💡┃suggestions-web

[hybrid crane]

Do not post double please. I removed it becaues you violated our discord #📜┃rules @devout sparrow

[devout sparrow]

It seems that version 3 is not pleasing, right? I noticed that the songs are a little lifeless, repetitive, as if they were all following the same melody pattern.

[devout sparrow]

I noticed that you make a point of deleting my messages, but you don't respond or offer any position. I am a content producer and pay for the use of this platform, however, I have observed, including in version 3, that it seems to be less capable of understanding and its quality seems to have regressed since launch. Please, I appreciate this platform, but it seems that if it continues like this, competitors will soon take its place

[hybrid crane]

I am a volunteer moderator who is about to go to bed. Not a dev.

[fallen egret]

In v4, it will be possible to drag and drop our own vocals?

[half cargo]

It would be nice to choose an image from one of the generations that you had in your song. For example, being able to choose if you want to go with the picture from Part 1 or 2 or 3.

sometimes my last generation's image is just a bird. but part 2 was 🔥 lol

and unfortunately nobody will get to see it.

[ionic skiff]

remove the idiotic 1 minute limit from subsequent generations (part 2 onwards) i have lost so many credits / chorus just because part 4 of the song generated a sound track of 0:58 seconds .. this is a stupid limit that spoils good songs!

[fringe hamlet]

I don't get it, your song is completely ruined because it's 2 seconds shorter than you wanted or what?

[ionic skiff]

the 2 seconds short caused the chorus to cut out at the end, subsequent generations from that point onwards entirely change the chorus

this is extremely infuriating, when you are deep into the song

and then have to redo several parts ..

losing a lot of good music in the process

[fringe hamlet]

Hm interesting in the past, I've had it very often do that, cut itself off mid sentence for example, but then if I just extend from the end anyway, it still "remembers" where it was and finishes it

[ionic skiff]

it sort of remembers, but it changes the way it was singing the chorus

i can show you the song, but i am unsure if anyone (except suno) can help tbh

[tame vector]

Not sure if this is brought up earlier, but insert between to points would be awesome. It’s so frustrating when your chorus kick hard and flows super smoothly but the pre-chorus before is lacking the same energy build, and you need to run an extend from before the pre-chorus and you never get the same dope chorus after.

[delicate cedar]

Some of us have been advocating for such. But there are higher priorities at times, like making sure everything else is running smoothly and ensuring that the next version of Suno will be even better than the ones before it. But this is a good idea and many of the mods have been suggesting it too.

[turbid ocean]

It would be nice if the devs got a community manager who conveyed development updates and progress. I think that would quelm a lot of negativity and give reassurance on what's being worked on

Udio is very good in this regards

[jagged fable]

This x100. They apparently give the discord mods some information on the roadmap and then those mods have told us things about potential updates. But this is a really weird way to go about it compared to actual direct transparency.

Most people understand it's hard to commit to concrete timelines and promises, but some type of general overview on future goals & updates would really be great.

[turbid ocean]

The need for it isn't even really a hard argument to make. If they hired a part time community manager, the increased transparency, communication, and complaint responses would surely increase subscriber retention and customer satisfaction rates. Responses from them can be as simple as "Were working on it." or "this is not planned at the moment but we'll take it into consideration". People just like to feel heard.

And to add onto that, people HATE feeling not heard

[jagged fable]

Yeah I also feel like it'd probably calm down the number of repeats with suggestions. There are certain features that are suggested dozens of times a week. If people know whether or not those features are something they're considering, it'd at least mitigate the clutter and repetition, making room for other suggestions

[plush scroll]

Suggestion:

I would love it if you could extend a clip backwards the way you can on Udio.

[obtuse tree]

we're doing this

[sage mirage]

sweet where do I send my glorious resume

so you guys can go "what the hell is this lol"

[turbid ocean]

Awesome to hear, and thanks for taking the time to give the heads up!

[indigo plover]

Hey, it would be nice, to see what was the voice used for the song. And the possibility to select or reuse the same voice for another song.

[tepid trench]

I expect that reaction with eyes popping out on anything interesting or controversial 😆

[misty fjord]

Hello! My distributor asked me to provide a license for the tracks I uploaded to the sites. Where can I get a license? My subscription is "premiere"

[hybrid crane]

If you're subscribed to the "premiere" plan, you own the commercial rights to the music you create. There's no separate license document.

[misty fjord]

@hybrid craneand how do you prove the distributor is right, a screenshot from the office?

[hybrid crane]

I have released several songs from Suno. They never asked me to proof.

[low solstice]

Dear Team, It would be really great if you could tell the transformer to only use what is specified in the prompt, you could make an exclusive botton and then you could enter tempo, instrument and harmony in the prompt box and e.g. 80BMP, Bass Guitar, F-Minor and the transformer would only use that, and not add everything randomly as it does now . If that were possible then you could make quality music as a professional musician without the guesswork and hoping. at the moment it's a great tool, but only for amateur musicians.

[misty fjord]

@hybrid craneuntil asked...And if they do, what will you provide?) I've been asked, and I'm in the middle of proceedings

[hybrid crane]

They only ask if they THINK it's already in their database. And that's mostly an automatic process, and always (so far) false flags. And I expect it to be false flags always since Suno makes unique music. So then you dispute it, a real human will listen to it, and it will be solved.

If they ask for a license, it's really weird. And I would move to a different distributor.

I mean.. asking for a license? I created a lot of songs from scratch on my DAW.. How would that prove anything? I cannot license myself lol

[misty fjord]

@hybrid craneCan you tell me how I can write to Suno technical support?

[hybrid crane]

[e-mail]

[misty fjord]

@hybrid craneI have a catalog of 5000 tracks and they asked me because they immediately recognized that the latest tracks were written to me by the AI. Anyone can be asked, it's just a matter of time and how much money you make

[hybrid crane]

There are distributors who do not want AI created. And that is their full right. So again. Consider to go to one who doesn't have a problem with it. Like Distrokid or Ditto.

As far as I know.. TuneCore and CDBaby are not taking AI music.

But Suno does not give licenses.

[soft meteor]

well... technically... one could just write a license for them self..
But I somehow must agree that a written license would be beneficial. It might even be attached to the songs.. These could be automatically generated and attached to songs generated during a subscription in PDF format or something with a digital signature.
Personally I am the opinion showing the TOS and a bank statement would be equivalent to showing a license because the TOS mentions granting such license and the bank statement/receipt shows the proof of subscription status. This however feels a bit complicated and could be simplified.
I don't know why suno does not want to provide a clear cut license at the current time, but I also do not care because I don't need one.

[hybrid crane]

Oof.. I am not a lawyer. So I personally don't know what steps Suno should take then. Also, all countries have different laws, so I think it could become very complex and very costly. But I am just a moderator as you know.. 😉

[soft meteor]

Well there are probably limits as to how many countries it can be valid for because different laws.
But a simple "This song was generated during a paid subscription and due to this this song is licensed for commercial use" would probably suffice for most use cases. But I am no lawyer as well.. so 🤷‍♂️

[hybrid crane]

I don't disagree.. it would be useful. But still so much to figure out in general when it comes to laws.

[frosty breach]

The whole licence thing from Suno could do with some cooperation from their side.

“They said it’s fine” isn’t the best defence.

[fading heart]

I’m still going to ask for a song feedback section for the discord 😊. Is it hard to set up an extra page on the discord?

[zealous trellis]

Is it possible to get an invoice for the paid version?

I kinda need it for my reporting, but can't find anything on the website

[soft meteor]

you should have gotten one via e-mail

[zealous trellis]

Thanks, found it

[tranquil whale]

First of all, let me tell you that the app is amazing, thank you.
I have some suggestions.
I'll tell you how I use it, I'm a music producer and I work at Daw. For me it would be of great help if I could create loops of a single instrument. Example:
I'm working on a house/pop song in Dmin and at 125BPM and I have made the midi of a rock piano for the drop. In that case, it would be great to be able to upload the midi I made of my piano and have the AI ​​offer me that piano with different approaches and sounds.
It would also be great if he filled in drum loops that I uploaded. Example:
I upload a wav that I have made in my Daw that only has a kick and snare, and the AI ​​would fill me in with toms, hi-hats...etc.
Thanks again for the app, it's incredible.
A greeting from Spain.

[young basin]

Please add the function: "EDIT"

• in the selected fragment we can - in the text of the song - add different signs, change words, add to it in order to influence the music in a particular part of the song!

