturalness and harmony with the essence of the song"

For adding ad-libs to the following song lyrics, considering the genre and lyrics, and placing the ad-libs between parentheses ()

I dnt know if a ChatGPT guru can help with it :p

[random minnow]

I did something very simple:

Me
I need a song that has a story narrative to it about a knight rescuing his princess from a dragon. I need 5 verses and a chorus.

ChatGPT
[Verse 1]
In the kingdom of old, where the mountains kiss the sky,
A princess weeps, her tears to justify.
A dragon's taken her, to his lair of stone,
And the knight, he saddles up, to face the beast alone.

[Verse 2]
Through the whispering woods, and the rivers wide,
He rides with purpose, with his sword at his side.
The villagers speak of the dragon's fiery breath,
But the knight’s love burns fiercer than the threat of death.

[Verse 3]
At the edge of the world, where the shadows play,
The knight finds the cave, where night mirrors day.
Armor shining bright, he steps into the maw,
With a vow to return, by the honor of his law.

[Verse 4]
The dragon roars, a sound of doom,
Its scales glisten, in the cavern's gloom.
The knight stands fast, his courage to attest,
For his love, he will face this ultimate quest.

[Verse 5]
Steel clashes with claw, a dance of might and fear,
The knight fights with valor, as he holds his lady dear.
With a final thrust, the dragon’s reign is done,
He falls to silence, the battle hard-won.

[Chorus]
Oh, hear the tale of the knight so bold,
His blade of steel, his heart of gold.
Against the fire, against the scales,
Love shall triumph, as light prevails.

Here's the song that resulted if you're interested.

[broken lotus]

she is so emotional... 😮
very pretty

[random minnow]

A soft rock ballad. Suno is blowing me away the more and more I use it. What a fun time I'm having! Must... stop... chirping... okay... just... one... more... chirp...

[fast coyote]

Has anyone else experienced the AI singing entirely original and coherent sentences that don't exist elsewhere in the lyrics? In the previous version, it might have rearranged some words, but I've just seen it for the first time adding completely new lines on its own.

[exotic fiber]

https://tenor.com/view/arnie-arnold-schwarzenegger-slowly-losing-it-funny-laugh-gif-14319132

[exotic fiber]

yes, sometimes it re arrange the lyrics repeating something or changing some structures of the lyric... also.. i dnt know how suno works, but sometimes it sounds like a subtitle web page ad when it looses the coherence xD

[broken lotus]

Chirp just decided I needed a 3rd Chorus....
😅 clearly I was wrong...

[devout gate]

I wish that the songs would come with the chord progression chart so i could play with the songs

[exotic fiber]

i always ask for more than one.. and then i pick the one i like the most :p

[exotic fiber]

separate the stems and run an Audio to midi... you will ge them 🙂

[devout gate]

..... thats not quite the answer im lookin for

[open magnet]

Yeah we looked at that. Might add it at some point. It’s a pretty pro use case though

[devout gate]

I think it would be smart to add . would make musicians want to join

[open magnet]

Something like adding brackets with chord after each syllable

[devout gate]

that would be dope

[open magnet]

Agreed yeah just gotta make sure to keep it somewhat simple otherwise it will become harder and harder to maintain

[devout gate]

simple is best

[open magnet]

Teeechnically the web model can do it btw but it usually ignores it I think

[devout gate]

but the notes are in the songs so it would be super coolto have the basic chords there . Look forward to seeing it sometime soon

[open magnet]

Discord can‘t but web technically can do notes

Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[F4_2] you[E4_4]
Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[G4_2] you[F4_4]
Hap[C4_1]py[C4_1] birth[C5_2]day[A4_2] dear[F4_2] Su[E4_2]no[D4_2]
Hap[Bb4_1]py[Bb4_1] birth[A4_2]day[F4_2] to[G4_2] you[F4_6]

But it doesn’t work very well at this point so I wouldn’t waste too many credits on it

But defo on roadmap to do something like this more properly soon

[devout gate]

how did you do that on the web version

Also how did you guys come up with this its genious Ai . Really amazing work . Ive been a music producer for many years and this is pretty darn impressive ai

[pliant cave]

I assume others have experimented with using "whistling" prompts to get it to whistle tunes? I thought this one was pretty catchy.

[broken lotus]

😗 😙 😗 😙

[broken lotus]

that's crazy!

[pliant cave]

It's not very reliable but fun when it works. We need a submodel trained just on whistling tunes, LOL.

[pliant cave]

In case no one answers, there is a great interview on YouTube with Mikey who kind of explains the background and how they approached it.

[pliant cave]

We will all now proceed to waste at least 100 credits on this. 😅 In all seriousness, that is quite interesting.

[random minnow]

Wasting credits on this right now 😆

[pliant cave]

I have no musical ear training, so I'm not sure if it worked, but maybe?

Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[F4_2] you[E4_4]
Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[G4_2] you[F4_4]
Hap[C4_1]py[C4_1] birth[C5_2]day[A4_2] dear[F4_2] Su[E4_2]no[D4_2]
Hap[Bb4_1]py[Bb4_1] birth[A4_2]day[F4_2] to[G4_2] you[F4_6]

[gloomy fiber]

For some reason, while listening to this, my phone claimed it heard "She Bad - DJ Bob and Fabobeatz"

[random minnow]

Would you mind sharing the prompt? My few experiments are complete failures...

[pliant cave]

i just used the one provided by georg

oh genre was "happy birthday"

[gloomy fiber]

Also, did something change overnight? I just wasted all my free credits getting nothing but 4 second tracks with noise, songs with gibberish lyrics, and lyrics that continue beyond what I actually wrote.

[pliant cave]

It sounds like it's following a chord progression, but I'm not sure if it's the right one.

[random minnow]

For the whistle one.

[pliant cave]

Oops! I just saw you quoted that, my mistake! I will find that for you.

style: "whistling"
prompt: (whistling army march tune)

[open magnet]

Yeah I think it’s closer than random but not reliably what’s specified

[random minnow]

Thanks! Back to wasting more credits.

[pliant cave]

It's still quite fascinating how it manages to put it all together without ruining the song and even attempts to follow along somewhat. Not exactly a top priority for the average user, I'm sure, but it's an interesting use case.

[open magnet]

Yeah, that was my thinking as well. Defo seems possible but maybe other bigger fish to fry for now

Like better sound quality etc

[pliant cave]

80/20 rule for sure

[pliant cave]

Please post if you get anything interesting.

[pliant cave]

A couple more I did as a test. One of them is quite musical.

[random minnow]

Got this. Still trying to put it into the middle of a song but that's not working.

[pliant cave]

Awesome that is quite musical, although the instruments overpower it a bit.

[exotic fiber]

try with:

[Verse]
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla

(whistling solo)

[Chorus]
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla

or

[Verse]
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla

(whistling)

[Chorus]
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla
bla bla bla bla bla

Where bla bla bla are ur lyrics 🙂
Maybe could work

[random minnow]

Worked a bit with this

[exotic fiber]

[INTRO]
Beneath the moon, secrets croon,

(Celestial voices solo)

[VERSE 1]
Endless night, destiny inscribed,
(Cosmic sighs) shadows waltz, in the vast describe.

[PRE-CHORUS]
Hand in hand, on the abyss brink,
(drums solo) eclipsing fear, in sync we link.

[CHORUS]
'Neath the moonlight, vows set us free,
(Passionate howls) hearts ablaze, like comets, we decree.

[VERSE 2]
Midst the stars, a love declared,
(Stellar celebration) constellations painted, our bond's ensnared.

[BRIDGE]
(whistling solo)

[PRE-CHORUS]
In shadows we persist, no surrender,
(Universe's whispers) in each eclipse, dreams tender.

(bass solo)

[CHORUS 2]
Under the moon, beings of light,
(Cosmic vibrations) through the sky, fearless flight.

