💡┃suggestions-web

[orchid forge]

WTF BRO u pay like 10$ for a freaking gamechanging Ai. People like u shouldnt be allowed to use this kinda things.

[sage mirage]

I know you feel personally slighted by the change, and I can't help you with that. All I can suggest is you don't renew, clearly Suno isn't performing to the standards you previously thought it was.

It isn't the Suno you liked.

[small raft]

For the record, I've dumped probably ~$300 into Suno thus far, but almost entirely stopped using it after the change.

A bit more than $10

[sage mirage]

You're already voting with your wallet then.

[orchid forge]

all i can hear is mimimi

[small raft]

Voting with my wallet would be submitting chargebacks, since I'm not getting what I originally paid for. But, as a business owner, I know that's headaches for the company. So I'm hoping the devs make it right.

[orchid forge]

Go pay a producer. Im a musicer with 30k monthly listeners i can tell u what real producing cost... I pay 2000€ for 1 song so be quiet man

[sage mirage]

Small business owner is right.
I'm going to bow out of this conversation, there isn't anyhting else left to be said. Lyrics are coming back in the videos eventually, they were removed temporarily to get v3 out asap.

You clearly aren't happy, sorry about that.

Related to this conversation actually, since around the time this originally came up the 2 generation thing also was discussed.

It's a GPU renting sorta thing

[small raft]

I've paid more for music in the past. Besides the point. This is a service offered, was advertised a specific way, had specific functionality, and that changed without warning in a way that made it double the price to work with. That's bad business. Period.

[tepid trench]

Looking at the subscription page, I seem to be getting exactly what I paid for.

[sage mirage]

It isn't any less expensive to generate 1 song vs. 2 as far as Suno is concerened

[orchid forge]

go whine

[sage mirage]

If Suno were to offer 1 generation vs 2, that 1 generation costs Suno the same amount as giving you 2

[small raft]

Super helpful!

[sage mirage]

Two generations per ~generate~ is the GPU efficiency sweet spot. According to pat, anyway.

and in-line with othe generative AI platforms

[shy relic]

oh!

[frail leaf]

I have a suggestion: Keep giving subscribers the same 50 credits per day that the free people get.

[shy relic]

that should probably be more publicly clarified

[sage mirage]

Suno is like 12 people

[small raft]

This isn't exactly linear, but there is an advantage to batching. It's not 1 costs the same as a 2 batch. It's more like 1 would cost 1.2 or so.

[sage mirage]

Maybe more now that they are big time what with the social medias and the rolling stones

[shy relic]

its not hard to like add a paragraph to the faq

[sage mirage]

It isn't, but lets be honest, nobody is going to read the faq.

The bigger change should probably be the language around how many 'songs' you get with your credits on the sub page

[feral aspen]

Let's lower the temperature in the suggestions room.

[sage mirage]

Since the wording reads, to a layman, that you get thousands of WHOLE songs

[small raft]

Either way, batching 2 doesn't mean they had to make the lyrics+style identical for both with the recent change.

[sage mirage]

Rather than a bunch of 2 minute clips

[shy relic]

sometimes i feel unique because i always read the faqs

[small raft]

Is there a better room for airing grievances? 😅

[sage mirage]

Same

[shy relic]

dms ig

[sage mirage]

You're right though, clarifying how Suno works for those that want to read it is easy enough

But I'm not Suno, just some dude

Suppose it could go in the wiki though

THat we can add to

[feral aspen]

It's totally fine to request features, changes, whatever. But Suno tries to use this room as more of an archive or list than for conversation (or especially arguments), to get a gauge of what suggestions are most popular etc.

[formal bear]

Hi, I will address this one box thing here this last time.

The decision is not based on cost -- yes we call openAI one less time, but hey generate lyrics is free for you in custom mode, so that call is almost infinite there. Clearly the call to openAI cost isn't the full reason to change it.

When the generations are with different lyrics, it is much more difficult for us to recongnize what you like / dislike -- is it the model, or is it the lyrics? As you know, lyrics provides a large prior on the music generated. We want to remove that inconsistency from the equation, so we can collect your feedbacks with one box generations better.

This is not a cost decision so please do not spread that theory again. We appreciate your support and would be very sad to see you leave if changing one box to same lyrics is the reason that leads you to do so. Yet, we believe that this will help more users in the longer run.

[small raft]

I'll point out (think I may have already) that this did completely break an extremely fun use case.

[formal bear]

I know, and I will def keep this in mind in the future designs.

[small raft]

I suppose an API would fix it.

[formal bear]

I know we take it out now, but when we get to mobile, etc, a lot more decisions are to be made. This multi-lyrics factor will be considered in the future.

And we do plan on making better lyrics auto-gen in the future as well.

[small raft]

Won't derail this channel more, but is it against ToS to... unofficially use API endpoints? I could probably get back what I was doing before going this route. The calls seem pretty simple.

[sage mirage]

My mistake on the cost thing, it's a much more nuanced user experience deal.

[small raft]

(by doubling the amount of generations I make, and ignoring one)

[formal bear]

yes it is kinda against ToS...and honestly the backend change makes it not so easy to pass on the requests. I saw someone trying to do that yesterday and it errored.

(yes we do watch who is accessing API)

[small raft]

Well, for the suggestions channel then: Please make an API. 😅

[formal bear]

but we will get the videos fixed first. 😄

[small raft]

thanks for the attention to all of this. admittedly the two unique to one unique thing just kind of caught me off guard, so apologies for causing a ruckus. I think an API would probably have me shoveling more money than I care to admit at this, though, so.... lol

[formal bear]

yeah we should have done a mini-annoucement

I asked someone that morning and we decided no just move 😦 -- will do a better job next time

[small raft]

No joke, I was helping my brother do some tile work a couple weeks ago, and just kept doing 10 generations of "[random genre] song about two guys working on some kitchen tile" over and over the entire day 😂

was absolutely hilarious. don't even know how many credits I burned through doing that 🤣

[formal bear]

sigh it is extra work on backend and frontend, but we could make it optional and you have to select two gen with different lyrics for this to happen.

that being said, we have a lot more higher priority things to ship / fix

you would want to hear v4 earlier than this right? 😄

[sage mirage]

[small raft]

maybe. wen v4 alpha? 😛

with an API I'd 100% make an interface to just keep making a playlist of new songs on a given topic in random genres and just keeping playing through them until I'm out of credits.

[sage mirage]

georg talking about inpainting is by far the most excited I've been for a feature in Suno. After the 'Continue From' one, clearly, for v4

[formal bear]

oh we also have radio on the map..

so

your wish may come just through another feature

[small raft]

excellent

[sage mirage]

Pipe in everything everyone else has already generated

[formal bear]

or public ones, but yeah

[sage mirage]

Right yeah, public. Still

Will have to have plays a part of songs even if they are off Explore. People will be stoked to see others listening to their stuff

[formal bear]

oh a lot more user features are coming as well

lol it's going to be very fun

[small raft]

Overall though, I've enjoyed this product a lot. Way more than any other AI stuff I've messed with (and I work pretty heavily with it in other contexts). So, keep up the good work for sure.

[formal bear]

TY. again, appreciate the feedbacks.

[ivory rivet]

Hey, Tony. I've been enjoying Suno for awhile. I appreciate you communicating with us and I'm very humbled for that. That is all I wanted to say and I'm excited for new things otw.

[formal bear]

Thanks! Talking to the users is one of the most interesting and exciting part of my day, even though I should spend the time writing code sometimes...

[ivory rivet]

😂 Yeah I get you. Thank you, Tony.

[tight beacon]

Is a Volume control coming to the Web UI?

[small raft]

You too? For my work I have to sometimes completely close all chat apps to get work done and stop talking to customers and procrastinating 😅

[formal bear]

should be there already if you refresh?

[tight beacon]

Oh they finally added it?

[tight beacon]

I'll be danged it's right tehre now

That was a big request on my list

I see they changed the Explore layout too

[formal bear]

yep that's @solar igloo 🙂

[languid bobcat]

Great stuff!My suggestion would also be a separate tool for the tracks😇

[sweet shell]

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

Would be amazing to have something like this but updated, like the last version

[bitter crescent]

Is it possible to download the subtitle file (like .srt) of the lyrics?

[lament jungle]

hi - thanks to Suno for turning on v3. I noticed that the "download video" in v3 does not include the lyrics in the image part of the video (only title and photo), which were included in V2 "download video". Is that coming back? Thanks, Tony.

[drowsy root]

I think he could add an option to clean up the slight roboticization of the sound after making the songs. In some places there's an echo in the sound when the song is going up. If you solve this echo or roboticization, you can do seriously good work, thanks to you, we can do very good work, win win.

[red tree]

suno needs folders! . i want to sort my music into genres, songs completed, in progress etc.. also id pay so much more just to have the ability download track stems and midi. just saying 😉

[drowsy root]

Yes true But I couldn't participate in the payment part anymore, I'm already making 4 ai payments 😵‍💫

[quiet warren]

Use playlists

Only after 10 tracks from v3. I wish I could go back to v3 alpha. V3 seems to be forcing or trying to get the full 2 minutes and is almost exactly the same if you re generate, also try’s to end the song on the first try. V3 alpha was better, I wish you didn’t take it away.

[cold forge]

how using man sounds

[rugged walrus]

If I created music using Suno with a free account, can I use it for commercial purposes if I pay a subscription later?

[odd glacier]

I'm not having much luck with V3 either. All of my song production so far has been "alternative, grunge, 90s" and V3 seems to have completely forgotten what that genre sounds like. Additionally, I've noticed that nearly every generation now uses "harmonized" or "choral" vocals instead of a solo lead singer like V2 & V3A did. V3A had plenty of issues; I can't believe they thought it was ready for an official release already.

[tight beacon]

Is Library sorting something being added in the future? Like sorting A-Z or Z-A.

[willow smelt]

Would love to have actual folders to clean up the library. The playlist doesn't prevent cluttering of the library as everything stays in the library anyway.

[gusty pelican]

Similar to the reason for requesting folders, would like a multi select option for songs to delete permanently from trash. Having to do each one individually is time consuming and hand cramping when I haven't kept up on it and have like 20+ files to delete 😭

Also would love a way to "save" custom input like lyrics, prompts etc on the web UI itself without having to waste a generation or use an outside program. Sometimes I get spur of the moment ideas that I don't have the time, focus or inspiration to follow thru on for a complete generation and would like even just a primitive notepad esque feature there to shove stuff into for later

[safe sand]

V3 I created most of a song missing just the last verse/chorus. I then left it and came back the next day and FFS, it won't create the final piece with anything like the original pieces. It's like if you don't create the entire song with all it's parts at the same time, you'll never get it to recreate the voice and/or music style. SO annoying wasting tons of credits.

[safe sand]

No

[mystic bloom]

can you hlep me fix it ?

[willow smelt]

You can use shift and control click to select multiple files and then move all of them to trash or add to playlist.

[rancid falcon]

API.. if not for free, then for $$.. Waiting.

[gusty pelican]

Did not know that, thanks. But does this work for deleting multiple songs permanently when they're already in trash? Because if not, that's the feature I'm describing I'd like
(I'm a bit OCD like that, makes me anxious when bunch of songs pile up in trash so I'd like easier way to mass purge them)
Ty for the tip tho, still helpful ✌️

Edit: Just tried this in the trash page and it works with mass deleting them, so thanks, issue solved!

[velvet vector]

It would be very helpful to be able to tell Suno after you create an instrumental which part is the chorus or the verse. It would be easier to repeat the parts

Also if you could add bagpipes to the instruments it would be super cool

[steady scarab]

Please add V3 alpha back to the selection list alongside V3 and V2. Users will have the opportunity to experiment with all three versions, and we can gather feedback to determine the preferred one after about a week. This approach will allow us to accurately assess if V3 is the suitable version for public distribution, especially considering the obstacles some users have encountered with it.

[sick marsh]

#💡┃suggestions-web, #🪲┃bugs-web, #1101325881059840040, should be combined into a single forum channel. Forum should have a pinned post of suggestions the team has already planned, & bugs that they are aware of. This would limit chit chat & communicates better in both directions....

[opal pulsar]

The generated clip under a specific amount of time are refunded. This can happened when the ask seems to have no sense for the AI ( it happened to me when asking during test a bridge after an ending fade) or when the output of the AI just bugged

[potent crane]

The thumbnail AI pictures are a criminal waste of resources, if you are going to bother with them at all, at least make them hi-res enough so they can be used as album art etc, at the moment they are too small and when resized they go blurry.

[tepid trench]

Also, less bird pictures for instrumentals please 😄 I know it probably happens because of "chirp" but it would be nice to have some more interesting visuals.

[gusty pelican]

Speaking of the AI art, maybe as an added bonus for paid tiers, on top of higher quality / more varied cover art, let us regenerate the cover art if we're not happy with it
(also totally an option for free tier too if they want to be generous!)

[swift parcel]

When I wrote the V3 alpha, I had a good understanding of the dream pop genre, but the V3 model doesn't seem to understand it at all.
I'd love to see it added back into the V3 alpha.

[tepid trench]

• Show creator names against public songs (e.g. on New/Trending but also on the song page itself)
• Per-song options to allow/deny download, continue/remix by others
• Creator profile (with contact info)

[leaden marlin]

Hey Guys, love the product. Any chance to get mass download of the songs as a feature? Such a pain to get them quickly.

[turbid pumice]

Would immediatly subscribe to suno, if there were the option to use paypal for payment.

[hybrid crane]

It's in the works 🙂

[hidden basalt]

It has probably already been mentioned, but the possibility of splitting the song into individual tracks => vocals & instrumentals or even more
Then you could fine tune or process the tracks further manually.

But not sure if it's possible since the NN could do it in one session

[hybrid crane]

For that we need higher and better quality first. The team is pushing the limits to improve it. It's a complex task, so getting to higher quality like 320kbps is still in the research phase.

[turbid pumice]

great news 🙂 hopefully sooner than later 😉
thanks for the info

[nimble wharf]

would be cool if suno would extend to something like a video editing interface in the future. where I can select parts of the song and replace it with other text or re-run the AI on the part for better spellings etc.

Also copying the style of another created song for a new one would be great, as one specific style sounds pretty cool sometimes but it's impossible to re-create.

basically similar to continue from track, but start from zero

[mint thorn]

(i think is my best idea)

it would be perfect if Chirp could produce .wav in addition to .mp3 and .mp4

This would allow us to have the instruments on one track and the vocals on another track, that way we can go into software like FL studio and add effects to the vocals, add delay, reverb, glitch, or pitch the voice too.
that would be crazy! 🤔 🤯

[tepid trench]

Quite a few people want this, including me.

At the moment though (with current quality), WAV apparently does not offer any advantage over mp3 and the songs aren't generated from separate tracks/parts so there's no way to have those separated at the moment, without using a stem splitter to do some simple separation.

For example I use vocalremover website to separate the vocals, melody, bass, drums, then I can adjust the volume levels of those and add some reverb to the vocals, do filtering etc.

[mint thorn]

This is precisely the suggestion I am proposing. That chirp gives us in addition to mp3 and mp4 the possibility of downloading the instrumental and the voice separately.

[mint thorn]

use UVR ^^ is free and not limited

and you can have the best of the voice 😉

[mint thorn]

i never speak about the quality wtf "chirp gives us in addition to mp3 and mp4 the possibility of downloading the instrumental and the voice separately."

[tepid trench]

This quote is about the separation of audio into different tracks/parts. Basically they won't do it until it's good enough.

[tepid trench]

You mentioned "WAV in addition to MP3 and MP4"

[mint thorn]

yes becaus is the name of the file lol...

mp4 = video

[tepid trench]

MP3 is typically lower quality than WAV, but in Suno's case, WAV does not offer any higher quality than it can generate.

[mint thorn]

Well, I think it's really hard for me to express myself.

my suggestion is:
Chirp could give us the ability to download both the instrumental and the vocals separately once the audio is finished.
So. Good day 💜

[hybrid crane]

There is only mp3. That does not excist yet. I posted a quote from a Dev to make the point that we need higher quality first, then Suno can offer stems, aka voice and instrumental separately.

[mint thorn]

ah ok, sorry I'm not English and I have trouble understanding sometimes. Sorry

[dusty dove]

Suggestions:

1. More than 1000 characters in the text box to enter lyrics for customized songs.
2. Automatic continuation to finish the entire song. Add "auto-continue" option.
3. Add "full song" create button if song continuation idea was used.