Thank you 🙂

in the Video - please REMOVE all words that are placed in "square brackets" [ ]

Thank you 🙂

[versed bronze]

just allow us to post edit the final lyrics

[dull bane]

Does Suno ever look to expand their input from the user? As some of us use Sonoteller to help get a particular "sound" maybe some of these could help dial in the songs

Being able to add Vocals, BPM, Language, Moods even outside the style would be interesting

[thick reef]

It would be cool if it was possible to put a sample and it recreate it in another style, or put a sample and it add it on the song

[weak bloom]

a $15/mo plan for 5000 credits
30/mo is a lot for a subscription and I dont think I'll even end up using 10k credits, and I feel like I'm generating a lot of content
15 for 5k is a strong middle ground
i think im more likely to stay on a 15/mo plan over the other two plans for the slightly above average use
ok gotta go feed my chickens

[versed bronze]

it would be also cool to have the ability to upload my own images to the songs I made
maybe do it only for paying users

[tulip merlin]

I would like a markup editor mode so I could paste the lyrics, style of music, and title all at once

[haughty kite]

Would be amazing if we can single out sounds that have too much reverb or are just attempting to be melody but are noise... One day I'd like to see Suno have the granular control it needs to propel it to the best AI DAW ever created

[unkempt plank]

Put "in:suggestions forum" in the search box in the upper right corner and we will find many similar requests. (case by me, here is) #💡┃suggestions-web message (No one seemed to see it, though lol.)
We can check if there is something we want to request or not by typing "in:channel name [blank] search word" in the search box like this, but whether they have been implemented or not, or whether there are plans to do so or not, it just says what everyone wants.
The bug channel likewise just seems to be a bland description of information that may or may not be a bug for everyone.

By the way, don't the moderators have a role in adding Discord channels, changing topics, modifying server guides, and other configuration changes within Discord?
If not, I'd love to see the developers assign those roles to the moderators to help maintain the Discord. (Or hope add a new role like Discord Admin and recruit for it...)

Translated with DeepL.com (free version)

[tepid trench]

A lot of people don't like the "choir"/vocal layering creeping into their songs with v3. I try to work around it by adding more specific vocal qualities to the prompt.

Unsure how feasible or reliable this would be, but when processing the style prompt, some appropriate (for the genre and singer gender) prompt tags could be automatically included if the existing prompt is vague about the vocals (e.g. just "male vocals"). Maybe offer this as an optional toggle? (Though I have no idea what you'd label it!)

[haughty kite]

@tepid trench Is it only vocals it does it with? Maybe that is what I'm hearing? As the song goes on, some generations just become super busy with noise that don't seem intentional

[tepid trench]

That's a different thing. "Get Whole Song" before extending can help with that sometimes. I get metallic ringing or just lots of noise with some extensions. The choir problem is where the vocals just sound like multiple people singing them in unison. It might happen immediately or develop over time. It's not unpleasant, just unwanted in most cases.

[haughty kite]

yeah i've noticed that too

Good advice on Get Whole to try and remove noise, i'll give that a shot

[craggy echo]

Yes, single instrument loops would be great. I've burned a lot of credits trying to conjure up bass or drum riffs I could jam to with my guitar. Another thing that would be extremely cool is to upload a track of my guitar and have Suno fill in other instruments (and even more cool if I could pick the instruments).

[worn atlas]

Yeah! The competition already has its forums for bugs and suggestions on Discord. Good communication is a criterion to consider if the tool wants to be seen as more than just a toy.

[potent crane]

Your wrong, I have uploaded hundreds through CD Baby without a fuss

[hybrid crane]

That's great it worked for you. But we had one person who had a problem with that. Let me see if I can retrace it

[potent crane]

https://open.spotify.com/album/7JJWqssFT6VVZhnaObCoYu?si=tJ2ixXjgQ82vnjI3CFvkCg

[hybrid crane]

#🗣️┃chill-chat message

[potent crane]

just showing proof

[hybrid crane]

@potent crane I believe you 😄

I don't know why they are picky.. Some are.. I guess your are not too AI obvious?

[potent crane]

I make no mention of it

[hybrid crane]

Right.. and of they don't ask, why tell? I don't either 😉

[potent crane]

indeed

[hybrid crane]

Then see it as a compliment in your case I guess.. 😀

[potent crane]

Yep, my stuff sounds "real" lol

[modest thunder]

[wooden fable]

should there be a link to the wiki in the #💬┃general-chat channel?

[vale trench]

sunono i have an idea

release a list of the most used words in the dataset

so we have an idea what kind of stuff to use in prompt

[leaden marlin]

What about to change the artwork of a song in web app?

[frosty breach]

neon

[soft harness]

In the meantime, you can do the following: #📚┃suno-school message

[potent crane]

It should be made clear that when you choose to extend from a custom time, say 52 seconds, suno doesn't progess from that point lyrically, it starts again lyrically from the top.

[soft harness]

You will want to combine the parts together (via Get Whole Song) once you are satisfied with the generated Part/Extension.

If you have not seen the wiki yet, check it out: https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[potent crane]

I know how to extend a song. I'm pointing out that although the music commences from your custom point, the lyrics don't.

Could we please have a bulk download option, similar to the bulk delete

^2nd line was a separate point

[crude sinew]

Suggestion
How about the possibility to add individual (hash-)tags to your songs?
This way it would be imo much easier to structure the song list... 🤔

[frank stream]

we've got an extend from. I'd like an extend to(creating a minute before the selected point in the chosen song), and an extend between (creating a minute between chosen points in two songs, or two chosen points in the same song)

[nova raft]

I have some suggestions, not sure if they can be implemented. Like being able to permanently delete music thrown in the trash in bulk, and automatically executing a cleanup action if it's been over a month. Hope the engineers can receive this practical suggestion.

Another suggestion I have is to create an interface where unfinished music, meant to be refined into complete songs, can be separated from the interface for creating new music. This would streamline the process for users who generate a large volume of creative work.

If possible, having pagination in the interface would be another good solution. It can make the operation more convenient, as navigating through the interface without pagination can indeed be a bit cumbersome.

[hazy hornet]

Suggestion for feature
Feature called "Remix Sample/Audio"
Where the Song Takes a Sample Or Audio from Whatever audio you want in other language remixing a existing Song with More Intense or Cool Sound for Example, They can Be a Cover or instrumental music with different tone
After you apply the sample you can generate the song with any Theme you want and the song will Created with The Sample of Song you Putted on
Sorry for bad english but hope you can understand this and add this feature

[glossy pumice]

I'd love to see this moved anywhere else so it isn't covering the playback controls. I'm constantly tripping over it.

[sharp zodiac]

Dear Suno Ai Team,

I'm writing to provide detailed feedback on my experience with your platform. While I admire the innovative approach, I've encountered several challenges that significantly impact usability and enjoyment.

Voice Quality: The process of finding a suitable voice is excessively difficult, with a success rate of approximately 1 in 1000 attempts. Moreover, most available voices exhibit a lack of clarity, often containing indistinct noise and a vibrating quality that detracts from the listening experience. This inconsistency in voice quality makes it challenging to find an enjoyable voice for content creation.
Lack of Continuity: In earlier versions, such as version 2, I appreciated the ability to use a preferred voice across multiple pieces of content. However, with the introduction of the "continue with time" feature, this functionality has become unreliable. This disrupts workflow efficiency and adds frustration to the user experience.
Priority on Voice: As a user, my primary focus is on finding a clear, correct, and enjoyable voice rather than on the music itself. However, the current versions of Suno Ai prioritize music over voice quality, which does not align with user preferences. Improving voice clarity and variety would greatly enhance the platform's appeal and utility.
I hope this feedback provides valuable insights for further development. Addressing these issues would undoubtedly improve user satisfaction and make Suno Ai a more compelling platform for content creation.

Best regards,

[plucky wave]

A simple search for examples of music styles would be nice.

[frigid pivot]

Change up the Suno homepage a bit more, some songs are getting obscene amounts of plays while new songs are difficult to find without a link and aren't getting a chance. Linked songs are always going to be more popular than unshared Public songs, basically.

Also consider creating the ability to listen endlessly to songs created by different users, with the option to shuffle popularity. Example: play a song with lots of plays and likes, then play one with very little plays and likes, then one with moderate, and so on. It would increase variety and help prevent user burnout, knowing your songs have a chance of being heard after the first 5 minutes post-creation (currently the "New" category on the homepage).

[stiff canyon]

Hi Suno Engineers,

you know it, we know it.
There are a lot of users (especially in Europe) who don't own a credit card but would like to subscribe. Could you expand the payment methods to include Paypal, for example - it would be worthwhile 😉

Thanks for your awsome work!

[urban sleet]

Search for examples? Search in my own library would be gamechanging 😁

[wraith grotto]

when i subscibed site whiped my 350 credits i had saved, would be good there is warning

[worn atlas]

Do you mean something differents from the metatags ?

[bitter pasture]

More payment options

[frosty breach]

I would rather have delete permanently be an actual delete, rather than just hiding it from you.

[worldly sphinx]

Any kind of search would be a game changer. Search on your library, search for users, search for genres/styles, search for lyrics even...