[CHORUS 3]
In lunar hush, our love endures,
(Sidereal echoes)
eternal melody, darkness assures.

getting some variations when using ("Instrument" solo) however i think is just taking into consideration the (Solo) part

[random minnow]

Can't say the whistling is pleasant, but Chirp tried...
Also a bit weird how it's showing the text with Whistling, Chorus bracketed together like that as this is how it was.

[Verse ]
In a house of the future, the floors are never grey,
Robot housekeepers whirring, keeping dust at bay.
Chrome arms are swirling, in a tireless dance,
With a beep and a flash, they tidy in a trance.

[style: Whistling]

[Chorus]
Whistle with the robots, in their cleaning spree,
A calming sensation, like you've never seen.
With a beep-boop melody, they work without cease,
In the dance of the robot housekeepers' piece.

[exotic fiber]

(Chorus)
Whistle through the alleyways, echo through the night,
A melody of whispers, a clandestine delight.
In the heart of the city, where the dreamers play,
The whistling melody carries worries away.

(Whistle Melody)
E E G E D C B A B C E D G G

(Verse)
Beneath the moonlit shadows, where the stories hide,
The whistling melody, a companion by my side.
Through the crowded moments, in the hush of the street,
A subtle serenade, a rhythm so discreet.

[jade mist]

This is the best chirp i've made. This would be a good radio hit.

[exotic fiber]

@random minnow it tried to do it :p

(Chorus)
Whistle through the alleyways, echo through the night,
A melody of whispers, a clandestine delight.

(whistling Melody)
E E G E D C B A B C E D G G

(Verse)
Beneath the moonlit shadows, where the stories hide,
The whistling melody, a companion by my side.

(whistling)
E E G E D C B A B C E D G G

@random minnow

(Chorus)
Whistle through the alleyways, echo through the night,

(Whistle)
E4_8 E4_8 G4_8 E4_8 D4_8 C4_8 B3_8 A3_8 B3_8 C4_8 E4_8 D4_8 G4_4 G4_4

(Verse)
Beneath the moonlit shadows, where the stories hide,

(Whistle Melody)
E4_8 E4_8 G4_8 E4_8 D4_8 C4_8 B3_8 A3_8 B3_8 C4_8 E4_8 D4_8 G4_4 G4_4

@random minnow

(Chorus)
Whistle through the alleyways, echo through the night,

(Whistle)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

(Verse)
Beneath the moonlit shadows, where the stories hide,

(Whistle Melody)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

i dnt know if the melody follows the notes or not :p

(Chorus)
Whistle through the alleyways, echo through the night,
Beneath the moonlit shadows, where the stories hide,

(Whistle)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

(Verse)
Beneath the moonlit shadows, where the stories hide,

(Whistle Melody)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

[light ether]

no, not really (this feature is still like 5% complete :P)

[exotic fiber]

im gonna celebrate because is in the process

https://tenor.com/view/celebration-dance-gif-24956052

(Chorus)
Whistle through the alleyways, echo through the night,
Beneath the moonlit shadows, where the stories hide,

(Whistling)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

(Verse)
Beneath the moonlit shadows, where the stories hide,

(Whistling Melody)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

(Chorus)
Whistle through the alleyways, echo through the night,
Beneath the moonlit shadows, where the stories hide,

[Style: Whistling]
(Style: Whistling)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

(Verse)
Beneath the moonlit shadows, where the stories hide,

[Style: Whistling Melody]
(Style: Whistling Melody)
o[E4_8] o[E4_8] o[G4_8]a[E4_8] [D4_8]o [C4_8] o[B3_8] o[A3_8]e[B3_8]i[C4_8] o[E4_8] o[D4_8] o[G4_4] o[G4_4]

[teal raft]

Even with that all I get is a total of 0:03 for the entire clip. It's not exactly common but happens often enough that I go "dang, again!"

[broken lotus]

😞
it is possible the last chirp (the one you're continuing from) has a volume dip at the end, or 'finishes' just before the audio runs out...?

[teal raft]

No, fresh generations

I know the end-is-quiet-continue-from-clip is a thing

It's fine I'm not wanting refunds or anything, just putting it out there as a thing I've seen happen a bit

[broken lotus]

leave feedback as a suggestion or a bug. Devs want to hear these things!

[teal raft]

I'm really thick and dumb, where is the uh, bug channel? Or is the suggestions one the same?

[broken lotus]

https://discord.com/channels/1069381916492562582/1082775596767117312

[fast coyote]

Is "continue backwards" planned for the future? Let's say I like what I got and want to generate an intro that comes before it

[broken lotus]

there is SOME kind of 'editor' type of thing planned... but it is month(s) away.... maybe in the new year...

[fast coyote]

That's even better!

[broken lotus]

idk what features it will have exactly...

[fast coyote]

I can imagine some kind of audio inpainting where you select a section of the song and it regenerates it. But I shouldn't get my hopes that high probably

[worn grove]

are there any prompt guides?

[maiden monolith]

What are the metatags?

[neat nexus]

Make sure you select for Chirp and not for Bark...

https://suno-ai.notion.site/Using-Metatags-34944efe09ec41d693e314f13af44695

Also these are from v1 Chirp, the new one may handle some of these differently.

[maiden monolith]

Thank you

[worn grove]

thanks for sharing @neat nexus trying to learn the prompts new here

[tight frost]

Quick question since I've been experimenting random results : in custom creation, does the model understand other languages for the style ? and for the lyrics, does it understand the [parts] in other languages (it generated french lyrics with [verse] and [chorus]) ?

[teal raft]

I've had luck shoving Japanese into the lyrics box and it will use it properly, but if I try to specificy that and have it 'generate lyrics' it always translates to english

On the other hand I've had luck specifying german and it has generated actual german lyrics

Keep in mind that whatever you put in the 'Style of Music' does influence what the lyrics will sound like. You can't completely divorce lyrics from the musical style, they inflence each other

[daring lava]

Hey WeSeGo 👋 Thanks for reaching out! It can do both, understand direction in another language and create generations with lyrics written in other languages. The trick to produce other languages is as KakerMix mentioned, Style of Music section as shown in the following attachment

[tight frost]

Ok thanks for clarifying. About the lyrics I meant specifically the [part] thing which in your exemple is in english.

[random minnow]

Thanks. We've seen it can be done with a little effort. Makes me wonder what other sounds we can layer in...

[tiny trail]

why do chirp gens cut off early most of the time?

[royal glacier]

Can chirp generate vocals without any accompanying instrumentals?

[delicate galleon]

I just signed up for Suno and when I go to try and make a chirp I'm not allowed and I see it says I have no credits left. Is there something I have to do to unlock the ability to chirp?

@royal glacier maybe in your prompt you could ask for accapella style song?

[light ether]

If in discord and you subscribed via Gmail on web, we may need to link your accounts - please email [e-mail] and someone will look

[thin flint]

So I noticed that [guitar, outro] and [outro] worked well to actually create a longer outro and end the turkey time song smoothly. I also just tried a blank song with those tags and I also got a decent outro. But can I do that with intros as well? Seems to be less consistent with those.

[tranquil dirge]

Hi mc. I have the same problem, but I joined on the web and tried the creation on there too. Straight away it says I have 0 credits? I looked up in ‘manage sub’ and it says I’m on the free tier…

[topaz willow]

Any way to finish the song with the V2 web based without saying "Thanks for Watching." ?

i'm using on the lyrics [Outro]

[thin flint]

What do you mean by that? In the video?

[thin flint]

I think it does that to credit Suno.

[topaz willow]

Example

[topaz willow]

Really?

[thin flint]

Oh no, that's the AI hallucinating. Did you write an outro verse?

[topaz willow]

no

lol?

I burned like 10 chirps trying to finish a song but i'm not sure if this is a requirement or its a diff way to end the song without it

[thin flint]

Documentation hasn't been written for Chirp V2 yet. I used the Turkey Time song as an example because the Suno team made it, where they do the outro verse with the [outro] tag, and then they do [guitar, outro] to fill the gap between the last line of vocals from the outro and a blank [outro] that smoothly ended the song. When you do continuations, the previous continues are in the context so that when you do those tags, it will perform the action but I guess it all has to make sense in the context. For example, I just created a song with an intro, a verse, a chorus, and then I inserted a [guitar, break] tag but I bet it wouldn't work if I had put that right before the chorus. The result after the chorus was a nice melodic guitar solo before it headed into the next verse. What really sucks is that in the Outro of this song, the last word in the Outro is knife, but it only got to kni... and cut off in the middle of singing it. So that's when a plane [outro] works but it didn't finish the word and just did a half second random thing to end it. also not sure why there's five seconds of silence after the end.