I believe this would be very welcome.

[echo flower]

Are you saying it would just process say 12 - 16 lines, then auto continue and do 8-12 more lines in a loop till all the lyrics are exhausted? Would it branch? 2 roots, then 4 2nd level, then 8 3rd level? That would burn a lot of credits (70 credits with a maximum of 4 minutes) and never give the user a chance to steer.

[rigid eagle]

I've just noticed that the video exports no longer show the lyrics in the video, I kind of liked that, since it showed you the exact prompt that was used to create that song, maybe it can be added as metadata in the video or something? I'd love to keep that.

[hybrid crane]

it will return.. overhaul taking place 🙂

[rigid eagle]

Nice, thank you! 🙌🏽

[sharp zodiac]

Add feature like ChatGPT's image generation, where you can refer to a specific image seed? This way, we can easily pick the voice we have discovered and like without wasting time.

[worldly falcon]

A seed display, so that we can reuse the same seed of a song we made previously to keep the same vibe and BPM

[hybrid crane]

Hey Cristazio! Unfortunately, Suno doesn't use a fixed seed for song generation, so you can't reuse a seed from a previous song. Every generation is unique, even with identical inputs.

[sharp zodiac]

So add it, if you don't add you competitors will.

[echo flower]

you can continue from a clean part of a song you like, then use external software to chop off the part you don't like then modify it. @hybrid crane made a video the other day talking about a workflow that helps you keep the things you want by starting with an instrumental.

[sharp zodiac]

where are these videos?

[hybrid crane]

First of all.. what competitors? There is nobody doing it and sound like Suno... And it actually works like it works because it doesn't have a seed.

[echo flower]

That's not how seeds work on most models. you don't get a "feel" you get identical output with an identical seed/prompt/model. Trying to nail down a feel probably wouldn't use seeds in that way.

[echo flower]

https://www.youtube.com/watch?v=5uJw5kb05as

[acoustic pike]

thank you for 2 minutes extra time limit.

[autumn crown]

I'm not advocating for implementing the feature to let the user set the seed, but either there is a (pseudo-) random component to the generation and thus a seed, or no random component. The latter doesn't make much sense to me. But maybe I'm missing something.

[hybrid crane]

Hey saddswwsdds! You're right that Suno's generation process involves randomness, but it doesn't offer user-accessible seed settings. This randomness is integral to the AI's creative process but isn't something users can control or replicate.

[strong rampart]

Not sure if you have a roadmap, but allowing export of stems would be an amazing feature

[hybrid crane]

asked.. many times..

on the roadmap and are researching.. for that they need to improve quality first

[strong rampart]

Yeah, the mastering is a noticeable issue.

[autumn crown]

As far as suggestions go: It would be amazing to be able to upload our own music snippets. I have literally hundreds of song ideas from 10s to 2 minutes, and I would love to generate a dozen ai-ideas for each of them.

[hybrid crane]

I get where you're coming from. Implementing a feature to upload personal music snippets to Suno would be amazing. However, it's tricky because of the potential for copyrighted material. The Suno team would need to ensure that anything uploaded is your original work. It's a complex challenge, but it's great to know you're keen on this idea!

[red tree]

bring back v3 trial! i just wasted so many credits trying to get it to change genres mid song, i was doing it successfully on every try before, it wont do it at all now. its also giving me songs for genres i didn't want or even come close to asking for. its also outputting nearly identical sounding songs instead of giving 2 different ones, i used the same prompt 3 times in a row and received the same song so 6 tracks all the same. i think i need to take a break for a month cause im over here spending to much money trying to recreate the magic i was getting. everything i enjoyed about suno is not working now. 😦

[kind vale]

Second what @red tree wrote. The latest update ruined all the magic. Also the sound quality did not get better, but worse. But most annoying is the general mushy noisy "style". Instruments and especially vocals were a lot clearer in alpha.

[marsh jasper]

I would love for Suno to label what was V3 Alpha and what was V3 & to bring back V3 Alpha in addition to V3

[marsh jasper]

Has “consistent voices” been asked about? Would be cool to have the same voice for a series of songs that I am making 🙂

[kind vale]

First of all we'd have to get any usable vocals back 🙂

[red tree]

i use moises or stemroller to separate vocals, its about the best you can do for now. unless you make the instrumental and remix it with lyrics. or visa versa. not ideal, but i imagine it will eventually get to that. i want complete stems not just instrumental and vocal tracks. separate it all. maybe have an extra tier of membership that allows this. honestly id pay a few dollars per track as long as it includes midi. it would be nice to change instrument sounds using midi . i mean i dig the violin being used as a guitar for my technical death metal tracks. buuuuuut guitar would sound better lol it would make it possible to stop the ring on notes this way. those have killed at least 50 good tracks of mine.

[modest trail]

Style of Music needs more than 120 characters:
BPM
Meter
Keys
chord progressions
instruments
style
common adjectives
music score specific adjectives
etc

time signature

[delicate cedar]

Have you tried to use any of those in the style prompt with any success yet?

(Within the number of characters that are currently allowed, I mean.)

[modest trail]

Keys, chord progressions work, instruments, work

staccato seems to work, first attempt on music score specific

[delicate cedar]

That's all news to me. I know that some things work and people have tried other things with inconsistent results. I did not realize that those were sucessfully used.

[modest trail]

https://app.suno.ai/song/c24e4a69-fa47-47d9-90ab-f2ae344862cb

[dull sandal]

any reason why we cannot browse new and trending compositions past the first page? can you add some pagination or next/prev at least ⁉️

[timid heart]

Is there a way to slightly edit lyrics without changing anything else in the song?

I did this once, changed a few words, hit create, and an entirely new song was created with the slight change of lyrics. I was hoping to only change the few words, not create a brand new song.

Also it’d be great to be able to add my own art from my saved photos if I choose to (unless this is already possible but unavailable on smart phone)

[topaz prism]

Infinite Radio When?

[hushed oasis]

like streaming?

[warped ember]

Integrate samples into music, analyze it and make music based on it.

[mint thorn]

they will talk to you about copyright ^^
I already asked for the names of the artists (similar to have sample) to get some inspiration from the vibe, they told copyright
It's sad because samples are appearing more and more these days. plus it brings back a note of nostalgia

[gusty pelican]

Only way I know of so far to kind of accomplish this is to continue the song (either from the very beginning, 00:01, or from whichever point in the lyrics right before you want to change) and regenerate from there. It should keep the style of the song although you won't get a 1 to 1 copy of the original.

[somber dove]

idea:
DJ Mode
Have some variables (i.e. rock, breakbeat) that you can adjust throughout the song live. Every minute will spend 10 credits. And yes, this was inspired by the MusicFX that Google made.
Here’s an example of what I’m talking about:

[vocal portal]

Just another request to have the censor dailed back:
Theres an irony to the fact this has come out as one of the best songs Suno has created for me. I am very much going to miss this app when my sub expires tomorrow.
https://app.suno.ai/song/42feeb07-90f5-4687-b26d-0bc024bada11/

(i mean it broke the end again, by resinging the last verse... but thats kinda normal)

[balmy zealot]

Why the decision to remove lyrics from the mp4?

[tepid trench]

This is a temporary thing. They're working on adding it back I think.

[timid heart]

Thank you

[unreal ice]

Eve, Stellar B;[';lade, Artifi;cial hyperiii;ntelligence;, w;ife of Maciej Nowicki

[blissful mauve]

just an idea, possibility to separate the tracks of music and vocal cus sometimes you need to mix in the vocals better via a DAW that would be AWESOME

[blissful mauve]

thats why it would be awesome to have the vocal track separate if you wanted that

[hybrid crane]

asked daily.. it's called stems.. if you look in search you see how popular that is.. but not easy for them to create

[blissful mauve]

i see

no im sure its not but well perhaps in some future version who knows

im trying it but man its not easy

you know make a track with as less possible music so you can filter out the audio better

and then make the music mix it in and add your melodies

like this 🙂

did some work on this

just in the end its bit muffled 🙂 did so much trying to get it right with the izotope tools

all i can do is rewrite the song 🙂

[proud tendon]

Take it into https://vocalremover.org/splitter-ai

[blissful mauve]

oe sweet

[proud tendon]

Be careful there's a limit upload before you have to donate or wait 24 hours

[blissful mauve]

aah okay thats cool

[hybrid crane]

Great share guys.. but please keep this channel on topic about suggestions.. Feel free to continue it in #💬┃general-chat

[proud tendon]

Suggestion:

1. Generate only vocals
2. Generate only specific instrument e.g. only bass guitar for a Jazz track
3. Voice to voice
4. Train your own vocal models
5. Use reference track(s) as inspiration

[hybrid crane]

1. try "a cappella" in the style of music or [a cappella] before sentences in your lyrics
2. try the instrumental option and specify for music instruments in the "style of music"
3. no idea what you mean...
4. Suno has to be sure it's your voice due to copyright so they are still investigation how too
5. Suno is about making unique music, not copy catting and copyright infringements..

[fallen loom]

@red tree Look how they lobotomised my toy

[neat plinth]

Hey there!

First off, I gotta say, the launch of V3 is fantastic! However, I've run into a few issues and have some suggestions that I hope you'll find useful.

1. When creating songs in Chinese, the first character of the lyrics often gets mispronounced, and it seems to happen about 80% of the time.
2. During song playback, sometimes parts of the lyrics get skipped. This seems to occur in roughly 30% of the cases.
3. When creating new songs, two songs are generated at once, which burns through my credits pretty fast. Is there any way we could switch to generating just one song at a time, or maybe have an option to choose?

Keep up the great work!

[shy relic]

dont have an answer to 1 and 2 but for 3 - apparently it's the same amount of processing power needed to generate 1 song as 2, so generating 1 song would cost the same amount in terms of credits

[fallen loom]

Add me to the pile of people asking to roll back to V3 alpha: unable to get anything interesting out, despite any changes to prompts, no matter how dramatic. Just keeps plodding along, or wraps up prematurely, still suffering from deterioration of vocals over time, same weird glassy synth wash intruding. To put it bluntly, it's fucked, and lost most of it's value to me.

[hybrid crane]

let them know what you think of V3 in the poll: #🗳┃mini-polls message

[fallen loom]

@hybrid crane Thanks!

[neat plinth]

thanks!

[fallen loom]

@hybrid crane it seems I'm in the minority😮‍💨

[hybrid crane]

Still your vote does count. They are not stupid. They know it is not perfect.

[fallen loom]

@hybrid crane 👍

[errant grove]

language filter on trending charts would be nice - or just add pagination with more results 😄

[lethal cypress]

Would Love to train this AI for Intro of Music and More Complex Genre Blends.

I have a suggestion related to music. It would be interesting if we could give a structured suggestion for the music. For example, we could start with an intro, followed by a break with glitch sounds, then a fast-paced build-up, and a hard techno rumble kick at the drop. Then, we could have a second break with dark pads and melody, another build-up, and a dark hardstyle drop. Finally, we could end with an outro. I hope that clarifies what I meant. Also, if possible, it would be great if the tracks could be mixed and mastered.

[young valve]

I need a tool to analyze the uploaded song and tell me the style

Also it'd be nice if I could fine tune the song, decide when to add effects, what effects, etc

or what words to stress, etc ... more fine tuning please

[hybrid crane]

https://rateyourmusic.com/ can help you

[young valve]

might also have a page that will show an example of all styles of music that the model is capable of creating

so that I have an idea what I kind of might like

There are occasional misspellings - I'd be nice to have an option to rerecord the same song with a fix for the misspelling

[hybrid crane]

Team is research a new feature.. paint in doing just that 🙂
Hopefully in V4 😉

[young valve]

Thanks - I'm just voicing ideas ...

I also would like to specify the speed of the words

as the same kind of style of music can produce varous speeds

[blissful mauve]

small suggestion for payment options 🙂 not everyone has a CC 🙂 but a debit card, for south american countries for example paypal wallet or any other payment option would be awesome, im sure there are thousands of people that would love to use this Gem

id suggest paymentwall

Instantly connect to over 150 payment methods globally to reach customers locally in 200 countries and territories

PLEASE 🙂 lol

[steady scarab]

Cyanite also a great tool https://app.cyanite.ai/

[steady scarab]

Cyanite also a great tool https://app.cyanite.ai/

[solar holly]

you can add structure like that. go to suno.wiki and look at the song structure tags

[pseudo surge]

I was hoping by now we could see with the highlighting, or a better way, which parts go with which in Library tab. I get that we have these tabs on the right side that drops down to show what was used but still that is hard to tell without playing each part it shows. Not to mention it does not show this when you're clicked on full songs, at least it is not for me right now. Actually what would make things better in this sense and many others is if, just how like in the playlist section, we can grab and move every part. Whether we are in create or library. That way we can keep each part together how we want to. I know the Suno team is working with a lot but the ability to organize and keep track of our tracks should be, in my opinion a priority close to the top of the list. I know for me, I would upgrade my sub once I could keep better track of everything I create instead of just having page after page, knowing I would be wanting to create so much more.

[neon harness]

WHAT DO YOU THINK? A system in which AI analyzes other music that already exists and, using it as a reference, creates new music based on the same rhythm.

[sick marsh]

Can we please have a button to a "Full Song" so we can see if we like the transition? I throw them in playlists, but that's tedious.

[young valve]

yeah - making the final full song would be a good feature

also a quick fix - I would like to "scale" how much I like something ... like I like this x5 and that only x1

[sick marsh]

What about doing something like this for album art on Suno? #💬┃general-chat message

[frank sequoia]

I suggest a different tab in the website where you can put like a word or a sentence to see how the AI pronounces the word or sentence written the way it is, and it should cost less credits to do that (or none)

A lot of my generations come from trying to get the AI to pronounce something right

And it takes quite a few credits

[boreal saffron]

Hello , In summary, aside from saturation on the vocals and an overly compressed output format, encoding, limiting, or other issues, we're on the verge of perfection. I was wondering, given that Chirp generates the drums and Bark handles the melody and arrangement (I'm not sure about how you generate the vocals), could we dream of having 3 high-quality .wav stems in a version 4 alpha for export after generation—drums, vocals, and all the rest? (This would be without using an AI splitter, which introduces phase issues and artifacts).

[tidal wharf]

Add saving downloaded audios under their uuids instead of names, so it’s easier to find them later by url

[tired canyon]

My only real feedback for V3 is it seems the vocals get drown out by the instrumentals - would a fix be on the way?

[broken zinc]

Too many songs are coming out with large vocal ensembles, especially in the choruses for harmonies. It's great that Suno is better at choirs, but i don't want a choir on a country song. Just 1 or 2 or 3 voices at most would be fine. Thanks. The algorithms are getting too heavy on the choir vocals.

[lethal cypress]

Thanks 😊

[spiral acorn]

When the "painting in" feature is added, is there also a future possibility for deleting parts or moving around parts?

For example is the following possible with how the "paint in" feature would work:

• You delete a part in the middle of the song. So now the song is shorter. You now create a small selection across the 2 previously seperated sections and click "Paint in" and it kinda smooths everything out.

[near garden]

Hello Suno team. Nice work! Curious is there a way to accurately condition an output sound over text prompt? e.g. [Drum] and [Raining white noise], beyond lyrics

[tired frigate]

There should be a way to describe the AI how the song is supposed to develop. For example a prompt like: "starts tender, accelerates to upbeat mid-section, ends slower". Is currently completely ignored. In order to produce songs that are more unique this feature is very important.

[unreal flame]

Add a a "SEARCH" tab for private library - essential. I´m new here only 4 days in now and Im already 16 pages deep in my library and find it difficult finding previous songs. thanks

[topaz prism]

yes, infinite streaming of some song.

[sick marsh]

You can find some ways to manipulate it, but we're all still learning! I'd love to give you confident advice on how to drive the pace & mood of your songs, but all we have for right now are loose things that work most of the time. lol

[MetaTags] are important, & can shift things.

I have very good luck using [Key Change]

You can also continue a song & change the style

[hushed oasis]

suno speed rap

[lunar bolt]

If I generate an instrumental via LIA with the subscription, can I sell them on BeatStars, etc., or not at all?