[peak vale]

I think the stems, and EQ of each one, will be the most greatful update

🙂

[crude sinew]

Yep... afaik metatags are used within the lyrics, right?
I'm rather talking about hashtags to structure your own songs or songs-parts.
E.g. a hashtag for your "#SongA", another hashtag for your "#SongB", one for "#cooldrums", one for "#favorites" and so on.
When you click on one of these hashtags you only get the songs in the "Library" (or even on "Create") shown that have that hashtag.
Wouldn't that make structuring much easier?
Or maybe I haven't understood the current concept yet? 🤔

[rocky jay]

On download, can we have the file append a time stamp. so for instance, if the title is "Song Name", can the export be "Song-Name-MMDDYY-HHMMSS.wav"

[lyric cedar]

What about an option to choose how many songs you want created?
It seems like a pain that you have 5 credits wasted every time.
Why can’t we just create 1 song at a time instead of 2?

[silent surge]

NEGATIVE PROMPTS.

[west swan]

Spanish:

Soy músico, así que mis sugerencias van sobre ese aspecto de generación en base a creaciones propias.

• Supongo que ya lo habrán dicho por ahí, pero me gustaría probar subir una canción acústica grabada de estudio y que suno generase la producción instrumental.

• También podría estar bien que, lo mismo que suno puede seguir una canción generada por él, lo hicese con canciones no generadas por él (subidas por el usuario por ejemplo)

• En oto aspecto, habéis probado o contemplado que se le pueda dar a suno una secuencia de acordes o notas de aromonía en la que basarse?

Saludos!

English:

I'm a musician, so my suggestions go to that aspect of generation based on my own creations.

• I guess it's already been said, but I'd like to try uploading a studio recorded acoustic song and have suno generate the instrumental production.

• It could also be nice that, just as suno can follow a song generated by him, he could do the same with songs not generated by him (uploaded by the user for example).

• On that note, have you tested or contemplated giving suno a chord sequence or aromony notes to base it on?

Greetings!

[deep lintel]

Select - female or male singer !

[quartz steppe]

Small suggestion for web application:

• Find links to playlists, that contain the track I have selected. Then add them to track details component, that is used in Create and Library pages.

I use playlists to organize my tracks, but the problem is that playlist content is visible only in Library page under Playlists tab. So, when I just explore my library or making tracks in Create page, and want collect them into playlists, I don't know which are attached to playlists and which not. This leads to risk of attaching same track to same playlist twice ( this case is unhandled ), or attaching the track to several different playlists, or skipping some track that I want to add, etc. Basically, absence of references makes things extremely inconvenient, hard to navigate. I have to make screenshots of playlists as a workaround, but it doesn't work well when playlists grow over 20 tracks. So, UX would be greatly improved, if track related playlists are requested whenever it is selected. It shouldn't be difficult to implement.

[worldly sphinx]

The mobile version of the website needs some love (testing on Chrome/Android on a Pixel 7 Pro):

• multiple selection is not possible (no shift/ctrl key available)

• the "Create" UI forces the user to continually switch from custom to simple mode simply to be able to listen to the 2 clips I'm creating (in custom mode the "custom mode" bar is sticky on the top of the page, covering the tracks). This had me accidentally create "simple mode" songs simply because I forgot to switch back to custom before hitting "Create", wasting my credits on some random stuff.

• adding stuff to playlists seldom results in duplicate entries on the playlist I'm adding to and/or in the song being added to more than one playlist

• audio sometimes stops playing when the browser is in the background, reopening the browser makes it resume the playback. This may be an Android issue but other websites that can play in the background don't seem to have this issue (e.g. Soundcloud, YouTube Premium)

[sage mirage]

I see your last point as well, also Android

I don't create on mobile, just listening

[floral laurel]

Ability to change thumnail art via generate button with different description box.

[worldly sphinx]

a working, sure way to end a song at a specific moment, one that does not require credits and "tricks" like putting [outro], [fade out] or [end] and hoping for the best

[modest thunder]

[spiral acorn]

It'd be nice to have the ability to set the end, and also a fade duration that defaults to 0s.

Those options would allow you to choose if you want an instant end or a faded end, or even a kind of "quick" fade which might be useful in specific situations.

[worldly sphinx]

Either way, I'd like to stress the "no credits" bit. I don't like wasting credits trying to end a song

[spiral acorn]

yeah agreed, i mean even just an "instant end" would help in most cases. The fading just makes it a bit more flexible

[sage mirage]

Letting us clip a song, no matter what comes after in that single generation

[final drift]

Some generations are completely bad. In my opinion You should have it only charge credits for rendering that you actual like, extend, or download. and or give credits back for anything a user labels with a thumbs down. I have too many generations where Suno will completely ignore a few bars of my lyrics and ruin the song. Not to mention all of the times for no reason at all it will go back to a previous time frame in the music and start over from there and completely waste generation time. This still seems like a lower Beta program, I do have faith with continued Training Suno will be great, I just hope you treat the people who are using it with dignity so they will choose to continue doing so.

[ruby stag]

^^ Similar to this. On delete, get at least 1 credit back. Easy.

[sage mirage]

It still costs the same amount of (expensive) GPU cycles to summon up any generation

The idea is nice but we're not in the realm of if we like it or not, more like fuel burned

You aren't going to get very far trying to get your fuel back if you didn't like the destination after you drove there

You burned it, it's gone

Suno isn't getting that GPU time back

You don't get refunds with MidJourney either

[ruby stag]

This is for suggestions, not giving your (frankly wrong and flawed) opinions on others suggestions.

[sage mirage]

I'm just explaining to you why that is a big ask

Frankly, flawed, etc

If you understand why how Suno (or any generative LLM) works, it should give you a better undserstanding of what you are actually asking for

[rapid siren]

I'd like a song search and maybe a way to follow people on suno.

[fervent idol]

this is something i'd also like to know. As a musician and artist myself, i want to make sure that the training data used was fairly collected via either royalty-free, commisioned, or properly licensed.
does Suno have any documentation about where they've gotten the data from, and if they've gotten consent to use it?

[rocky jay]

Suggestion, at times when you hear a clip, you like something in the middle and want to disregard the beginning and end. Would be great to be able to extend based on a range. So for example Start: 00:22 End: 00:50

[unkempt plank]

I think the site we access from Suno.com Help (https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448) is too old. If it is the official help, I think they should be properly maintained. Currently there is (probably) an unofficial suno.wiki, but I think the official one should publish official documentation on official site.
Or , if the suno.wiki is official site, I think they should be change help link on suno.com.

[abstract garden]

MY BIGGEST HURDLES WITH SUNO
**- I keep hitting the limit for lyrics entered

• Can't redo early parts of the song and keep the rest**

[cold remnant]

Without reading thru all suggestions so far. I'm not sure if it's even possible but
Able to enter inside for the model to recognise where to put in drum or guitar solo's as well as adding in things like back vocals in Chorus etc entered as ( Chorus: female backup vocals )
Then extending from that for eg (verse 1: male voice hiphop )
( Verse 2: female voice raping )
Or even to the point of increasing the tempo or style mid song

[cold remnant]

Yeah same, it should allow to enter and the over limit to be in a different colour perhaps then atleast one can highlight and cut/copy to place in ' extended ' part to make the song full

[sage mirage]

You guys aren't trying to put your entire song in each continuation, are you?

Each section you generate you only put in the lyrics you want in that section

When you 'Get Full Song' you'll get the entire song with all the lyrics

[abstract garden]

My rap songs speed through the lyrics and at the 1:00 minute mark it just does instrumental because I couldn't enter more lyrics

[sage mirage]

Ah I see

Then yeah that is a problem

[twilit arrow]

When I free plan you gave me 50 credit everyday

But when I pricing plan not

[acoustic pike]

Music not generate the magic 😵
What happening?? I don't like the result.
Ah! Suno's new music style is not good for me. It's like, all are same.

[acoustic pike]

Yes! It's complete plan. 😂

[visual brook]

could we get more updates as v4 gets trained etc ? loss curves or anything ?

[swift pond]

I would like to have an option to edit lyrics while keeping the same voice, music and tempo. Often I have to generate dozens of songs until I find a gem, but if just one word is badly pronounced, it's all lost. I would like to be able to edit the lyrics to spell the word differently (phonetically) and regenerate the same song... For me this is a must have feature... So first I generate song to get good voice, music tempo and then I can tweak lyrics to make it perfect...

[wicked karma]

there should be something where you can insert a mp3 file and it can edit it for you. that would be cool.

[hearty star]

I would like to be able to choose v3alpha in the list of models; the newest iteration is not producing what I have come to expect.

[proud shadow]

Longer song duration

[dusty river]

you can make songs over 5 minutes if you wanted but lose quality

i would like to suggest some form of search capability so that we can listen and enjoy others fantastic work.

[calm sonnet]

generating songs from midi files would be ideal

[tidal isle]

A prompt [xyz] that suno should use only the lyrics exactly as I write them without any extras

[steel latch]

make intro or interlude in proper way please

[north dune]

After burning through over 20k credits since V3 released, I still feel like V3 Alpha was better. Can we bring this back? What exactly changed since Alpha?