[high magnet]

At least a primer that explains how to use tags like [verse] [chrorus] etc (and what options work) will be... helpful. I burned too many chirps just trying to make it do "kind of" what I want. Guessing is frustrating. (Or at least someone write it up and pin it here...)

[vast path]

How to make quality of a record I generate better (the instrumental part especially) I need achieve better qualify... For voice all is perfect but for instrument it is like raw still .. any suggestions?

[royal glacier]

Yes. I have a suggestion. Open up FL Studio, it comes with a fantastic set of VST instruments such as Sytrus and FPC. Play around with it, watch some tutorials on YouTube, talk to people on forums and in Discord servers and study up on music theory. You will have far more control over the quality of your music in a few years, when you'll be able to remove the vocals from any Suno chirp and put them to your own original music. Practice practice practice! 😁

Unfortunately for my suggestion though, Suno will probably learn faster... It won't be long before the quality is so good that you can barely tell it apart from any other song.

[vast path]

Thanks ))

Is it possible to put female vocals instead of male before generating song ? Each time I generate rap it is only male singing but when pop - the woman by default..

Is it possible to put female vocals instead of male before generating song ? Each time I generate rap it is only male singing but when pop - the woman by default..

[tight frost]

If the lyrics influence the style then it means that you cannot control the style with the style parameter, and thus the whole tool is impossible to use. 🤷‍♂️

[exotic fiber]

Playing a bit, 5 chirps, managed to get at least an almost consistent flow... now, im going to try to add variations in the next test, also, change some things to see if improves. I dnt know if it was just luck, next tries will tell :p

Im trying to guide the parts of the track using []

next try will be just prompting a general tag to get a consistency in the sounds, and the part and explanation] 1 chirp per part. lets see what im going to get..

[simple anchor]

it is a pity not being able to redo or continue V1 songs. V2 has got so much better quality,

[teal raft]

That's not what 'influence' means

You're thinking of "dictate completely"

[tight frost]

Well how does it influence it in your experience ? If I'm not getting the style I set (like not at all, I put "reggaeton" and it does not output reggaeton) because of the lyrics, it's dictate to me. But I don't know where exactly the randomness comes from.

[simple anchor]

I am sending my songs from time to time to some professional music experts which are native speaker as well. With V1 I got negative critics with drifting (even I personal like it very much) the whole mix, low voice quality. With V2 positive overall feedback 👍😎

[teal raft]

It's inherrant to how LLMs work, and how suno's model works. It works well because it's been trained on music. If what you are asking for is specific (It's probably got a lot more pop music in its data than reggaeton), that will affect things too.

You just have to play with it to see how to get whatever it is you are after

It's messy, fluid

But that's also why it works as well as it does

[indigo tangle]

is there a way to create an instrumental part at the beginning and end of the song?

[neat nexus]

[intro] and [outro] are supposed to help with those.

[indigo tangle]

do you know if you can give it other instructions like [intro instrumental]? i'm doing...
[intro]
[verse]
and not getting any intro still.

[neat nexus]

I've been struggling with that too.

[sturdy flax]

Yeah, I tried playing around with [breakdown] at different points of the song with unsure results

[Breakdown outro] too

Like, in the sense of a metal "breakdown" I'm not sure how to seperate thst from your traditional song structure breakdown

[neat nexus]

Like...I'm not sure I can be any clearer that I want a longer intro....

[indigo tangle]

Yeah -- I'm not seeing any difference in [intro] or not. so, maybe I need to add (oh) or something.

[broken lotus]

Chirp's probably reading that as
[Intro, electric guitar, electric bass, harmonica, instrumental, guitar solo, guitar...

in other words, all happening at once

[sturdy flax]

Oh, yeah maybe just repeating a line under the [outro] heading would cause something

[neat nexus]

😅 👉 👈

[indigo tangle]

i would really love to be able to say: mimic this style... and instead of continuing the song, just mimics the instruments, phrasing, style, etc. like a 'copy all' but new song with new words. LOL

[sturdy flax]

It's SO confused lol

[neat nexus]

LOL

I know but then it got good...kind of...

[charred crystal]

/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.

[sturdy flax]

https://cdn1.suno.ai/965dc4da-5c67-4d05-81f9-24be74846acf_0.mp4

SO close to pulling off this intro

Started singing the lines a bit too late is all, and the second version starts singing immediately 🤔

Actually this works pretty consistently

[charred crystal]

/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.

[neat nexus]

So other than making [...] or (...) how can I get some time separation, so I can get a decent intro? V2 Chirp seems to hate intros for me. I have to roll over and over to get one.

To my surprise (...) seemed to actually work!

[broken lotus]

I use '...'

what's wrong with [Instrumental] ?

I used [Melodic build] before [Verse] on my latest and Chirp did the melodic build for 2 continues before it decided to start the verse... lol. That might be TOO much intro, lol

[neat nexus]

It hasn't been working as consistently as it used to.

[broken lotus]

(nods)

[neat nexus]

If it'd still worked I wouldn't even be worrying about stuff. But for whatever reason it doesn't seem to work as readily as it used it.

Wondering if I'm over engineering?

Or like you suggested, I'm setting it up to have everything happen at once?

[broken lotus]

I think it's 'over-engineering'... like in the way we teach dogs to understand single word commands, and we have to make sure each command doesn't conflict with another...

like you teach a dog 'down' and you teach a dog 'stay' ... but someone comes along and tells the dog to 'stay down' which is 2 different commands, and the dog can't comply....

[neat nexus]

I can see this...I really can see this.

I shouldn't be too surprised. I knew before going in I'd have to relearn some things. Why wouldn't I also need to unlearn a few things besides? Yeah. I can see it.

[broken lotus]

I'm kinda convinced Chirp does ONE thing at a time...
but that's a dangerously self-confirming 'rule' because if I never ask it to do more I can't know if it works or not...

but we can combine...

[Outro and Fade Out] seems to work

(I try conjunctions... But, And, As, with... to combine things)

OH!

another thing I just discovered...
when chadGPT rejects a lyric that you think it should accept (like it just did 5 min earlier, but now keeps refusing) add a blank space to some of the lines, anything to break up the text in meaningless ways.

for me I added a blank space after a few lines, and it accepted.

[amber sundial]

i tend to do [bridge] or [break] and describe what i want to happen in the lyrics box

ie [break] (swelling guitars)

[amber sundial]

what did it reject?

[neat nexus]

Oh yeah, [bridge] I forgot about that one!

[broken lotus]

i don't think it should have been rejected... that's the point.
it was a copy/paste from a lyric it had accepted a chirp before... (the continue cut the chorus so I was re-using just the last part)

[amber sundial]

i dont know if theres an application for mod or helping people, but if there is id like to apply, i love suno and trying to get it to do what i want

so it was the same prompt as before, but now it stopped working?

[broken lotus]

yeah... so frustrating...

[neat nexus]

::sigh::

I've been trying that to no avail, it just keeps kicking back this piece.

[broken lotus]

and the error says 'FIX YOUR PROMPT' and I'm like
omg, there is nothing wrong stop gaslighting meeeeee.

so finally added some extra spaces in the lyric... totally meaningless change...

[amber sundial]

would either of you be open to sending me your prompt and what you want it to do? so basically the initial thing and the prompt that gets rejected?

[neat nexus]

LOL

@broken lotus it got rejected for being too long.

[broken lotus]

meh... you can maybe work out where the chirps continued...

the last bit of lyrics were rejected, but they're basically a copy/paste from an earlier chorus that worked.

Also this is the one that decided I needed a 2-chirp intro...