[near garden]

this works for some yea

[hybrid crane]

Yes more people suggested that.. But Suno is about creating new music, not copycatting someone and then end in a copyright law suit. 🤷‍♀️

[hybrid crane]

They are working on improvements to make things more clear...

[hybrid crane]

Hey Mari.. Great idea but sadly it doesn't work that way. Each time you generate a song or speech with Suno, the AI uses a new "seed," leading to potential variations in pronunciation, even within the same language. This can result in different accents or dialects being used, like Dutch being pronounced with a Belgian accent. It's a characteristic of the current stage of AI music generation. But every version will become better.

[hybrid crane]

Hey there! Absolutely, dreaming big is what drives innovation. Currently, Suno primarily uses Chirp for all music generation, not Bark. While the idea of high-quality .wav stems is super cool, it's not available yet. Suno's team is definitely exploring ways to improve, including better quality exports. Keep those ideas coming!

@tired canyon @broken zinc please use this poll to let the staff know what you think about the current V3 model: #🗳┃mini-polls message

[hybrid crane]

Hey PlayerOfGames! The "painting in" feature is still in development, so its full capabilities are yet to be confirmed. But, the idea of deleting and rearranging parts, then using "paint in" to seamlessly blend sections, is definitely a cool concept. Let's wait for the final feature roll-out to see how it evolves!

[hybrid crane]

Hey GodKingDarius! Thanks for the kudos. For your question, yeah, using specific metatags like [Drum] and [Raining white noise] is a neat way to guide Suno. This should influence the sound to include those elements. Keep experimenting with different combinations for best results!

[hybrid crane]

To craft a song with a dynamic flow, you’ll want to approach it in stages. Think of your song as a story unfolding. For the tender start, use metatags like [soft], [gentle], or [mellow]. Mid-section, switch to [upbeat], [energetic], or even [fast tempo]. For the ending, transition to [slow tempo], [calm], or [reflective]. You’re painting a picture for Suno, so the clearer your vision, the better it can follow. Remember, Suno's AI still can be unpredictable. We don't have full control yet!

Why should Suno adjust that? It's up to the streaming services to accept it or not.

[hybrid crane]

Hey Mira, quick answer first: Yep, if you’re a PAID subscribed user, you can sell your Suno-generated instrumentals on platforms like BeatStars. Free user are not allowed to monetize their songs created with Suno.

[lunar bolt]

thx

[limpid zinc]

When you "Continue from" it should autofill all the details from the song you're continuing from

[hybrid crane]

I agree

[spark geyser]

is the reverbing, choral/muffling of lyrics intentional, is it part of the watermarking to recognize what was created in Suno? So this will not change in the future either?

[marsh jasper]

I’m seeing this in a lot of generations as well

[tepid trench]

The watermarking is apparently not something you can hear.

I suspect the excessive reverb/chorus may just be due to how the model has been trained. The watermark was there before, but the devs just didn't tell us about it until recently.

[spark geyser]

Now many of the lyrics sound like an evil possessed cyborg army growing in strenght the further the song goes 😄

[marsh thistle]

I think a mobile app or an app for all devices would be good, there should also be a searching bar so you can search for songs. instead of it being random songs we want the cumuinity to vote to get it trending.

[fiery geyser]

why cannot i change my subscription plan?

[delicate cedar]

Please do not doublepost. Posting in #bugs was fine. This is a warning.

[fiery geyser]

I would like to upgrade my plan, but it shows i cannot do it,can i change my bank id card to make it?

[near garden]

Hmm, it's not actually working for me though

[daring vapor]

Hi, 2 suggestions:

1. Would be nice to include paypal as payment method for subscription plans. I think Stripe even has an inbuilt integration for that. Personally I always hesitate adding credit card numbers to online services (and so will do others potentially).
2. An option to buy just a bunch of credits instead of monthly subscription would be highly appreciated, even if it would be more expensive. Subscriptions are often a headache to cancel, which makes one-time-purchases way more comfortable for casual users.

Otherwise, keep up the great work, it's really amazing!

[near garden]

Feature request: Allow generating ~5 seconds simply to test prompt. Should be easy to implement! I just want to know if a prompt works, and it shouldn't burn 10 credits -- should not burn as much as when I want to create an entire song

[hybrid crane]

It's not consistent yet indeed.. Sometimes you have to try a few times or tweak left and right. A good idea of what is possible is going to the Explore page on the Suno Webpage where you can listen and see how other use their metatags

[ashen robin]

Once a song is done. It would be great to be able to edit the lyrics. When doing song continuation from an early spot in the song, it doubles up the lyrics. This is minor and has nothing to do with the awesome AI your team has built. Just a minor request. Makes sharing the link make more sense.

Feature Request: Show Song time down to milliseconds (and then also allow for continue from to use that same precision. ex: 00:32.123). Better for cutting off fast lyrics at a more precise point.

[frank sequoia]

In my personal experience, it has been pretty deterministic on how it pronounces something. Like, sometimes you want to repeat vowels to make the AI speak it longer or at a certain tone, but then it speaks it in a different way

[hybrid crane]

Sadly everyone has this (with foreign language). According to the Devs there is nothing we can change about that atm. Only by writting it as it sounds as a workaround. Which I do frequently.

[frank sequoia]

I understand

[craggy stratus]

有汉语吗

[dire harness]

I would like to be able to download my V3 songs

[potent crane]

use lala.ai

[worn pelican]

Hello,

I have a suggestion that I believe could improve the user experience when generating continuations for audio tracks. I try to share as many of these suggestions as possible with the development team.

Currently, when I generate a continuation of a track, I need to manually review both generated options to determine which one is better. Sometimes the choice is clear, but other times it requires careful listening to make a decision.

To streamline this process, it would be helpful to have a simple preview feature that plays a short segment of the original track before the continuation point, followed by a short segment of the generated continuation. This preview could be around 10 to 20 seconds long on either side of the continuation point. The transition between the original track and the continuation is particularly important to me when evaluating the quality of the generated audio.

At the moment, reviewing continuations is a bit cumbersome. I have to go back to the original track, fast-forward to approximately 10 seconds before the continuation point, listen to that segment, and then switch to the generated continuation to assess how well it fits.

Implementing a preview feature as described above would greatly enhance the user experience and make it easier to compare and select the best continuation for a given track.

Thank you for considering this suggestion.

[grim cedar]

I agree with that, sometimes it's hard to tell if a continue works but when you 'get full song' there's an abrupt jump in the music. That's happening less in V3 than the alpha though

[wise scaffold]

Would be great if we could have subtitles in the video exports again. Especially since the vocals are often noisy in V3.

It also seems like vocal clarity deteriorates with each step of continuation. Is there any way around that?

Thank you

[frank anvil]

5- Ist's not about copy catting a song or music. Ist just a chance to describe what a prompt can't tell. An influence, a special rhythm, a guitar sound, an atmosphere. INSPIRATION not copying! Things every band is doing, saying: "Listen to this, dude, it's the new record of XXX! Let's take that feeling and make our own song in a similar style!" That's the way music always worked. Says someone singing in bands for 40 years. And if people wouldn't be creative that way, we'd still be drumming on empty trees. That's called progress not robbery. Cher used autotune in "Believe" - then Zillions of musicians grabbed that effect! In your Suno songs too. Are you violating copyrights? No, you don't! Is it copy catting? Maybe - and for sure if copy catting is the way you defined. Even naming and using a genre is copy catting!

[hollow oak]

I don't know if anyone have said that before but:

A search bar would be really good, at least trying to find a certain music by her name, or even by the gender of music, it would really help to find musics and help find musics by musical gender

[hybrid crane]

That's totally fine! But you have to do that with text. Not letting an AI replicate the style 🙂
A good point to start is looking at the genre the song or artist is about. And websites like these are a great help: https://rateyourmusic.com/

[rare blade]

🎵 Amazing work, @fresh shard team. Next step: please teach the model to pick up chord notations in the lyrics.

[round zephyr]

one thing id love to see in the suno.ai website: Clear trash (permanantly delete everything in trash instead of having to select one by one)

[split flame]

I know it's been said but a search bar would be so great for this. There are some songs I have seen others react to that I want to listen to but it's impossible to find them at all right now

[crisp furnace]

Is there a way to compensate v3 beta testers with a ceratain amount of free credits for assisting you? I lost a lot of credits that the system did not catch and, therefore, did not get any credit refunds. On that note as well, please could you refund lost credits that are reported under the "bug" section. I am still experiencing unrefunded credits for silent song generations. I norally report them but I am never credited. I have lost a good percentage of generations due to this issue. I love your product and realize that you are still working hard at improving it. Thank you for creating it. I am making these suggests so that customer can get what is due them, and that Suno ai can also benefit from happy customers. Blessings on all your endevours!

[obsidian leaf]

Just jumping to say +1 for API please

[wise scaffold]

Suggestion: Split style and song description boxes.

If I try to describe a style, it often ends up being something the AI decides to sing about, which is very annoying.

[tulip monolith]

Don't automatically scroll down the page when you delete a song, add option for bulk delete (maybe checkboxes)

[runic pumice]

Suggestion:
Specific parameter options such as ...

• Vocal Style
• Timing
• Tone/Theme
• Genre(s)
• Key or Scale

Suggestion:
Additional community viewing options for public songs. Let us filter and search for songs, genres, tags. This allows us to view other prompts so we can better refine what works/doesnt work. It allows us to find things to remix on better. Comments or a seperate description for public view would be great (think Character.AI). View Trending, Most Played, Most Liked, etc

[hexed spade]

The volume slider is linear, but it should be exponential.

[leaden marlin]

Desamor

[sacred patio]

Suggestion: Suno Creator Studio. Once a song has been generated, show it on a timeline like you see in Audacity or in video editing software. Editing options would include cutting, moving, and deleting segments, and also using Suno AI to generate music to fill in gaps (with the ability to guide the AI with instructions you give it via checkboxes or a textbox). Generation needs to allow for finer control, like the ability to specify a chord progression or choose the voice for the vocals from menu of voices. Also, Suno Creator Studio should have different output options, like the ability to split the vocals from the instrumentals during export (and generating sheet music from instrumentals would be huge). The main reason I want this feature: Sometimes a song generated by Suno has an amazing 10-second part in it, but the parts around it are all wrong. It would be awesome to be able to extract that golden part and then generate a song around it. (The current Continuation feature can only kind of do this, and it's not always possible to get what you want.) I'd personally be willing to pay $20/mo for Suno Creator Studio if it works and is easy to use. Lastly, I want to let you know how totally mind-blowingly awesome your product is. You should be really proud. I'm using it in an educational setting to create a musical from scratch from ideas the kids have. Suno helps use come up with the music, and then the kids listen to it and learn how to sing the vocal parts. We can create cool stuff for super cheap. This is opening up creative pathways that were never possible or at least practical and affordable before. TY! 🙂

[round zephyr]

And a bulk permanent delete in trash too!

[spiral summit]

Provide options to describe the voice, such as soprano, mezzo-soprano, contralto, tenor, baritone, and bass 😆

[tulip monolith]

Option to hide songs from create that are in playlists

[fossil plaza]

Respond to your emails its been a week since I emailed suno for help.

[shy relic]

whatd you need help on

[fossil plaza]

I want to know why they do not give the daily free credits for paying customers

[shy relic]

they give it all at once and then add 1000 on top of it

[silk cargo]

new generated videos doesn't include lyrics

that was a good feature

[pearl glen]

Please add more generations, I find myself burning through the credits to get the type of male/female voice and beat I am looking for even though I explicitly put it in the music style

[tiny aspen]

re-title songs once created, sometimes whatever is generated doesn't actually fit the title

[feral aspen]

That is already possible, click on the ... Menu

[astral gust]

Suggestion: unlimited relaxed mode (as they do in Midjourney) 🙏

[hybrid crane]

Creating audio is way heavier than making an image... GPU's are expensive ya'll... 🤷‍♀️

[dusty dove]

Hello,

I have a question about creating only instrumental music:

1. Andean flute sounds
2. Tibetan flute sounds.

Please, can anyone tell me what the correct description would be for the musical style?
Thanks.

[hoary charm]

i'm sure this suggestion comes up a lot, but i'd love to be able to make a playlist from the songs that weren't created with a license and are owned by suno, i'd even be willing to listen to ads since y'all are awesome

[formal bear]

sora?

we are not an OpenAI company 😄

also I hate ads

[hoary charm]

Sorry lmao, I’ve just watched a million YouTube videos and everything’s jumbled

Also idk what y’all studying, I’m a computer science major and suno v3 is straight up amazing

[formal bear]

physics

[hoary charm]

Shoulda known

[tulip monolith]

Don't show the red bubble/move to trash warning above the scrobbling bar

[silver turret]

Oh yeah that's ok. You can just release the model and we'll run it on our own hardware

[hybrid crane]

https://tenor.com/view/melting-green-liquid-gif-15736941

[silver turret]

Nah, we can handle it

[hybrid crane]

https://tenor.com/view/bill-hader-snl-try-not-to-laugh-can't-help-but-laugh-breaking-character-gif-9801257541091161092

[grim geyser]

special tags in the vocals prompt that tell the AI where the chorus, bridge intro and outro are.

[devout echo]

I would still like folders to enhance organisation. Thanks.

I'd love to see the chord progressions/riffs used.

I'd love to be able to download vocal and backing track as a midi file.

[grim geyser]

also I really think that artists from the 16th century should be allowed as tags, everyone knows Mozart Bach, Beethoven, and Handel are all dead.

[feral aspen]

There are tags like [chorus] and [bridge] in case you didn't know. They aren't 100% reliable but they can help guide the AI. [chorus] and [verse] are pretty solid.

[errant pewter]

Is there any way that we can edit the time of the music generated? Sometimes I only need a very short piece to see if my prompt works well, but now I have to wait for a kinda long to wait for the 2 mins version😭

[brisk loom]

Hope editing midi and dividing by layers or instruments can be given soon. Really looking forward!

[devout echo]

I don't like the way it puts the song style into the lyrics. For example a stoner rock song and in the lyrics it put something like stoners haze. A rebellious rap song it put it might put rebellious rap or rap something in the lyrics

[feral aspen]

You can switch to "custom mode" and still generate lyrics according to a topic. Here is my response explaining to someone else: #💬┃general-chat message

That will let you retry until the lyrics don't have that in it.

[shell mirage]

It would be nice to be able to pull a High-Res Image of the Clip Image from Suno directly.

Also, I know Lyrics are coming back on the videos, but a built in Lyric timing editor would be nice if needed.

[cerulean sequoia]

No full download feature?

[worn pelican]

https://app.suno.ai/song/a977353a-9430-4d51-97ae-691a0bdf7986

Yes, capybara inspired

[shell mirage]

how are they censoring? Bad words or something?

have not experienced that, interesting

[worn pelican]

Yeah I don't really get it, I mean I made a opera song using just the f word like 2 weeks ago

Although just for my private amusement (it was beautiful though)

I do hope they are liberal in what they allow moving forward. I do worry about heavy handed censorship. But yes, my first test of it not long ago worked just fine. I guess they changed something? I haven't tried it again

There are some genres where cursing is very common, like rap.

If you look at most popular music today, it features major adult themes. And yet often they aren't even age restricted by Youtube

But the line must be drawn somewhere. You can't have "no rules".

Its a liability issue as well. What if someone makes threats towards an individual in the lyrics?

If you think deeply about it, this is an incredibly difficult problem to handle

In rap, yes its common to actually make threats against others. But suno doesn't want any part of that kind of thing

You must understand why that is.

@river glacier Perhaps an ethical responsibility. But I think liability is an issue as well

Well, its actually been untested in courts

This is all being worked out now

There are questions as to if these models are even legal in the first place, since they were trained on copywritten data

So its uncharted territory, this is why many companies are playing things safe

In a heavy handed way

An LLM can detect threats. It can understand word play.

It can also pass a Bar Exam

Maybe some things will get through, but as they improve, trying to get around them will be impossible

Lol, I can't even get Claude to talk about vampirers

Vampires. And yet it has cursed in its lyrics

But again, the models will be improving. We have GPT-5 this summer

Yes, the censorship sucks. I am hoping they settle on pretty loose guidelines, with only a small amount of restrictions. There are some that I think should be blocked.

But basic cursing? No.

Not even an opera full of just the F word, I think that should be fine.

But thats just me, its not my company afterall

Yes you are right. When something that is clearly nothing major, its very annoying to get hit by a filter

But they read these chats and know that people are talking about, so I'm guessing they are also thinking about it quite a bit.

