#๐ชฒโbugs-web
1 messages ยท Page 2 of 1
Thanks yeah noticed that too. Gonna try to prevent in next version. Bit tricky cause technically not โwrongโ ๐
Hey, I keep getting responses where the language is gibberish. Then the same text, will generate it in Englishโฆ any suggestions?
Can you link the clip?
Which language are you trying
i am watching everyone making music but i can't๐ญ
Why when I ask metal or/and hard rock without lyrics so I ask instrumental i have some howls or screams on results. How to suppress that ?
replied in general, 'instrumental' is your friend i think
thanks a lot
So with the credit system, currently 2 days into my subscription and I am a little over half way through them. The problem I am having is, I made a song with an [Intro] and a [Verse 1], the bot only outputs the [Intro] and then stops at 9 seconds, The 9 second intro was the vibe I was going for so I figured, I'll just continue and start at [Verse 1] and add the [Chorus] to it too. The result was it made two, 4-second generations with no words. Before credits, when stuff like this happened, yeah it was annoying but it was free so can't complain. Now that I am paying, feels like I am kind of on a tight leash and every generation matters so when outputs happen like this, its very frustrating to say the least.
sry yeah that's not great, we'll look into it. obv we can't guarantee great generations, since you know...llm :/ but the 4s thing is obv not great, and also uses less gpu on our side than eg a 30s generation that's just not perfect, so will have a look, thx for headsup. do you have a link to the generation or is it in DMs?
yeah not going for perfect but just an output that shows the bot made an effort, I have the bot on my server, I can send you the generations in a DM if you'd like?
๐ yeah totally get it. yeah maybe DM me the clip. i need the uuid to try stuff with it but i can prob find it in the database if u send me the clip
might take a bit cause bot is currently crashing so a bit ship on fire
as if autoscaling ever works correctly :/
okay sounds good, will send it over. oh no hope you guys get it sorted! Could that be a possible reason this is happening?
would love to say yes, but feel unrelated, so wanna see why this happends, should be quite rare
copy that! just sent them over to you
tyty
Is it not working rn for maintenance or just too busy?
compute provider is having issues so we briefly put into mainenance mode so people don't keep submitting without knowing what's happening
hopefully back soon
sorry
looks like we are back!
Attempt to add bot to my own server gives me this error (and trying again with Continue ends up with same message again)
Yeah sorry, discord working on thisโฆ their support is slow in approving more servers :/
I like to start with choruses of songs first to make sure i have a good foundation, the continue feature is great, but i'll love a feature that lets me fill in the blanks before the chorus, like a continuation but one that precedes the clip
Try to add [instrumental] or [interlude] before the continuation text
A little bit like going backwards in time? Yeah, on our roadmap but some research to do
I don't know if this is a plact to suggest features, but it would be awesome to be able to request a specific key to the generation. As in key of the song, like select a specific scale the song will create in.,
yeah super interesting. in principle not hard to do... just gotta find the right types of knobs that are versatile and easy enonugh to use. technically you could even condition on chords or something
Getting this when I try to add Suno to my server. Help?
sorry yeah this is discord limiting us cause too many servers.. we are on it. hopefully they will lift the limit soon
Oh okay, no worries, thanks!
Is there some way to keep or repeat a vocal rhythm? That would be nice. each generation falls further away from the groove set in the first gen by the time i get to the next chorus... its a whole different tune. is there 'seeding' options in future updates? I think this would help with that and also having the ability to use a preferred voice on different songs as well. Thanks in advance.
You can try to use continue to keep the voice / tune you prefer for now.
I see, yeah. We are definitely thinking about how to do this so that you can do verse, chorus, verse ,and have those verses be similar - some research to do
yea thats what i do, the voice stays reasonably the same but each continuation creates a new vocal rythmn...
Thank you so much! That would be FANTASTIC!
Yeah, it's a little tricky so some R&D to do first but we definitely want to do it
Hello, It says supporting my language but I cannot generate the voice.
Hi, I think the videos are a bit bugged for Thai. You can try style of music "thai pop" though.
Change from Ramdom to Thai pop, correct?
most songs these days are just 16 or so bars, a bridge and then repeat. while the vocal rhythms stay the lyrics just change. maybe a command for the total number of bars to be generated before bridge and repeat? IDK... just throwin that out there if it helps. Thanks for this service!
yes, if it still is an issue we can take a look
yeah- definitely thinkign about things like that, thanks
Thank you so muchhh. I will try.
i don't know if you like rap, but also tried "rap" as a style before with Thai lyrics and voices came out (but I dont speak Thai)
Voices don't come out for both style ๐ฆ
I will get back to try again and again ๐
looking at yours, let me see if i can figure out some tips
I accidently blocke sumo_bot. Help me please to unblock it
I purchased your subscription and it's been 6 hours it does not let me create songs
DM'd
sorry discord is blocking us right now cause of usage, see #๐ขโannouncements
I'm receiving this error. Subscribed
Sorry yeah this is discord limit us. Working on it, hopeful theyโll life the limit soon
I can add Suno to my personal server?
Yes you will be able to ๐ there is a brief pause on that for the moment though. Discord has a process that prevents the mass spread of bots (over 250 servers) without approval. They are working on it now. SUNO brought this up here #๐ขโannouncements message
Once it is approved by discord though, check this outโฆ https://suno-ai.notion.site/Add-Chirp-to-Your-Discord-Server-997a3121b373431695b953ce164800b3
Iโm sure they will bring it up in announcements once more servers can be added
yeah very interesting. right now the context window isn't really long enough to pick up repetitions beyond ~30-40 seconds, so you have to do tricks like variations or multiple continuations of the same clip with same chorus and get different versions and then combine in ableton or something.
looking to add more context in the future though so that you can have more cohesion across verses
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
thanks for the feedback. yeah I noticed that too --particularly frequent for non-English
Hi guys , I am facing difficulty installing Bark infinity modle
Is there a simple way to download and use Bark Infinity
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
Another IMPORTANT feature request please: ability to upload my OWN track, then ability for your AI voices to sing/rap to my track...that would be so AWESOME!
In the #Continued from link can it please include which of the two? (Continued from track 1or2)
Language
It will be great to improve the output audio quality the gain is so high and add distorsion on the music and voice
Some feedback about continue 1and continue 2 - I feel this is QUITE clunky and limiting. I was able to do it, however, it seems like a band aid solution. Need a solution that enables users to create at least a complete song as ONE GO, not rely on rather piecemeal solution. If this gets fixed, this will be a really user friendly app, and reduce significant frustrations to others. When I create a full song with the current software, i can clearly see the where each patchwork occurs, which is not super useful and does not sound professional. Ideally, it should sound like a complete song, not pieces of a song stiched together....
Yesterday I tested Chirp for the first time. Here's my review.
- I created all the songs in Russian and I never managed to extend the song correctly.
When extending a song, I receive vocals in an incomprehensible language, not at all what I asked in the request. Also, most of these 30 seconds are music without words and my words only begin at the end, making it impossible to continue the song. - The same problem is present in the usual generation, not the continuation of the song. There is an intro without words for 20 seconds or more and a couple of words at the end. Sometimes words don't start at all.
- It is also unclear what can be used as a description of the genre of music.
Is it possible to make references to famous artists so that Chirp generates their style with a similar voice? What keywords or special characters should we use in prompts? - For some unknown reason, some words are not read at all; instead, something incomprehensible is pronounced.
For example, in my fragments the words โะฑะปะธะทะฝะตัะพะฒโ or โะฝะฐัะปะตะดะฝะธะบะพะผโ were distorted and sounded like noise. I redid the generation several times, and this remains the same in all variations. This is strange, considering that the rest of the text sounds quite good, even the accents are placed correctly almost everywhere. - Sometimes there are very good fragments, but I only got two in 25 generations. It is still very difficult to use the service. In the future, I might purchase a subscription if the length of the fragment is at least a minute and it will be possible to properly extend the song.
Thanks to the developers for your service, you have a great future.
yeah... hard to verify ownership there so we gotta be careful..
um... isn't that (1) and (2)?
yeah..... you should've heard the version from a couple of months ago ๐ definitely working on improving!!
yeah, right now that workflow is limited by the model context window. kinda like how gpt 8k vs 32k etc comes at very high cost. so we gotta make a bunch of improvements (or win the lottery) for extending it. definitely in the works though ๐
thanks a lot, super helpful. so hard to test in other languages that you don't speak :/ the lyrics issues is definitely something we are working on improving. hopefully we can set up some language specific channels or something and ask help from the community for giving feedback. we are a tiiiny company for now chirp is 50+ languages so we need some help from yall ๐
as for genres, here some hints (metatags) in #๐ชฆโdocs-guides-resources
artist names is tricky for obvious potential copyright issues...
Separate channels for language testing is a good idea. I will definitely help you test Russian ๐
If we can verify ownership could we do that?
My question is: I want my own music with Suno singing or rapping over it ๐
Possible?
It doesnโt say which of the two songs it is continued from. โ๏ธ
It can be either, they just have to match. If you heart 1, then you would want to continue 1, and so forth.
If you are trying to trace where it came from, I would just go back and listen to the two that you began with. The difference should be recognizable
"should" ๐
Hello, how can I remove a song created with suno, because I realised the text I wrote is not correct for public
yeah, this is a good suggestion
DM me a link to the song (we are implementing this so you can do this to your creations)
Thanks Georg. Just wish to super clarify what I meany by OWN track. I was referring to a music track that I create outside Suno, perhaps using another software app. The use case is where I need a rapper/singer to sing/rap to a beat/music that I create outside Suno. In any case, just my two cents here.....cheers Jose
Ah got it, yeah thatโs definitely something we are thinking about. Basically overpainting vocals ontop of existing backing track
Another consideration - unique music , unique voices vs Copyright: I came across another app that boasts about ability to generate unique music tracks/singer and rapper voices, which implied the resulting user created track/music can never be subject to copyright infringement. If this is a feature possibility here, that would be good to have, so that users will have piece of mind when creating and using their creations in social media, for example. I did come across an article where Spotify has banned/complained about certain websites as users where spamming them with ai generated music...not sure why that would be an issue, assuming there was a solution that guarantees uniqueness....not sure if that is technically possible with audio AI at this point in time....
I was creating a song with this just for practice but the voice skipped over the last part so it didnโt record the entire thing eg the first part of this video
Was created using the chat gpt option.
Having the option to customise the lyrics would be cool as well especially if they could add animations to it
sometimes when i create a song there is only the melodie but no the singing
Could you share the prompt you are using? A screenshot maybe?
yeah one min
Ok great ๐
"piano & singing" "piano with singing" "sad piano with singing"
these are the different versions i have tried
Iโm assuming this is what you put in the โgenreโ section. Do you add anything to the lyrical input section? Or do you use the ChatGPT function?
๋น์ ์ ์ฌ๋ํ์ง๋ง ์ด์ ๋ ๋์ผ ํด
๋ด ๋๋ฌผ์ด ๋ฉ์ถ ์๊ฐ ์์ด
์ฐ๋ฆฌ ์ฌ๋์ ์ด์ ๋๋ฌ๋๋ด
์ด์ ์ธ์ด๋ ์์ฉ์๋๋ด
์ฐ๋ฆฌ ํจ๊ปํ ์ถ์ต์ ์ง์์ง์ง ์์
๊ทธ๋ฆฌ์๋ง ๋จ๊ฒจ๋๊ณ ์ด์ ๋ ๋์ผ ํด
๋๋ฅผ ์ฌ๋ํ ์๊ฐ๋ค์ ์๋ฆ๋ค์ ์์ง๋ง
์ด์ ์ด๋ณ์ ์ํ๋ง ๋จ๊ฒจ์ ธ ์์ด
๊ทธ๋์ฌ, ๋ด๊ฒ ๋ค์ ๋์์์ค
๋ด ๊ฐ์ด์์ ๋ด๊ธด ์ฌ๋์ ๋๋ ค์ค
ํ์ง๋ง ์ด์ ๋๋ฅผ ์ก์ ์ ์๋ ๊ฑด๊ฐ๋ด
๊ทธ์ ์ฌํ๋ง ๋จ๊ฒจ์ง ์ฑ ๋๋ฅผ ๋ณด๋ด์ผ ํด
that was the lyrice
and yeah i put "piano & singing" in the genre thing
is the space in the lyrice the problem?