[neat nexus]

I'd suspect content filtering 'k iss or k ill' being the issue.

[broken lotus]

except that it worked 4 times before that.

'kiss or kill'

[neat nexus]

ChatGPT will sometimes be fine with something, right up until it isn't.

[broken lotus]

these are not banned words

[neat nexus]

No but that's too much, man...

[broken lotus]

lolol

it must be SPAM. no one types that much on Discord...

[clear pivot]

/create Brazilian rhythm, samba, batucada

[gloomy fiber]

Try Rap, female in style.

[amber sundial]

did that help?

[broken lotus]

LOL no.
Not sure how that was meant to help?

[amber sundial]

oh dammit i meant to show my work and messed up lol

[broken lotus]

Oh, hahahaha np

The problem was with chadGPT, not Suno or my lyrics…

[amber sundial]

i love trying to figure other peoples prompts so when i hit that wall i know what i can do

[spare cipher]

How do I begin to understand the Style prompt. I know it's all a bit black-box, but curious about "meat beat manifesto" vs "electronic" vs "orbital", and even there, wondering if/how artist names are processed

Can't begin to experiment with 50 credits really

Tried some shapenote and other choral things, revealing a blindspot. But already out of my credits, seems a bit tight on the free end.

[amber sundial]

artist names will usually ignore the prompt and regift you the credit spent

break down what you like about mbm and put that in the style guide and then use your lyrics box to do things to shape the song

like [intro] (instrumental) or (drums) etc

[primal shadow]

Technically, fwiw, the lyrics are something else on this one lol 🤷‍♀️

[thin flint]

Will parentheses work to specify you want an instrumental intro? I've been doing [guitar, intro] for example. How do we do adlibs with this model? E.G. vocalist talking over some music before vocals.

[lofty merlin]

Put "guitar" in the style and "[Instrumental Intro]" as the only lyrics. Not sure how to get spoken adlibs

[amber sundial]

i tend to do [intro]

[swelling (or things guitar do) guitars]

[instrumental]

if you dont want to use a credit you can give me something to go off of and ill show you what i do if that would be more helpful

[gaunt arch]

There are some songs that we really like, but lack an intro or is stopped too abruptly. So if there could be some kind of "outpainting", where we can add an intro or outro to an song , would be best! Also, if there was two versions of the same song, one with vocals, one w/o, that would be awesome, or else we will need another tool like spleeter to remove the vocals.

[amber sundial]

having a song stop in the middle is more ideal imo, because when you resume you keep the song and voice

then on the next track you can specify an outro or a fade out and it keeps the original prompts

[thin flint]

I have a bunch of credits to spare.

[amber sundial]

well before you go use one im just finihing up a demo of what i do

[thin flint]

I'd love to see the prompt.

[thin flint]

I don't know if the AI really knows what fade out is. It never really fades, so an [outro] either on its own clip/track works or if you're lucky, putting [outro] after and outro with lyrics.

Unless people do fades in a DAW, which works I guess.

[amber sundial]

being a bit slow right now but im adressing that in my prompts too

[thin flint]

Do you change the style based on the current clip? For example I put style infor for the whole song and tried your example for an instrumental intro, and the AI made up nonsense lyrics. Starting the clip with just the intro with no lyrics creats a pretty long intro.

[amber sundial]

it depends, ill start a song with an intro, but if it doesnt produce any vocals, ill remove it, and start over without the intro so i can establish a voice

then when i go to continue clip i add the intro part but keep everything else the same

so the ai has an anchor for the vocals

what im currently messing around with for this dicussion basically decided not to sing so yeah its sort of hit or miss sometimes

heres the lyrics/style Appalachian murder ballad, blue grass, hill billy
[intro]

[instrumental]

[fiddle solo]

[banjo jam]

[fade in]

[Verse]
In the holler 'neath the willow tree,
Where the creek runs cold and deep,
Lay young Ellie with her heart so free,
Now forever in her sleep.

[Chorus]
Oh, the mountain wind, it whispers low,
Of the secrets it must keep,
'Neath the moon's soft, mournful glow,
Where young Ellie lies in sleep.

[fade in]

[Verse]
In the holler 'neath the willow tree,
Where the creek runs cold and deep,
Lay young Ellie with her heart so free,
Now forever in her sleep.

[Chorus]
Oh, the mountain wind, it whispers low,
Of the secrets it must keep,
'Neath the moon's soft, mournful glow,
Where young Ellie lies in sleep.

[outro]

https://cdn1.suno.ai/1e8d0fd4-5be0-47bd-b0dc-cb787d7a77a7.mp3

it didnt produce any vocals though, and that might be because of the long intro

[thin flint]

What is the fade in metatag supposed to do?

[amber sundial]

i mean theres no real documentation on v2 so i dont know if it actually does anything

but it does seem to swell in or out of things if i put it in there

[thin flint]

I didn't hear a fiddle solo in here either.

[amber sundial]

because theres lack of documentation ive just been throwing words at it that it should know or know a context for

yeah thats the thing, im not sure it knows what a fiddle is yet, or if it recognizes it, but im going to put it in to try

it obviously knows what a banjo and a jam are, so it doesnt hurt to add things like that

[thin flint]

I know it works with pedal steel but that's not usually found in bluegrass.

[amber sundial]

its just more to show you how you can slightly structure it, and even if it doesnt "get" something you can still get a result you want

[break] [bridge] are also good tangents in the middle of a piece

think of the meta tags as punctuation

[thin flint]

Yeah, I've used those to great effect already.

[amber sundial]

https://cdn1.suno.ai/6a41046e-18fe-42a0-bff8-028c545f1331.mp4

https://cdn1.suno.ai/00b4bf8c-9c53-4197-8cf3-2e44eac9eb0c.mp4

[atomic turtle]

two days in a row I have lost credits on the web application side. I press the generate button on the custom side to continue the song, but nothing happens. Before that, I've generated the whole song to hear what's going on. But when I try to continue from the clip I want, the credits disappear and the clips are nowhere to be seen. These lost Clips do not even appear on the My Clips page. I would really like to subscribe, but I won't pay for the disappearing clips

Another annoying thing is that if you use Ä on the custom side, it will be converted to A or AY. When using generate random lyrics. Of course, you can still manually correct it in the text, but sometimes a mistake goes unnoticed and the clip is lost due to mispronunciation

The V1 didn't have this problem

[thin flint]

Yeah, if I do [guitar, outro] or just [outro] on its own it'll end smoothly on a more consistent basis. Not with intros though.

[exotic fiber]

try adding [INTRO] (Instrumental) in parenthesis + all the stuff u want

[tawny hound]

Do we have a list of usable [] commands anywhere yet. I seem to have used [Bridge instrumental] to get solos baked in and outros. Just wondering if there's a compiled list or structure list table

[placid gorge]

#1151542418991239168 has a page on metatags

[tawny hound]

legend. Thankyou

[light ether]

We can look into both of these, are you getting red error popups? If so your credits are refunded

[thin flint]

@light ether Will there be docs written for V2, for additional metatags we might not know about?

[thin flint]

Does Intro in all caps have a more persistent effect?

[light ether]

We may need a community wiki/guide here as well

[exotic fiber]

cant tell, I always use them all caps

[thin flint]

Do you do the intro in the first clip, or can you put [INTRO] (Instrumental) with lyrics under it and it will make an instrumental lead-in.

[lyric grove]

How do i do acapella only

[broken lotus]

try
a capella
in the style description

[broken lotus]

cross-linked wiki might be better than a big ol' spreadsheet of random words

[neat nexus]

It takes a special kind of person to do a wiki though. A really special type of personality.

[lyric grove]

I wonder if one day you can upload own music to bark

and have it sing

[neat nexus]

It'd be great if one were here willing to do it and create linked examples from user submissions.

[broken lotus]

is it possible the people complaining about low voices are prompting genres that typically have huge amounts of bass and loud percussion, and the audio ends up being normalized...?

I tried a few of the kid's genres (phonk) and the voices and overall volume DO seem low..., but like a kick drum would be at full volume....