OpenAI has said that they haven't ruled out the ability for Sora to create videos...with no clothes. So we might be seeing some loosening on restrictions moving forward

The good news is that most people will never deal with filters because they are writing totally PG music.

The ones that will be upset about it will be a significantly smaller group, but still important to consider

Yes. You are right about that

I think that is why they will loosen censorship. Because they want people to be able to make music that will resonate with the masses

And its not hard to see what resonates with the masses

Just look at the Billboard as you mentioned.

I'd say down the line, there will be an open source model

You'll probably need a great GPU though, probably one that isn't consumer grade

I can tell you are for pretty much for 0 guardrails with AI in general? Just give everyone powerful open source models.

But you must understand the danger in that as well

Thats true. I just am sure that Stable Diffusion has been used for great evil already.

The people you are trying to attract, will often be the people that won't be thrilled about that

I think in general its important to be able to let people express themselves however they would like to, as long as it doesn't cross a line. Like the rap stuff we were talking about. I think that is actually considered protected speech. Even if they admit to crimes or make threats in their lyrics.

It cannot be used against them.

Not a lawyer however. But I read about that coming up in a trial not that long ago.

Eminem wouldn't be able to use Suno

I just realized we are in the suggestions chat. We are in the wrong place 😦 We should be in off topic. But I guess our conversation still falls loosely under the umbrella of "suggestions".

Anyways I am glad that you have shared your feelings about it because I do think its very important to be discussed, and hear from the users about it

I think their CEO just quit, like yesterday. Stability is in free fall

He is trying to make decentralized AI now. But if you want to chat about it just send a message. I like what you have said so far because its how I feel as well

Yeah thats true.

[real sapphire]

Please accept Paypal as a payment option

[zinc ridge]

Hello! I'm curious, if I subscribe, can I access a more extensive sound library?
I tested two songs already, and mentioned "Mozart, chorus, four-part, cello, flute, hymn, classical, strings, trombone, harp, bass drum," but noticed that cello, flute, strings, and drums are not included. Additionally, the style feels very modern rather than classical.

[fiery linden]

The hard cut off at 1 minute is really credit draining. If it cuts off at chorus, it's impossible to get it back. Always have to waste so much credits to make it somewhat work. Please make an update so it doesn't cut off any word. Just let it go for another 5 seconds.

[dusky charm]

give subscriber the choice between v3-alpha1 and the current V3 please 🙂

[long torrent]

I would love to have some sort of sorting mechanism, like folders were we can store our creations, it gets cluttered fairly easy...

[zinc crystal]

My get flagged by Moderation?

what can i do?

[dusky charm]

your lyrics got flagged?

[gusty pelican]

gotta change some words, there are certain topics GPT doesn't like
cannibalism, overly sexual stuff, overly violent stuff, etc

[zinc crystal]

yeah

[dusky charm]

Write the words acoustically

[zinc crystal]

English:

[Chorus]
Mitti's member on the top,
Emily takes it, doesn't stop,
Emily's expecting, it's no game,
Mitti's member stays the same.

The condom ripped, what a sight,
Mitti had a little fright,
The original horizontal frame,
Of Mitti, not so pretty, claims.

[Verse]
He gets up and heads straight to the shop,
Finds a perfect one, doesn't stop,
Buys a giant one, quite a toy,
Emily loves it, oh what joy!

Mitti holds her tight, no regret,
Emily's test, she must get,
And between his legs, what a nest,
Emily's parents, they detest,
After that, he needs a medical rest.

[Chorus]
Mitti's member on the top,
Emily takes it, doesn't stop,
Emily's expecting, it's no game,
Mitti's member, he's not to blame.

Mitti's member, it was too small,
Fitted thrice, not fitting at all,
Emily wanted him to fall,
But there's still room, it's not a brawl,
So he uses the plug, as a call.

[Verse 2]
She doesn't like him, because he's large,
Then he finishes, it's a barrage.

Mitti doesn't fit Emily's type,
His passion's moonwalk, it's his hype.

[dusky charm]

"I'm going to kick you dcwnstairs" (flagged by GPT) ->"I'm going to kik you downstairs"

[zinc crystal]

what is false

[dusky charm]

up to you to find it out

[zinc crystal]

ouhhh

[dusky charm]

(sorry for the post I just removed. I was in the wrong channel)

[gusty pelican]

GPT also takes context into account... there's a lot of this that could cause it to get flagged if there's enough there to imply its about sex

"condom", "between his legs" are likely the big contributors

[zinc crystal]

ok

and this? :

[Chorus]
Mitti's member on top,
Emily takes it, doesn't stop,
Emily will be pregnant,
Mitti's member remains always small.

The condom is also torn,
Because Mitti messed it up,
The original horizontal body structure
Of Mitti doesn't look so pretty to others.

[Verse]
He gets up, and goes straight to Emily's house,
He goes in perfectly and buys a huge member,
He puts something in her behind,
She licks Mitti's little eggs.

Mitti holds her tightly,
Emily needs a pregnancy test,
And Mitti has between his legs, a boar's nest.

[Chorus]
Mitti's member on top,
Emily takes it, doesn't stop,
Emily will be pregnant,
Mitti thinks that's very fine.

Mitti's member was too small,
It fit three times.
Emily didn't want to be with him,
There's still enough space,
That's why he uses the anal plug as an addition.

[Verse 2]
She doesn't like Mitti anymore, because of weight,
Then he doesn't take a step because his leg breaks,
And Emily laughs in his face.

Mitti doesn't match Emily's preferences,
Because Mitti's passion is moonwalk dancing.

[hybrid crane]

Just go to chatgpt and post the lyrics there with the question, why is this text flagged for moderation...

[zinc crystal]

ok

[hybrid crane]

@zinc crystal In your case you would get this:

[zinc crystal]

what can do?

can i do?

[hybrid crane]

Either change your lyrics, or write certain words as they sound...

[zinc crystal]

k

[vivid magnet]

Regarding @feral aspen 's message in the bugs section about bugs not being the same with model feedback, I agree but suggest having a different channel to track model feedback. The poll can't capture all kinds of issues and not all nuances of each issue. Personally, I refrained from voting (eg, V3 is better in some areas, worse than alpha in others, so is it better or worse overall? I don't know). So, to summarize, could there be a dedicated channel for this?

[hybrid crane]

I voted on all.. that is a possiblity too 😉

[carmine quest]

Suggestion : add a function to remix the tune with different lyrics, in addition to the current remixing which uses the same lyrics with a different tune.

V3 is good quality music, but it still lacks more ways to give information on what we want exactly, and it also has a tendency to not follow closely instructions, and hallucinate.

[south axle]

Please bring Grunge back into V3. V3 alpha had it locked in. Now im getting all kinds of weird harmonies on every vocal.

[formal bear]

Hi, I know there are a lot of you folks asking for V3 alpha access. I will write here to clarify on this issue.

We appreciate the feedback and we understand that V3.0 is not perfect. We are working on upgrading V3.0 to a future V3.x version. Meanwhile, it doesn't help us much with brining back V3.0 alpha -- that's actually in some sense a step backwards. So rolling back V3 alpha access is not possible at this moment.

If you are experiencing difficulties with V3.0, help us improving it by sending in feedbacks! Dislike the clips that you think have issues, and like the clips you think that are meeting your expectations. 🙏 Thank you!

[carmine quest]

Suggestion : instrumental switch should not remove lyrics field, as it could be used for giving instructions on the tune composition and articulation.

[hybrid crane]

Thank you Tony. Can you please pin this too?

[indigo sage]

I second what Demethel suggested about additional Remix option.
I had quite a lot of times when I was experimenting with semi-completed text or even with a completed one, finding some good sounding tune I'd like to use, but later finding that the text has some issues or even if I want to change something.
But by then, it's almost impossible to get the same tune later when the text is updated, so an option to use the same tune, sounds, etc. while changing to a different text would be great.

I makes sense that it might be weird if the text changes drastically, but if you only fix some mistakes or the changes are minimal, it would be a great option.

[loud canyon]

If I could make a brand new song but point to a song that I previously made to match style and vocals, that would be great. Right now, I've gotten pretty good and consistently reproducing my intended sound but its a trial and error where I generate ~10 clips and then pick the one that sounds close

[pseudo crane]

Give me an option to not generate art with each song to limit the power used by the AI so it can focus on making songs. I love to see most of the artwork but I dont use it

Oh and an option like "Start a new song from XX:XX" instead of "Continue song from XX:XX" would also be cool

[upbeat rock]

option to edit the track art?

[upbeat rock]

yes this. Remix should keep the song mostly the same but let u edit the lyrics and maybe some sounds. Remix is more of reroll/retry and those should be different

[frank anvil]

True! It helps to structurize the instrumental - perhaps for recording own vocals in another app!

[upbeat rock]

also cant wait till we can "in-paint" certain parts of the waveform

[carmine quest]

^Yes!

[tight beacon]

A "looping" feature would be great for making tracks that loop so they can be used in games.

[covert jolt]

I would challenge this. It is unfair to whoever played with V3 Alpha and kept the sub expecting the same level of quality moving forward. In my case I upgraded to Premier so I bought the credits and am now in a position where they might expire worthless. I can't wait for you guys to update the model to V3.x and even then I am not sure it will fit my use-case because from what I see, it depends on song genre and other variables.

I just question the sudden pull-out of V3 Alpha. You could have run the two in parallel and get feedback first or charge us more if we kept using V3 Alpha. That would have worked for me but the sudden change is something I wasn't prepared for and neither do I have an option to pause this subscription until it gets sorted

[upbeat rock]

seems like generating a full song is within scope so why does it make us do so in three steps and not one just to share something complete?

[pseudo crane]

Same here, I feel kinda cheated why tf is it worse now than 1 week ago

[pseudo crane]

Whats the timeline?

The sounds now are far from as crisp and clean as they were before

@formal bearIf we wont get the alpha back can we get refunds? I started using a product and paid more for it and then you just remove it, that is so f'ed up that I dont even know how you can run a business like that

[mint void]

Have to agree, I was using the alpha for a while and got such good results with it and now with the V3 model I feel really cheated as the quality dropped by a lot.

[pseudo crane]

100% same here, v3 is a noise maker and alpha was a music maker

[pseudo crane]

@echo canopyCan anyone in the team adress this?

[delicate cedar]

I encourage anyone seeking a refund to email [e-mail]
(Personally I think v3 sounds great!)

https://app.suno.ai/song/b34d54ae-640b-4205-af3d-cee21c9f7ea8

Literally just made this, this morning.

[pseudo crane]

Thanks and v3 is shit compared to how amazing alpha was, and I paid for alpha only to have you remove it. This is no way to treat your customer base fr.

[warm zephyr]

I'd have to agree, alpha v3 was better, v3 keeps adding background singers making it sound even more robotic

the quality also doesn't seem better

[crisp furnace]

All my over 1,000 credits that I purchased are gone becuase it's the next billing cycle????!!!! This is not okay, I feel cheated off of my purchase. I was not planning to buy any more credits until I was fully done with what I had previously purchased. Please, how can I get them added back into my account? Please do not take away remaining credits that were rightfully purchased, it's not good business and as a customer I feel completely cheated.

[hybrid crane]

It's says so on the website @crisp furnace

Suno AI needs to predict and manage its computational resources effectively. Monthly credits are a way to forecast usage. If users were to accumulate credits, it would lead to unpredictable demands on their system, complicating resource allocation and potentially affecting service quality.

[wooden fable]

hmm, i'm actually getting slightly better results from v3 than v3alpha so far (only just started testing it now because this month's subscription just started, giving me credits again)

@crisp furnace that is why i generally stay far away from subscription stuff. suno is the only exception for me atm. (i got tempted)

[crisp furnace]

This needs to change because you are going to loss customer once buyer find out that they loss all their remaining purchased credits on the next billing cycle? Why not allocate a section folders for the previous month remaining credits? Yes, it will take effort on your part but that's the cost of having an ethical business that does not take away from customers what is rightfully theirs. I was looking into purchasing the premium plan later, which offers 10,000 credits, but at this pont, I will not be doing that as it's just going to be a loss on my pocket. Please do not settle for these terms in your business, it's not worth it. With ai businesses pop-up everyday, this may be the area where your next competitor takes your customers away from your platfrom.

[pseudo crane]

Thats so shitty wow

[hybrid crane]

Please remember to be respectful, as our community guidelines clearly state. You can review these rules here #📜┃rules message. I'm taking a moment to clarify a specific issue which, as you might have noticed, is effectively managed in similar communities like Mid journey and Suno. This isn't a hidden practice, but a standard approach. Due to this situation, a 5-minute timeout will be implemented. Thank you for understanding and helping us maintain a positive community space.

[pseudo crane]

Jeez christ... Okay its not shitty its horribly bad and I dont care if "others do it too". Thanks for showing your level of customer service.

[hybrid crane]

Any more concers can be send to [e-mail]

[pseudo crane]

Cool so now I have 10k credits that I wont use because you removed the functioning version and the credits wont even be saved for next moth.

[hybrid crane]

I've been clear

[fair turret]

I love this having a lot of fun. That being said I do have a few things that would be helpful if implemented. 1. Much Longer Prompts, 2. Much Longer Lyric Area. No room for multiple verses bridge outros intros etc. 3. Persistent Sound and Vocalist. 3. Ability to target certain areas of the song for change and save what you like.

[pseudo crane]

I loved alpha, this is so sad.

[fair turret]

I kinda figured they wouldn't roll over. It would be nice if they did but I understand their point on forecasting. Maybe an option to pay a bit extra for rollover credits with a max accumulation?

may be a way of compromising

[hybrid crane]

You can buy top up credits who stay perpetual

[fair turret]

Ok then they kinda already did what I am suggesting if there are perpetual credits. I Think they should put a warning in so user know when they can expect to lose them. I know they already know from their plan but a tiny warning that is seen may help so you don't get surprised. Maybe they do this as well I am very new

[crisp furnace]

No warning was sent via email or even on the website, all my roll-over purchased credits just disappeared.

[pseudo crane]

Tbh I wouldnt mind it if as long as they kept the functioning version. Now they remove the whole product I was using so I wont be able to use the credits, makes no sense to me as a customer. Seems like a fast way to kill a businesses reputation and client base 🤷‍♂️

Fingers crossed for that refund, which from the responses here I doubt is likely to happen

[fair turret]

Im confused I thought if you bought more they were extra credits and perpetual based on what @hybrid crane posted? Like I said I am new so I may have the wrong end of it

[hybrid crane]

Thank you all for sharing your thoughts and concerns. It's important to keep our discussions focused and constructive. For any specific issues regarding billing, credits, or account-related queries, please direct them to [e-mail]. This helps us ensure that your concerns are addressed appropriately and efficiently by our (Suno) support team. Let's keep our community space focused on relevant topics. We appreciate your understanding and cooperation in maintaining a respectful and organized environment. 🌼

[hybrid crane]

Extra credits are perpetual not your monthly credits. For more info see https://app.suno.ai/account/ or mail [e-mail]

[fair turret]

Yeah that is what I thought I did read it right 😛

[crisp furnace]

@hybrid crane I no longer have that option on my account as a previous customer, not even the extra credits that free users get. The only option a past custoner has is to buy another plan under the same account. That basically means the custoners hand is been forcing to make another purchase, and if not, they cannot continue improving on their last music generations. So basically the customer has to endure lost roll-over music purchsed credits, and the inability to improve on purchased songs, if they wish to discontinue. Please, I implore you, do not settle for this kind of business, even if midjourney is doing some thing similar. Please seek higher ethical business terms, and set yourselves apart, it will pay of later, I assure you.

[hybrid crane]

I point back to: #💡┃suggestions-web message
I have no say in this, I merely help guide to the knowledge that is open to all. I just happen to have a good memory and been asked by Suno to help, I am a Volunteer Guide.

[crisp furnace]

@hybrid crane I understand. Please pass our concerns to the higher-ups, because this is quite disheartening. I was routing for Suno.ai, but the fact that I have lost roll-over credits and I am also been forced to make another subscription to continue improving on my previous generation (as I have no free credits rolling in like free users), is just too much. This a two time loss for any customer who wishes to no longer purchase a subcription. I am basically been forced, or rather, manipulated, into remaining a subscription buyer. This is no way to reward the very customers that are helping suni.ai grow.