Here letโs move this to this channel and we can work on it together https://discord.com/channels/1069381916492562582/1102237570924040212 that way we can keep it all together ๐
Using continue will most likely result in other languages
What language are you beginning with?
chinese
There has been recognized discrepencies with continuing languages outside of English. But let's see if we can hammer away at this and maybe find a possible solution. Could you meet me in https://discord.com/channels/1069381916492562582/1102238006804500580
Hi all, my suggestion: ability to upload your own track and have the AI continue
Dance Diffusion and Jukebox can both do it... I'd like to know what Suno is running?
This can go off-shelf in so many ways (copyright, mal-content, data privacy), and we currently do not support this.
i see... a bit disappointing, I was kind of expecting it.... so what is Suno using under the hood?
On the lyrics front, giving folks the ability to tweak the words before generating the track could prevent some of those janky pronunciations. And creating channels for specific languages where people can give feedback on pronunciation seems like a solid move. Plus more data and spelling fixes for languages that need it. Maybe even showing the lyrics while the track plays so you can double check it matches. Lots of good ideas there.
When it comes to length, no doubt we gotta expand that context window so the AI can remember more and make longer, coherent songs. Building up tracks verse-by-verse could be a good workaround for now too. And being able to generate sections separately then combine โem seems clutch - like doing verses on their own then hooking โem up with the choruses later. Ooh, and a timeline view to arrange everything would be fire.
Finally, on the audio front - really tuning up the model with higher quality data is key. Cleaning up noise in the output could help too. And letting people pick the audio quality if they're willing to spend more compute? Big win.
So in summary - solid suggestions all around. Giving more user control, expanding the context, upgrading the audio, and targeting feedback by language seem like the big ones to me. Appreciate you laying it all out! I feel you on the structure issues. It's tricky getting all the pieces - verse, chorus, bridge, etc. - to flow together nicely right now. A timeline view to arrange sections would be clutch.
And I dig how you mentioned "seeding" motifs and styles. Being able to plant little musical or vocal hooks that carry through the track would make a big difference. The songs would feel way more cohesive.
Bottom line, we gotta get that context window bigger so the AI can remember more of what came before. Right now it's got the memory of a goldfish! A longer attention span would let it stitch together longer, more sensible songs.
Iterative verse/chorus generation seems promising too. Almost like building up a full song line-by-line. Might be a workaround for now.
Oh, and we can't forget about improving the raw audio quality and fixing any janky pronunciation. Cleaning up the vocals would really level things up.
So yeah, those are some solid ideas for making the whole experience smoother.
Would love to see the ability to add a clip of music previously generated back into a continuation set, using identical prompts and seed to flow back into a chorus between verses without variance in how the chorus sounds.
Ex:
Generate [Verse 1]
Generate [Chorus A]
Generate [Verse 2]
Generate [Verse 3]
-> Insert [Chorus A] again.
Generate [Verse 4]
...etc
I'm new here so if this is already possible apologies in advance, I didn't see it.
โI would like to upgrade my plan. However, when I try to subscribe, nothing happens. How can I do this?โ
did you figure it out? "/subscribe" should do the job, did that fail?
I have click on the message account from the chat bot. When I do this, this date you can subscribe. If I do this, nothing happens. I need a type the comments in the chat."/subscribe"I have clicked on the message account from the chat bolt. When I do this this date you can subscribe. If I do this, nothing happens. Ore doI need type the comment in the chat.
hmm let me ask @muted flame
Hello, it seems there's a trouble each time we want to create a RAP song (with french lyrics but I don't if this point is relevant and related). It doesn't work (and I've wasted all my credits while trying over and over again). We get songs without lyrics or songs in a different genre
Yes, I come now by a webpage where I can pay. But it only except credit cards unfortunately.
Unfortunately, in Hungarian he often doesn't sing the lyrics, he just makes the music for it. Other times, he doesn't start singing at the very beginning, he only starts singing in the middle of the text. Could this be due to the smaller data set? Unfortunately, my tokens run out quickly in this way.
hm yeah so the late starting in non-english language we are aware of, bit annoying i know, but should work with continuation. like if you give it the lyrics from where it left off in a continuation then it should work
for not singing lyrics at all it helps to put 'vocals' into the music descriptions. since the model can also do instrumental it will sometimes choose to do that instead of following the lyrics. but "vocals" often helps
yeah sorry that's the only option for now
@muted flame I'm starting to getting issues where there isn't any vocals on my generation's. To answer questions yes I have l enter in lyrics in the correct box. I have examples if required
Want to DM me some examples?
Will do thanks, just wasted so many credits
If you allow me a piece of advice, it might be better to use the credits weighted for the non-English language due to quality differences, e.g. Hungarian language 1 credit, English language 10 credits ๐
I want to propose a refer a friend perk to where just as the name implies, If you get a friend to join the discord then you get some credits for generations. ๐ Keep doing a great job devs
haha will consider it, but the problem is they eat the same GPU :/
definitely working on that, thanks for recommending!
I was hoping that my subscription would allow me to delete gens. Is there a trick I'm missing to get rid of old useless generations? It's getting messy up in here! Thanks!
That isn't possible at this time but it is something they are working on
Looking, will DM
I second this request. Definitely need to clean up my chat w/ Chirp.
Did happened to me too a few times within this week
Hi, I can't make a private channel, i'm a paid customer
Feel free to DM and we'll figure it out
Hey, this is now live. You may need to refresh Discord. Just right click on a Chirp message that you made and hit Apps > Delete Chirp
Hey, this is now live. You may need to refresh Discord. Just right click on a Chirp message that you made and hit Apps > Delete Chirp
Amazing! ๐ค
yeah it's on our roadmap. for now you can also try to get it to do "vocals only, acapella" or something like that, but then you of course don't have the backing track
It seems like when I continue, it's starting at the beginning rather than where it left off.
My last few have been like this.
Could you possibly post an example, maybe we can figure this out ๐
@old wind #๐ฃโchirp-beta-7 message
Okay so I think I can help. Check this out... So it generates 30 seconds at a time, that being said, I would split the entirety of your lyrics into parts and add them into different continuances. The bot is taking everything inputted very literally when continuing. So if you input the entirety of the lyrics when you continue, it thinks it needs to start from the beginning of the input. Try doing 4-8 lines per generation (sometime you can squeeze more depending on the tempo). In the case of your example, it continues at the second verse. So maybe do this...
1st generation
1st verse ->Chorus lyrics
2nd generation
2nd verse -> end
Again, also prepare to break it into even smaller sections for now being that the limit is set to 30 seconds to squeeze the lyrics in. Could potentially give you more instrumental parts (Intros/bridge action/etc)
Hope this helps
@old wind
Ok, but, that wasn't how it was working yesterday morning.
And when Ive tried that, I unable get it to finish the line they were in the middle of. It just cuts off and starts the whole song over. New into, beat, new tone. Every 4 - 8 lines it starts the song over no matter where it is on the chorus or bridge. Or what lyrics i post. I end up with a partial chorus only two lines in, and then it'll give me some guitar solo and into start where it left off halfway through, with a brand new song with a different beat.
Sometimes I'm lucky when it restarts because the two songs sound good enough, but there is zero consistency throughout the song. There is no flow.
I get it, for sure... very early days with all of this stuff. Remember when all the text to images had like 12 fingers? ๐คฃ
Still very much in the early, testing phase to figuring out the best workflow and even then as it becomes further developed that workflow will change, becoming more intuitive, comfortable, consistent and user-friendly.
We walk an unchartered path together, friend.
Tell you what though, keep testing. Try breaking down the parts you can so that they squeeze in as they can without lines being cut off. To my knowledge that works the best thus far.
Also, this is definitely a group growing together kind of situation. So if you learn anything along the way that stands out and makes it a little more consistent. Please share โค๏ธ
@old wind I appreciate the good customer service and the positive attitude, but I can use this in this state. Any practical advice to get things working like they were? Maybe wait on an update or a restart or something?
That I don't know. Better minds than mine hold the keys there. But if I find out, I'll let you know for sure๐
@old wind Thank you for the time man!
Very welcome, happy to help ๐
hey sorry just saw
super weird, nothing should have changed
so you generate a clip, continue it and it sounds like a restart?
was this all a continuation from the same clip or did you see this from multiple different ones
cause technically it's possible that a generation is seen as 'complete' by the model and it will just start a new one. but that should be very rare...
Hey, welcome and thanks for being an early supporter - I'll add to what DAViD said - one thing you if you haven't tried could do is: if your generation get cuts off, when you do a continuation, start the new lyrics at exactly where it got cut off. That's our intended workaround rather than putting the same long chunks of lyrics in every time. We are aware of the fact that stuff can drift over the course of many continuations and on our roadmap to make that a lot better
My Suno keeps ummm..... talking about over throwing the government
Is joke?
No x.x
I typed in "Aaaaaaaaagghhhhhh" and she starts singing about the governemt
Wanna DM some clips?
That's hallucination mode ๐ฆ
when we find a vibe and vocal tone we like we can save it like a preset and keep re using it to generate acapella. we can continue but i find its hard to start a new melody because it uses song structure
it's a great idea, working on this, a little R&D to do
you could try to continue and use [Verse, acapella, vocals only] or something like that and then continue from that. no idea if that would be reliable but could try ๐
@old wind something I was trying to test it out.
love the typo in style lol, seems like it worked though ๐
I see what youโre saying, it seems the slower vocals really stretched it out. Try continuing and picking up from the last words used in the generation ๐
Aimi studios is another AI music platform that will release hopefully later this year that I am really excited about. Seems to be chirp donno how to spell it xd
You mean with entering the last words in the lyrics?
Hello has someone created a private chat on discord with the Suno Bot? is it possible like Midjourney?
I just saw it from the "msuic" lmao
Yep you can DM the bot that's what I do
Yes exactly, in the lyrical section ๐ and if you are using ChatGPT, I believe you can still add influence to that as well and it will add it following the ones you inputted directly
Wont it just be using that as lyric or can I say like continue from:"..."
Here you go ๐
Not sure I follow, could you explain differently
Nvm I think I got it I will just try see if it works
hi, i tried to add the bot to my server, and i get this message "this bot can't join more servers as it has not been verified or is requesting gateway intents it has not been verified for. Ask the bot developer...."
Discord is blocking us to add it to new servers now. We are trying to resolve it. Stay tuned.
Needed more electronic genres, like Midtempo Bass (REZZ, Blanke, 1788-L) and Complextro (beetween dubstep and electro house) (examples: Skrillex - Rock n Roll, Virtual Riot - Energy Drink etc.)
I have a suggestion, extending generations backwards
Sometimes clips may start at a part where it sounds like some stuff has happened before, so this could be useful in those cases
Yeah, it's a good idea, we have to figure out some time travel for this one ๐
Ahaha
The continuation of the song works very badly.
https://discord.com/channels/1069381916492562582/1150883962617593937
best I could do xd (with some video editing)
ugh yeah, i think there is a little bug, working on it sorry
was the credit reset date changed?