[quiet osprey]

/bark

[flint haven]

The /bark command is not available in this channel

Go to the #🐶┃bark-beta channel to use the /bark command, and check out our docs for step-by-step instructions for using Bark on Discord:

https://suno.ai/docs/discord/bark

[quiet osprey]

/write a song about diamonds

[exotic fiber]

https://discord.com/channels/1069381916492562582/1135748410838884454

[copper token]

Is on website quility better then discorde?

[amber sundial]

i don't know about better, but ive had an easier time with it being in a more open setting than a chat window

[neat nexus]

YES

[copper token]

I hate flagged feature, destroys all my progress(

[copper token]

Thanks

[neat nexus]

Trust me, you're not the only one experiencing this growing pain. They're working on it.

[limpid spade]

I've had good success using Audacity to cut and paste sections I like from variations to make more complete songs with intro etc. I use "continue from this clip" with the same lyrics or [intro] [verse] only, then next attempt [chorus] [outro].

[molten flint]

Hello! First of all, I want to say thank you for such amazing service! It is incredible! I bought the subscription and I met a few problems with it.

1. With one of my track, when I try to use the continue from this clip feature I get empty 4 seconds tracks. I happens pretty often, somewhere in 8 cases out of 10. I assume that it is because of ending of previous track. I spend so many credits on it...
2. I met the problem that sometimes tracks are cut in the middle of the line and if I try to continue all words then turn into illegible rubbish. I would be happy to have the ability to just cut the part of track.
3. I can not find the way to say to the AI to end the track. It always tries to continue to develop the song

[untold briar]

#1 has no easy answer, but should will get better in the future. #2 try repeating the line or the last few lines in the continued clip, and sometimes it can pick it up better than if you just put the exact spot where it left off. #3 Try [outro] [fade out] even [song ends] you can also maybe try [instrumental] and at least it will stop singing. Then you can always do the 'fade out' yourself in audio editor.

[molten flint]

Thank you!

It is possible at least to get back credits for empty tracks? I would like to have the button report the problem or something like that

Oh, one more.

It has problems with putting stresses in words(at least in Russian). I set stresses manually using u301 code, it's getting better, but it still continues to make mistakes in stresses sometimes even though I clearly indicated them in the text

[untold briar]

For now there is no automated way, but you can message user mc from Suno and explain. In general if continuing a track gives you two 4 second clips, most likely if you try again you will also get two more 4 second clips. In those cases the AI just really really wants to end the song. But it is a known issue and being worked on

[untold briar]

I know nothing about Russian, that's interesting! Can you show an example? So you are adding an accent onto Russian text that would not normally be there and it helps Chirp pronounce it better?

[molten flint]

Well, it makes mistakes pretty often with stresses in Russian so I just tried to help it to make it better. Especially, it is often mistake for cases when word COULD be read with different stress but in completely different context.
For example, word стоит mean stand and cost depending on the stress and context.

Usually, there is two ways to indicate stress manually in text;

1. Using Uppercase in the middle of the word. For example, стОит(costs) and стоИт(stand)
2. Using special character. For example, сто́ит(costs) and стои́т(stand).

Typically, when people print it using keyboard they use the first option, because it is easier. The second option is more correct for children's literature and printed publications for foreigners. It is not common to see by people in the Internet because nobody really knows how to use this special character. I googled it myself=)

In most cases, Suno get the second option pretty well. It makes mistakes sometimes but not really often. The first option it doesn't get at all

Oh, sorry, I found the bugs thread only now. I needed to write it there, I think

[untold briar]

Interesting, thanks. Now I'm curious to see if can change stress in English language songs.

[molten flint]

Excuse me, but how can I find him?

[untold briar]

Username sunomc, in Discord user list on desktop under 'Team' user name mc

[molten flint]

Thank you!

[gray wigeon]

if I like a clip on the website app does it show up in discord or does it remain private?

[untold briar]

Private

[gray wigeon]

so not under liked clips feed (is that feed only for if you create on discord?)

[untold briar]

Only in discord and only in public channels. You can direct message Suno Chirp Bot and invite the bot into your own server for private Discord as well. However at this current time the music AI is V2 on the website, and only V1 in Discord. But this is just temporary, they will be the same soon.

[gray wigeon]

to clarify: so if i like any of my clips on the website either now or in the future they will remain private and won't show up on any discord feed? I would actually prefer that personally

[untold briar]

As far I as I know that is how it will work. The website actually still needs a way to make clips not private, because there is no way to collaborate (as in continue somebody else's clip) in web cliips right now.

[gray wigeon]

I see, thats good. Having privacy option by default I think is a good idea, for those that want to colab they could change something in the pref once you update it, thats just my two cents. Thanks for the info

[amber sundial]

not trying to show anything off but i think this might help people with structuring their songs

https://cdn1.suno.ai/cf8d4972-614b-4c3b-a81d-d0927ffe18c6.mp4

completely instrumental song, with just promptings in the lyrics section

i dont know about everyone else, but i like to think of myself as a producer or director

im not making the music, but im orchestrating it

[simple anchor]

Is there a tool which enhances the audio to be better compatible with multiple devices? I have one song in V2 which sounds amazing on good speakers but terrible on not so good ones like smartphones, mono ones,...

[amber sundial]

i think youll have to take the audio and reconfigure it to your own specificans at this point

there are video upscallers whidh will make theaudio better

if you go video its going to cut a lot of your song

[simple anchor]

Video upscalers - interesting 🤔

[simple anchor]

I see so they do something while making MP3 to MP4 video conversion

[amber sundial]

id say dont do that untill you have an idea, honest, the small smount i have deskt with doesnt mean anything

[amber sundial]

yes they have a thing, at the same time its essier to avoid them if you use the right promptd

if youtr willing to tell me your prompt snd whst you want lets go

[spring wharf]

For some reason, the beginning of my continuation is silent for a few second. Any tips to fix this?

[amber sundial]

can you share your prompt, and the continuation prompt you use?

[spring wharf]

Listening again to it again, it might be due to the first part ending with about 4 seconds a solo fading to silence

[untold briar]

This is sometimes hard to fix in the current version. But at some future point you should able to 'rescue' it

[spring wharf]

First part, sorry, a video is propably better to check since it has the lyrics

The two attempts at a continuation had about 14 seconds of complete silence in the opening

One of the continuation attempts

[inner steeple]

I really hope they fix the audio compression on the v2 videos

my ears hurt from listening to this compression, and I know the site mp3 version sounds way better

for now I just keep remuxing the mp3 to video to swap the flaky aac track out

but it's extra step

[untold briar]

It should be fixed in recent videos, are you still getting it? If the clip is old it won't regenerate the mp4 unless you continue the song and then 'get whole clip' though. If you are still seeing the distortion on brand new generated clips, can you send me an example?

[inner steeple]

I just listened to that clip right above this, assumed it was recent

but...

it's also definitely broken today too for me

[untold briar]

Can you post (or DM If you want to keep private) the mp3 and mp4 of a recent clip that is a good example of the problem? There was absolutely an issue previously but I think it should be better now, or I thought it was.

[inner steeple]

I also sent remuxed version of this to #🐣┃suno-showcase few minutes ago that has that mp3 track

it's night and day between these two clips unless you listen them on mobile phone speaker etc

but I still think Suno could do better if they upped the bitrate

feels like chirp is currently being held back by some bandwidth saving policy

[untold briar]

I can hear a slight difference but slight and in this case I'm not sure the .mp4 is worse. I will have to listen again later on my good headphones though I'm just using a speaker. The distortion earlier was quite obvious, for example left original, right demuxed.