[hybrid crane]

I am sure if you mail them they will be empathetic. Suno is good people.. but you have to understand there is a reason why it works as this.. Again see: #💡┃suggestions-web message

Just mail them.. They will make it up to you some how @crisp furnace

[kindred carbon]

Damn, I need a volunteer guide this dedicated for my biz too!! I think V3.x will address some of this and I can't see SUNO being predatory, so far it seems like good people who listen to feedback. Seeing this many suggestions daily though - I don't know how you guys juggle it all.

[warped ember]

It could be good to import a desired melody in the form of a mid file and be able to generate music from that

[crisp furnace]

Will do, thanks. We appreciate your service to the community as well. Many blessing.

[tepid trench]

The subscription page used to be less clear about the subscription credits not rolling over. They did at least change this around a month ago following feedback (to what it is now.)

[raw cave]

https://app.suno.ai/song/61609e2b-d937-409e-afd3-68ab4910cde8

[spiral acorn]

Oh yea, I just remembered a small UI suggestion I had. When you're on a Playlist it'd be nice if it showed the Playlist in the browser page title:
"My Playlist | Suno"

It already does this kind of thing when you're on a song page:
"My Song | Suno"

Just makes things easier when you have a few tabs open. I'm sure this is the intended behavior anyway, but I'm just mentioning it in the case it wasn't fixed later 😆

Edit: Also probably makes sense to do this for the Create / Library / Explore pages. Again, I'm sure this is something that was almost definitely going to be done later, but just mentioning it regardless

[tepid trench]

Also, regarding v3 and v3 alpha: I have also noticed v3 seems to be more "chorus"y but am personally not that bothered by it. However, I would agree that having access to the old model would be nice if possible. Chances are that whenever you take something away, a few people will be unhappy 😉

[pseudo crane]

Suggesstion for future releases and versions, maybe have a focus group test it out before swapping versions out? This alpha to v3 seems to have been not so well received and maybe it can be prevented from happening again with a group of beta testers?

[thorn comet]

We need ultra clean Stem seperation & Stem single and combined download

[atomic bloom]

First, this is not a promotion in any way. I am sharing the link to my original album in an attempt for Suno to underestand this odd kind of music. Much of this album uses Harmonic Minor. Suno doesn't seem to understand Harmonic scales. Not really. So... I am sharing this album so Suno team can download these tracks as training. You have my permission. All should be downloadable and creative commons which means for training you the Suno team have free reigns. I have tried to get Suno to generate something similar to these tracks but no dice lol. Maybe not much training? Anyways, download away Suno team. Throw these tracks into your training models.

https://soundcloud.com/allen-polley-official/sets/a-fools-prophecy

[potent crane]

I am a heavy user of both V3 and alpha, it's not only less than perfect, it is demonstratably worse than alpha - I have been on suno for many hours every day for months and today is the first day in all that time that I am downing tools and walking away, regretting my premiere sub - my creation has ground to a halt for two main reasons - lyrics all over the place, missing lines, duplicated lines, missing words, and secondly, no longer do we get solo singers (regardless of prompts) we get a chorus of singers, sometimes sounds like a whole choir.

[raw cave]

v3.x

[red ember]

Search feature in the Library

[formal bear]

If you like v3 alpha and don't like v3 as much as you want to have a refund, just email [e-mail] and we will refund you. Thanks.

[pseudo crane]

Thank you Tony I did send an email about that, I will await reply and looking forward to your next versions

[grim geyser]

https://www.youtube.com/watch?v=SDJwg1JoPtY

[tulip monolith]

Better ordering for the trash can

[pseudo crane]

@formal bearand sorry if this has been asked a million times already but is there really no chance to keep alpha open? It seems a lot of people really enjoyed it as much as I did and found it very useful

I wouldnt mind if it was slower and took a minute or two or more to generate songs

[formal bear]

what if it takes 4x the cost

[pseudo crane]

That would be too steep for me now but in the future I could probably cough that up. Is it using too much power? Could it be limited and all work put in a queue? But I guess you already looked at all the angles, I just loved alpha but I understand if its not feasable for you to host it

[formal bear]

I hear your voice, and all of your complains 🙂

again I appreciate the feedback -- that's what we need to focus on and improve

but I don't see any world that we launch and immediately roll back in any sort. Just wait for the update. It's not like v3.0 is going to be the product for the rest of the year.

And who knows, if there is any inference tricks that we find can help improve, we will roll them out. But all of these take time to be figured out.

[pseudo crane]

Theres no shame in going backwards sometimes to regroup. 🙂
Cool cool, will eagerly be looking for next versions. For my part, scrapping the art generating wouldnt make it less useful, if that part is draining a lot of power maybe give a button to opt out of it or something, if that could help lessen the power usage

[formal bear]

well if the poll has inverted result we would have reverted. but see #🗳┃mini-polls message.

I know y'all hate it but there are also folks liking it more

[pseudo crane]

Oh I see, I wonder how and what genres they use it for

[delicate cedar]

Genre matters more than I though, I have to admit.

Of course, I've been through these transitions twice now. So I understand that there's an adjustment period after every release to relearn how to do things.

[earnest leaf]

I noticed that Hardcore Punk took a hit with this update, it sounds more like power metal now. Full chorus of singers instead of just one solo vocalist.

[pseudo crane]

Adjustment like how? Did you have to change all your style inputs at new versions? Since I do instrumental I dont think I can adjust so much but that might be a skill issue on my side

[upbeat rock]

wait what am I missing. V3 is amazing to me

[pseudo crane]

Which genres do you create?

[upbeat rock]

ive tried everything from funk, dark ambient, pop punk, hip hop, house

i just found the tool yesterday and dont know about alpha

[pseudo crane]

Oh ok hehe

[upbeat rock]

so i dont know what im missing maybe

[pseudo crane]

Yeah for sure

[upbeat rock]

im like blown away by v3 and ur telling me it gets better?

[umbral scaffold]

alpha imo was much better

[pseudo crane]

I mean yeah they are developing it constantly, just this change wasnt a good fit for the genres I make

[umbral scaffold]

I can totally see someone first experiencing v3 being blown away cuz I was blown away with alpha

[pseudo crane]

Yeah

[upbeat rock]

which are what - just so I know

[spiral acorn]

Would it be possible to have a download ZIP option for playlists? Even if it was limited only to the owner of the playlist? It's probably an annoying feature to add lol, but it's something on occasions I would find useful.

Once I tidy up my playlists I'd use it to make sure I always have the latest version of the tracks. Though I totally get if a feature like this isn't added.

[pseudo crane]

If I dont misunderstand the question: synthwave, retrowave, etc

[upbeat rock]

copy. Ive done synth funk well, was gunna try vaporwave next

[pseudo crane]

Enjoy it man dont let my experience make your good experience any less 🙂

[upbeat rock]

REEEEEE

[delicate cedar]

Absolutely.

And I completely lost several styles from the v1 to v2 transition.

Plus I had to learn how to lean into the clip context window and fill it up so it didn't hallucinate. Before I could basically compose stuff live. Do some instrumentation, then a verse. Then do another verse or add some instrumentation.

It was a change that I had to get used to.

[frigid timber]

Don't have to roll back, just allow us to model select V3A, the declining results are heavily genre based. So if people are sticking to edm, or light hearted genres they would never see this or at least not very often. Or if possible only allow users who had V3 Alpha access prior to be able to model switch to it. Don't see how it is a step back for Suno, as you keep ALL your customers happy and dont have to deal with the influx of refund requests

[umbral blaze]

I imagine it's not free to hold another instance operational

[pseudo crane]

Just hack Amazon and host it there, ez

[umbral scaffold]

@formal bear said "what if it takes 4x the cost" so I'm assuming v3 is cheaper on their end than alpha. Tbh I WOULD be ok paying 4x credits for alpha generation. It's a business decision on their part I think. Maybe they could release Alpha as V3 Pro a new model that costs more credits)) idn

[frigid timber]

may not be but then youre going to have customers not using the service until it is fixed

[formal bear]

that's just metaphore -- plz don't take the numbers seriously

[potent crane]

You mean they like disordered and missing lyrics, choral singing when prompted for a solo voice, and blues that can't count to twelve bars? I would love to hear their music.

[pulsar hornet]

Chirp/

[pseudo crane]

Lets have patience and let the team work on the next version that hopefully will blow us out the water with how good it is

[formal bear]

thank you -- the more time I spend arguing on this the less time I have coding so...

[umbral scaffold]

Yeah I'm hopeful for the project. I'm just super glad I didn't get an annual sub))

[pseudo crane]

Understandable, sorry for all the excessive complaining today. I have eaten food now and am much more sensible lol.

[potent crane]

...and in the meantime I have 3000 credits that I need to burn up by 31st March because they don't roll over, so I waste them rather than lose them. Sophie's Choice, methinks.

[formal bear]

lol

[hybrid crane]

https://tenor.com/view/im-hangry-hangry-grumpy-grumpy-cat-angry-cat-gif-14749953160056315192

[umbral scaffold]

also idn if this has been aswered

[pseudo crane]

https://tenor.com/view/commercial-snickers-cranky-gif-5478685

[umbral scaffold]

is there a reason v2 is still up?

do a lot of people use v2?

[frigid timber]

Ive been trying to scroll through the responses my bad if its sounding like a broken record at this point, just never been in this situation with suno before so im eager to keep creating

[formal bear]

again we take ur complains & opinions seriously, but let's try to put it in a way as:

• I wish v3 get improved in vocal quality / sound quality like v3 alpha ✅
• I wish ABC gets improved
  not
• I just wish a rollback / another option. ❌
• I want the old days back.

Let' all eye on the future.

[lofty sonnet]

I think I am ready for the option to set length and primer song structure style, you could default first gen to 2 mins and continuations to 1 but the advanced option to modify gen length based on cost, so give even the potential to be able to select a song length end window of like 5-6.5 minutes progressive house track if wanted... select output type (rough genre based formatting), sound fx, podcast, etc and total generation length suggestion 1:30-4:30 for pop music or something, and this generation length as defined has the cost of the generation in credits sliding when adjusted. Could be helpful and make you money too if people generate more expensive tracks and then have to go in and inpaint certain portions of their masterpieces

[umbral scaffold]

uhh but you kept v2 when you released alpha and also v3

[pseudo crane]

But v2 is for free to use only I think

[kind vale]

Support sent a reply and suggested to use V2, if I don't like V3. But it also said they are working on improvements. Pretty generic, but as they seem not to have AI answering their mails, I'd rather have every human working on a fix than writing to hundreds of customers 😉

[umbral scaffold]

well anyway

hopefully V3 improves and is as good or better than alpha

[kind vale]

We're not complaining, we're trying to emulate the chorus effect of V3 🤪

True. But the previous flaws didn't keep me off making music. But as it's been pointed out: we just have to be a little patient. The future will be brighter and eventually we'll have Sora to create videos to our songs and unlimited creative freedom.

[lofty sonnet]

@formal bear and any devs listening. Could generation length "window" option be something Premier users have access too? I don't know what max limit would be enough.. 7 minutes.. 30.. 2 hours.. all night? I have some great prompts, and I would love to give SUNO the ability to craft a much longer format track/album/set/ "complete full song" in one go for a sliding cost of generation, based on these great prompts. We could always go back and continue from points if need be, but the off chance you are lucky and the whole track is just genius?! Incredible potential. Could make use of some drop down window of rough style "formats" (song, album, live set, EP, etc.) Thank you so much for all of your hard work and efforts in making this revolutionary tool of sonic creation. People would BURN through generations, you will be rich (and continually improve) and we will be very creative!

[formal bear]

I have seen bugs from users try to concat > 1hr long audios, and yes that bug is fixed so you should be able to generate 1hr long audios if you concat.

[lofty sonnet]

concat?

[formal bear]

get whole song

If it is per generation length, you are looking at the limit 2 mins right now. Any control we'd give you would be shorter than that, and longer time windows are still R&D. (And trust me they don't sound that great yet).

[shell mirage]

Are there plans to implement a Mastering tool within Suno?

[formal bear]

yes tbd when

[shell mirage]

AI ?

[formal bear]

we don't want to give u sth not good

[shell mirage]

I assume it will be plug and play, automatic AI?

[lofty sonnet]

OH I see. Thank you. Looking forward to the advancement and R&D of longer generations not concat, stuff like a perfectly structured house track, or even the simulation of a longer set of different house tracks of specified style, etc.

[formal bear]

no I will literally go in and edit every bit of the sound file you choose. 😄

ofc it will be AI...

[shell mirage]

Bruh, nah, I just didn't know how involved it would be on the users part.

But glad to hear it lol

[spiral acorn]

when you add the "paint in" features, is there any possibility of deleting selections of music, and moving selections of music? Would it be something like you delete a section and then you could "paint over" the connection area and smooth that out? Is that kind of thing possible?

[tulip monolith]

Grouping in trashcan (collapsible by continuation chain)

(Sorry, I am assuming I can just drop notes here?)

[formal bear]

yes

[lofty sonnet]

how far from first tests of inpainting are we?

[shell mirage]

@formal bear you're the GOAT. Higher resolution clip images coming at some point, with availability to download?

[lofty sonnet]

Would we be able to stitch two different tracks together selecting their "crossfade" or ai gen inpainted section?

[formal bear]

no I am not plz... many more others on the team are actually hard working now rather than chatting 😄

[lofty sonnet]

this bit is important to guide their work 🙂

[formal bear]

yeah images, should be easy, we are just a bit worried about latency, but should be very very easy...

[shell mirage]

Dope

[spiral acorn]

strange request, but would it be possible to download all audio tracks in a playlist as a ZIP file? it's a niche request, so I can get if it doesnt make sense
edit: I don't think anybody else wants this, so it probably doesn't matter lol

[lofty sonnet]

this would be huge when we can inpaint two tracks next to each other in a playlist and get whole playlist

THAT is the old school definition of mastering

[spiral acorn]

lol, turn the playlist into a concept album ??

edit: ah yea, i see what you're saying by connecting everything into a single track so they can master uniformly

[lofty sonnet]

live dance music sets... albums, continuous mixes, connected ambient tracks, idk, its a dream

[spiral acorn]

i see lol, well id be happy just to be able to download the the audio tracks of the playlist in a ZIP. I was mainly thinking so I can easily download the latest versions of all my tracks

edit: again though, this may be a feature that nobody else wants

[pseudo crane]

Btw where can I follow Suno to get notifications when new versions are released and such?

[tidal wharf]

simple button to rerender v2 with v3?

[tidal wharf]

#📢┃announcements

[shell mirage]

lol

[lofty sonnet]

I bring up this inpainting/transitioning/gluing two different tracks in a playlist together over time because I've had some awesome bridges and breakdowns and turnarounds generation and in could be cool to stack on top and regen or give space between and regen over time to transition.

[tidal wharf]

I imagine this would be an exorbitant load on the server. Syncing with google drive though? Now we'd be talking...

[spiral acorn]

i did have a feeling that could be the case 😆
yeah, the more I think about it, it's probably not a great idea especially on really large playlists

[tidal wharf]

me wondering if this button means "show my deleted tracks" or "delete my entire library" 🤔

intrusive thoughts were high that day

[spiral acorn]

surely it must be show deleted tracks (and thanks now I know where this is)

[shell mirage]

That's the Self Destruct button

[tidal wharf]

let me know once you check 😄

[spiral acorn]

i will click it... let's see

[tidal wharf]

BOOM

me clicking through to the 82nd page of liked chirps because there's no search field (sad face)

[spiral acorn]

haha it did go to the Trash page. I was going to make a joke about what happened but I couldnt come up with anything 🤣

[tidal wharf]

me trying to guess which of these (n)'s refer to which generation

[tidal wharf]

should've asked chatgpt! but thanks!

[tidal wharf]

use case: I'm trying to convert all my v2's into v3's, but as my finished tracks are heavily cut-and-glued, I need to find the original chirp for every wav

[tidal wharf]

@formal bear it would be super-helpful if, when downloading a track, it added the generation uuid to the filename when downloading audio. (It did in the very beginning when track names weren't a thing, but maybe we can have the best of both worlds?)