We had some new features so gave everyone a few more credits see #๐ขโannouncements
I tried to use the >> continue for my song but it just kept repeating the beginning of the verse, now I'm all out of tokens.
What I'd like in the future is more consistent continuations. In addition longer songs and more lines allowed. Also I have no way of continuing this song because I ran out of Tokens so I'm stuck.
Hi, point me to where you tried the continuation and failed -- I can try to extend it
This is the original, successive attempts just led to it repeating the opening verse.
What's the prompt you entered for continuation?
The exact same Genre and prompt that I had previously used word for word. I tried to continue the 1st one 1>>
k let me try
For continuation, it is good to start where you left off: see #๐โv1-songs message @granite warren
Thanks! I'll keep it in mind!
In the ai it puts several songs over one loop, and it also doesn't do promote too well (atleast not with artists)
For legal/copyright reasons, using artists as meta tags/descriptors isnโt ideal. That being said, hereโs a compiled list of some of the meta tags that carry over pretty consistently https://suno.ai/docs/metatags
Metatags guide prompts but arenโt verbalized themselves. Like stage directions or parentheticals in screenplays, you can use metatags to describe a specific part of your prompt (e.g. [laughter] and [piano interlude]) or to influence your entire prompt by describing speaker characteristics like [whispering] or music genres like [rock].
๐๐ป
Hi! I have made a chirp I really really liked, but unfortunately it didn't sing the last word of the lyrics... Is there a way to continue/regenerate the music clip for that specific chirp, without changing the ~29 seconds it already created nice music for? I.e., just extend the already generated one to sing that one missing word, but using exactly the same style/music/voice parameters it used for the 29 secs?
I know of the continuation feature, but the generated clip from there seems to reshuffle style/music/voice parameters instead of using the ones from the previous clip
Hey there! Try another continuance and this time put whatever your genre is + instrumental in the genre section and [outro] in the lyrical section
or
Try another continuance and this time put whatever your genre is in the genre section and the lyrics that were cut off in the lyrical section
Hope this helps ๐
Thanks for the super quick answer! That's already what I tried, unfortunately it doesn't seem to pick up on the original track's style at all. While it still fits the genre, it seems to use a different random seed or so for the new generation. It doesn't sound like it would have actually continued extending the previous track. I made sure I click on "Continue 2" to trigger the new prompt window, and did not re-send a "/chirp" message
Very welcome, still very early days with this stuff ๐ Still learning new tricks everyday. Thank you for the feedback. As soon as I see something that tackles this, I'll let you know for sure ๐
Thank you! ๐
Hey, this is probably doable in the future, just need to think about how to do this on our end!
Greetings. Apparently chirp cannot do islamic nasheed from what I saw. It's a very interesting and unique genre. Can you guys at Suno train the next versions of chirp in Nasheeds? Thank you
๐คฏ๐คฏ๐คฏ๐คฏ
When asking for a MALE voice choir it just gives strange womens voices
we don't control gender, artists, etc, it is kinda pointless to prompt them
ohh
I think gender specifically isn't totally hopeless. Not nearly as consistent as music genre terms, but things with female vocals or woman singing in the description field do seem to have an influence, on average. But I should double check.
Will that change ?
I agree with Fly -- gender sometimes plays a role but definitely not like the difference between rock and jazz
I've had pretty good luck with "Girl Group" "60's Girl Group" "Girl jazz singer", etc.
suggestion: allow the use of specific voices like bark can do - also allow the tone of the voices and the speed of the voices to be adjusted, and be able to specify if it's one voice, or multiple (at various times during the clip)
pretty sure this is known, but bug report anyway. at times, chirp totally misses lyrics. in this generation, it missed the entire first stanza #1152402964573597837 message
thx yeah this is a continuation bug. i believe this is already fixed in code, but gotta wait for next deploy...
passing in the wrong history so it get's confused on what it has already done
the AI is actually, actively, avoiding that stanza. i tried multiple times after posting this to get it to sing it. I pinged tony on this message #1152402964573597837 message you might take a look at it then scroll up and listen to all the generations with that stanza in them
thx ya chatting with him now
we'll prob have to look tomorrow. annoying little buggy bug
getting some silent generations with just ambient noise occasionally, or generations when i use lyrics that either A) only have them in the first or second half, B) dont have any, or C) start them and then end the generation as soon as its done craming the lyrics together.
this is a great technology but it could really benefit from bar/measure markers i.e. [1] lyrics for the first bar [2] lyrics for the second bar, as well as a [refrain] command such as [1] lyrics [2] lyrics [3] [refrain] [4] lyrics
How about creating a channel exclusively for English music only? I'd like to check the English creations, but I'm getting lost in the mix of different languages.
Hey @stray olive, we've been talking about adding a question to the onboarding flow about your primary language(s) and using the responses to auto-add members to language-specific channel(s). Is that what you have in mind?
You could just made specific channels. Then the user can go them them themselves. Some ppl are Multilingual. being sort of forced into one channel might not be good . So just let the user pick .
Hey @ionic edge, weโre looking into this!
could it be that it returned part of the dataset to me during generation?
You should be all set now, put some extra ones in as well
its look like get me back the original music
I think this is just a bug where it is hallucinating
If the original music is within the last 10 seconds-ish of the song you continue from, sometimes it just goes back and repeats that, yeah. I once thought Chirp was translating languages because it was doing that, but it was actually just repeating lines from the previous clip. Songs repeat a lot so it's probably not too weird of a thing to do for the model, if I had to guess.
i reran it, hope this one is better? (sorry don't speak Hungarian..)
marketing is an interesting style term, seems to kind of be like a jingle sort of, if it's having an effect. Could just be a coincidence too.
yes ,it is hungarian thank you
Suggestion* Increase the paying customer's tokens until bugs are worked out. It takes around 25 chirps to make a coherent song. At pro that's 4 songs a month. Not sustainable for social media users which is pretty much everyone.
not sure if this is something the devs are aware or not. but based on what i'm doing - if i start more than one clip at a time with continue, chirp gets very confused and starts pulling in data from all the other clips I've given it
also, with continue, i've started giving it 3 stanzas of four lines each at a time. it will spit out a few garbage words, ignore the entire first stanza, and sing the 2nd and third almost all the time.
At the rate I've been using it - which I do NOT consider excessive -- I'm going to use up my "Monthly" subscription in less than a week.
How do we get a smooth song ending? When I try to get it to make an ending it gets leaving them with an abrupt cut off.
Try prompting just [outro] or just 1 or 2 lines of lyrics, then [outro]. Also can try [outro music] or [song ends] -- though endings are a bit hit or miss, in general.
Ah, OK. Thanks so much. It is amazing. I can't belief how epic the music can be.
Hello, I purchased a sub and credits wont appear. Please help.
Hey Mysterier, itโs not a silver bullet, but try adding [Outro] to the end of your prompt. You can also play around with variations of โla la laโ at the end of prompts, which can extend or wind them down in interesting ways
Hey there, sorry for the inconvenience. Weโre looking into this, and weโll get back to you ASAP.
I really messed up. I didnt realize i had a different discord under my email. The account i purchased it under was Containedviper#0986 and i do not use that account. I'm so sorry.
No worries, we need to make this a little easier to tell. I'll move the sub over.
Thank you so much! ๐
Figured it out - you should be all set
Thank you!
Hello again. I had a banger of an intro going and I did three follow up gens in a row that stopped between two and five seconds. I thought I should inform you guys.
was there a prompt in the clips that ended?
(the part where it happened 3x in a row suggests it's not random, but something in the prompt, or in the previous prompt)
For the onboarding, it'd be better to remove that first sentence as Suno bot doesn't accept DMs.
Hi !
i have a question for Staff to ask the team Suno :
Are you thinking of making an unlimited subscription? For example, pay โฌ30 per month and have unlimited credit? or much bigger than the 1000 we have? for example paying โฌ30 for a subscription and having 10,000 credits? or 8000? because 1000 is really little and since there is only one subscription we have no choice :/ Or maybe an unlimited subscription at โฌ50/month?
When we have well-formed ideas about the music we want to make, the credits go quickly, that's why I ask haha
I think if you did more different subscriptions I think it could work more. (for those who really want to create dozens and dozens of music)
If I take my example, 1000 credits for me would be roughly a day's work. so I would need around 10,000 crd to make 3 qualitative sounds per day over 10 days. so for the month you would need a subscription of 20,000 credits minimum. What do you think ?
Hi Thanks for the feedback -- we are definitely planning other subscription plans actively. For now, you can still buy top-up credits.
it would be great to use for commercial use for the paid pro plan as a NFT artist. The only thing stopping me from committing to a subscription.
Yes I saw but I find it quite expensive in extra credit, I would rather wait for a better "plan" offer I sent an email to have a custom subscription I am waiting for the answer ๐
and i'm also wait to an update voice in chirp to pay more ๐
I have too many ideas for music to make that even 8000 extra credits wouldn't be enough eheh, I'm crazy creative
There wasnโt any indication that the song would end or was ending. I tried a rock/metal style and it was having issues so I changed the type of music to random and went the edm route instead and it worked fine after that.
anywhere we can see a list of upcoming features/changes?
good question, already working on the next version. still have to do some testing on what will make it into the next release and what not, but we'll post an announcement as soon as it's a bit more clear and we have first results
we're small, so... all of it lol. ML though, not software eng
cool. may i ask, if you guys hired another ML person, what skills would you ask for? (I am a software dev gone back to school for math/stats/ML ultimately to get an ML job)
(sorry, I know this is not the purpose of this channel, but I am very passionate about the subject so I had to ask)
npnp, thanks for the interest, feel free to send us a resume. no hard-cut requirements, but very small team for now (kind of on purpose) similar skills needed as an openAI/midjourney etc
yes that was what i suspected ๐ thanks. And I get the small-team-on-purpose thing, for sure
Hey Georg, Will there be a collaborator program to join like Pika Labs for creators who can help spread the word of the AI model with large social media accounts?
definitely would be cool. in the meantime, feel free to send us an email and we can see if we can work something out
Great thank you ๐
which is the best email? can only find billing email
honestly they all go to the same place, just helps categorization. maybe just use "hello at suno.ai"
The collaborators in Pika dont get an discount or such as its still all free for now
will sales be outsourced?
I am hoping they get marketing help. The pricing plans are unsustainable for creators.
I've already talked about it, it's actually quite expensive when you want to create content with it. and not just make 3 videos for fun. I even did a little calculation above.
here ๐
yeah, we are looking into better plans for lots of creations. the main problem is that is scales linearly with compute cost..
gpus are expensive ๐ข
Drop free to 10 generations. Would free up 15 per for paying customers. The ones you want to begin with. Easy compute calculation. Give less away for free and gain more paying customers that will stay.
What you have is solid gold. I should know.
I don't know if it's possible to do that with chirp, but, when I worked on RVC or SVC (an artist voice training AI) we were able to code (not me) an app on paperspace, which you will be able to make 2 types of subscription. 1 that would be chill is on discord. and the second which would allow us to go to paperspace and pay for the GPU ourselves. which would cost you less and earn more by making a different subscription ?
A request, is it possible to specify key and keep the music produced in the same key and tempo even if key and tempo not specified?