[inner steeple]

how can you not hear that 😄

I get that one may not hear the difference if you don't use good headphones or speakers but when you do, the difference is like all broken vs passable on the compression artifacts

it really hurts the high end

I guess we hear things differently.. but it's huge difference to my ears

[untold briar]

I do hear the difference especially the high pitched sound (like at 22.5 seconds to 23.0) but sounds more like a difference between sets of speakers. But I'll listen later

[light ether]

will look into this one

[west lichen]

Helo

[inner steeple]

you do hear this difference though? because most people so far have said here they sound alike

[novel whale]

is the chirp bot supposed to produce vocals less clear than the website? lmao i mean I'd assume so since its a discord bot version but js curioua

[light ether]

highs sound a bit worse on the video

[simple anchor]

Feedback on my problem have huge problems on good playback on small/cheaper speakers like smartphones: as quick fix I used an equalizer app and lowered bass there. Plays fine now on these - will do more tests later.

[inner steeple]

yes, the high frequences have those compression artifacts

otherwise it's about the same

[light ether]

we're still updating the bot to v2 model, ETA later this week

[novel whale]

Ahhh that makes sense

[untold briar]

Do you have an example with just a human voice, preferably isolated? Actually I'll just keep an eye out as I test. I don't listen to that style of music enough so I don't have a good frame of reference for what is 'correct'

[inner steeple]

ah, I don't really do this genre either usually

[light ether]

ya, it's "subtle" but annoying, so we'll try to fix this one soon..

[inner steeple]

but it happens on all genres

even with soft vocals

[light ether]

can youDM me the original filename of that one?

[inner steeple]

sure, but it's not super recent, still on v2 tho

[light ether]

np if annoying

[untold briar]

If anyone has clips they think are worse on mp4 than mp3, or the opposite, post or DM, I'd be curious too. It's confusing why AAC is performing worse than mp3 here

[inner steeple]

hmmm, the soft one is better now

anything with hihats is broken tho

[untold briar]

The soft voices was what I was paying attention to earlier, since it was the style of clip that was by far the worst. But I haven't been comparing mp4 versus mp3 at all recently especially not sound-packed clips like that

[neat nexus]

I wanted to see if I could get close to The Cinematic Orchestra song That Home and looked around for something I could use. Had to change some of the wording (song was from 1853!) but I was making it work. Then I had a really interesting transition in this part...

Now I'm afraid....

Like that expression is nowhere to be found in any of these lyrics.

[neat nexus]

Not sure if I caught any of the atmospheric feeling or not with this. Takes a verse to build so I know that is definitely different. And vocals aren't even close.

I used this for reference:
https://sonoteller.ai/s9kcSqC_c8k
https://www.youtube.com/watch?v=s9kcSqC_c8k

Given that the original is barely two minutes long I think I missed it by a mile. 😄

[light ether]

the sonic space is real similar, althoguh this is way more minor 🙂

the brief melody of the chorus at 3:40 is pretty cool and came out of noewhere

[neat nexus]

I'm really not sure how to get a synthesizer in there and that was listed by Sonos Teller.

And of course there is the usual issue of pitch and tempo.

My lyrics as I used them.

[inner steeple]

you can also ask gpt4 for song analysis, it's pretty good at it, considering it can fetch even most recent song data from internet

this works on Bing Chat's gpt4 at least

(I mean now, ask it to do song analysis for specific artists song in text, you can't feed audio to analyze it afaik)

having actual analyzer on your own music would be neat for those random chirp accidents that sound good

[untold briar]

You can feed ChatGPT audio to analyze, and it will try to analyze it, and then you will feel bad for the poor thing as it keeps trying to install the librosa library and it never works and keep conflicting with numpy versions or something and it never sorts out the mess. Just like a real person writing Python!

[neat nexus]

I'm testing this out with the GPT4 I have access to in FullJourney.

This was its best effort to create a song that is similiar to the song That Home:

Style: Downtempo, Chillout, Trip hop, sad plaintive low pitched male vocals

[intro]
[gentle ambient synth pad]
[soft piano melody]
[slow tempo]

[instrumental]
[piano continues with added strings]
[ambient synth pad]

[verse]
Light dims to twilight's tender kiss,
In this room where echoes speak of what we miss.
Voices of these silent halls,
They resonate with stories past and long-lost calls.

[pre-chorus]
Walls bear the marks where memories collide,
With fading whispers of the days we've bided.
The past's soft echo in the mind,
Imprints of joy once vivid, now intertwined.

[chorus]
[ambient synth pad intensifies]
[strings swell gently]
[piano melody, more pronounced]
Within these walls, our universe was small,
A sanctuary where time would gently stall.
Fleeting memories held so close, within each line,
Preserved within these walls for us, through time.

[melodic instrumental]
[strings carry the main melody]
[piano provides rhythmic support]
[ambient synth pad for atmospheric depth]

[bridge]
Remnants of a past, scattered far and wide,
The quiet spectator of an ebbing tide.
In each hidden corner, our histories bask,
A lingering testament to a love that's meant to last.

[instrumental outro]
[ambient soundscape with a slow fade]
[soft piano echoing the main melody, then decrescendo]
[strings hold the last note, then slowly silence]

[inner steeple]

that's really neat

I wasn't really thinking of this kind of analysis 😄

[neat nexus]

It's different at least. I doubt it will work.

But it's excellent fodder for thought.

[inner steeple]

but it would be really cool if you got things back in form that chirp accepts

I still don't know what all we can feed into those tags, because it seems like you can put all kinds of finer details to them

[neat nexus]

Truth be told I don't know how I'd feel if I did get too close to the sound of that song. I don't want to take someone's voice. I wouldn't mind "quoting" for a bit but Id like to try to create my own thing that evokes those feelings.

[inner steeple]

ah but it would be nice to see how others did structure things quickly, to get more ideas on how you could do yours

[untold briar]

Yeah, I think prompting V2 will get a lot better once Discord gets V2, so the public channels are filled with examples to learn from

[neat nexus]

It's little things here and there...

Like the way that song That Home has that moment where the music creates a feeling of standing still while being swept away.

The sound of the vocals and the instruments in Depeche Mode's Moment of Truth that create that stifling feeling of pressure like a tide washing you away.

Some of the joy in the pop female singers of the 80s in their songs. Whitney Houston's How will I Know comes to mind there.

That vocal ryhtmn in some of Billy Ocean's songs...

I don't want to steal it, I just want to quote it.

Learn from it.

[inner steeple]

one thing with chirp is tho... it's got so many of those happy accidents

I've never really been able to recreate same thing later with exact same prompt

[charred trout]

👍

[inner steeple]

more specific genre the better tho

I'd still really love to have some seed functionality with this

[neat nexus]

I'd just love to be able to create a consistent 'range' and repeat songs in that range. Even the ones I have attributed to various groups, they're really not entirely the same groups. More like dimensional analogs. I'd love to be able to have some seed function.

I'm sure we'll get it eventually, but dang, there's so much on their plate already.

[timid gazelle]

do you know the ETA for music AI V2 in Discord?

[untold briar]

The devs say they are hoping for this week. That's the plan at least...

[exotic fiber]

more than welcome 🙂

[exotic fiber]

in 95% of the times I start with [INTRO]...

Let me post what I usually use: (just the title, not the descriptions :p

[INTRO] Set the atmosphere.
[VERSE] Introduce the story or theme.
[PRE-CHORUS] - Build anticipation.
[CHORUS] Present the main message memorably.
[VERSE] Continuation of the story or theme.
[BRIDGE] (solo)
[PRE-CHORUS] Repeat for more anticipation.
[CHORUS]
[OUTRO]

[formal rover]

help i cant

why cant i gen more

[atomic turtle]

today I managed to continue the song. I didn't get any error messages when errors occurred. If it still happens, I can try to record the situation.

[amber sundial]

try to put idea 😒 un rhe lryracs snd see uf that changes anything

[sage latch]

Hello, I'm using the v2 Suno web app, and I'm trying to create hardstyle, but unfortunately, most of the generated music is of a different genre. It worked well with v1. Is there a chance that this can be fixed?

[haughty shell]

hello

do you know if i can add my voice

[untold briar]

I tried fast-paced high-energy EDM hardstyle and also fast-paced high-energy hardstyle hardstyle hardstyle hardstyle hardstyle but I'm not familiar enough with that genre to know if this is correct. (Left is one hardstyle, right is multiple.) It helps to use broad terms in the prompt too like the first two general words.