[tidal isle]

A fix prompt [xyz] that sonu do exactly the same Chorus as before

[weak bloom]

modifying things in the waveform, dropping sound effects and clips in and the AI incorporates them best it can, i see this like giving midjourney image inputs and region editing, same concept different beast

[hazy hornet]

Suggestion for feature
Feature called "Remix Sample/Audio"
Where the Song Takes a Sample Or Audio from Whatever audio you want in other language remixing a existing Song with More Intense or Cool Sound for Example, They can Be a Cover or instrumental music with different tone
After you apply the sample you can generate the song with any Theme you want and the song will Created with The Sample of Song you Putted on
Also if you add a sample or audio of already existed song like in the internet the song that you Generate will have same melody or Similiar melody if you decide to remix the song or make a different music with the referenced Instrumental in some part
Sorry if i have bad english but hope you can understand this and add this feature if the dev want to

[empty geyser]

shit, why does the sound seem more fibrous? Synth sound burst, like compressed but scattered

[grim carbon]

seed prompt

so it have a same exact song with exact melody but with different lyrics

[stable spindle]

negative prompt would be nice

[twin bear]

Create genre specific channels

[ripe horizon]

you gotta improve on Metatags understanding, i put [Male Rapper] and is a girl singing eurodance, seems like its stroggeling by changing a style on a song. 1990s eurodance involved rap and eurodance

the rythm is exact, but it miss this touch of ... lets say, Real mccoy and ace of base

[shadow crater]

We need a blacklist, this stupid reverberate in the music destroyed every second part in electronic music. This part are so beautiful and its complete wasted with this disturbing noises. Please fix this or give us tools by hand to forbidden this.

EVERY SECOND song destroyed by over echoed.

[plucky wave]

Would be nice to have a search prompt to find examples of music style: search: Afro Pop 👉 get examples of style Afro Pop like in the Explore section. Or you get a message like “Not Available”

[plucky wave]

A search prompt in the Explore section.

[hybrid crane]

Good suggestion. In the mean time I found a solution and showed it here: https://youtu.be/kB3W5xLza_k?si=lF2sPwlQGIHYWanc

[stuck steppe]

Give the possibility to rewrite/adjust the lyrics to final songs maybe? It becomes a mess if you start extending like a mad.
Or add a second lyrics display like "Final Lyrics" that you can modify whenever you want

[potent crane]

I have a 36 minute song, that long enough?

[thin kettle]

should be good to have the possibility to build a library or a project folder to make some "order" to the songs, or album something like that

[forest basin]

Can something be done to fix cover art that doesn't match the song? I just made a song about Jesse Owens in the 1936 Olympics and the cover art that came up is a cycler instead of track athlete.

[weak pendant]

love suno the next thing you can add in suno is automatically making section of like English song, Japanese, hindi because i generate many different language long it would be nice if they build a section were you can see generation based on language

[mighty sun]

Kiedy powróci możliwość generowania piosenek przez discorda suno ai

[rocky jay]

Under "Extended From..." can we please provide the time from where the clip was extended from?

[coarse hound]

The Voices ( Male and Female ) can be better.. it sounds very compressed

[cold remnant]

Deeper understanding of sound effects for example reving f1 engine, didgeridoo, directing it to do bpm has various results too,
Custom/upload own cover art

[wind crest]

yeah more sound effects training would be outstanding. i wanted to put a studio laugh track in a song but the best it can do for now is just a regular laugh. training on every possible sound effect is a pretty broad scope though haha

[plucky wave]

Next step: AI Music Videos 😉

[cold remnant]

It doesn't do live crowd very well either.. for example capturing the buzz and atmosphere of a concert/rave like tomorrowland where the crowd sings with Chorus or claps at the beat etc .
Early days still this is still the best Ai music gen so far I've been toying with so I hope they can take it to the next level.

[wind crest]

i've gotten live crowds by complete accident a couple of times at the start of generations that didn't specify them, but it refused to add them intentionally haha. tried to put instrumental warm up noises at the front of a track too without luck. yeah suno is outstanding, absolutely no complaints from me even with the current quality issues with the vocals, and at this level of feature request we're really asking them to gild the lily

[plucky wave]

Maybe a “Play sample ▶️” button in the Style of music section?

[plucky wave]

I know you want to earn money and the more credits I waste the better for you.
But really I would like to hear a sample before I create a song.
I don’t like to buy the pig in a poke 🤓

[real night]

Would really hope subscription accept another method, like via Google Play. Coz I got difficulties to pay from Indonesia since even MasterCard debit card is also declined. IDK it comes from my country regulation or what

[nova raft]

I'd like to suggest adding more musical styles like traditional Japanese music with instruments like the shamisen, as well as incorporating diverse international genres. For example, unique singing styles from Mongolia and theatrical styles like opera. Thank you.

[kindred prawn]

I'm confused on how to command a fadeout .. sometimes I'm very successful, other times I spend a billion credits to try finish the track but it just starts another drop and cuts off after a minute 🙈 maybe a "wrap it up" button on the extensions?

and/or a crop tool, haven't found one. but feel bad if a nice track ends abruptly 13 seconds after a potential finish

[flat sphinx]

negative promts......

[wicked hound]

I would love to get checkboxes instead of each song being highlighted when you are selecting multiple songs. I keep accidentally adding multiple songs to playlists, especially on mobile.

[delicate cedar]

My friend let me tell you about Noisee....which is just that... 😄

https://discord.gg/ZUcZVPsqwN

Of course just like with Suno, the more you put into it the better your results will be...

[cunning pecan]

Midi files. I would love to take songs to another stage.

[jade pendant]

splitting the song into stems would be great! that would allow greater flexibility for editing and give more control to the producers

[sage mirage]

You can do this already using various tools. Suno will probably have this ability eventually, maybe even soon

But it's going to be the Suno-> Master -> something else -> stems path

Suno doesn't use stems at all

It outputs the master directly

[jade pendant]

i know, but those tools are not that great tbh .. and if you get the ability at source, i'm sure it would have better results

[sage mirage]

Most likey. One of the devs specifically mentioned they aren't fans of having people use outside tools

But that they don't have it to where they want it quality wise to offer it themselves

Which leads me to beleive that they will offer a stems output. They also might be doing all sorts of stuff that they don't talk about

[jade pendant]

also a group feature in library would be great .. right now one can create playlists to work around that, but library tends to get very cluttered very quickly

[sage mirage]

I am positive that they are pretty social-focused

Just today we got stealth-updated with the ability to follow users

[jade pendant]

i meant grouping song parts into a "folder"

it makes it easier to stay organized

[sage mirage]

Yeah

That stuff comes with it though

They also read these discords, especially this one and #🪲┃bugs-web

[jagged fable]

A sliding scale to change the model's stylization bias for each output is greatly needed. Even if it makes a song less conventionally "good", there needs to be a way to make the model more coherent and listen better when it comes to atypical song formats.

[urban sleet]

Please stop making UI worse and worse. Bugs where page just scrolls by itself, missing menu options in certain places, failing to copy all the data with "reuse prompt" (missing reuse prompt in menu) and now I even see there's missing "extended from..." feature from withing the song page..
Why are you doing all that?

[noble yoke]

Hello, I want to download Hi Quality WAV file from voice and stems from a generated songs, I can pay a new plan to get this files, thank you!

[sage mirage]

Suno doesn't use stems, so there are no stems to give you. It outputs the master directly, as-is. Likewise, they could offer a .wav file but it would sound exactly as the .mp3 you currently get, because that's the quality Suno outputs which is driven by the model itself. Presumably in v4 the quality will go up as classically this is how the progression from v1 -> v2 -> v3 has went. The quality is what Suno outputs, it has nothing to do with the file format.
For Stems, you have to use something that breaks them out from the master. There are lots of options already.

[cold remnant]

The general jist I get from many of the suggestions is that mostly wanting greater input/power over the production of the music sounds/voices/tempo/solos .. timings, bass drops and etc.
Other options I'd like to see is to select how long a song will take as it is 2 minutes for 10 credits is reasonable but what if I just Want to fo a whole song,/tune for 4 minutes minutes it should let that be an option but pay more credits. Saves the ' extending it '
Also the explore screen needs to have a button or link so the discovered style can be used immediately by forwarding g to the full create screen with the style already in prompt box

[tardy coyote]

What if you were able to take already created songs with lyrics and make instrument versions to make people sing then on their own or make a record button?

[urban sleet]

There are AI's doing that. Free ones too. (I don't know the names, but it's an easy search)

[stable spindle]

v2 should cost less credits imo

[sage mirage]

I think one of the devs mentioned that v2 actually uses as much or more resources than v3

But I'm probably very wrong

[fathom acorn]

Hello

it would be nice if we could choose that generates instruments and vocals in 2 separate tracks so we can master and edit those

(this is my most common wish every time when i generate something good)

[sinful anchor]

You should add a hidden [Please] tag to each user request.
Neural networks get tired. They need support reinforcement.
Imagine that you work hard and a lot, but your work is not paid.
Unsurprisingly, the quality of the songs gets worse over time.

[kind mirage]

inpainting!

[sage mirage]

They call in 'infill' and maybe with v4

georg, one of the devs, has talked about it a few times

[kind mirage]

Looking forward to it, that would be epic. Is there a timeline on v4?