It's nice to listen to what people made and Idk if this is a acceptable suggestion but maybe a seperate generation channel for english? and other languages if possible.
would it be possible to make the AI has actual start and end to the song? not just randomly starting in middle of the song and ending all of a sudden
(i know its related with length but just saying) ty keep up the great work ^^
Yeah, working on that ๐
yayyy
Hello team, first of all kudos for creating such a great tool. It's brilliant and has the potential to be unique at the moment. I am a professional music producer with a decade of experience and I must say I love your work. I'm sure someone has already asked about these features, but I just want to add a few comments. Is it possible to develop a feature that allows you to get isolated vocals? There is a bunch of software that allows you to split the track into stems, but it would be great if this was built into the bot. That would be extremely helpful. Next, it's about bpm and key signatures. These are very basic things that needs to be controlled. Right now, you stand out in terms of melody creation. The Chirp bot is a great resource for getting inspiration for creating a song, but without the basic controls, it's still quite a challenge to bring the "prompted" song to life in terms of professional production. I understand it is just a beginning and super excited about the further development. Thank you!
Thanks a lot yeah for sure things for our roadmap. For stems can you elaborate? Is the idea just to get vocals by themselves or to eg add vocals to an existing track. We could of cours just add stem splitting to the output like you said but quality wouldnโt be better than things like demucs
Thank you very much for the answer. The most basic feature I mentioned is the ability to render only the vocals out of the generated song for further production. Simply to save time and not use third party services for this.
The more advanced features:
-
The ability to upload a lead melody as audio input.
-
The ability to upload a chord progression as audio input.
-
The ability to have some sort of "remix" feature. 2 situations: I like the vocals, but I want to check how the same vocals would sound in a different genre of music. Or I love an instrumental piece, but the melody is completely out of play. Possible realization: I have a "Remix" button and in the basic variation I choose between "Remix vocals" or "Remix instrumental". "Remix vocals" creates a variation of the melody or maybe a change of gender within the same instrumental. "Remixing instrumental" will generate another song but the vocal will remain the same.
I think these 3 options will definitely change the evolution of the game.
And it would also be very helpful if you could open a channel where you can post your plans for the next release. Just like midjourney do in their "office hours".
Thank you for reading.
These are great ideas! Thank you for the feedback ๐ There will be a lot of further development along the way. Goal is to give people a comfortable way to make and create. Again, thank you for sharing your ideas
Thank you for creating such a wonderful tool. I am following almost all the current developments in AI music and I must say that you have taken a big step away from the others. Suno is becoming my first tool to start with when it comes to creating a song. Thank you!
All the gratitude goes to the team! They'll love seeing this. Very happy to hear you are enjoying it so much ๐ Stick around, it's going to get crazy cool
What else will be added?
The team will be notifying you of further developments via the #๐ขโannouncements channel. A lot to come though that's for sure. Very early days here โค๏ธ
Cool thanks!
Happy to help ๐
can you link me?
What
can you send me a link to the post
Will this person be banned?
Thank you for bringing this to our attention. It has been removed and is being dealt with. Moving forward, the team is very hard at work trying their hardest to build a great tool for you all. Feel free to ping any of our Mods first about things that come up. Definitely want to keep a close eye on making sure that this server remains both safe and healthy for all. Again, thank you for your help
Are there any plans for creating a UI for this tool once it gets more advanced features, or is it going to stay exclusive to Discord like Midjourney?
There are a lot of things in the works right now ๐ Very early days here. But definitely a lot of plans on the road map. That being said, any and all suggestions appreciated as the team further develops these projects. Would you prefer a UI for this tool? Or would you prefer it remain Discord only?
Thanks for the answer! Currently it works perfectly within Discord, but it sounds like you have big plans for the future and many upcoming features. My concern is that it might become a bit cluttered (already starting to feel it with Midjourney) after adding so many options for a simple Discord interface. In this scenario I think a separate UI will be helpful and much more convenient.
Great feedback! Will be leaning a lot on our user's guidance along the way to ensure that the experience is intuitive, user-friendly and covers all the bases for anyone that wants to make a great song! You brought up some great points ๐ Thank you
it's already possible with the metatags inside your text. You can start with a line and write [intro] and for the end [outro] you also have the [bridge] [hook] verse and chorus to help the bot
for example I managed to ensure that my first line was a beatmaker TAG.
I dont understand what they did
same, Allah is just the name for god.
for the Suno folks, it seems like the subscription thing is turning people off, and the quality doesn't seem "there" yet to demand such high subscription fees. i suggest returning to your roots like you did with Bark and release the weights and pipeline to the community. assuming it's just Bark with some custom attention mechanism to tie the lyrics to music..
with stuff like this consuming credits, it's hard to justify
@devout geode @ocean finch It wasn't that they used Allah. It was because they were being crude and offensive. We appreciate and encourage members to explore the depths of their creativity. In a tasteful manner. When you are saying things like "Allah f bomb Allah f bomb" it is crossing a very easily identifiable line clearly stated in the #๐โrules . It's important we keep this environment healthy for everyone.
ah its ig another language thing. sorry for being clueless.
hey so tried out generating in gregorian and orthodox chant styles
didn't hit the style at all
orthodox chant was like vague pop with synths but completely bugged out
gregorian chant was like vaguely gregorian in back but the front vocal was female pop
loll, yeah probably not one of the main genres ๐ had some success with gregorian chant in the past but i'm also not a connoisseur..
hello everyone .im trying to sitch songs i gave like but its like it assimlate very fast and sometimes im likin 1,2 ,3 and on the 4 it doesnt assimilate ate the song. can have a delay , like 5 to 8 second on it so it gave time to us go back and fort?
or theres a better way of doing it?
I don't know if this has been said, but I think I have an incredible suggestion. Except that it was obviously said! x)
Should we make sure that we can put our instrumental on Chirp, for example upload an .mp3 (or a ytb link idk) and then be able to make Chirp work to sing/rap on it?
Or if you have a bot like this I'm interested ^^ but given the quality of Suno, I think it's great to add this kind.
maybe a 3rd bot after Bark and chirp? or just add an option to chirp I don't really know the possibilities and what is less annoying to code.
yeeeeah
anyway if you want i can point you towards examples of orthodox chant if you'd like to widen the dataset
And the clip generated when we click the hearth should also let us stich with another ones. Like that we can not only increase music time but be more coerent in the song( for instance using same chorus etc)
Another suggestion. I'm trying making a metal song and I use a curse word and doesn't allow me. That can be a limitation , but for metal and rap songs maybe review that
you can import the clip in a logiciel like FL studio or something like this... you cut the chorus and copy it to have the same in song ๐
Thanks for answer. But the mix/blending wouldn't be that good. At least for me that I'm not a pro
I seem to remember sometimes asking how a user might prefer to prompt Chirp with audio. I've got a thought. It would seem ideal if I could hit the microphone button in discord and scat or sing a small sample which would then be added to the Dialogue Box. Alternately, the box could have such an option or just an upload. Then, there would be a prompt specifically associated with that clip that describes to Chirp what that sample represents within the clip. Like "this is a drum part" or "wailing chorused guitar." The rest of the prompt structure would remain unchanged.
that's why people are using the credits system - just follow the lead of the datacenters, charge compute time
Being able to "like" a generation and hone a particular sound from that would be pretty awesome - I'd say maybe like 1 in 8 generations have a sound I like and I would love to be able to focus in on that
Thanks for suggestion ๐ Great idea
Check out Adobe Audition, it does stitching with AI!! ๐
Thanks! Hopefully something that could be done ๐ I'm absolutely loving Suno, gonna pick up the paid tier on payday!
Wow. Thank you so much! I will see it!
Hello congrats for the amazing Suno!! Another suggestion, let's say we choose to stich several pieces with the likes/earth. Would be great to have a command (like enter or something) in the end to Suno app know it as to stich those. Exemple. I made 20 pieces of music. I want to alternate some in one. Give like on the 1, 5 , 7 and again 5 and 1 ( example) now I give command ( enter or whatever ) then Suno knows it as to sitch all of that. And voila . Salud
Interesting! Good feedback!
Thanks. Like I said I'm all in. Hope your app go to the moon. I see lot of potencial! So the more feedback the more the project can grow! Let's go!
My friend , it's any YouTube video or so, explaining that AI feature? I never had that software
All the congratulations go to the amazing Suno team! They are quite amazing. Very excited for this coming year. Been waiting all my life to see something that could comfortably help people make music. Thanks for the suggestion โค๏ธ
My last 4 generations have been double 2 second clips. Is this a bug?
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
Very possibly, what genre are you trying / are you continuing?
Will DM as well
Hi @muted flame I have had the same problem
my last 4 generations have been 2s clips and 6s clips no sound
I had it while trying to continue a song with genre deep house
Thanks, yeah, certain genres + continuing seems to trigger this. Especially when trying for instrumentals / no lyrics. Will DM
I'm putting back my suggestion because I don't know if it's good to take or if it's really not possible :/
Should we make sure that we can put our instrumental on Chirp, for example upload an .mp3 (or a ytb link idk) and then be able to make Chirp work to sing/rap on it?
Or if you have a bot like this I'm interested ^^ but given the quality of Suno, I think it's great to add this kind.
maybe a 3rd bot after Bark and chirp? or just add an option to chirp I don't really know the possibilities and what is less annoying to code.
(this can allow beatmakers to put a topline to give more ideas about their instrumental, or just artists who find instrumentals on ytb and pay them to rap on them)
Something wrong. I'm making new songs. I'm putting likes to generate all song. And it's like he's getting an old song, not the ones I liked
I didn't know you need credits. How do you get credits?
That'll depend on exactly what you need to do, but see what you can find! https://www.youtube.com/results?search_query=adobe+audition+ai+tools
"/subscribe"
Thanks gonna look
The AI doesnt seem to be able to pronounce ale correctly. As in Ale the beer.
Definitely not disagreeing that there may be some hiccups, especially with pronunciation. In cases like this, try writing it how it sounds or leading into it if possible with a word that rhymes. That can help.
Very early days here though. Lots of room to grow, develop and become the user-friendly musical tool it's destined to be ๐
Hi,about 85 percent of my generations the text strats in the last few seconds (maybe last 10), in rap >I use music suggestions like hiphop classical chinese music and sometimes added other music prompts.Whats happening?
Are you adding any prompts before the lyrics, like [Intro]...?
no normally not. I guess I have to?:)
No, that would potentially delay the lyrics...
Just trying to guess how your prompts are, and if they are influencing the timing... It's just a guess though
(since you are not telling us what the actual prompts are)
what prompt are you using? sometimes this issue happens with non-english. you could try adding 'vocals' tot he style, that might help bias it a bit towards spitting out lyrics earlier.
Hi, It would be cool to be able to be able to use custom uploaded image(s) with the generation to ensure a more cohesive aesthetic or possible to be able to give tags for the image generation. (especially for parts with no lyrics)
i am trying out bark. i used a random voice. the output is perfect. but how do i know which voice did bark use??
thanks yeah we might add some more image/video support, but audio is our focus of course. (images are just sdxl) maybe just use the audio and generate the rest ousite of discord to get it just right? the videos are mostly meant as a quick nicely sharable format. could even consider opensourcing that part of the code to make it easier for people to make sharable tracks that look fun...
@summer wraith yes, but it would be nice to skip the editing part ๐
hehe yeah totally down to do it. discord makes it a bit difficult to add too too many inputs but we'll get there. midjourney has had good success slowly introducing more complicated inputs, so hopefully we'll get there as well ๐
I think it is understanding that things inside [ are not lyrics] so maybe it could analyze this to also use for image generation?
right now image and audio are totally separate models, but ya we could do something in that vein
also it would be really nice to be able to use the same seed, if you have this kind of concept or be able to keep the same style in consecutive clips, not sure if you have suggestions for this?
Currently I have just kept the same style prompt and tried to modify it with some success
we could defo fix the seed but as soon as something is modified i think the seed won't matter
yeah, you mean the lyrics etc.