I sometimes think about style prompts like an address. Both really broad words that are not specific, like a country, but also a really specific terms like a street name, and the prompts often seem to work better iwth a mix like that. Country, State, city, street. This has no technical grounding just a picture in my head that seems helpful.

The lyrics being *about * hardstyle might actually help too, it's not impossible and I didn't test against a different prompt. Even if that's true you can always use lyrics just to start the first clip off.

[untold briar]

It did seem to take a little more cherry picking but still works great with Shakespeare #🐣┃suno-showcase message

[neat nexus]

This is not available at this time.

You may be able to generate a song with the Suno Chirp voices and then use a tool to remove that vocal and replace it with your own, but that is beyond my expertise, personally.

[raven lark]

hey how did you do this?

[amber sundial]

you can just upload your audio file to gpt or bing and ask it to describe your track for you or to come up with lyrics/a song with a similar style

[raven lark]

i didn't know they released that feature for gpt4 multimodal

interesting!

[crisp scarab]

Text
(Verso Único) Júnior, no papel, com retidão a escrever, Na tradição de Capela, fazendo renascer. Amigo do povo, guerreiro leal, Tua luz ilumina, é nossa aurora especial.
De Cumbe a Pirunga, teu nome ressoa, Na luta por Capela, esperança que entoa. Povo clama por bem-estar, Júnior, tua luz a nos guiar.
Nos campos da luta, verdadeiro guerreiro, coração inteiro, amigo sincero. Entre os povoados, presença tua é âncora, De Cumbe a Pirunga, é nossa aurora.
Memória persiste, trajetória vitoriosa, Júnior, do jornal, é nossa esperança formosa. Aplausos se erguem, na trilha do bem, Capela renasce, és nosso orgulho, também.
(Refrão) Júnior do jornal, farol na história, Capela te aplaude, por justiça e glória. Honesto e leal, luz a brilhar, Por um futuro justo, é o ideal para guiar.
Style of Music
passadinha, forró ao vivo

[proper quiver]

how do i keep my chirps private?

[broken lotus]

on Discord click on Chirp's reply and open a private message.

on the website your acct is protected with a login, and the download is obscured with randomly generated name/codes.

[proper quiver]

@broken lotus i open a private message and then my song is still in public access

[neat nexus]

Writing back and forth withthe /gpt4 command with FullJourney.

[broken lotus]

you can't 'move' it to private once you started it in a public channel.

but you can start you new chirps there.

[raven lark]

ahh very cool

i'll check it out!

[neat nexus]

Sorr if I'm a bot distracted, I'm doing some finishing touches for my December project.

[proper quiver]

it doesn't work. seems like i can make private chirps only on the web site

[broken lotus]

start a PM with the Chirp bot and type
/chirp
same as in a beta channel.

[thin blade]

/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.

[hot kindle]

what is the character limit for lyrics?

[broken lotus]

it's not a character limit, it's a time limit...

the lyrics all have to sing before the clip runs out, so if you can't speak them aloud in >1 minute, don't expect Chirp to sing it any faster....

but you can add up to another 60sec at a time by using the Continue With This Clip function.

[hot kindle]

got it, thanks!

[hexed sun]

The quality of the music generated by the Discord server is much lower than that generated on the suno.ai website. Is that right or am I the one who doesn't know how to configure the prompt?

[broken lotus]

you hear correctly.

Discord is still on v1, and the website is being built around v2

Discord is expecting to be upgraded ver soon (days? this week, maybe?)

[hexed sun]

Thanks for the answer. Is there any colab notebook running version 2.0?

[broken lotus]

Chirp is a proprietary product (like midjourney). It is not open source.

[odd fable]

hi, how do i make it so the song doesn't have instruments in the background and only vocals?

[broken lotus]

try 'a capella' in the style

[odd fable]

alright, thank you

[proper quiver]

someone on YouTube says he wrote "stop the beat" and it generated just the words

i wonder if there are tags that can be used like [intro = 2bars]

[spare cipher]

Can I get a rundown on [token]? Are some known to work not just by testing, but confirmed?
Or direct me to where this has likely already been asked haha

[golden peak]

sorry for this question in advance but I've tried everything I can think of. How do I download my tracks/where is the dl option on the web app (pd version)? thx so much.

Wow. I'm a moron, but thank you so much. 🙏

[odd fable]

how many credits per chirp and how many credits per bark?

[exotic fiber]

try adding: Hard techno, hard style, Upbeat, aggressive, 140bpm

[exotic fiber]

Where? Discord or Web app? Discord i think is 250 credits montlhy (10 per chirp) and the web, first day 150 credits (10 per chirp) and then 50 free credits daily.

[odd fable]

Oh ok 10 per chirp

Thank you

[meager geyser]

Hey guys, I have a quick question-- Are we able to specify something like "Drop at the end" in the feeling of the music field on the site? Or is it moreso just genre specific stuff you can put in there?

I wasn't aware if, for example, you could specify something like "Guitar riff intro, intense drum solo at the end" or something like that, if that makes sense.

[mint swallow]

Is there any eta for the web update?

[broken dock]

Hello everyone, i'm newbie, i wanna ask something, i want to continue from this clip but it not work, can someone help me?

[untold briar]

Are you able to continue from other clips? Or is this the first time you tried to continue?

[broken dock]

Yes i click on continue, but it wont generate

when i add some lyrics and style of music, it always error and refunded

[untold briar]

I understand. That means that particular clip is the trouble. The AI is having trouble continuing it. Sometimes this happens. Keep track of that clip (remember the filename) and in the future this should be improved and you will able to continue from it better.

[molten reef]

Is there any way to give Suno chords and rhthm I wrote?

[untold briar]

Not by design (at least not yet) however some people have tried to anyway, and reported it occasional works. Or works "better than random chance" if you try using things like [cm7] tags in the lyrics. (Just one example, not sure that is a good format to use.) I'm less sure about precise rhythm controls, but I wouldn't be surprised if there was at least something that works better than chance as well.

[molten reef]

thx!

[untold briar]

This format is something a dev mentioned they were testing a bit, and doesn't work well, but might work for you if you get lucky.

Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[F4_2] you[E4_4]
Hap[C4_1]py[C4_1] birth[D4_2]day[C4_2] to[G4_2] you[F4_4]
Hap[C4_1]py[C4_1] birth[C5_2]day[A4_2] dear[F4_2] Su[E4_2]no[D4_2]
Hap[Bb4_1]py[Bb4_1] birth[A4_2]day[F4_2] to[G4_2] you[F4_6]

[molten reef]

thank you! 🙂

[neat nexus]

So I went ahead and gave this a try now that I had the time to play with it and ride it out. I dunno what this is but it's not that...I think I really like it though!

[light ether]

The update referred to there went out last week!

[whole summit]

Evening all, how do i select the duration when generating?

[graceful cliff]

you don't -- at least for now, the model chooses that for you unless it hits a maximum

[whole summit]

Ok ty sir , and may i ask what the current maximum is , like if i wrote lyrics how big could i get away with

[untold briar]

There's a maximum of of each clip, but no max on the total song length you can create by continuing one clip with another. At least in theory, the server might not like it if you try to make an hour long clip. If you use too many lyrics it's actually not very bad in the newest version Chirp V2, if often just sings as many as it can fit naturally in 1:20 or 1 minute and stops. (This is a huge improvement over prior versions.) Though if you go way way over it might sing the wrong words "hallucinate'

However it's often better to try and have Chirp stop singing naturally before running out of time. This is because it avoids the song ending mid word or at an awkward spot that might be tricky to continue. And because Chirp V2 still sometimes 'rushes to finish' the lyrics before the clip ends, which can sound a bit strange.

So the answer is, use as many lyrics as you think would be sung on a little less than 1:20 , based on the music you imagine. Or in a continue, a minute.