[sage mirage]

~soon~

If it's like any of the releases before it will just show up one day

[solar drift]

Could you add something like a section underneath style of music for the instruments

[stiff canyon]

Perhaps it is a good consideration to create an additional form field for the title image when generating new songs in custom mode. 🙏

[wanton pilot]

Will it be possible to download individual audio STEMS (voice, drums, instruments)? Through separation by third-party methods, the quality is sometimes average

[robust jewel]

Soon™ is the only thing we get until they are very sure of a release timeframe, then we suddenly get a "this week, probably" or something like that. I actually prefer it that way, so we don't get a hopeful date that has to get changed as things need more work.

[pliant rose]

Please allow people to retroactively buy the rights to songs generated with the free account 🙏

[prime stirrup]

Folder structure in create would be a life saver. I have so many great ideas but get lost fast in my process, ADHD isn't helping much with organizing everything in one folder.

[cold remnant]

If you mean remixing already commercially available songs by actual artists I think it will go against the terms and policy .

[weak fern]

Recently, I utilized Suno to craft an instrumental piece. However, I noticed there were numerous instances of noise within the composition. Specifically, while the bamboo flute melody is crisp and clear, the accompanying guitar or piano seems somewhat obscured and resembles noise rather than a discernible accompaniment. I'm wondering if there's a way to enhance the clarity of these accompaniments. Any guidance on this matter would be greatly appreciated.

[rose ice]

Has there already been a suggestion or, is it already in the works? Stitching/Merging an extended track of a current project and a new track if you decided to take the song in a different because the extended track kept disregarding song prompts? Also, would it be possible to offer custom background tracks like Foley tracks to support podcasts and ttrpgs streams?

[pearl ice]

It would be cool if you could run singular lines back through the AI but maintain the rest of the song. Where you could click refresh to have AI try again on a lyric but look forward and backward in the song to keep similar structure. You could also have a "like" button where you're telling AI that area of the song is perfect so it knows to use that structure again if we continue forward. Also a button where you could rewrite the entire lyric in that spot and have AI re-sing that spot.

[tepid trench]

The "in-painting" feature likely coming in V4 should help with this (should let you regenerate part of the song)

[pearl ice]

https://tenor.com/view/lets-go-gif-drake-gif-26307972

[unborn marten]

No idea if devs read this or not, but do we really need the Trending stuff to be forced on us in the right hand side of the webpage when we're viewing our song with a view to extending it or whatever? We already have the Explore option for people who are into that.

I'm very positive about being a Suno subscriber but I hope the GUI won't be ruined by forcing 'social' functions on us all like Spotify have done with their horrible updates.

Rant over 😄

[teal barn]

I know that this suggestion is nothing new but I think it's really important that there will eventually be a "download stems" feature, or atleast "download vocals".
Stem seperators exist but the quality isn't the best. I think this will make it much more accessable for creative people that really want to make the songs their own, similar to how you'd do remixes of songs.
Then you could mix & match stems, add your own melodies etc much more easily and with higher quality results.

I think this would be the only way Suno would really be accepted by musicians.
Suno creates inspiration and ideas, the musician adds the soul.

Just my 2c 😁

[leaden flame]

Suno Feature Suggestion: Really need the ability to create folders on the Create screen so we can organize our music creation projects into their own folders and subfolders. Currently with all 'Create' screen songs and extensions appearing all at the same level, it very quickly becomes quite a mess to navigate through and to find specific parts of songs. Automatic creation of a new Song folder every time you create a new song for the song parts and extensions for that song to automatically be placed in when generated would already be very helpful for keeping things much more organized and under control for song creation.

[still parcel]

We hear you, and we’re working on improving this feature based on the community’s feedback

[cold remnant]

How about being able to add own custom album/playlist/single image?

[sage mirage]

Still would be excellent to be able to type a number here

Since I still have to click those arrows a whole lot to get back to stuff I made months ago

Before playlists were a thing

[tepid trench]

You know the cheat/workaround for this, right?
(Press button, hold down space or enter)

Pagination go brrrrrrrrrrrr

[sage mirage]

More like br...br....br....br....br

[tepid trench]

I would like to thank a heavy book for this discovery.

[sage mirage]

oh

Space is slow

but enter IS brrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr

Ok yeah while typing a number would be a nicer way, that suggestion now goes way in the back for 'things I would like'

[tepid trench]

Being able to type a number is needed really, this is kinda obscure xD

[hot cradle]

Hello. Batch download is on the list of upgrades, perhaps?

[rapid siren]

Maybe add a thing to tell us which songs were made under subscriber and which were not.

[dim snow]

I'd like to be able to download a song, edit it locally with my own tools (cut out the parts that are bad, mostly), upload it back to suno, then continue generation from the place it left off, with a new prompt.

[sage mirage]

We are probably never going to get uploading to Suno like that. There is no way for them to know if you are uploading music you own or copyrighted to someone else music

Audio is audio is audio

Legally a huge minefield

[hybrid crane]

https://tenor.com/view/minesweeper-mine-sweeper-fast-speedrun-gif-24760449

[sage mirage]

Except you explode before you can even play

[worn atlas]

Unrequested or non-essential features of Suno such as 'following' appear first, while highly demanded features like 'search" are still missing, making the suno web interface a joyful mess.

[sage mirage]

You may have a skewed view on what people were asking for

Since from my POV 'following' was always requested a ton

[worn atlas]

More than 'search'?
To have a better view on what is asked suggestion should use the discord forum style

[sage mirage]

I'd say so, but they can do more than one thing at a time. Search is probably a heavier load than following.

[worn atlas]

All developers know that and i am a développer myself. It is matter of priority.
Some features are 'Nice to have' and others are 'must have'.

[night agate]

suggestion: avoid chat on this channel, set CD for 12 hours and let us send real feedback

[frosty breach]

When holding shift, it would be neat if all links become unclickable. Was doing some cleaning and opened several extra windows.

[deep monolith]

I would like to suggest Suno Library have a Search box, so that we can search our songs by name. I think it would be very helpful to a lot of people.

[deep monolith]

I would like to suggest a learners corner where Suno videos show new people how to add the lyrics section by section to the different parts of their songs. Also it would be a good place to put up other videos showing how to use Suno. I think with each new version things will get more complicated and people will need how to videos.

[soft harness]

You mean something like this that is pinned in General?

https://youtu.be/lj8utBd_IUM

[deep monolith]

Yes exactly but more advanced in details. I want to watch videos made by the Suno guys to guide me and new people. I think version 4 will be more complex and people will need specific videos to help them.

[sage mirage]

I would honeslty take user's advice over the Suno devs at this point

and I suspect they might even agree

I don't speak for them though

But building and using are different beasts

Which is why the 'Home' screen is all users

And if it's anything, 3Daizy is a moderater and is closer to the devs than us users

[deep monolith]

I would like to echo or repeat what some users are already asking for. I want Suno to give me high quality waves files when downloading a full song. I would gladly pay a higher price to get wave files. I don’t need them to be substantially higher or too big but higher…also I would pay for a stems feature when downloading a full song.

[sage mirage]

There are no higher quality to give. The sound quality isn't a file format thing, it's a model thing

Suno also doesn't use stems, at all

It outputs the master only

You can get stems breaking it apart afterwards though, and if Suno offers Stems in the future it will most likely be an internal version of that

Master -> stems

If they offered .wav you'd have the exact same quality

Just a larger file

The fidelity just isn't there 😭

However every model version has jumped in quality

v2 we got stereo

v1 was mono

Maybe v4 will get to the point where the file format does effect quality, we'll see

[waxen fox]

When I produce the song, it automatically gives it as 2 minutes. I want to do this in 2:30 minutes. how will I do?
When you shoot something for more than 2 minutes, the button turns red and gives an error.

[soft harness]

https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

[grim geyser]

negative prompt? it takes up so many of my credits going through examples that aren't working.

[sage mirage]

Not too long ago one of the devs was asking if we'd like negative prompts even if it cost more credits

Those of us there at the time all said "yes"

then the dev said "interesting" and posted the Homer-Fade-in-the-Hedge.gif

So to your point Trance, yeah, absolutely I'd like it even if it did cost more because as you say, if I could negative prompt I might still come out on top in regards to cost

[visual brook]

negative prompt should be able to be Direct Preference Optimization (training method) into the current transformer model. once fine-tuned or trained in that way negative prompt would have no extra cost (beyond the admittedly significant Fine Tune data collection and training itself)

LLMs work that way, Bark would be easy to proof of concept on too, maybe i try that sometime

[grim geyser]

but I just get a lot of random stuff I don't want.

[sage mirage]

According to the devs it doesn't work like that

But I'm no LLM-ologist

Same reason why we don't have seeds

[grim geyser]

Reee! lol

[visual brook]

have you ever used ChatGPT and has it ever also answered stupidly? because this will always be a thing for these models

[sage mirage]

I know, but Suno has a temporal aspect to it

Like video

Like I said though, the technical stuff is way beyond my grade

[grim geyser]

also some of us are actual musicians and when we use terms that are specific to certain concepts the AI ignores that.