I made this but the style is getting a bit more different in terms of vocals from the beginning
you mean v1 of suno? I used that alot also ๐
haha awesome
Suno is a great product for sure
I have a bug, I try to generate promt and it says that its fail iteration
me too, but I see others generating music
Sorry, try again now
it's high load on server?
Just had to restart a few things but normal load
Any reason for the two step process? Forms are definitely neat but I think it'd be nice to have something quick MJ style
Sorry what do you mean with two step?
First /chirp and then submitting the form. As I said, the forms look pretty neat but something like this would make it easier for users who can't manage to figure out /chirp and just type it as a message in the chat
Ah yeah, mostly cause of new lines
Pretty important for song structure and in a 1-step they get swallowed
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
Definitely open to suggestions that are consistent with newlines . I'm also tempted to put "Remix" label on the regen button just to give users more eye candy to click on
Unsure if I follow
Can't seem to do this on desktop (could be user error) and if you paste in newlines on desktop they get collapsed too
Shift+ enter
Doesn't work ๐ฆ
Interesting, I think you can add dashes -- to indicate new lines but I guess that defeats the point of easy one step process
@muted flame would be nice to add a link to the command instead of just text to that message, like </chirp:1136027103767318712>
This is great - FYI @delicate iron. Thanks!!
Oh this is great. Thanks, @tiny vale!
I'll change it for the responses in the commands work so we don't just make it easier for people to get another Woodhouse message ๐
Can also make these messages ephemeral so they don't clog other channels
ะะดะต ั ะตะฑะฐัั?
Great point. I'd love to delete them and send an ephemeral message, so it's all private. Looking into whether Dyno supports this!
its not giving me more chirps even it is announced that everyone will get 100 free credits i have not used it even single one and its saying i have used all chirps
most AI cost 10 times less to use than suno per GPU minutes?
is this a bug, or a feature?
needs to be a way to seperate vocals and music also
song minutes isn't the same as gpu minutes unfortunately...
for stems yeah we are thinking of adding something built in, but until then you can just use demucs or other automatic source splitters
Hi, I have noticed since you introduced the feature that puts your newest generations lowest in the screen and pushes everything up, people are liking the wrong things and even continuing wrong songs because the display is updated just when you will click. the old system was better
the command /chirp doesnt work for me
eek interesting thx
mostly affecting when many generations are going at the same time
yeah makes sense. i suppose public rooms are always a bit of a mess
Public generation channels is a 'public good' in a sense. It lets other people see the effects of prompts and styles, it gives people a a base clip to spin off their own songs. I was worried everyone would go private but so far plenty of activity in public, love it.
Most of the chirps I generated today either had no lyrics, missed lots of words, or repeated sentence fragments. No I have no credits left.
Sorry ๐ฆ some prompt are more likely to give background music rather than vocals. It helps to explicitly add โvocalsโ to the style description
Moderation in 50+ languages is a heck of a challenge, yeah. Suno folks mentioned it's being worked on. You can flag real bad cases and msg somebody from Suno with the link (or me but I just forward it to them anyway) and they will look at it and over time the system should get better.
The AI completely mispronounces "Eurovision". Works fine if I add a space though.
Just a funny thing, in the Suno Bark model I found "Eurovision" was some kind of super word almost like a tag in itself. Just add that one word to a prompt and Bark was like 10x more likely to create a random voice with some European languages accented English. TOO POWERFUL
Thanks for the reply. I must remember this for later and put this superpower to good use!
CC: @iron pelican @trim sonnet
Hello. I'm trying to buy more credits and doesn't work. First I even tried to renew ,said was done but nothing. So I tried the credits and appear error
And now I don't know if supposedly I paid for the renewal because appeared done , but it seems didn't renew
I will reach out to folks about this, stay tuned
Thanks!
Can we get extra chirp credits for things like this that only last 3 seconds? I've had a handful #๐ฃโchirp-beta-1 message
โค๏ธ thanks, yes all of those are in the works ๐
i believe sub 5s doesn't count towards your credits cause it's usually the model being a bit too liberal lol
Gm everyone. Still can't have the credits
If you could, please email billing@suno.ai and explain in detail what is going on. Someone from billing will reach back shortly and help with this. Sorry for the trouble โค๏ธ
thanks. i will do it.salud
The ai doesn't recognise Fado and Pimba sadly
You really likes Portugal right? Lol
I'm Brazilian
fr ? nice
Escuto algumas mรบsicas portuguesas
quais ?
Thiago Nacarato รฉ meu favorito
nunca ouvi falar
Bug: I cannot save voices generated when using bark feature.
As you make songs that continue, the song becomes distorted and the volume of the music becomes extremely loud.
I think there might be a bug related to the one I had before, this one seems to take the form of producing silence. The "fun" part of this one is that it can trick you, because I tried what I did before and went backwards and avoided using the style prompt section and cleared it out--completely removed the text from that field. But somehow Discord was still inputting that text back in when I hit generate.
Why am I getting this message? "This prompt has been flagged. Try again with a different prompt." What is the problem? ...or is it a censorship filter that is not working properly?
I have found that if you pick a clear beginning voice to continue from it has less issue than if you pick a simi distorted one from the start.
subscribing seems not working ๐ฆ
ะัะธะฒะตั
ะฏ ะฟะพะฟัะพะฑะพะฒะฐะป ัะดะตะปะฐัั ะฟัะพะดะพะปะถะตะฝะธะต ััะตะบะฐ ะฐ ะพะฝะพ ะฟะพะปััะฐะปะพัั 3 ัะตะบัะฝะดั ะธ ะฑะตะท ะทะฒัะบะฐ
ะะฐะบ ัะตัะธัั ะฟัะพะฑะปะตะผั
Were you able to get it to work? Please DM me if you need any help!
@here How soon will the new version of Suno be available?
I'm having a hard time having the song with the full lyrics. Sometimes it spends 20/30s with almost nothing at the beginning, then the song begins but it doesn't have enough time to say all the lyrics. Is there a workaround for this issue ?
The only thing that's worked for me was either trying different metatags to goose it to continue, trying different genres to push it forward (or alternately if you're already doing this in every continue, not doing this) and as much as no one wants to hear it...
...sometimes you just have to start over and try again at an earlier phase of the song's development, using metatags or genre shifts (or again not doing those as may apply) and hoping for the best.
It can be painful when you get stuck somewhere after so many good segments.
Oh another possible thing is if you're doing too many words in a single attempt. That got me a few times. It's better to break it down by syllabols (and develop a rhythm) than to put too many in at once.
guys, when i try to get me a full chirp from the segments i worked on with monthly gaps due to fast free credit consumption.. it just wont happen, and im devastated ;p asked for a full clip 10hrs ago and still nothing
Hello I noticed that sometimes the text appears in the middle of a musical passage or, for example, when generating an opera version, at first a voice can sing, for example, in German, switching to the desired language, I hope this remark will help make the neural network even better. Thank you!
oh interesting, how many months do these span? there could be a problem if 2 or 3 as some versions changed..
if it's not in DM, can you link the message w/ the last chirp and say which one (1 or 2) you hearted?
phew, it worked. i tried to love few earlier chirps and smehow it "unblocked" itself - went like 7 chirps back and started to heart them one by one
final chirp is 10m33s
Whoa! Are you willing to share the results with the rest of us?
I'm just beating my head against the wall lately. Do the [verse] [chorus] [male] [female] tags no longer reference previous parts of the composition? Do they have lower weight than when they were first introduced? I feel like it is very hard to get a transition any more and it just feels like I'm being ignored a lot of the time. Other than continuing whatever sound is happening at the very end of the preceding generation, I feel like continuing has very low continuity. I've been trying to finish a song for hours and hours. Kinda burnt out.
Does it matter, if the beginning of the song is from several days prior?
I've backtracked and started from previous spots a few times. It just keeps getting stuck in ugly ruts. Or, maybe I'm driving it there. I dunno. Gotta go touch grass. Lol
Not sure if adjustments were made or if I changed my prompt strategy...
It really seems like my prompt has more weight now.
Got it done.
Hi everyone, what causes the fact that I'm trying to generate a song with chirp but it generates two silent and very short clips?
Short lyrics prompt maybe. But also just bad luck sometimes - AI is unpredictable.
well ofcourse it sounded better when i checked the single clips heheh, but sure!
I'm amazed. You had one transition somewhere in the middle where it kind of stopped or dropped for a bit which threw me off but overall you kept it going and you did it all with an input in the song style input, mostly. And the instruments towards the end...nice.
And all of it towards a ten minute song. Without losing coherence. Impressive!
I am stuck! ๐ญ
did it work itself out, or still stuck?
I kept trying... now it says it finished, but I see no generated videos....
That screenshot looks like a stuck one (the finished part is referring the previous message).
Able to try clicking the previous button where "interaction failed" again?
I never got that clip to regen..., went to it's previous clip and Continued instead... resulting in the one that is stuck.
Sorry! If something gets stuck for 15 minutes or more safe to retry as it won't come back or cost credits...
really struggling with contuinity of songs. I got first good part, and I give new lyrics, but the latter parts are short and start from the middle lyrics. ๐คท
Have you tried stitching (with the heart) to see how the parts are lining up? I've been told that Chirp will kind of roll back a bit to maintain its focus window and remember what it's doing in the context. It might not be breaking in the middle so much as repeating it. It's good to check for that sometimes.
Argh...I just ran into this. It ended at just before the last word in the chorus. I stitched it up to confirm. Nuts.
I really liked that segement so I'm going to try to see if I can put it in the beginning of the next segment and hopefully it will blend properly. Damn thing is I also liked the continuation of the next segement. ๐ฆ
hey guys, just renewed my sub late last night, but no credits ever came to my account
Hm that's strange. Can you send us an email at billing@suno.ai, @eager shadow? Looking into this!
sure will do! thanks
sent it over
yes, I just made new versions to overcome the problem and somehow it started to work better
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
On the suno website , when I click subscribe to the subscription, it doesnโt do anythingโฆ idk what to do
I tired on my phone , and on my pc with multiple browsers
thanks for the report - we'll look into this
I can go to the manage account website but once Iโm there when I click subscribe it doesnโt do anything there
Could you try logging out and logging in again, via the bottom left corner menu? Sorry! We'll get this figured out for you
if it has a green Active that just means you're automatically on it (free)
I tried , it has a new updated page but now it doesnโt even have an option to subscribe , it only shows the enterprise info now . And now it shows a monthly option and a annual option but no button to subscribe now
The option to subscribe is missing for me
I'm running into a lot of voices in my attempts at a country intro.
Like. There is no dialogue here. Shut up.
thank you for advice... I hit the same issue.
I'm not able to stitch my song together beyond 11 clips (5:25).
Hey wetcircuit, are you willing to share your song via DM? Would love to take a look and try to figure out whatโs going on there
I shared it with you in DM, thank you!
(the song isn't important to me, but glad to help debug)
I try the following text, crips wont read it in chinese even if i specify genre as chinese rap:
ๅคงๅฃฐ่ต็พไธไธ็ฉบๆฐ็ธ้
๏ผไป
ๆฌกไบ็ฉบ่ฐ้ซๅ้
็็ฌฌไธๅคงๅๆ๏ผ๏ผ๏ผ
ๆ้ๆๆ๏ผๆฐ้ฒ็ๆ๏ผ้ฃ็จๆฒน๏ผๆตท็๏ผ้ป่กๆค๏ผๅคง่
1๏ธโฃ็ฉบๆฐ็ธ้ ้้บ2ๅฑ้ก็บธ๏ผ็ถๅๅจ้ก็บธไธๅทไธๅฑๆฒน
2๏ธโฃๆ่่ๆพ่ฟ็ธ้ ๏ผๅจ่กจ้ขไธๅทๆฒน๏ผๆ็ๅ้ป่กๆค๏ผๆๅๆพไธ้ขๅคง่
ๅทฎไธๅค่ฟๆ ท๏ผ็ถๅไธๅๅฐฑ็ปช
3๏ธโฃ200ๅบฆ5ๅ้ใไนๅๅฏไปฅๅปๅไธช้ฎไปถ๏ผๅทไธช็ข๏ผๅทฎไธๅคๆถ้ดๅฐไบๅๆฅ๏ผ
4๏ธโฃๆ่่็ฟป้ข๏ผๆ็๏ผ้ป่กๆค๏ผ200ๅบฆ3ๅ้!