[pseudo crane]

My 5 cents is let other AIs do images and graphic art and make this one all about the music and focus all money and energy on that 🙂

[red tree]

how many upset people will it take to get v3 alpha back. can yall just add it to the list of versions we can pick from, call it 2.5 maybe. i used alpha i know it works. as cartman once said "whats the big fing deal b**h"

[pseudo crane]

Well it probably uses a lot of power($$$)

[cinder willow]

I had a very nice outro to my second prompt that I would have liked to use at the end of the song I just created, but I can only choose where to start from. Maybe we could get a cut'n'paste feature? Below Continue from this song, perhaps we could have a clip part of song feature and some way to paste it back in when it would fit the overall song better?
I got inspired to write an extra verse and fix a bit of wording in the chorus and that meant the final song would miss the nice outro I got initially.

I realize Suno.ai may not have in mind to have a social component on the website, but I'm a little sad that I can't follow the creators of the songs I like on there, or even really see anything about them.

[hybrid crane]

@cinder willow The feature you're referring to, known as "inpainting," is in development for Suno and should allow users to selectively regenerate specific parts of a song. Additionally, the option to follow other creators on Suno is being planned and should be available soon.

[grim cedar]

A good point brought up in #💬┃general-chat can it be optional to allow people to 'continue' your songs, I see it only causing problems down the line and people are already worrying about it. Seems like it would be easier just to stop the problem before it truly becomes one

[cinder willow]

A very minor nitpick: When I glued my song together, it kept the Chorus that I replaced in the Lyrics section and there seems to be no way of editing the lyrics. It's not really a big deal, but I like to follow along the lyrics and if I hadn't created the song myself, I would get confused.

[hybrid crane]

The concern about the 'Continue from this song' feature possibly causing issues in Suno has been raised by other users as well. The developers are aware and discussing how to address this best for us users.

[raw cave]

Two words: Neon Logo

[hybrid crane]

In-painting is again the solution.. So stay tuned 🌼

[grim cedar]

is there a road map of confirmed future updates and maybe some in discussion?

[red tree]

the suno logo looks like this, oddly enough its another ai music app www.staccato.ai

[cloud tangle]

MIDI export option for paying users?

[fervent spade]

I don't know if it's been recommended before but some more control on what the ai generates would be useful with sliders, and genre switch ups in the same song would be nice. Also if possible to have a whole outline made to make a complete song would be nice ngl

[sick marsh]

Hey, @grim cedar was referring to me. I know they know. I was just curious about it as well. Wanted to hear some more thoughts on it.

Thanks for looking out for everyone.

[zenith gorge]

Genres -> Showcase as thats mainly what its been used for recently anywho?

[torn barn]

hey will there be paypal, amazon pay or google pay available in the future? I would like to get the sub-version

[vast sun]

search in website ;c

[torn barn]

what do you want me to search there?

[formal bear]

can we frame this 😄

[wooden igloo]

It would be nice to have a "session based" create system, instead of all the library appearing in a list.
I'd be able to work on multiple songs...

And loosing custom mode settings after a refresh is really easy to solve.

[vast sun]

Site Suggestions:

1. Global search bar for a track by name and similar title.
2. Pages with music for training and new pages.
3. Track name scrolling if it does not fit in the bottom panel.
   4) Displaying your Liked tracks from Trending - New pages in the library.

[gentle cypress]

Having a negative music style prompt input box would be nice. Terms/styles to try and explicitly exclude from a generation

[tulip monolith]

Labeling multiple generations of the same continuation as
"Part 2-1"
"Part 2-2"
...
"Part 2-8" etc.

would be useful.

I'm getting lost sometimes on which generation I prefer to go with.

Similarly, when "getting full song" and continuing generation from that, if the previous version could have a different label. Maybe something distinct from Full Song, or simply "Full Song-1".

[simple cave]

Oh man inpainting would be amazing. I would also like a "save this singer/ band" for when you get vocals that really rock.

[hybrid crane]

https://tenor.com/view/man-cant-believe-shookt-surprised-mind-blown-gif-8714188

[simple cave]

Haha. Yeah I am totally going to do my best to find the pennies to sub fort my birthday next month. I have sensory disorders that pertain to sound, so finding new music is a big risk. Having the ability to input what I want to minimize that risk is mind blowing

[thorn thorn]

remove the silly moderation like seriously it's fn music have yall heard mainstream music lately? its getting absurd and unusable.

[sudden valve]

I would really appreciate more precision for the "continue from" timestamp. Like 0:30.5 or 0:30:50 to generate from 30 and a half seconds.

[median kernel]

That would be a good idea but I don't know if algorithms could handle that.
The following would also help:

Basically, being able to select what notes are used and what's the root note.

[smoky galleon]

Alright, suggestion. you guys need a much larger context window for the song generation.

Here is the reason why. I am trying to make memes using this service. Think "someone did a politics debate with a live streamer. Make a music video based on the debate. Here is the full transcript of the debate".

And right now the context window is like 199 CHARACTERS!

I don't need to input novels info this. But please, give me a larger than 1 like box, and allow more that a couple sentences.

This is for the lyrics generation mainly. I don't think a larger context window would matter much for the song part.

Or, another funny example. Someone writes a reddit comment, and I want to turn that reddit comment into a music video.

Stuff like that. But 199 characters is way too small.

Or, maybe I missed a feature. The custom tab doesn't allow you to use AI to generate the lyrics in that larger lyrics box right?

Instead thats just manually putting in lyrics?

I could generate/copy lyrics with my own prompts I guess, and just put them in there. Not sure how good their lyrics model is.

[feral aspen]

It does let you generate lyrics. Type a topic in the lyrics field and hit "Make Random Lyrics" and it fills in the Custom mode Lyrics field.

[smoky galleon]

Ahh, interesting. Will have to try that out.

I wasn't sure if there is a context window limit to that also.

[lilac jackal]

I'd love to be able to put lyrics/melodies over a generated instrumental.

eg. Generate instrumentals, then to get melody ideas, I can continue from a part in the song with lyrics to get melody ideas.

[worn briar]

Dear Suno Team,

I hope this message finds you well. First and foremost, I would like to extend my sincerest gratitude for the incredible work you have done with Suno. As an active daily user and a big fan, I can confidently say that your product has significantly enriched my life, bringing endless joy and inspiration through music. Your dedication to creating such a fantastic tool does not go unnoticed, and I am genuinely grateful for the opportunity to experience the magic of Suno every day.

Having immersed myself in the vast creative possibilities offered by Suno, I would like to propose a few suggestions that I believe could enhance the user experience even further. My aim is to contribute positively to the community and ensure that everyone can make the most out of this wonderful platform.

1. **Guideline Creation with Writing Templates and Tags: **One of the features that I find exceptionally valuable in creative tools is the availability of guidelines, writing templates, and an exhaustive list of tags. These resources can significantly streamline the creative process, helping users to generate the best possible output music. It would be immensely helpful if you could create a comprehensive guideline for Suno that includes writing templates and all possible tags. This addition could serve as a beacon for new users and a helpful resource for seasoned creators, ultimately fostering a more inclusive and productive creative environment.

2. **Voice Customization in Music Generation: **An exciting frontier in music generation is the ability to choose a consistent voice for the music or, even more intriguing, the option to train the model to use one's own voice for music generation. This feature would not only personalize the experience but also open new avenues for creators to experiment and express themselves uniquely. Could you share any insights into when we might expect to see such a feature? I am eager to learn more about the possibility of customizing the voice in music generation and how this could be implemented in the future.

3. Launching an API Service for Third-Party Development: Finally, and perhaps most significantly, I strongly encourage the consideration of providing an API service. This would enable developers to build applications on top of Suno, potentially ushering in a wave of innovation and creativity. Offering an API could attract a diverse range of new customers and investors by enabling the integration of Suno's capabilities into a variety of applications and services. The potential for collaboration and expansion is vast, and I believe this move could significantly contribute to Suno's growth and success.

Your team's passion for innovation and excellence shines through in Suno, and I am thrilled about the potential these suggestions have to further elevate the user experience. I understand that implementing new features involves careful consideration and significant effort, so I am curious about any developments in these areas and how the community can support your endeavors.

Thank you once again for your dedication to creating a platform that has truly revolutionized the way we interact with music. Please know that your efforts are greatly appreciated by myself and the wider Suno community. I look forward to any future updates and am excited to continue supporting Suno as it evolves.

Warmest regards,
ILIA

[dark notch]

I love Suno, keep on guys! My two suggestions would be: 1. Search function. It´s a pain to go through the entire Library if you are looking for something. 2. Enhance the quality of the audio. Often there is always some background noise and crackle.

[hazy grove]

Really need suggestion 2

[solid sage]

Try mastering the audio. It removes most of it for me

[rocky hornet]

Not just a search function but need the ability to group things, either into folder, by tag, by song title (when your spamming creations) just a whole bunch of UI stuff needs to happen to make it a comfortable UX imo

[peak pasture]

Please let us choose the singers ourselves in the future. (by Land and Gender) ❤️

[pseudo crane]

You can do that already

[peak pasture]

How?

[pseudo crane]

In style write: Female vocals, etc. Brazilian female vocals, not sure about country but gender 100% can be chosen

Sorta like this, but I duno if country is applicable:

[peak pasture]

No, i know that, but i mean by same voice for all.

[pseudo crane]

Ooh sorry then I misunderstood

[peak pasture]

np 😉

So i found i good voice, and i want this use for more Songs,

Or ist this all generated only?, so we cant use the same voice for all?

[feral aspen]

Using countries like "Swedish" often gives you music that the country is known for, instead of changing the voice. Not always but usually. You can still try it, it might work. Also maybe try "Stockholm" I have used "Berlin" and got German accent singers

If the lyrics are IN Swedish, you don't need to write Swedish. It will just work. This is only for English lyrics.

[peak pasture]

And plz add Paypal for Subscription!!! ❤️

[peak pasture]

german works fine, but i want a voice that i can use for more than 1 Song.

[feral aspen]

There isn't a specific feature for that (yet? this is suggestions so you can suggest it), but you may able to get close using a different method. If you can find relatively short clip (probably <10s) with a singer, that when you continue from it, you get new clips with the same voice but not the same melody or music. This is possible with some luck and patience.

[peak pasture]

Yeahh i know and it works fine, but only for continue this one Song. What i mean is a Voice that i can use for more than 1 Song ( not Clip). ❤️

[feral aspen]

I can explain a little more. If the clip with the voice you like is very short, then if you continue from the short voice clip multiple times with different prompts, each time can sound like the same voice but the song will be different. The very beginning of the songs will be the same but if you download you song you can use regular audio editor to remove the beginning part that is the same, and just leave parts that are different. This is something that can happen, it does not always work like this. So the first clip is the tree trunk, and each new song is a branch. Is possible, but needs some luck and patience to find a clip that works well for this.

[peak pasture]

ohhh ok, thats nice, thx. But would it be better we can save this generated voice for all Song we want?

and name it by our own.

[feral aspen]

Yeah for sure, voice feature would be cool and is a good suggestion. This is just a crude method for when you want to try and make it work even now.

[peak pasture]

i will test that next Day ( no credits more) 😄

[heady goblet]

An ability to use a command to get Suno to try and use a specific sound Sample, a example could be
[sample - sonic drop ring sfx]

That is my suggested feature

[peak pasture]

Ok , finally here is my suggestion: An "Add ur Voice" Button, that we have created from a generated one, so that we can use this Voice for all Songs (created with SUNO) we want, as default or renamed as preset " Voicename". Then it is also easier to clone your own voice with another tool. ❤️

[feral aspen]

likely a coincidence (only on this one clip) but listen to 0:34 and 0:39 ... 🤔 🤨 ?
https://app.suno.ai/song/e3169a9e-71d8-46ef-87e3-90f66a329ec5/

[heady goblet]

That is interesting

ya know what, I also wish I could look more into Suno's thought process when making certain noises or a certain string of noises in particular

The Guitar riff at 0:22 of this song is so perfect, then I continue from it in other songs and it doesnt sound the same at all whatsoever

https://app.suno.ai/song/598a7f7c-40e4-4164-afbb-e571de21059e

go to 0:22 for example

vs in this at 0:37 https://app.suno.ai/song/a4c56fb8-e8e4-4ad0-9cd4-78c54fba7e88

continued from the first song

its so goofy

[feral aspen]

Interesting idea. Maybe not an easy feature, but maybe there could be a colorful visualization of what the music AI is thinking over time. For example maybe you could change color based on how "confident" (under the hood the token probability or distribution) the AI was over time and maybe that would be interesting or even useful to see. I'm just wildly speculating no idea how Suno music actually works but this kind of thing is done sometimes in AI stuff.

[heady goblet]

Yeah you're right that might be too much to ask, a more reasonable way of adding what I'm trying to get at is maybe being able to "reward" or "punish" the AI by clipping certain parts of a song you did or didnt like and use that to revise the first one until its good enough for you and sounds right

Idk I just think theres so many ideas flowing through my head on how this AI could be made better and im fairly passionate about it since I've been begging for this for a while since AI became popular and Suno popped up out of nowhere, sorry for blowing up the chat 😅

[pseudo crane]

Ooh I see, tyvm for the input! 🙂

[south axle]

Dont you hear that it sounds like a chorus of people singing instead of just a solo vocal? Can you understand that having a weird harmonic chorus on every vocal does not work with a lot of music? Thats why we are all complaining.

[radiant bane]

Are there any plans where I can structure the song on verses and choruses and maybe tell the ai where I want specific parts to be?

I'm hoping for a stem export feature in the near future. I got really decent generations and really like the quality of V3

[tribal tartan]

A suggestion: when requesting a solo instrument, such as 'drum solo. 1970s prog rock' it would be great to get a solo instrument rather than a standard song with full instrumentation.

[fast jetty]

It would be SUPER helpful if the songs include metadata about the KEY and TIME SIGNATURE and BEATS PER MINUTE.

[tropic prawn]

Ability to change the "album art" to set all songs of an album the same or alter it if the art doesnt quite match

[raw cave]

https://app.suno.ai/song/61609e2b-d937-409e-afd3-68ab4910cde8

[hidden terrace]

It´ll be great if we could also make vocals separate (like instrumental) since stem separation isn´t always really clean. Probably got suggested as well before but also having individual stems from the instrumental would be great too.

[tepid trench]

Yep, this is suggested on pretty much daily. I'd like this too.

[hidden terrace]

haha ok xd

When did Suno get published? It seems pretty basic so far. hopefully we some some major features coming soon. *Im excited to see what the future brings

Also the biggest issue i´ve faced so far getting an actual good song structure. The prompts in bracket dont really seem to help. A lot times it thinks the verse is the chorus and vise versa etc. Its also skipping some lines sometimes

[dusky charm]

make the 'thumb down' also a 'send to trash' (maybe toggleable). Right now giving a thumb down means the track is sent to the library, and it's already tough enoug hto keep some structure in there without having all the rejected clips too

[hidden terrace]

We also need more options to continue on a song. sometimes i only like a certain part in the middle of a generated song. and so far i think we can only continue at the end of a clip? Im new here so i might be missing something i didnt see yet

[raw cave]

https://app.suno.ai/song/427697d1-b7e6-45f4-9d47-6d6a471569bc

[pliant canyon]

you can continue from anytime you want

[hidden terrace]

how? i dont understand lol

[hybrid crane]

@hidden terrace ^^

[gusty pelican]

rather than having to manually select each clip in trash I would like a button that is just a 'delete all permanently' button to empty trash out. recycle bin can do it, so can suno!

[hidden terrace]

oh theres a timer box i didnt see lol ty 😄

Need a second option tho. sometimes when the clip is 2 min and i only like 1:30-1:38 or smth like th at so it would still keep the rest i dont want. but i guess i just cut in my DAW ^^

[delicate cedar]

Unless I post it in #📚┃suno-school to do a public dissection and explain what I did and what failed for me to have a post mortem (and maybe get suggestions to kick a song back to what I was hoping for) generally speaking if I post a song, I like it. I'm happy with it. Warts and all. I thought it sounded good.

Then I started doing some work on some rock songs and saw some of what you guys were talking about. But I'm still not a hundred percent convinced it's not new model growing pains.

[hybrid crane]

What you describe is coming. They are working on a "paint in" feature. They hope to be able to develop that for V4.