[sage mirage]

Yeah

People asking for stems or .wav

Suno doesn't work like that because lmao LLMs

[grim geyser]

Riffusion does stems

[sage mirage]

and Suno does not. It's master only

[grim geyser]

it's jsut that it can only do 11 seconds of audio at a time

lol

[visual brook]

ah yeah, it's not trained on enough music theory dataset for sure

#💡┃suggestions-web message had this suggestion back April 23rd

[sage mirage]

Again, the devs when they do talk (which they do and I assume I reading this conversation right at this moment) the training data and how the model is built is both why there aren't stems nor .wav

[grim geyser]

Id like to say Minor key and not get peppy 4 chord pop music

[sage mirage]

the .wav is simply training data. The quality is unrelated to file format

[grim geyser]

then Suno pulls off miracles now and then though

[sage mirage]

Yeah

It's just a fundamentally different musical tool than what (I assume) most actual musicians and producers are used to

Like getting a demo tape and being asked to 'produce' that

No more data

[grim geyser]

would also love to tell the vocal synthesis where the rests are so I don't get clunky musical execution

[visual brook]

v3 si more like a plain-english-to-music model, and music theory is a special domain of language almost like math with very specific sub-domain meanings

[grim geyser]

so if I try v2 is it better for that?

[sage mirage]

No

[grim geyser]

shoot

[sage mirage]

It's just how Suno is itself

I mean you can give it a try

Same with V3 you can plug musical terms into it and it will do something

It's just jiggly, fuzzy and unpredictable

[visual brook]

no, all the music datasets they trained on and have likely (until v4+v5++) are plain language and not music theory

[sage mirage]

v4 is ~soon~

always soon

[grim geyser]

also could be as simple as that they don't have a musician on the team who can write language sets for those terms

[sage mirage]

The feature with v4 that's been talked up is inpainting/infill

[sage mirage]

They are mostly all musicians afaik

It's data

You need so much

so so much

Though winds have shifted in the LLM space and now it's less 'more' and more 'quality'

Keep in mind I'm still speculating. Besides the infill part, they've talked directly about that being a part of v4 probably

[grim geyser]

well long ago in computer science circles according to Mike Ferarri there were no artists around when the computer scientists came up with EGA color.. so I thought it might be something similar

[visual brook]

it took plain text language a semi-equivalent 15 trillion words to begin to reason close to humans (meta llama3 et al)

i doubt there will be enough music theory data to be passable for music-theory-text-to-music for a while - i haven't seen a dataset on huggingface like that

[grim geyser]

also I think that using public domain music examples could help a lot

[visual brook]

but suno also surprises me, i suspect their datasets are still barely 1 trillion audio tokens and it's awesome so far... so maybe

[grim geyser]

I mean.. using public domain music titles in the prompt could be incredibly helpful

[sage mirage]

I assume they are already in there

[grim geyser]

I've tried.

[sage mirage]

Well Suno isn't trained like that

You have to describe the music itself

[grim geyser]

maybe some kind of token weighting system could help?

if I put the title of Bethoven's 5th in my prompt it just flags it and won't generate anything.

[sage mirage]

Yeah

That's a legal thing

[visual brook]

if you're interested enough to study, you might look at state of the art methods for image production/Large Language Models https://scholar.harvard.edu/binxuw/classes/machine-learning-scratch/materials/stable-diffusion-scratch

stuff like this is a true post-doctoral level of potential study tho

[grim geyser]

it's public dmain though

[sage mirage]

Yeah but asking them to go in and reasearch what is and isn't public domain

Then that becomes a human cost

Much easier to just block it all

[grim geyser]

I see

[sage mirage]

and tell the user 'figure it out'

Technically Suno is probably capable of all sorts of wild stuff

Legally though: lol and also lmao

[visual brook]

you can use public domain like "Holst Mars" to influence your gens tho

[grim geyser]

chat gpt can tell you what is and isn't though

[sage mirage]

Never trust LLMs

ever

[visual brook]

i do that sometimes, Holst inspired and sounds like John Williams Star Wars

[sage mirage]

I mean ha ha and everything but I'm serious

ChatGPT wants to give you an answer

no matter if its true or not

[grim geyser]

I was trying to get that anger rap from the late 90's and it jsut won't do it. It did it once and I just can't replicate it

[visual brook]

LLMs are a nice data point to add to grounded truth from elsewhere

everything is a multi-variable analysis

[sage mirage]

LLMs are fantastic in addition to

[tepid trench]

Riffusion pretty much just incorporates a stem-splitter AI (doesn't use true stems afaik - doesn't sound like it anyway.) It would be nice to at least have that convenience here.

[visual brook]

add "emotional anger 90s rap" to your prompt (no commas)

[grim geyser]

I understand that. it's neat to see it was integrated

[sage mirage]

tony (who I haven't seen around lately hope they are ok) mentioned they want to internalize stem splitting

[visual brook]

use more adjectives 🙂 instead of seperate comma nouns

[sage mirage]

But it isn't high enough quality yet to apply 'Suno' to it

[tepid trench]

Yeah I just swooped in on this channel to go "ackshually...", don't mind me 😅

[grim geyser]

see the only reason it's an issue is that I pooched the lyrics, and so now the music is perfect but the phrasing is clunky due to some typo

[sage mirage]

infill will be your solution, probably

Maybe

We'll see how it works but ~in my mind~ it's pretty much THE tool I want more than anything

before this it was 'select when to continue from'

As that hasn't been possible the whole time

[deep monolith]

Try...."Rap Dirty South" as a style

[sage mirage]

"Atlanta" also seems to have an effect

Think more vibes and less music theory

[grim geyser]

I'd like to be able to put it in terms of beats bars and ticks like an SMPTE timestamp, minutes and seconds are useless in music unless your music is fixed at 60 beats per minute

SMPTE

lol

also MP3 encoding always puts so much silence at the start

that way I could slice in based on what the AI thinks the tempo is

one of the other things I noticed is the spring reverb effect seems to come from a short burst of noise.

@deep monolith this is the prompt I used, but it won't make the same sound I'm targeting.
I left you an example in the showcase channel. also something for you too @sage mirage

[deep monolith]

I got a song titled Dirty South that I placed in the public domain it was so good, it’s only that I specified Rap and it gave me Country.

[leaden marlin]

2 hours later after rendering like 30 songs in a row that are not sounding at all like what im looking for i would like to suggest some sort of "analysis link suggestion promp" a way Suno can identify what "ballpark" of sound your trying to create.

so in other words.. "a way to show Suno what your trying to create"

[ionic skiff]

i dont know if this was suggested before, but .. lyrics extender in Suno would be awesome!

[digital knoll]

please change the censorship of lyrics guardrails, it's flagging lyrics and i cannot find what it is flagging.

Either you do em right there or at em in the head, son <----it flagged this for my rap song

[robust jewel]

It's probably considering "at em in the head" as violence and may be getting overly cautions because of the "son" being added on. If it thinks that's talking about violence involving children, it will block it pretty quickly. The censorship is basically just ChatGPT refusing to deal with a lyric for whatever reason.

[leaden flame]

Suggestion for addition to the Suno Create page:
One of the biggest sources of confusion I have been seeing for many Suno newcomers appears to be newcomers to Suno not being made aware that the use of meta-tags in the 'Lyrics' text entry box and all keywords entered in the 'Style of Music' text entry box are taken as "suggestions" only by the Suno model. At least currently, there is no guarantee that Suno will closely or even partially follow what is entered as meta-tags in the 'Lyrics' box and in keywords which are entered in the 'Style of Music' box. Nor is turning on the 'Instrumental' button a guarantee that Suno won't try to generate some vocals in the song part it generates. Likewise, Suno can sometimes also ignore some or all of the lyrics a person enters in the Lyrics box, or otherwise mangle them in some way.

You could probably spare new Suno users a lot of confusion and frustration if you placed a menu item or link on the Suno Create page in a prominent location labelled something like "Important Note on Using Suno", which would then pop up a window or similar that clearly explains to the effect of what I have written above. I know it took me a few days of using Suno to start figuring that out on my own, even though I had skimmed through the Suno FAQ shortly after I sigend up. At that time, at least, this important information was just not stated clearly anywhere that I could see. A fact is many may not read through an entire FAQ/Help page unless maybe if they are completely stuck, but many probably would click on a button or menu item on the Suno Create page labelled "Important Note on Using Suno". I know it would have been a big help for me when I first started using Suno about a month ago. 🙂

[digital knoll]

thanks.

[formal iris]

¯_(ツ)_/¯

[royal venture]

A suno showcase for instrumentals, so people can choose whether or not they want to listen to one. You cant tell before hand if a song is instrumental or not.

[ashen ridge]

show us what words are censored in our songs

[tepid trench]

I always use custom mode, and usually want an instrumental. Other people may have other preferences. Would be good to be able to set a default (in profile settings maybe?) for when logging in, and then have it use whatever settings you last used each time you open 'Create'.