5๏ธโฃ้่3-5ๅ้ๅณๅฏๅบ้ ่ฃ ็๏ผๅนณๆถๅๆฌขๅmedium rare็ๆๅๅฏไปฅ็ผฉ็ญ็ฌฌไบๆฌกๅ ็ญๆถ้ด
ๅณ้ๅฃๆ็ๆฏไธ่พ็ๆ้ฆ๏ผ๏ผ๏ผ
ๅฎ็พ็่๏ผๅคๆฑ๏ผ
ๅฝๅฝๅฝ๏ฝ๏ฝ๏ฝ้
ไธชๆฒๆๅฐฑๆฏๅฎ็พๅ้ค๐ฅฃๅปๅป๐คญ
Hi first chirp is amazing tool however I have been trying to continue a song I started yesterday and a lot of my attempts resulted in songs with no lyrics or one time just random gibberish if you want more details or would like to see the output just do me
I have used all my credits now so unfortunately canโt try more generations for you untill tomorrow
The censorship is inconsistent, flagging the exact same prompts that had been approved minutes before.
Is the bot not working for anyone else?
overloaded.. wait a few minutes to catch up
Rhyme no longer works great in Russian. It used to be much better, about a month ago
I noticed that when ever you do an [Intro] first then Continue, the algorithm has a hard time saying lyrics and it only makes 30 second long clips nothing longer
For example, I do 3 [Intros] and no lyrics, then continue, and the songs afterward are messy and not coherent...lyrics have a hard time catching up and songs are never past 30 seconds
how to create an intro?
[Intro] usually works,
DMd. Thanks, Barefoot!
Continuations are maximum 30sec.
Suggest adding a description of singer to the start of your genre prompt on the Continuation.
Description of the singer?
'Male' 'Female' 'Operatic' 'Melismatic' 'Jazz' etc etc
alternately a description of the 'vocals'
Oh, OK. I do something like that now sometimes. male vocalist, smokey voice in some of my earlier attempts. I thought maybe there was something else that I'd missed out on.
Thanks for explaining.
Feature request: take over a style of an created song.
Like continue but starting a new song
Feature request:
Possibility to add parts in the beginning, sometimes I would add some more instrumentals in front of my song
I'm new to this.. I saw (chorus) and (verse) but how can I specify 'Male', 'Female' in the prompt ? Thx
Male Vocal
Female Singer
...
consider putting it in the genre description, even repeating it a few times
Male Lounge Singer, Male Lounge Singer, Male Lounge Singer, Male Lounge Singer, ...
Like a 'start from here' button? That could be interesting.
why are continuations always in a different style? Am I missing something or it's a bug?
What button are you pushing to continue?
either "Continue 1" or "Continue 2"
OK, those shouldn't be switching styles on you.
Are you keeping the voice within the last ten seconds when you continue?
Are you adding the same genre to the Continue...? or the same description, with rearranged words?
It can go off and do it's own random thing if it's only relying on the 10sec sample from the chirp before... (especially if it was a percussion break or something like that)
I left the genre empty and just added some additional lyrics text to continue on
the voice is in the last 10 seconds
Hey, when I type /chirp nothing pops up ๐ฆ
Can someone write an example of /chirp inline ?
Hey @sharp root, the /chirp command isn't available in this channel, but you can use it in #๐ฃโchirp-beta-1. See #๐ฃโhow-to-chirp for more details and instructions.
I still am not able to subscribeโฆ is anyone working on this ?
If not , I guess it just wasnโt meant to be
thanks anyway โ๏ธ
if you click on that /chirp option in the menu, and press enter, what happens? You may also want to try DM @storm vale
Can you try clicking on this menu and logging out and then logging in again? You may also need to verify your Discord email if you haven't yet? That's supposed to show your email if you're logged in..
Yes I did that again today before writing the message, it shows the same thing after re logging in
And dm the bot is what I have been doing lolโฆ
Sorry that was a response to a different message ๐ฆ
No problem
I tired emailing the enterprise contact info as well .. idk whatโs going on
๐ฆ
just try again if this happens ๐ฆ sometimes discord or us has a temporary spike
lohan3626 spamming bark, please make him stop
why is the continue song button not working the song wont work
I keep asking for male vocals and it continues to give me femal vocals
Channels are missing sometimes.
there's a server setting to show all channels
Thanks ๐
I can't make dsbm or death metal :/
I suspect this is a bug more on Discord's side than Chirp's but here it is.
Sometimes I'll try a musical style and dislike the result and when refreshing I'll decide to remove it and try again without the style I added. Discord will put the style back in the submitted form somehow and it gets added back in when I specifically removed it from the field when submitting it. Annoying.
lulz
Yeah it came back...didn't matter much. The song was just doodling and went no where.
Still it made me stop and look up lyrics by some popular groups and determine that my crap lyrics fit right in. I just need to stop thinking too much about it...
๐
bug
Hi all, we're seeing some hanging/failed generations from Chirp. We're looking into this now, and we'll post here once resolved. Sorry for the inconvenience, and, as reminder, you will not be charged credits for failed generations.
All set
trying to manage my account and subscribe, but it get stuck
DMing you
I need help with trying to subcribe is there someone i can DM ?
i want to add these songs to spoyify, etc ๐ซ
Please help
You may have to wait till tomorrow, it's middle of the night EST for most of Suno.
thank you, hey check out the song i made with suno in the #๐ฃโsuno-showcase i just posted it
i love Suno !!
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
Can someone help me , I want to be able to monetize
I think you got unlucky weekend not many Suno folks who help you around, but try messages mc or keenan from Suno if you haven't. I can tell you the most common problem people have is they subscribe but use the wrong discord account (you login to Suno website with a discord). But I can't fix anything like that myself, sorry
I definitely logged into the correct account, the button just isnโt there anymore lol it was before but wouldnโt do anything, hopefully someone gets back to me on a week day maybe
@muted flame in case didn't see and around
I just want to be able to post these songs on Spotify and Apple Music ,etc
I can't speak for Suno but can see you've been *trying * to sub for a couple days, hopefully can work it out
Iโve been trying for a couple weeks
I donโt want to post it and get sued lol
Hey, we're continuing to look into this from our side, but could you double check that you have an email in your Discord account and that it's verified? Thanks.. You can go here: https://support.discord.com/hc/en-us/articles/213219267-Resending-Verification-Email#h_01GSG5S5DTDCNVJS1Y2FFNSJMZ
Have you recently signed up for Discord, but now your verification email has expired? No worries, you have the power to resend the verification email right to your inbox.ย
What does this article co...
It wasnโt , I just verified it now
But I just tried again and thereโs no subscribe button
refresh one more time
Ok itโs working now , I logged out and logged in again
Sweet !!! Thank you ๐
Sorry about that. Give me a sec before you sub. I'm gonna send you an email
Ok
Thank you so much !! Iโm so excited
Is the enterprise plan the only one that allows you to post on Spotify ?
No, all plans but free/basic
Sent u an email!
Wow ๐คฉ THANK YOU ๐ , is that a one time thing ?
I freaking messed up bro , I couldnโt find where to put it in , and paid the whole thing ๐ญ
We got u, will email
Thank you ! I canโt believe I did that ๐คฆ๐ฟ
cant clone voice
every time shows network issue but i have a good network connection
as musica la tudo certo nรฉ com 40 segundos, sรณ que o bot comeรงa a canta com 30 segundos mesmo, e naqueles 10 segundos restantes nรฃo da nem metade da musica
Olรก Dark, esta foi a sua primeira geraรงรฃo ou uma continuaรงรฃo?
as duas coisas
Rerunning a continue causes model quality to degrade immediately, this does not happen if I go to the original start snippet and click continue but rather after doing that going to the new snippet that is near perfect and clicking regenerate. This has probably cost me 10 to 25 generations to figure out
Rerunning from a reran continuation worsens quality further as well
Some evidence of a problem occurs, clicking continue leads to continue pop up, rerun on that continue leads to a chirp pop up
In retrospect probably more because each step the worsening compounds for some reason. This problem occurs on the first continuation.
So arguably you wouldn't notice it right away just a slight decrease in effectiveness but whatever, but as it compounds the problem worsens, thus I would argue and rerun of a continuation is not using an optimal version of the model unless directly clicked from continue button.
Fixing this should lead to a noticeable improvement in quality
Lol if true this could greatly decrease what we have been calling hallucinations which people have noted seem to occur in grouped blocks of time, while going long periods without as many.
This should be easy to repeat.
- Generate some snippet.
- Click continue on that snippet
- Generate continuation
- Click rerun on that continuation
- Generate rerun of first continuation(technically a continuation of first snippet)
- Repeat 4 and 5 until model worsens to noticeable effect
Sorry 5 should be: Generate rerun on the previous "continuation" (which is a continuation of snippet from 1.)
We'll look into this!
Please let me know if I was onto something
it's normal bark spent more than 4 minutes for this phrase: jungle, the mighty jungle, lion barks tonight?
maybe bark it's broken
yeah, it's broken
i guess it's broken only for mobile discord
@cyan garden thank you for the hint, I'll try to avoid rerunning continutations then! ๐
Yeah if you scroll back up to the original snippet you wanted to continue and just click continue from there it should be fine @south hedge
it's not delivering a liked clip for #๐ฃโchirp-beta-2 message. Have waited like 20 minutes. It just stays "stuck on "getting clip...""
Discord is the easiest way to communicate over voice, video, and text. Chat, hang out, and stay close with your friends and communities.
I have 50 points but it wonโt let me generate anything
I havent been able to renew my subscription
Try to catch @muted flame from Suno if he happens to be online today, if not should be around Monday for billing problems. I think it auto renews btw by default you may not have to do anything special
Glad itโs not just happening to meโฆ Canโt use Continue 2 on any song, including previous ones.
My song is coming in another language
Are you both doing this in a mobile browser running Discord?
Iโm on the discord app on my phone, yes. Itโs worked fine for weeks up til now. Suno, I mean.
It's the app. I thougght it looked like a web browser. My mistake, sorry.
This is on my iPhone SE 2023. Are you unable to get to this point or is the issue after that?
I will say it was tough getting the button to tap because of how it broke the formatting of the page.
Before and after, depending. I just canโt Continue 2. Continue 1 works fine, and Iโm sure /chirp works fine but my good generation landed on 2 and I canโt use it now. Any of the 2s. The messages moved to where the button is actually between every message and canโt be clicked on.
I meant to click Reply before typing. Sorry.
Good catch. I think it was one I tried. Let me test it out again...
OK, wow. That is attrocious. It completely is unable to be selected. Good catch.
Yeah. Itโs pretty jarring. I stopped creating immediately once I noticed it. Still trying to wrap my head around what would cause thatโฆ Hopefully the admins see this.