[hidden terrace]

ok cool good to hear

Also having preset voices that are customized to choose would be cool. cuz im having also hard time getting the voice i kinda like

[hybrid crane]

Totally get where you’re coming from about preset voices. The thing with Suno, though, is it doesn’t keep "seeds" from previous sessions. So, each time you generate, it’s like starting from scratch, which makes keeping a specific voice consistently kinda tricky.

While preset voices aren't available at the moment, a potential workaround is to note down the specific metatags and settings that have yielded voices closest to what you're looking for. Then, use these as a starting point for your new creations. This method isn't foolproof due to the inherent variability in AI generation, but it can help guide the system closer to your desired result.

[hidden terrace]

Yea i was thinking about the seed thing. We kinda need that xd

Tbf Suno seems to be one of the first kinds. when i was looking for text to sing Suno was the only option ive found. Any other site i found was text to instrumental basically or voice to voice

[hybrid crane]

I understand you would want that, but the way it makes music so unique is because it doesn't keep seed. So according to the Devs this is nothing that we are supose to set.

[compact pulsar]

Is there any way to see what is causing Suno AI to prompt moderation, when writing lyrics?

[hybrid crane]

paste your text into chatgpt and ask why it is flagged for moderation

[compact pulsar]

Oh! thank you. Didn't know that was an option.

[hybrid crane]

The moderation is mostly done by chatgpt so the gatekeeper should know best 😄

[compact pulsar]

I shall do so immediatly!

[upbeat rock]

"Continue This Song" feels like an incomplete feature. If I am working on another song and I click "continue" on a different one I have to clear out the lyrics and style of the first song, and then recopy the style for the song I am continuing. That feels wrong

[tepid trench]

Minor thing: I'd like it if the scroll resets to the top of the page when I switch pages.

[delicate cedar]

Are there a lot of people working on more than one song at a time?

How do you keep them all straight in your head? I think I'd get whiplash!

[prime widget]

Is there an ability coming to make intros? It’s so hard to get intros to happen before singing starts and I’d love to apply intros to most songs I have made #💡

[quiet walrus]

/chrip

[still stirrup]

A parameter would be good within "Create" where I could specify the number of songs to generate.

[hidden vault]

----------------------------------------------------------------- 💡 IDEAS to add in the WEB💡 ---------------------------------------------------------------------------

1. In Pc when you delete a song, the MESSAGE TO UNDO is over the music player, making you wait to move or change songs. Move the undo message up and that's solved!
2. The **ABILITY TO ADD INTROS **to clips that start directly in verse. Shouldn´t be impossible cause you already have the ability to continue a song using the style of other clip and you can stick 2 clips together (to make the full songs) keeping style and making flawless transitions. You only have to figure how to do it backwards, but it would be the same as creating a continuation clip calling for a complete silence [silence/mute] and then call for an [intro] and the rest of the song, stick all together and then cut everything before the silence.
3. Saying this: the ability to TRIM CUT clips, and to try different clips sticked together like a puzzle making the AI do the transition between them would be CRAZY good. So a simple editor would be awesome.
4. You could add to every track generation info like BPM or key. You could use or partner with free software like https://tunebat.com/Analyzer. It´s been a great thing for me to learn about music and have more control off what i want. It´s probably not difficult to do and would be awesome!
5. ✅ Add a VOLUME SLIDER! for the web page. ✅
6. Sometimes you use different instruments or styles in one prompt. We don´t have ANY info whatsoever on how the IA reads it. Example: i COULD not generate an accordion instrumental solo (50+ generations), but i saw in community songs a hardstyle accordion song... Teach us how to communicate better with the AI PLEASE, or let people apply as beta testers (I would love to)!

(Answer this with more ideas for the developers and i´ll add them!, Let´s make Suno better together 🔥 )

[fierce cargo]

will it ever be an option to generate seperate stems?

like a seperate Bass track, vocal track etc that are individually downloadable

[upbeat rock]

While a couple are generating I'll go back at relisten to others and be like oh yeah this has legs let's expand this. and needing to remove the lyrics is tedious. The default option when continuing a song SHOULD be to populate the prompt boxes with what the song already used in the first gen

[idle herald]

I'm wondering if suno can have a tree view instead of the flat one, so that relationships between generations can be clearly shown

[rain peak]

A toggle option for this would be great too.

[idle herald]

Yes, both cannot replace each other

[minor walrus]

Just do continue and also click on remix to get the text from the song before, and you're good to go

[void oxide]

an statuspage, to see if their is an problem or not.

[pseudo flame]

oh, one suggestion would be to keep the user's generated stuff somehow accessible by a limited backup system when doing a routine maintenance to most of the system... [edit: except it seems this isn't a routine maintenance but an unexpected downtime]

ie. if i had had downloaded the song clips i just generated as a reference, i could keep working on the lyrics (based on how the AI dealt with them), but sadly I wasn't clever enough to download the generated clips before the maintenance break.

(not a big deal though)

[vivid magnet]

I would suggest to have a dedicated page or discord channel for the server status. This would allow for users to easily check if there are known issues, ETAs for fixing etc

[void oxide]

this what the i meant 😂

[vivid magnet]

Actually read a couple of messages to see if others already suggested it but didn't notice yours, don't know how 😂

[void oxide]

ah ok, no problem! 🙂

[pseudo flame]

i've done this a couple of times, yes... and yes, it gets a bit confusing. 😅

[south axle]

This is what I was making in V3a and with the same prompts I cant even get close to this anymore. https://www.youtube.com/watch?v=_4UPMDiZ3mQ

[pseudo crane]

When I find a song like that I just add it to my playlist called WIP and then I continue making my current song ,once that is done I continue the other song

At least that way things dont get mixed up and since its saved in a playlist it doesnt dissapear

[lofty sonnet]

Suno should Film a chat with Elon, or ask him about his thoughts on it what they've got so far

[gusty trellis]

Hi team, absolutely loving the v3. Would love to see a library option where I can sift through all my "Full Songs". That should help clear up the clutter I think.

ps, i'm playing a dj set of some of my Suno tracks in a gallery in France this Saturday!

[woven pilot]

are my credits going to roll over when my bill is due ? i have the highest plan . its extremely unfair if they dont

[hybrid crane]

Hey goodmoneybunchy! About your credits rolling over, they actually don’t. It's mentioned on the Suno website too. This policy helps manage server resources and service quality. Any concerns? Best to email [e-mail] for more detailed info or to discuss your plan.

[woven pilot]

if i pay my fees i dont see why my points dont roll over or why im not atlease credited

[gusty trellis]

• 1 more request, in the "Continued from..." dialogue, it could be nice if there would be an option to "continue from here" next to each track, rather than it opening a window (imho)

because if a continuation is not to your liking, it is a bit of a hassle to scroll to wherever the original generation was

or simply a "redo" option (redo the same continuation)

[frigid timber]

It basically works the same as midjourney, any credits you receive as part of your subscription do not roll over, any extra credits you buy will roll over

[hidden sparrow]

Is there an option to search others songs on suno site?

[still stirrup]

ahh yeaaa we need search!! also the ability to search for genres would be beast!

[rancid marsh]

Is the suggestion channel managed by a moderator? My suggestions seemed to be not answered. Additional the mp3 quality is realy poor for the moment and a handicap for a payed account. So the base system is ok for ordinary users but if you like as artist to work with results I personally see just quality issues. The question is it just a question of time or is this a basic problem. Inside cubase pro, Nuendo, Davinci Resolve or other daws the level is not good enough.

[delicate cedar]

Suggestions are regularly combed through by members of the Team. They take in the suggestions as an advisory for what they are already working on.

Making a suggestion does not guarantee a personal response.

Or that it will be implemented.

[gusty trellis]

I would like to 2nd the mp3 concern

[hybrid crane]

I recall you seeing in a long conversation with one of the Devs there. It was addressed that better quality is not in the cards atm. But will get better in the future.

For the record "Tony" is a Dev

[wraith quiver]

I would love to have a audio editor like i can say Add Part 2 again after Part 12 or so.

This would be allow to create real Refains.

[gusty trellis]

or maybe bridges / morphs / generative fill

[wraith quiver]

Maybe that, but a start would be maybe like "Add to current track" thing like we have with "continue from this song".

[bronze crescent]

Is the author there? I don't have a credit card. Can I use Alipay, WeChat, PayPal and other methods to pay?

[delicate cedar]

Please do not post the same message in multiple sections. I have responded in the other channel you posted.

[short path]

Is there a way to make a small change in lyrics and redo with exactly the same music and style?

[hybrid crane]

Sadly no... The Suno team is working on a feature that will be able to, and it is referred to as the "paint-in" tool. They hope to be able to implement it in V4.

[potent crane]

Would like to see the option to add instructional prompts like "end" when doing instrumentals.

[pseudo crane]

You can... Just open lyrics and write:

[Instrumental part]

[Drop]

[Big Finish]

[End]

[potent crane]

How can you open lyric in instrumental mode? Are you saying take it out of instrumental mode?

[pseudo crane]

Yeah

[potent crane]

k

[mossy agate]

Help: can anyone suggest to me what music style I should choose for my 6-year-old girl's birthday song?

[raw cave]

Polka?

[mossy agate]

not working well, I types Male Vocalist and output gives me constant female voice versions only

[solar holly]

the term "bouncy kids music" give me kids bop type tunes. example https://app.suno.ai/song/dac37e59-3b52-4ae3-bc74-dce419feb446/

[hybrid crane]

try this Upbeat pop, playful melodies, children's choir, light percussion, joyful harmonies, bubbly synthesizers, whimsical

https://app.suno.ai/song/fb4cab74-4223-4b65-95ea-b97d3133f172

[mossy agate]

How can I make the same music and sound but with male vocalist?

[hybrid crane]

add male vocals, male singer

[mossy agate]

I did almost 4 times used all credits in last two days but it comes with female vocals only, in two days its her birthday so came to get help here dont know will be able to do or not

[hybrid crane]

AI results can be unpredictable, sadly this is the current state. Sometimes you have to try a few times. Sometimes it helps to also mention it in the lyrics of your song. Like [verse, male singer].

[hardy falcon]

The ability to test the words pronunciation before generating the song PLEASE. 🙏

[mossy agate]

Great Feedback and suggestion, Suno should try this feature will help lot of creators save time and credits I guess it will reduce their machine consumptions as well.

[raw cave]

https://app.suno.ai/song/6497f6e8-1e07-4e3c-a940-fe1b4bade9ef

[midnight badge]

I've been enjoying using Suno, but it badly needs training on more genres and eras of music. Getting it to make anything that sounds vaguely '80s is incredibly difficult.

[boreal jewel]

I like the images suno makes for the cover art but sometimes I get a really cool image on a song generation I don't want to use and wish I could make that artwork be the artwork for my final song. I really really would love the option to change the cover art work

[midnight badge]

it's also obvious that in custom mode it discards a lot of keywords you put in the style prompt. It would be nice to see what it actually is using to generate the song, so we can better tailor our prompts

[hybrid crane]

Hey Serqetry, glad you’re enjoying Suno! You’re right about its current genre range. AI music is still in its learning phase, so nailing specific eras like the '80s can be tough. About the custom mode, yeah, it doesn’t always use all keywords. Learning effective prompting is key. Keep experimenting with different prompts to see what sticks!
For more tips see the community wiki here: https://www.suno.wiki/
And I myself made video what my workflow is: https://youtu.be/5uJw5kb05as?si=gGiiztZWEX9rFmV-

[midnight badge]

I tried the style prompt: "dark satanic groovy electronic '70s sleaze disco" and I'm pretty sure it is ignoring my entire prompt except maybe "dark" and "electronic". it sounds like modern pop music or synthwave

thanks for the reply and the links

[balmy zealot]

Metal polka

[balmy zealot]

Lol sometimes it incorporates them in the lyrics

Like itll aknowledge some instruments but not others and read them as the intro

[midnight badge]

oh that's the worst when it does that... but that doesn't happen when you provide your own lyrics. the automatic mode is impossible to use because you can't differentiate between lyrics and style in one prompt. when I first started using it I did it that way and tried to make a song with violins, it gave me a song with no violins but the lyrics said something like "here come the violins"... ugh so bad lol

[balmy zealot]

i can confirm it does. I never use auto

[midnight badge]

I'm talking about when you provide your own lyrics. it doesn't change them so it won't put words from the style prompt in the lyrics

[balmy zealot]

Oh lol i was talking about reading the style words as lyrics

[balmy zealot]

What should that sound like? 🧠😍

[midnight badge]

Like the band "My Life With The Thrill Kill Kult"

yeah it needs a separate prompt for describing lyrics to generate. The way it is now where that is part of the style is just no good.

[upbeat rock]

are playlist links not shareable? I know there is a share button but it doesnt appear to make an embed playlist but instead prompts login

[balmy zealot]

💡Please do a “recently used” to put the first few playlists used at the top of the list when clicking “Add To Playlist” 👌

[tidal wharf]

oh wow, you can now continue with v2 from any point too?

[balmy zealot]

Right, one logs-in first to play it

[tidal wharf]

I find v2 sound still "fuller" than v3 in a way. They got rid of noise, but also got rid of something else, something good.

[pseudo crane]

Would it be possible to get maybe style specific channels in the discord? Like one for synthwave or so

Or that would prob result in too many channels

[quiet star]

Ayo, Im not sure if this channel is just for Discord suggestions but it would be nice if one could upload an own instrumental or even a song and the ai can coninue to the song/instrumental and if the AI could add vocals to the uploaded instrumental, wouldn't it?

Im not sure how it is with these copyright shit, but yeah.

[plush scarab]

Suggestion for the Discord: Make a channel for people to showcase their creations in separate from #general so people can ask questions without being flooded

[hybrid crane]

#🐣┃suno-showcase you mean? 😉

[marble canopy]

Add other languages than english

[hybrid crane]

@marble canopy there is... if you don't see them make sure all your channels are on "show all"

[plush scarab]

Oh, huh. I guess I just wish there was a dedicated help channel that didn't have that problem

[marble canopy]

I meant for the ai? Can the ai make a song in swedish?

[hybrid crane]

why wouldn't it be possible? Have you tried it? I personaly make Dutch, German, English..

[marble canopy]

No I thought it wouldn't work but thanks for telling me

[hybrid crane]

😄

[marble canopy]

How do I make it generate a song in swedish then?

[marble canopy]

Do I just type the style in swedish or do I say make the song in swedish

[hybrid crane]

Best is to click on "custom" and put in your own lyric in swedish.

[marble canopy]

Okk

[quartz basin]

I really would like a "Notes" field for my songs, so I could write down ideas for them.

[marble canopy]

Btw if I have the free version will I get 50 credits everyday?

[hidden terrace]

I wish so much we had more allowance of characters in the Style of Music textbox ^^

[hybrid crane]

yes... Free users get 50 credits per day. Keep in mind, the music you make under the free account cannot be used commercially.

[marble canopy]

But commercially will that allow me to upload it without earning any money on it right?

[hybrid crane]

Depends.. if you use it professionally for a twitch stream for instance, you have to use the paid plan of Suno. Even if you twitch stream doesn't make money yet. For more information please see the website of Suno: https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448#f0f1441180484d6094206e84e334ba36

[plain cosmos]

Tony, every clip I generate has the same problem, and I have provided a lot of detailled feedback on it. I dont feel like generating more because its useless. Given the dozens of users complaining about the same issues here, I guess Suno is at the point where they are about to lose their loyal customers. For your users V3 is going back from V3 Alpha..

[marble canopy]

Well I thought I could upload it to youtube without earning money??

[hybrid crane]

Please check the url for more clarification.

[marble canopy]

Okay

[hidden terrace]

Quote: Commercial use includes any use that generates revenue. This includes monetization on sites like YouTube and uploading songs to music streaming services like Spotify and Apple Music.

[marble canopy]

Yeah then I can upload it if I don't earn anything

[pseudo crane]

I really think it would make a lot of sense to scrap the art-generating part to save power consumtion and focus all energy on the music generating part. Theres a ton of AI to make art with out there. But maybe it doesnt use a lot? Or at least make it optional, I would opt out of it to let the AI focus on music

[onyx hamlet]

any way for creating a complete song? First generation is 2 minutes or so, then I can ask for a continuation but it adds another piece and sometimes it is truncated. is it possible to ask for song ending in some way?

[upbeat rock]

please please plesae just an "end song" button

[upbeat rock]

at least I want to edit the art

[vestal isle]

paypal as pay method

[dark notch]

@solid sage how do you master your audio to remove the noise?