[worn lotus]

Is there any estimate on when V4 will be out? 🙂

[drowsy dove]

idk but V5 bouta go CRAZY

[sage mirage]

"soon"

[drowsy dove]

forgot the TM --> SoonTM

[robust jewel]

Soon™

[winter agate]

i use chatgpt foor lrics

[latent flare]

how can i get refund?

[soft harness]

(Question already addressed in #🪲┃bugs-web message )

[pseudo surge]

Will we ever be able to separate the songs by stems in the future? Or do more like editing to the songs inside Suno? I know with a lot of creations I have generated I get this fuzzy static sound that makes the song sound so bad. Just would like to be able to get rid of this or get stems or something of the sort.

[sage mirage]

You have to break it apart from the master using other tools. If Suno offers stems in the future it will be this way as well since Suno itself doesn't work with Stems, it only deals in masters

and usually because it's the other question that comes up, output quality is also a model thing and not a file format thing

There is no point for .wav because the quailty would be the same. You'd just have a bigger file

[pseudo surge]

I haven't found a tool yet that does it well enough in my opinion though.

[sage mirage]

that's what one of the devs was talking about why Suno doesn't offer it yet either

because they can't get it to a place they want it to be quality-wise

[robust jewel]

I expect the AI to get better at not having noise. Cleaner audio being a clear focus from the beginning.

The way the AI currently functions, there aren't stems to get. It generates the complete piece.

[pseudo surge]

I hope it happens soon. I know there are other new AIs coming out like Udio that seem to have cleaner generations in some ways but I still like Suno better

[robust jewel]

@fringe knoll @feral aspen

[high shoal]

audio input, to create similar audio

[quasi nest]

Instead of being able to import choice audio, let us use existing songs on Suno as reference points. As being able to import your own audio for reference would result in people using copyrighted audio to generate a song.

Additionally, members should be able to toggle if their songs can be referenced, as per Suno's policy those belong to the members.

[sly furnace]

Any news, when v4 beta with improved sound-quality will arrive? I really would like to come back to Suno.

Some teasers would be nice

[lament carbon]

Would be cool if it was possible to get the chords to a song within the suno platform 💜

[robust jewel]

Soon™

That's the answer even if literally nothing has been said about it, honestly. We got a clue that V3 was about to happen about five days before it released. They don't set a timeframe very far in advance.

[drowsy dove]

the pizza is done when it looks golden brown, until then we wait ❤️

[sly blade]

Will outpainting/inpainting also include uploading your own files? for example if i have a piano melody or song and want the ai to finish/improve it.

Like image2image -> song2song?

this would be awesome

[plush lily]

iwould love it if we could keep parts of a song. ive made so many songs where i only like a part of it and not the rest, so it could be cool if we could rewrite (or something else) the rest that we dont like but still keep the part we like (i have part i like but its bc of how its sung, not the lyrics)

[drowsy dove]

extend from x:xx, but yeah doesnt work if you dont like the very start

[plush lily]

yeah the part i like is the outro of the song unfortunately. thats why i thought id suggest that 🙂

[tepid trench]

Would be nice but I think it's unlikely. People will drop copyrighted material in there. So they'd need to be able to detect that.

[devout haven]

hi, where can i generate?

[hybrid crane]

On the website: https://suno.com

[simple delta]

Id love to be able to upload self recorded voice tracks and have suno create a song around it.

[sage mirage]

Almost certain to never happen because of the legal issues around that

You'd upload self recorded tracks, someone else would upload Michael Jackson

[thin hinge]

I don't know if it exists, but having music like [intro music] and [outro music] for the beginning or end of the song would be great.

[quaint valley]

Need a way to spitfire rap like Eminem and Busta Rhymes.

[simple delta]

could we get SRT files soon?

[neon wren]

there should be a way to create full song without using all of the last part (something similar to the "extend from" input but instead "extend up to")

[robust jewel]

You can try [Intro] and [Outro], but it often ignores those at the moment.

[robust jewel]

An "end at" and a "fade at" would be great.

[neon wren]

would be great if we could use a separate prompt for the album art instead of having it determined by the style of the last clip (which is often something non-representative of the song like "end")

[desert condor]

They should allow you to add lyrics to the instrumentals you create

[leaden marlin]

I want to edit the order of playlists

[ionic skiff]

empty trash, if this option is not already there

[ionic skiff]

that's good, BUT - it would be better if Suno could keep generating lyrics itself!

[glass heath]

I would love to get MIDI generation from the tracks. I feel like that would be easier than stems, and allow for lots of adjustment and fine-tuning...

[soft meteor]

nice idea but that's harder than stems as suno is generating audio directly so it can't create midi for the different instruments

[quaint depot]

Why on earth is the optional prompt size for extending a song shorter than creating it?

...and why does it not default use the same prompt?

...and why is it not the same instrumental setting if you've already chosen that?

Really seems like put the burden on the user UI engineering.

[robust jewel]

The extension length is shorter than the first clip, so the prompt box is smaller.

[outer swan]

i mean i was joking but it could be nice actually

[craggy moon]

I would love to be able to get (at least close) to the same voice so that I can make albums.

[solar cliff]

The ability to edit the shown lyrics would be fantastic, especially when it skips a section of lyrics in the song, makes the end result look a little messy when there are a few lyric lines missing!

Currently got a missing part that says [OUTRO] in the lyrics, lol, but it is definitely OUT of the song completely, lol

[sage mirage]

and give the option to show what the actual generation is if you'd like

Since I suspect the whole reason they show everyhting as-is is for maxmimum sharing

It presents bad though

[solar cliff]

EXACTLY my next though, have edit'd with an "original lyrics' option

[sage mirage]

Yeah

Since I'll snip 15~ seconds from a totally unreleated song but the box shows that in its entirety

[solar cliff]

Full disclosure is always nice

[sage mirage]

Even though it isn't in the song anymore

[cold remnant]

This is more of a suggestion for the Discord server.
The showcase channel is effectively just a spam link dump room Rn.
So I suggest categorised genre channels for people to post their works in as well as a ' connect ' # where profile links can be put up and connect with others with similar interests.

And one for the suno Web app is to allow Hashtags so music can be discovered that way .

[quaint depot]

That does not explain it.

[soft harness]

From what I remember others explaining it, when the AI extends, a minute of the processing time is spent analyzing the base generation like ~1 min from before the specified continuation point to get the style of the song as the AI does not retain any of the information it used to generate the initial track (e.g. seed, etc.).

Thus a minute of the original two minutes is already used, leaving the other minute for the extension generation.

If it the continuation point (Extend From) is short, it can be very different. Changing by a few seconds may be enough to change the tune to what you wanted (or to another particular part such as the start or end of a verse)

See https://www.suno.wiki/faq/making-music/extend-from-time/

[meager beacon]

an AI generated midi file would be cool as an option for downloading the generated song.

[formal cove]

It's really disheartening when I create a long musical style prompt that puts out an amazing hybrid genre, only to have suno retag the musical style. For example, "viking, ethereal, opera, rave, darkjazz" made a really amazing track and then replaced all those with generic tags. Sometimes I remember my exact wording, and sometimes I don't. And when I don't, it's frustrating AF and almost a deal breaker. It's so bad, that I screenshot before hitting create. Leave our original prompt alone! If I hadn't already paid a year in advance, I'd already be using Udio instead.

[untold shuttle]

I always write down my prompts in a notebook before I add it to Suno…

[lament mantle]

Im Loving This..!! even just on the web browser for a couple days, this has brought
tons of inspiration..! as a live instrument player with never-ending writers block...
i can't wait to see what this can Really do... Suggestion, when generating on-line,
give both fresh generations, an A & B label... and a way to chose what gen.s in order
are used for the Get Full Song option... Killer Stuff...!! Totally Rad..! Super Cool...

[lament mantle]

open an email to yourself... copy prompt, and paste onto open page...
I lost a few of my first ones...

[vestal swift]

does suno know suona？

[quaint geyser]

Never have this feature, it will ruin SUNO and other AI music platforms

[covert oar]

I've been using Suno's top-tier plan for eight months, and during that time, I haven't received a single thank-you email. Despite creating hundreds of songs, none have been showcased on the platform. Even after a competitor like Udio, which is free to use, entered the scene and pushed out major updates within just three weeks, Suno remained complacent. The business model is flawed, and communication is lacking. I've observed the Discord chat and noticed that user feedback is largely ignored. It's as if Suno relies on repetition to keep people paying, even while they're dissatisfied. There is no option to downgrade only cancel or renew. I'm done with this.

[livid ruin]

Suggestion: A 3rd paid option for $20 a month. It could have 5k or 7.5k credits. I'm having problems using all of my credits with the $30 plan but the $10 plan has way too little credits. Considering dropping to the $10 and using other free music bots while I wait.

[covert oar]

I feel ya. I still have 9,000 credits and it would have wiped them out in 6 days due to no roll over.