Speaking of which, what do the second buttons do? The heart 2 and continue 2 ones
Heart 2 stitches the 2nd generation with the ones you continued from before it and puts it together as a single song/video. Continue 2 saves the 2nd generation to your timeline and allows you to add onto itโฆ
Whoโd you find out about this from that didnโt even tell you how to use it? lol
No offense meant
Nice reminder to turn on show all channels
๐ Thing is I know that link exists, because someone also gave it to me...
You're not alone.
again, problems with getting full clips ๐ญ
Discord update yesterday
Still okay on my Android Discord. What if you put your phone in landscape mode, that should put both buttons on the same line.
gahah\
My app doesnโt support landscape mode, so I still have no Continue 2 option. I have one more thing I can try but if it doesnโt work, I donโt know.
Nope. That did nothing. I reduced my text size to the lowest to see if it would โscooch upโ but still the same problem.
Are you completely unable to access in a PC or a Laptop?
I donโt have a PC or laptop, no.
I see. Only a phone?
Yes. Sorry. I donโt make a lot of money. I canโt afford a new computer. I couldnโt even really afford this phone, someone got it for me and Iโm slowly paying them back. Even if I could get a cheap laptop, should I spend 200-300 dollars just to use a free version of Suno? :/
I see. Yeah, that's understandable.
Chirp is not open source unfortunately, youโd only be able to run Bark for free.
i still havent been able to resubscribe
Is it ios? seems like Discord broke landscape in older IOS a few months ago, annoying https://www.reddit.com/r/discordapp/comments/16alqs1/about_the_landscape_bug_sure_i_will_update_my/
I pinged someone but they likely won't be around till tomorrow. The most common problem by far is people accidentally logging into Discord and the website using two different Discord accounts, or using a Discord account which isn't 'verified' which means they need to find an email from Discord and clicking on it.
I'm not sure how I managed this, but it was only 24seconds and started in the last two verses. I'm assuming it didn't understand the instructions as instructions #๐ฃโchirp-beta-1 message
I think it's mostly the very long sentence in the brackets tripping it up, though this one started in a different place for one sample #๐ฃโchirp-beta-1 message (and next one down I removed the line and it started at beginning)
oh that's interesting, in the one you linked, the second one generated, it just kept singing the chorus over and over for 40 seconds. Not a bad result, not the intended result, but not bad
the second one, on the second one down, it skips an entire verse, but I actually like it
I'm going to have to do more testing tomorrow when I get more tries, but I want to see if I can figure a way to influence the sing style. My first attempt using ( ) kinda worked, but then the next time trying parenthesis it just sung them. 
some more goofy results #๐ฃโchirp-beta-1 message Probably my bracket use again, but also I'm guessing it has to be a real language to do lyrics
Thanks. Iโll try updating.
note to self: clean keyboard and screen AFTER you turn a computer off to avoid stroke-like posts of nonsense
I updated my iOS to 16.7 as recommended and I still donโt have landscape mode on discord. Is the โContinue 2โ format problem just not affecting enough people to deserve a fix?
Itโs really depressing to think that I just simply canโt use Suno anymore all of a suddenโฆ
Appears to be an Apple bug in Discord.
I just loaded the official app o my old phone using WiFi. (Usually on Android I am a fan of Aliucord since it uses the previous, non-Apple version of the app for Android which doesn't break as often. Now I see that the Apple code breaks on Apple devices too.)
Nice. Well yeah. Idk. Thereโs so many moving pieces, I donโt know whoโd have an easier time addressing it. I doubt Apple would care. Iโm checking out Aliucord right now.
IIRC that's Android only. Just so you know.
The story I heard was Discord went with portable code based off of the existing Apple codebase with the latest releases, so Discord itself was mostly the same codewise between platforms now except for the parts that are compiled to the respective systems. All I know is with the newer Discord on Android I kept having menus pop underneath the navigation bar.
Hi Mate ive gpt a track THAT I CREATED LAST WEEK AND IYS MY FAVOURITRE. HOWEVER THE PROGRAM WONT LET ME SADD ONTO THE FOIRST AND OINLY VERSE IVE GOT
I seeโฆ Well I guess Iโll just let go for now and hope for a miracle someday in the future. Til then, back to the drawing board.
I appreciate this but still haven't had anyone reach out ๐ฆ
You can also email billing@suno.ai
Maybe you can you open the web discord https://discord.com/app from an IOS browser, Safari or Chrome? (I don't know if this is possible, just that at some point in the past it was.)
"Getting Full Clip..." doesn't seem to resolve in several cases
didn't want to flood it so thought I'd mention it
Haha, yeah that occurred to me too. I tried it but it just said I had to get the app. I appreciate the suggestion though.
yeah, ouch. i wonder if there is some convoluted way to use IOS accessibility options to do it - I'm thinking of a voiceover mode i've seen on IOS where you can swipe through interactible elements on screen one by one, so maybe that could bypass the visual block. but it's getting a little silly even that works lol
Hey just to confirm, I'm on iOS 16.xx which version do you run?
lol yeah. I mean, thereโs gotta be a way to do it. Where thereโs a will, thereโs a way. I did scroll through accessibility but maybe I missed something. Iโll keep looking into it.
I believe Iโm on 16.7 something. It did say 17 was available but 16.7 was the recommended one when I went to the software update screen so I just chose that.
I'm in the same place. This works, don't "fix" it.
I wish we could put out a call though to see if anyone with iOS 17 is having the issue too just to eliminate that from consideration. Because you know that sooner or later someone is going to suggest that as a solution.
So.....not it....
Well I did some digging yesterday and what I found out isโฆ people have been complaining about there being no landscape mode for discord on iOS since version 7. I saw a thread with a similar comment for multiple iterations including 16. Which is why I stand behind my assumption that Apple doesnโt give a fancy crap about it lol.
Can you just rename the continue buttons 1 and 2 without โContinueโ that way we can click them. They extent beyond the clickable area. iPhone 15 here
I have no experience with iphone but maybe you can do an equivalent of CTRL+"-" (minus symbol) to make the discord display smaller like you would do on a pc
I love that you're thinking outside the box but I don't think that would work with a virtual keyboard.
Learn how you can set the Safari Page Zoom to a larger or smaller size on all websites on iPhone 12 / iPhone 12 Pro.
Gears I use:
Velbon Sherpa 200 R/F Tripod With 3 Way panhead https://amzn.to/2IuyFGa
Rode VideoMic Pro+ Compact Directional Microphone: https://amzn.to/36w0pme
For camera I use the Galaxy S10+ https://amzn.to/3prIjKv/ S20 Ultra p...
looks like there is a settings option for zoom
in Safari
although I agree with @velvet peak that the word "continue" doesn't necessarily need to be written out if it fixes the UI problem
?
That's not a bug, that means you ran out of generations for the day
When on the phone I'm not using Discord in a browser, I'm using the discord iOS app
As a style of music, "Death Metal" should not be censored (had my prompt flagged today for that. This must be new as the other week it worked fine)
The highlighted text was what I submitted. Chirp was hallucinating trying to continue the previous segment so I gave it up and removed that part and submitted. Then I got this.
Are we no longer called to call people stupid?
How about Doody-head? Can we still say that?
๐คทโโ๏ธ
Awww c'mon!
OK, I'm done for a bit now. Chirp is hallucinating like crazy. So who know what it thinks it's hearing when it's this far up its own rear end?
I submitted generations and nothing was generated
I am sorry if this has already been reported. I used random voice on /bark, really liked the voice, but I can not save it or determine which voice it used.
#๐ฃโchirp-beta-1 message Why did this almost entirely hallucinate in both versions?
Can't subscribe!
How it didn't happen if i submit the generations?
the servers are down?
I received a message saying I have many jobs in queue but nothing is currently being generated.
Is there any way to clear my queue?
I keep getting interaction failed when attempting to use /chirp or even /help.
Queue should auto-clear after 15min of things stuck - sorry about that.
my generations are often stuck like this, no response
is this normal? like it's busy right now or something? Or is there something wrong with my account? I didn't have this issue when I was on the free plan
Sorry, some intermittent issues right now which we're looking into (hope to resolve in 30 min or so)
okay thanks
IOS safari I had success typing "discord.com/app" on iPhone without Discord the app installed on the phone, even with text size really realy big. Maybe that could work for you.
Wow that actually works. Thank you lol. Guess now all I have to do is wait for all the bugs yโall are experiencing to pass on through lol
After I want to go back to my previous songs, I canโt press the continue 2 button
As a workaround for now, you may be able to uninstall the Discord app and using Safari type in discord.com/app for address, and use Chirp in mobile web browser #๐ชฒโbugs-web message
'Thanks for watching!' ๐
I've heard this hallucination 3 x in different voices, usually(?) at end of chirp.
not just me, others heard it too (meaning they've heard the same hallucination multiple times for it to be worth mentioning)
Makes one wonder how much was scraped from ||YouTube|| for it to be in the training, one way or another?
I have the same problem
Canโt continue 2 from phone
bot not worked
Ran out of chirps after 3. I used 2 from the previous day at around 11 pm est. When exactly do credits renew? I thought I was being clever by working from 11-1 lol. -_-
I paid to use the bot on discord and it keeps telling me I don't have any credit... how can I fix it? ๐ค
email billing@suno.ai
Most common problem is the website Discord account is different than the Discord account you are trying to Chirp from, maybe double check that just in case
I've just checked, but I'm using the same account.
the Suno account email must be the same as the discord one?
I'm not sure, mine is but I can't remember off hand if I even manually entered it or it pulled from Discord when linking. email the billing@suno.ai account they'll get you sorted
ok, thank you for your reply ๐
Hello, when i go to the page for buy prenium, the page load all time, i can't buy
Thanks for reporting - we will take a look
ok it's just me idk why because my friend can buy, but not me ... i don't understand
For a quick fix, could you try re-verifying your discord email (we use it for payment flow) https://support.discord.com/hc/en-us/articles/213219267-Resending-Verification-Email
Have you recently signed up for Discord, but now your verification email has expired? No worries, you have the power to resend the verification email right to your inbox.ย
What does this article co...
ah ok thank's man
LMK if that doesn't work and we'll look more!
i just bought the 10 monthly plan but my credits (which are showned in my subscription management page) do not appear as available in discord meaning... on discord it says i have not enought credits
any idea how to solve that?
same account
email billing@suno.ai
ok!
nvz
Hello
I registered for subscriptions, but I get an error asking me to register for subscriptions.
Can you email billing@suno.ai they can sort you out
Hello, it seems to say โprostituteโ after โthatโs what itโs all aboutโ.
It's a hallucination, it happens. That's a surprisingly coherent one actually, often it's not even a recognizable word
๐ฐ
Iโm worried bad words might be wispered in the background of songsโฆ
I've seen a lot of hallucinations and most make no sense, it's like "Simlish" the fake English from the Sims, and when they do make sense bad words are not common. In your case the word might have been more likely because of that specific music genre and the lyrics, and probably very unlikely in other cases.
Your style prompt is pretty unique, voguing twerk, might be even related to that
If that song is a continue it might be an echo of a line in the previous segment, if it's not that specific word
Getting a lot of errors today. Like half the time I push a button it errors out.
Can take a look, which buttons were erroring?
Keeps trying to charge my card. I don't know where account Access is. Im older than all this second millennium tech (discordisms). Assistance por favor.
Thank you
Chirp and retry mostly. Just kept erroring out. Even had some generations say they were generating but never completed even after an hour.
Today it is working perfectly! ๐
i got a bug where it keep saying the previous verse instead of the next one
Some issues with our workers so bot will be in maintenance mode while we investigate. Apologies
I spoke too soon it appears.
Should be all set now (we just had a brief blip). Sorry!
need to think about how to fix this as this is a bug in Discord itself ๐ฆ
I wonder what setup the folks at MJ use?