[upbeat rock]

u can try to prompt for [outro] and [fade out] it takes some trial and error but eventually it works

[solid sage]

Moises, bandlab, and soundtrap studio

[solid sage]

I use moises to seperate the vocals, clear them up in sountrap, and then export them to get mastered in Bandlab

[molten orchid]

never, you cannot be forced to give money without giving away you credit card info, and that's how subscription services work...in a scammy way (google does that as well)

[true coral]

is there a way to produce just one clip at a time?

[tepid trench]

No, you always get 2

[midnight badge]

Wow your rockabilly workflow video is really awesome. I didn't realize it could make a seamless consistent song with a proper beginning and end, I was doing it all wrong. Thank you

[hybrid crane]

thank you.. this song is my best song in that genre: https://app.suno.ai/song/2c3f700d-011f-43a3-82a3-7f27021cede7/

[upbeat rock]

this is funny to read for someone who joined right as V3 dropped. this is the most impressive generator there is. I believe you that things may have shifted from alpha, but if you believe it then you should still believe the team will deliver.

[midnight badge]

That was incredible!

[hybrid crane]

Thank you... with the right prompts and lyrics Suno really wants to preform 🙂

[solid sage]

Something like this

[acoustic pike]

Aa bhi jaa,
Aa bhi jaa...

Ae Subhaa...
Aa Bhi Jaa.

Aa bhi jaa,
Aa bhi jaa...
Ae Subha...
Aa bhi jaa.

Raat ko..
Kr vidaa.
Dil rubaa...
Aa bhi jaa.. Aa.. Aa... Aa...

Aa bhi jaa,
Aa bhi jaa...
Ae Subhaa...
Aa bhi jaa..

[dawn hamlet]

could thye bring back the lyrics on the suno generated songs? the lyrics dont display anymore and they were great before

[hybrid crane]

Thye will be bringing back the lyrics on Suno songs soon, they're just upgrading them right now! 🌼

[abstract wadi]

They could put the Alpha version for those who pay and v3 for those who use it for free, to continue refining it. 🤷

[tepid abyss]

For song endings, try the [outro] tag.

Oh, ha. That was already answered.

[tepid abyss]

This has likely already been suggested, but being able to stitch together continuations would be groundbreaking for composing. As is, one is forced to assemble song parts via third party DAWs and such.

[hybrid crane]

You can? Just click on the last part 3 dots menu and choose "make whole song"

[tepid abyss]

Yes. That's not what I'm referring to @hybrid crane . No worries, I was afraid I wasn't being clear enough and I should have elaborated.

https://app.suno.ai/song/43c203c1-e143-4963-bdb8-8d1c9fc856f1

[pine tree]

v3.0 added diversity to the first track generation, but I think it skips the lyrics of the next track. maybe all the continuation parts after the second track should be v3.0 alpha.

[kind bridge]

A list of all the trained words in box we can click on that influence the music style

[abstract wadi]

Let's be clear, for the people who could test the alpha version we know that this is better than version 3 that was officially released, after spending 700 credits I only had 2 good songs, neither in v2 I had to spend so many, also in some musical genres the structure and sound part is completely bad, compared to the alpha version, although most of the people voted that version 3 is better than alpha, could they really test it? How many of these could do it, and even if it is half of those who voted, 20% of the community is saying that version 3 is worse than the alpha, they should listen to that part too, even if they don't want to do a rollback because it will affect in the part of keep improving the version now, why not have feedback from the people who have this for free? (besides the people who have bought a subscription and will use this version), why not just please this percentage of the community? In the end we will all be fine so to speak, besides being able to continue improving suno with the feedbacks of the other users. (I miss the alpha version 😭 )

[frigid timber]

I dont think they should remove current V3 just add V3A to those who had access prior, Im in the exact same boat as you though, the vocals deteriorate for me so quickly. Every time I come in with a fresh mind to try again each day, I always just end up frustrated and shutting it down

[abstract wadi]

I didn't want my message to be to remove the new version, no. just like you said add the alpha version and make it as a plus for those who pay and also keep the community happy, version 3 will be bad so to speak, but it must improve, so the feedback can come perfectly from users who do not pay Suno and also from people who pay the subscription and want to continue using this one, besides they can perfectly put the alpha version while version 3 is being improved or simply for the future all the feedback can be taken to the alpha version and make it a more complete version than it was, but let's wait what the devs say, although as I said 20% of the community is asking for it, they should listen to us too, besides what's the use as I said before to make that survey of which version is better? There are people that maybe didn't try the alpha version and don't know how it sounds, they should have put a video of how the alpha version and v3 sounded to make a comparison.

[frigid timber]

yep i agree, just hope something is done relatively soon, ive used this service nearly daily since they released V0, so being stuck in this waiting "limbo" has me going nuts

[abstract wadi]

And finally, I know that the criticism parts affect a lot to the developers and with my other comments I do not want to discredit what they have done or anything, but rather I would like to congratulate them, since being such a small team they have done this in such a short period of time is impressive. But I really feel that this version was made very fast, I know that it could be because the community was excited about this version, but I would like you to take the time you need, if you need a year take it and if the community pushes you, haven't you thought to make in a period of time like for example every 1 or 3 months a kind of video or event to tell the community your vision about the next version or show small new features, also make small updates to make the community feel more satisfied(?).

But again I want to congratulate you for all you have done and don't let yourselves down, I know you will do in the future or next updates something great, but as I said if you need time, take it 😄

[formal bear]

Hey I hear you

Honestly the feedback is just vocals are bad in v3 right? I get it.

[thorny wedge]

The poll itself will result in misleading results because a large % of the community never even used V3A.

[formal bear]

Well, plz tell me how to do user research then?

Honestly asking.

if you have better suggestions I'd listen

[thorny wedge]

Does the discord say which accounts are linked to people who paid for the service during the V3A testing?

[formal bear]

No but free users also matters.

[thorny wedge]

Vocals are the main issue. Though the general sound quality is also worse.

[formal bear]

I agree with you, I listen/

[abstract wadi]

For what I do, which is more synthpop (chiptune), variants of rock and ambient/jazz. Of all the songs that I have generated, it does not compare with the alpha version, I feel that the structure of the songs is very strange, the dynamism part is the same and I feel that the sound quality is worse than in the alpha version (this is more in the chiptune, before it sounded much better).

[fallen loom]

@abstract wadi 👍

[abstract wadi]

sorry, I mean they are not dynamic, the songs are very flat, they don't have changes like they could have before in the alpha version

[thorny wedge]

Sure, but they don't have anything to compare V3 to other than V2.

I don't know what exactly said here can lead to changes but my honest suggestion would be to add V3A to be available for all accounts for the next week or two and then run a poll off of that.

[formal bear]

I think the general attitude toward the v3 issues is a bit off. I know haters is going to hate. I don't even know why I bother spending time explaining here every day that v3 alpha is not going to get back. But I am going to say it again that v3 alpha is not going to get back.

Just above the polls you also see all the issues v3 alpha has. There are very large amount of feedbacks on song structure that we addressed in v3. It is not going to be perfect. Nothing will ever be. We will try to get you the service that are more likely to get you an ideal generation. That being said, maybe the right solution is not v3.x but v4.

So please redirect your feedback / suggestion as yeah we wish future vocals / sound qualities are better. Bring depreciated versions is not going to help us build a better future. Again, let's all eye forward.

[thorny wedge]

I mean, call it whatever you want. Either way the problem is what have been said. So if that's being listened to then all I can say is I wish you guys luck on your efforts.

[robust jewel]

That feedback is helpful as a user too, since it let's me know it isn't just me getting rough vocals. I'd been feeling like I might be doing something wrong, given how much better the VDay thing did. Was wondering if I prompt really poorly or something, but seeing that other people have issues with it let me know it's probably something that I just need to wait on.

[odd glacier]

I'm fine with not bringing back V3A, but my question is what happened between V3A and V3 for the major difference in quality?

[tepid abyss]

For what it's worth, I used V3A extensively and I haven't noticed the issues being talked about. Reading this discussion has actually caught me off guard. They're clearly issues prevalent enough that others have noted them, but I have not myself observed a drop in vocal quality.

Not discounting anyone's experience, just sharing mine.

[formal bear]

It is really genre / music type dependent

Folks, next time, when there is an alpha, treat is seriously, and label your likes and dislikes, that's all I can say. Otherwise you may not get what you want.

[frigid timber]

I posted my comparisons in the bugs section, not sure if that helps, Im sure you guys have seen most of the differences
#🪲┃bugs-web message

[thorny wedge]

Idk. I get choirs coming in on all my songs from jazz to japanese pop to romantic piano to rock etc. I don't think choirs should be coming in on piano sax romance songs 99% of the time. But that's my main issue anyways.

[robust jewel]

I would. I enjoy taking part in AI alphas. I just don't have the money to join in. Sorry.

[abstract wadi]

couldn't they make a hybrid version for the future? for example, there are some genres that are weak in version 3 but these are strong in the alpha version and vice versa, so improve from here the following versions. i don't know if i explain very well

[thorny wedge]

On the other hand, the whole choir prevalence issue has made duets easier to make when you leverage it to your advantage.

[formal bear]

I like this idea a lot honestly, but that's not a switch right, that's personalized creation. That's sth we should be working on rather than arguing v3 alpha.

[tepid abyss]

Ah. Gotcha. I've been rendering mostly the same music from alpha, and I used likes/dislikes extensively (admittedly, mostly to help with organization and finding good renders).

I'm interested to see/hear comparisons. It's been a bummer reading this discussion, as I thought V3 was being well received.

I love it, at least.

[tepid trench]

I don't know how feasible this would be (I guess it takes up more resources?) it'd be nice to keep pre-releases available for future versions, even if that's just as a subscriber-only perk. Personally I'm not THAT bothered, just thinking what would help to avoid people being unhappy in future (though I guess there's bound to be someone who complains that the pre-release version isn't perfect and why isn't it perfect omg etc.)

[thorny wedge]

This is something that OpenAI does with its GPT API. They depreciate models about 1-2 years after being outdated to help customers keep their existing products on the same quality/provide similar results.

[odd glacier]

No, it's everything about V3. Vocals, lyric comprehension, genre understanding, ignores tags, mispronounces words, ignores genre keywords, tends to produce a smattering of notes rather than a rhythmic melody, garbled/distorted sound quality, slurred vocals, song begins with one singer but continues to add additional vocal layers with each bar until vocals under just garbled voices, hallucinates vocals, repeats verse/chorus segments or sings them out-of-order, etc, etc.

I honestly don't care what the poll indicates -- V2 would produce a damn good result by it's third attempt. V3A would do so by it's 10th attempt. I've now produced 80 attempts for my first song in V3, "Fleeting Elation" and it still hasn't worked.

[formal bear]

ok just use v2 then right...?

[tepid abyss]

The disparity between my experiences and others is really interesting.

@odd glacier, what genres do you usually render?

[frigid timber]

i tried using V2, its the opposite effect, vocals are good but instrumentals are not, just trying to get my song finished lol

[thorny wedge]

V3 also skips lyrics the more continuations you do. Which is a real problem when trying to structure a song verse by verse through continuations. By the 4th continuation I have the AI skipping entire verses at an annoying frequency.

[odd glacier]

V2 no longer works correctly now, either. Ever since the V3 release, V2 suffers from similar problems:

Ignores tags, ignores genre keywords, tends to sing vocals with a Jamaican accent, starts garbling lyrics, etc.

[formal bear]

I don't know what's correct then honestly

[odd glacier]

Alternative/grunge/90s/angsty

[formal bear]

see this is why I trust polls more, at least those are numbers

[abstract wadi]

I hope you take into account this idea, maybe people will be happier, besides the next version if they apply the best of each of these (alpha and version 3) can leave something amazing, plus the improvements that will take into account for version 4, but as a user also know that I should not demand much came out 4 days ago this version, I know it will improve over time. But as I said if you need all the time in the world for the next versions, do it, I believe that the community will be happy with what I told you, videos or conferences of how will be the next versions and features that these will have.

[eternal jungle]

Is there a way to convert the songs into tabs?

[tepid abyss]

I don't believe I've ever tested or experimented with those genres at all. Curious to now, though.

[odd glacier]

Polls don't mean the issues don't exist. I think we've established that the problems we're experiencing are genre-specific (though I'm surprised that even jazz is experiencing them), but unless Suno is going to cater to a specific set of genres, these problems still need to be addressed. Polls be damned.

[formal bear]

thx man I apprecaite it

[odd glacier]

Just a little comparison:

This is a recent attempt in V3 for a song called "Fleeting Elation." Notice the poor sound quality of the vocals ...

https://app.suno.ai/song/9f77902f-0141-4ff1-bb16-9d5208d2bd7f

And here's one done weeks ago in V2:

https://app.suno.ai/song/16591e8a-4103-4553-a1ca-17dae63d6f02

[vague salmon]

could you not judge user satisfaction by looking at the like/dislike ratio of song creatied in v3 vs v3alpha? dont know if that's possible, but just an idea

[formal bear]

we don't

[frigid timber]

Can we do a poll between V3A and V3 that only paid members can see? Will take some structuring of the discord but I think if only people who used V3A could vote will give more accurate data, since this poll seems to be weighing alot on decisions.

Just want to say also, I do appreciate all the work you guys have achieved thus far, this technology is amazing regardless of the model number, nobody seems to be able to come close to doing this besides suno, so kudos to that

[formal bear]

poll is not the only way to measure, and what i can say is it agrees with the poll, that's all I can say

[thorny wedge]

Yeah. V3 sounds like it's been overmixed and had its dynamics crushed.

[formal bear]

I am not going to spend more time on this...if there are no other suggestions, good luck folks. May the suno generation god be with you.

[abstract wadi]

@formal bear can you do one of these days another pool, but in this case, which parts of version 3 can be improved, as you did with the alpha?

[formal bear]

yeah

can do

but lol I think the answer is obvious...

I know it and you know we know it. So.

[tepid abyss]

Yeah, I definitely see the difference. Thanks for the comparison. I was completely in the dark as my own renders have been fantastic.

The Suno gods favor me this month, I guess.

[abstract wadi]

yes, but in order to have a record, even if it is obvious, I prefer to have everything in order so the community will be calmer with the messages of the new version.

[formal bear]

fair point

may not be me who posts it, but sure

[formal bear]

I have been explaining on this issue for a week now

and it just keeps on coming back

I have to say...I am kinda glad that it is so focused on vocals quality / sound quality and nothing else...

at least other aspects are not totally off the rail

[blissful cedar]

Hello, I need help, I made a song that I really like, but I would like to change 2 lines but keep the same song. Is there a way?

[abstract wadi]

I said it because I have nostalgia for the alpha version :,(, but as I said it is a matter of time to improve and also exceed the alpha version, but I think that after the survey will not be so many people who comment the same, plus I know we are all looking forward to future updates

[tepid abyss]

Again, not to discount anyone else's complaints, but I'm exceptionally pleased with how far Suno has come along myself. The only suggestion I have is more continuation/song assembly options, as mentioned earlier. And perhaps folders for songs.

[formal bear]

you can try to continue from

[blissful cedar]

the problem there is only an end part 1 minute but the beginning remains with the old Lycris

[abstract wadi]

the other thing I told you, is that some genres (chiptune in this case) the dynamism is flat and also some or most of the songs I have generated are similar. but with what you have already said, sound and voice quality, there is nothing more to add, in my case.

[frigid timber]

try continuing and use the exact same lyrics with the changed words, should hopefully nearly repeat itself

Then just cut out the beginning after you download

[blissful cedar]

I tried it, unfortunately it doesn't work but thanks anyway

[livid ruin]

Suggestion:

Ability to save a singer from one of your songs. You can then name them with an @ symbol. That way we can add the signer by their @ name.

That could be like GPTs from ChatGPT and people could trade and share singers. Lots of potential here.

[icy stag]

idk if this is something already aware of but on the mainsite when i extend a song the entire lyrics of it get put into the description when i combine them. So often there are lyrics which i didn't use in the description..

perhaps letting us correct the description's lyrics would be nice.

[formal bear]

yeah I like this a lot actually...

ty

[formal bear]

yeah it's a bit more complicated than a feature but we are thinking about it, yes

[livid ruin]

Yeah @tacit idol and I have been talking a little about it here and there.