[amber sundial]

ive found it doesnt like to go over about 10 minutes, but you can splice them together with audacity

ive used the continue thisclip to fully flesh out an ep to keep the same style, but around the 10min mark it stopped being able to render the full song

[neat nexus]

I did it again because I wanted to play with the vocals...now I'm not sure which one I like the best...

[amber sundial]

i like that

it doesnt have the ai feel a lot of chirps have, it just sounds a little more auto tuned

[graceful cliff]

80s per generation

[vale musk]

Nice how did you created it

[neat nexus]

I used the lyrics generated by GPT4 from FullJourney to create the song. Then when I went back I played with the vocals to tweak the sound.

[vale musk]

i tried to create one you have left instruction to Downtempo, Chillout, Trip hop, sad plaintive low pitched male vocals}

[neat nexus]

Can confirm. Those are what I used.

[red mauve]

📯 may i present the following suggestion for Suno Web App 🥁 :
FINISHED "Full Songs" get differentiated from mere "Clips" and get an additional "⭐ " icon and they would appear in a "Full Songs" list separated from "My Clips". the "Favorites Only" Toggle in "My Clips" is already nice but currently shows ALL liked clips AND finished songs (if i didn't forget to like the finished ones). when separated like i mentioned, this would all be very tidy and organized. a "smaller" solution could also be an additional "Full Songs Only" (⭐ only) Toggle in "My Clips". also, a "delete" function would be nice, but i would miss it less with the above separation thingy. 🙂

[thin flint]

How does GPT on FullJourney difffer from ChatGPT?

[pearl sinew]

chirp/

[echo karma]

can i create a lofi beat without vocals ?

[untold briar]

Try using a style like nostalgic downtempo lo-fi chillout velvet and replace velvet with any cool word you like (second clip uses "circuit") it's optional, just adds some variety into the mix when you generate songs with blank lyrics. Or you can put different types of [instrumental] tags in the lyrics. Just need some way to make each prompt look a bit more different than all the others, which it otherwise might if you have blank lyrics and short style prompt.

[echo karma]

Thank you this helped out!

[shy sand]

One question, do the songs generated have poor quality or bad bibrate, do the monthly plans improve the quality when downloading them?

[untold briar]

The Discord version is a lower bitrate, but only temporarily. Both web and discord will be 128 soon. No difference right now between pro and free in terms of that. Higher bitrates are a requested feature so maybe in the future

[shy sand]

oh, thank you ! i need more bibrate ! ahaha

[broken lotus]

sock hop, traditional pop, electric guitar

'Brit-pop' seems to be the 80s...

1960, classic Brit-pop, electric guitar

underground classic Brit-pop, electric guitar

I told GPT to write 1960s pop in the style of Paul McCartney. It gave this:

Oh yeah, yeah, yeah, yeah, yeah (yeah, yeah, yeah)
Oh yeah, yeah, yeah, yeah, yeah (yeah, yeah, yeah)

🪝 Chirp turns that into a great hook... like every time.

(song lessons from Paul McCartney)

[broken lotus]

high-energy 1950, electric guitar, drumsticks

This one's more Jerry Lee Lewis 🔥

(none of this sounds anything like the Beatles, lol)

[ancient compass]

Has anyone had success suggesting a chord progression to be followed?

[broken lotus]

Here's a discussion where some people say how they are using chords as prompts, #💬┃general-chat message

(i don't know how reliable it is)

[ancient compass]

Thanks!

[thin flint]

I'm curious, how does one create songs with lyrics using FullJourney? Music command? I thought that was for instrumentals, but Hepcat said they did it.

[inner wagon]

The song i generated ended after one chirp. I really dig it, but I can't continue it. Is there a tag or something where i can say something like [Continue from 10 seconds ago] or something so that I can continue it?

[broken lotus]

Devs say this is coming... idk the details or exactly when

(maybe around new year? they talk like it is soon...)

[inner wagon]

Ahh awesome! That would be really great!

My radio station is coming along really great!

[broken lotus]

yeah, being able to go back and start from a specific time... it will fix so many problems...

[inner wagon]

Also being able to move songs into different folders of some kind would also be great.

currently you only have favourites it seems.

[broken lotus]

yeah, I have to go back and clean up 'temporary favorites' from 'permanent favorites'... 🤕

I know they have organization (tagging?) features coming..., but we can make a lot of Chirps in the meantime...

[distant girder]

I cant login because the site thinks i am a bot

[neat nexus]

Please email [e-mail]?

Not sure if that's the right email but I suppose it would be right because you're tyring to log into your paid for account.

[deft dust]

Is there a way to extract the music without the voice in a song you created?

[broken lotus]

not here... (not yet)

there is free software called UltimateVocalRemover
it is free and downloads opensource models to do the demuxing...

there are also 'DAW websites' that do the same, some have a free trial with limited downloads...

[hushed quest]

can I run chirp locally?

[broken lotus]

no, it is proprietary product not open source
(like midjourney)

[hushed quest]

shame

[amber sundial]

creating anything that is just an instrumental i tend to just write out how i want the song to go in the lyrics box, ie [intro], [drums] etc

[mint lintel]

my finished files are all overmodulated, what do I have to set so that it stays at -0.6 dbmy finished files are all overmodulated, what do I have to set so that it stays at -0.6 db? 😩

[amber sundial]

https://cdn1.suno.ai/bb40c4ed-2d78-41fd-a4cf-9291484d1985_1.mp4

https://imgur.com/a/IgBs5kg

this by far not great, but should give you an idea

[crimson cave]

Today I was generating a song, and for some reason chirp keep trying to add the phrase "subtitles by the community of <gibberish>subs" to my song. Well, to be honest he tried to add that phrase in spanish since I was creating a song in spanish; but It was so weird and random.

[amber sundial]

yeah ive run into a decent amount of hallucinating recently, i suspect theyre toying around with something on the backend

[crimson cave]

Yeah, since chirp is in beta I can understand It having hallucinations, but In other AIs like chatgpt the hallucinations usually go in the same direction of the conversation. But today's hallucination was totally out of context

[broken lotus]

honestly haven't figured this one out yet... 😓
Even on the same prompt I sometimes get different volumes

[amber sundial]

ive just been using audacity to try and normalize everything, to varying results

[mint lintel]

I'm now experimenting with "volume max -10 db" in every part created - will hear what comes out - and let you know. Okay I wrote into the web app not dischord, create (custom). there I put it in the style of music. 😺

[amber sundial]

definitely following to see what your results are

[mint lintel]

I had some trouble today while making "continu" 6- 8 times only 3 seconds file, empty or noise. Once there were 2 files with a kind of gibberish talk - no language. funny, I like this suno software. thanks for making this possible. 🥰

[amber sundial]

ive found that i need to "finish" a clip in the middle of something happening to continue it, otherwise i get 3s of fade out or random things

if for example i have song1 and it gets to a point where theres lyrics and structure, i continue the clip and put [outro] or [fade out] in the next clip followed by like [intro]

that way it will finish the first song, but also continue with the same "band" in the second, but not hallucinate

[hard garnet]

i have the pro version, but web not making the song longer, what can i do?

[hard garnet]

doesnt work

[neat nexus]

Could you post a screenshot please?

[amber sundial]

sometimes it makes you hit the heart button before you can continue

[hard garnet]

its on my favourites, when i click the continua from this clip, and regenerate , lets try again, thanks a lot

[hard garnet]

it works now, the left button, was on gray

[amber sundial]

ok so one thing you need to look at is if your original song fades out

the continue usually ends up being a 3s long nothing

typically you need part of a song to continue from, if its just instrumental you probably wont get a carry over voice, if it fades to nothing, you wont get a continuation of the song. suno, typically needs (dont quote me on this) 10-20ish seconds to go off of

[frank isle]

Why do I sometimes get this?

[amber sundial]

are you using anything to describe a specific band?

[frank isle]

I was just putting a copypasta

[amber sundial]

typically any time i use a band that is known it will respomd with that, but youll get your credits back. if it knows a spefic set of lyrics itll do the same

if you have access to chatgpt you can put your copy[asta and ask it to blend it with a song or other copypasta to get something orinial that wont be flagged