I used to manage discords but its been a long time so idk how these buttons are even done lol
Is it a custom made bot?
One workaround for now is: uninstall the Discord mobile app from your phone, and then going to "discord.com/app" in Safari browser instead #๐ชฒโbugs-web message
Letter ร is converted to letter O when using app.suno.ai
Just had back to back generations come out in a different language. Not even sure which one but wow.
When you speak 50 languages sometimes you get a little mixed up.
Well a little mixed up and now 4 generations in a row are different things.
Does not matter what I do it continues in what sounds like German.
In all seriousness though, I'm sorry this is happening to you. I know it is frustrating.
Unfortunately all I can do is commiserate with you since I don't know how to fix it other than what you're already doing.
I assume you've moved text around, added instruments, changed the listed music style and other tricks already, right?
It was just continuing from the first generation which I really like. Sad face! 
Oh I hate when that happens.
I've also had it do a complete change on me in the middle of something, altering everything--and sounding great!--but it's useless to me because it doesn't fit at all. ๐ญ
so sad!
Nice intro piece...and you say subsequent parts are all in German or something like it?
Yep, all 6 tries were in another language. Even when I restarted from the initial generation.
Had to just start all over and pick a different start.
there is something wrong with the continuation button 2 with iphone discord app. I cant push that button at all, only the first one ๐คท
Yeah, this is a bug with Discord ๐ฆ there are some workarounds if you scroll up...
The save voice button always says โthis interaction failedโ Is there a way to fix this? Itโs happening to my other friend too
Can u help with this glitch?
Sorry, this is Bark right? Will take some time to fix..
Yes it is bark
Ok
Try emailing about this to billing@suno.ai to see if they can get this fixed for you.
Suno AI webapp always makes everything pop music, even when prompting other genres. This isn't consistent, but one time I prompted christmas Carol and it said "energetic pop catchy"
An example
I clearly asked for kazoo and bagpipes but it generated "pop energetic quirky"
your syntax is wrong, give it the theme like this : "genre subgenre subgenre, your text" .... e.g. "house electro pop, love you girl"
Yeah but on other prompts I used the correct syntax and it still said "festive pop" instead of the "Christmas carol" I asked for. It's closer to what I asked for, but still far from what I wanted.
We'll work on making the "imagine box" more convenenient but we'll also have a more custom form next week !
Hyped for it!
When signing in with a google account, I had to log out and log back in again for the free creds to show up, just wanted to let you know
Hi, small problem here: my account created first with gmail appear to be linked with my discord on the website but here it don't detect my subscription. Any help would be amazing
Yeah there's still a few hitches Google account ssometimes. If you've tried logging in and out again (ie, turning it off and on again) then email billing@suno.ai
Yeah just tried to logging in off and still the same I make an email thank you
An annoying bug with new Suno on web version:
Sometimes the AI makes pure instrumentals for no reason when thereโs clear lyrics, which is a waste of credits. (Plus itโs more annoying that the instrumentals are generally better than the vocal ones)
Ha, sorry about this. We're looking into how to make this less frequent. Any particular genres you notice which do this?
I first noticed it when I made longer lyrics when I told it to do multiple voices like "Person 1: lyrics
Person 2: lyrics"
But now its so frequent in any genre, like 8-bit, orchestral, and electronic
When the web Chirp has the fuller controls matching the Discord Chirp, can probably try and find style and lyric formatting techniques to minimize the chances at least. Soon....
Last one i tried ended up as "chiptune fast-paced" and even that ended up making the instrumental onlys
My guess is that so many chiptunes have no lyrics that simply using chiptunes in the genre field acts as a fairly strong signal against any kind of singing. You know, I'm curious, let me poke at it a bit this morning and see if I can find a ChatGPT compatible workaround. Also in the future if you get just one with voices you wil at least be able to use that as a base segment to continue from.
I was able to make multiple chiptunes before that was really good, plus the first time I experienced the instrumental bug was with rap, and I highly doubt rap is a common instrumental, especially when its just trap and stuff
Names for YouTube channel
You could try adding extra description of vocals, voices, or even both. Seems to singing more likely though sometimes the singing is chip-tune-ified too โค๏ธ A welcome addition to my robotgarden.
ChatGPT prompt: a fast-paced chiptune song with catchy vocals and voices
In general the songs seem less chiptuney though. You could try stacking chiptune prompts but sometimes ChatGPT ignores the instructions when that specific so maybe wait till full release with more contro.
I donโt know how much the gpt prompt actually matters with the AI generation, I get lyrics but the AI itself doesnโt care about them, and I dunno how much the gpt prompt really matters to thatโฆ
I can't seem to find messages about it, but I am getting this high pitched noise drowning the voices, then the music more and more and can't seem to find a workaround, sometimes, a different voice segment will actually come out good and restore the quality, otherwise, multiple generations of the same segment this time don't seem to resolve it, which is real sad. Unsure if anyone else is experiencing it?
Yeah that's an issue in the current version. The version incoming that you can preview on the website has a longer time for each section so it should be better, in addition to being higher quality in general.
Very nice! And was just on your GitHub before coming here funny enough, nice to meet you and thanks for the intel!
The lyrics sometimes have the genre in it
like if I put hyperpop future bass as the genre, it puts those words in the lyrics sometimes
or tries to incorperate it there
i'm using the web version
example
this is just one but it happened multiple times
it's kinda annoying cause it's using up my credits
but it doesn't happen all the time
still annoying though
The web version is a preview/beta of the next Chirp model. In the future you will have full control of lyrics just like the Discord chirp. Because yeah the lyrics AI isn't the brightest bulb right now.
Oh so it really is a bug
my bad man
Still the best ai in existence rn
I love yall
Hey @muted flame lil bug i kept forgetting to mention is that Web and Discords credits only sync when the Webpage is refreshed so theres an chance where you end up with an extra 50 credits to use.
peak
thanks, it's checking the same credits under the hood but looks like the web should refresh more often.
I have a bug or idk if it's intended... I can't see the generations before a certain point, it just stops scrolling up and I can't look for certain songs which makes looking for the lyrics kind of hard. Also in the "My Clips" tab I'd like a way to see the whole lyrics, since now it just displays half of them with no way to pop them up or read them whole. Thanks ! The service is incredible and I can't wait to see it grow.
Like here I can't see the whole lyrics , actually in none of the songs
The first ones intentional the 2nd one will be fixed soon in the mean time you can regain it through the inspect source tool
Yeah I figured the my clips tab is the way to see the lyrics of all the songs but ran into that issue, will try that !
Right click console and paste document.querySelectorAll('.css-1nvzlwc p').forEach(el => el.style.maxHeight = 'none'); or add a 'bookmark for javascript:(function(){document.querySelectorAll('.css-1nvzlwc p').forEach(el => el.style.maxHeight = 'none');})(); and run that
Hey, my subscription does not work. I linked my Discord account but it does not work
Could you please email billing@suno.ai about this issue so we can get it straightened out?
I did, thanks!
Do you have any support working over the weekend?
Potential fix could be Every 10-15 minutes have this check if tokens have been used on discord or Web if so sync
Trying to confirm this now...
Just as a possibility, could you try logging out and logging back into to Discord to see if that works?
Sure. But I'm on my mobile, I need two hours to get to a computer
I just got a confirmation, we do have support over the weekends.
/chirp
Go to the #๐ฃโchirp-beta-1 channel to use the /chirp command, and check out the #๐ฃโhow-to-chirp channel for step-by-step instructions for making music using Chirp on Discord.
@solar grove Im having the same issues with mine as well, i subscribed however i cant use it in discord, i sent an email but just thought id mention it here too. Thank you
Thanks for keeping us in the loop.
Try doing just the command and enter.
Huh.
OK, I see what you're talking about.
Can you do it in the #chirp-beta-X channel and then use the labguage of your choice?
Imma try that
Please do. I'm curious to see if the result then shows up in the Espanol stream.
it doesn't seem to show up there, but it did generate what I wanted tho
That's just strange. I'm going to mention this to others to see if it's not working as intended.
it works well on the web but not working in discord.
ใ ใ ใ ใ ๋ฐฉ์ฅ๋์ด์๋ค
Is it something a Moderator may assist with?
์ค๋ญ๋ผ๋ญ๋ผํ๋ค์
If you mean "The audio quality is better on the web." then the reason is the web Chirp AI is a preview version of the next version of Chirp. So the web is a little newer than Discord. Next week hopefully both discord and web will be the same.
hello. i subscribed just several hours agao but it still says i need to buy more credits. anybody knows what's needed to do here?
email billing@suno.ai
if it only says you need credits in Discord, but not the web, then check you are on the same Discord account, and maybe trying logging in and of Discord. If credits are missing on both then just email
Iโm the iOS app discord the continue 2 button is overlapping the message underneath. I canโt press that continue 2 button this way.
Itll be fixed soon enough
As a temp workaround, if you uninstall Discord app and go to the Discord web page on your phone browser you can continue it https://discord.com/app
Godamn it
Hi all.
Ive found a bug , when i search to songs i created some days ago, i find them, but they arent showing options buttons , like the continue one.
guys, can I dm y'all with a serious bug?
Who can I talk to? Don't want to say here since it's a pretty big exploit
DM @muted flame
Alright done
Mhm the one where if your on the website and use discord chirps u have extra tokens? Because if so thats been reported here already..
not just discord...
but I messaged mc already
And i'm not combing through all these messages lmao
i am trying to create a song about my grandson, no matter how I word it it gets an error message?
Same issue here, will do as suggested
i.e. it only says I need credits on Discord, the web version shows the credits from my subscription
Yeah you may just have to email for that one, still some occasionally hitches linking accounts
Gotcha
could you DM me the prompt you're using?
Is there also an API available?
no
hello! How can I really end a song? I mean all the clips do not really end with silence.
Try [outro] to get an extended instrumental you can manually fade out in a DAW application, or try using [fade out] a few multiple times to try to persuade the AI to fade out the song.
Hi all! I bought a subscription for $10. In my personal account it says that everything is active, but in the discord it says that there are no funds although the balance is full.
Email billing@suno.ai to get it sorted
I got this bug now where the website version duplicated one prompt
I did listen and these there's just 2 unique generations here
of course it did spend the credits double here too for some reason
not looking to get my credits back, just reporting this in case it's more common issue on the web version right now
there was something weird going on this generation in general
if you look at the website image
but actual video version has different image
also on top of that, this gen seems to have different lyrics to what I got shown, but it doesn't seem like it fully hallusinating it.. but this part could be just a coincidence
oh wait.. video has the right lyrics ๐
and it did do two different gens, they are just slightly different, but the preview images look the same..
The idea is that you can pick the one that sounds best to you to continue.
so... from this image, songs 1 and 3 got same audio, but using song 1-2 lyrics.. and songs 2 and 4 got lyrics from song 3-4
no I get the idea of picking.. but this is a bug
basically I asked once, got 4 variants and they got all messed up on the list (50% of them show wrong lyrics and 50% have wrong background image + didn't prompt twice to get 4)
Oh. OK, yeah that doesn't sound right.
yeah this hasn't happened before, and I'm past 800 generations
Thanks! Probably quite rare, could be because job got retried. We will take a look
I got screwed over by ai. None of two versions of my song have any lyrics despite them being written
Not even chorus present
c
Could be the genre youre trying to use
It happens sometimes. You just have to throw those dice again.
I've upgraded to monthly plan, but still getting this message on every chirp
ideas?
Please email billing@suno.ai to have this issue resolved.
HI ! im new trying to use suno , but 3 of my last generation have no lyrics .. im loosing credit on that
what is the prompt your trying to use?