[livid ruin]

It's a "Goldilocks" situation for me

5k-7.5k would be would be 🤌

[robust jewel]

User feedback isn't "largely ignored". Last month, they started a more in-depth feedback process of asking users to fill out a survey. That lead in to doing long form interviews which may have wrapped up by now. They were even paying for the interview via gift cards. Not something you do when you don't care. It's not that anything is being ignored, it's that it can take time to implement things. Especially if it's the kind of features that require a new model to implement.

It also may relate to the fact that the same features get requested a lot, even if they aren't practical requests. The number of requests for stems or for higher quality downloads despite the fact that there aren't stems and the current download option is the highest quality produced... The problem is that a Discord channel may not be the best way to do suggestions, since it's not easy to stop duplicate posts.

[soft meteor]

Yea this is very true, from what I see they slowly work through some suggested features and their own improvements. They are definitely not lazy.
I must however admit that communication could be a bit more frequent when I look at #🚀┃updates and #📢┃announcements
I would suggest to announce small changes there. As nice as it is to be suprised with new features sometimes it would be better to know beforehand that something works differently now 🙂

[storm mantle]

Please train music like audio so it recognizes artists but you change an artists to various styles and please never sell to the music industry! Ai will free artists

[plucky wave]

A thumbs down 👎filter in the Library 📚

[tranquil citrus]

I think one of the biggest things you could do is allow for voice training and to be able to select the voice that you want to be singing from your trained vocals. This would be HUGE!

[uneven lagoon]

I always use custom mode and use my own lyrics. Sometimes I could get a very good melody/music but couldn't use it because of hallucinations. It would be nice to be able to regenerate the same melody until all the hallucinations are corrected.

[drowsy dove]

PLEASE

[drowsy dove]

like a "Regen vocals" option

[ionic skiff]

discriminate between plays of song by yourself and by OTHERs, we need to know the count!

[uneven lagoon]

i heard from the technical discussions chat that they're developing this for v4. It'll be great if this happens soon.

[drowsy dove]

or how many times OTHERS replayed the song would be chefs kiss

[lean pivot]

How to unbind a payment account

[visual brook]

bump #💡┃suggestions-web message

feature request: a simple RSS feed for each artist page

[plucky wave]

Mark/Select tracks in Library and 👎 Filter to move them all into the bin 🚮

[cedar mulch]

also, filter unused parts, anything not extended or used for "get full song"

[ember hull]

wav export for pro

[sage mirage]

Right now there is no reason to offer .wav, the file format does not effect the sound quality

It's a model thing. v3 outputs at this quality currently

If there were .wav you'd just have a large file that sounds exactly the same

[ember hull]

yeah but i prefer wav when i master

[sage mirage]

Likewise there are no stems, Suno doesn't use them. It outputs to master directly

You can always just convert it to .wav yourself then

It isn't like there would be a quality loss

[ember hull]

well then my other one would be making it so that it doesn't limit extensions to one minute

[tepid trench]

Being able to repeat things like choruses, or reuse voices, seems commonly requested.

Perhaps the ability to "reference" part of the same (or even a different) Suno song would be useful, to pull in segments or style? Add it to the existing context when generating/extending.

[floral laurel]

Suno-showcase channel separation of Instrumental and those with lyrics.

[tranquil light]

what if there was an option to bounce off stems from the tracks we like?

[sage mirage]

Suno doesn't use stems, at all. It's master only

You can break the songs apart using other tools though

Not exactly the same

[pale latch]

Ad something so we can edit the lyrics of a song generated

[lunar berry]

Please add search bar in library.

[visual brook]

user story: i want to be able to view the prompt video for a song on Trending home page area - similar to Suno prompt videos which naturally display in Discord when a song is linked here

[strong swan]

Playlist are basically unusable on mobile in landscape mode… the playlist itself is so tiny that it can’t even show one entry. I suggest to change that 😆 i use my phone quite often in landscape mode in my car to listen to music… and it’s super difficult

[wheat peak]

When you make a song and this comes up , “the prompt has been moderated”, it should tell you what’s being moderated so you could fix it.

[olive lily]

It will be awesome if suno had reference song option for example , i wanna create something in style of BURNS - beauty queen , i know its bass house song , but if i type prompt " agressive bass house , electric piano , fast rythm, have main drop and etc.
Its not understanding , so if it have option to reference song and maybe advance settings like " use simillar samples" or "use same speed" and etc ... it will be awesome , but i quess this is just a dream ... or maybe someone can help me make prompt that will make simillar songs and will have main drop ?? Because roght now everything is just repeating..

[ionic coyote]

Would love to see a filter option for full songs or disliked songs or songs without rating. Or a search option so you can search for the song name.

[hearty star]

https://tenor.com/view/fallout-windows98-vault-boy-poor-video-games-gif-18229261

Hey thanks for implementing my previous suggestion about user playlists; that was cool. Now I have a few more suggestions:

First: It would be nice to have a trim function for the beginning and the end of the song. Thus if a generation went too long I could trim the end to where it belongs; same for the song beginning.
Second: The Artwork should have it's very own prompt; I don't like how it just does what it wants based on the style/song name/lyrics. It basically never lines up with the theme of the song as it is.
Third: Regeneration of artwork, using custom prompt, so we can fix the old songs with incorrect images.
Fourth: The mobile phone version of the site, tighten up the distance between elements so we have more room for song list at the top. On my phone in custom I get 0 space for song list, without custom I get 1 or 2.
Fifth: Fix the 'Extend' and 'Reuse' prompts - Reuse clears any previously chosen song time and continuation song selection (frustrating!). Extend should NOT replace the lyrics like it does - it should just put a continuation time and song but not delete work I already did on the lyrics.
Sixth: Switching between custom and normal modes should NOT lose my lyrics, song style, song continuation time and continued song. Currently it's lost on every switch (very frustrating).
Seventh: When extending a song it should NOT replace the song style. So many times I click extend and it overwrites the song style I already typed out...

[sage mirage]

I agree with absolutely everything here 100%

No notes.

[hearty star]

You can run suno in landscape? Wow. Every time I go to landscape I get a white screen with an error message that I didn't bother to write down; some kind of exception. For what it's worth it happens most when playing a song. (Chrome, Android Phone)

[simple jungle]

I can see where this might be abused, but perhaps even having AI looking at the reported issue to reduce potential costs and abuse. There are times Suno will generate something so bad it refunds credits - but other times, you'll have an extension of 9-17 seconds (what happened to the 60 seconds?) of gibberish and you're left out of pocket with that. That's not even taking about the hallucinations, butchered lyrics, and so forth.

[final drift]

exactly

[silver dome]

Can we get shuffle as an option for our playlists please? (If this already exists someone please tell me how I turn it on)

[still parcel]

yes, soon

[tulip merlin]

let me stack random styles

[desert condor]

Create the ability to add lyrics to instrumentals you make

[willow storm]

Change additional credit added amounts to be the same as monthly credit amount.

EG: topping up for $30 only provides 4k credit where a new monthly account for $30 provides 10k.

Why top up when activating another account gives more credit? Even if the credit doesn't expire its pointless buying it over getting another account run for a family member/employee out of the same household/business.

[modest thunder]

Z is so ominus and I love it

[robust jewel]

Top up credits persist, monthly credits do not. You use monthly credits first. If you only need 2k extra credits to get through the month, you would have another 2k credits for the next month. If you got a second paid account, you could get the 2k credits that got you through the month and then the remaining 8k would lapse.

[royal venture]

I use a samsung zfold 3. If im creating songs in landscape, go to my notepad in portrait mode to copy lyrics for next part, then go back to browser in portrait mode, suno will crash. If i try to delete songs in the list past the middle of the screen, the delete button doesn't show because the ui prompt is off screen. If you can fix some of these ui issues i would appreciate it.

[willow storm]

This doesnt matter when your burning 20k+ credits monthly, its cheaper to use multiple accounts.

[robust jewel]

It would matter if you use between 20k and 24k, though. If you're using, say, 23K, then paying for a third account is less practical than two accounts and one top up.

[willow storm]

Less practical but much cheaper, if you needed 23k and only paid with top ups you would be out $150 with some left over song credit.

If you paid for 3 accounts your out $90.

[tulip merlin]

Extend an instrumental and have it start at 00:10 and paste in the lyrics

[robust jewel]

And if you did 2 accounts and 1 top up, you'd be out 90 but have 1000 credits for the next month. Every fourth month, you wouldn't need to spend the extra $30.

[willow storm]

Yeah I get that but we shouldn't have to create multiple accounts for the same value is my point, with what your saying we're still using two accounts and a top up.

Just make top ups match or exceed the account sub value for each tier.

[robust jewel]

There is an option to contact them about custom plans. You could just get in contact with them about paying double for double the credits every month, so you could use a single account.

[clear turret]

Great stuff SUNO, I've been having a blast.
Some Suggestions:

1. Trim Function called out by @hearty star
2. Ability to update/modify lyrics without the alteration of the song
   - Its frustrating generating a great song and then noticing there's a typo in the lyrics.

[solar aurora]

Is this channel worth posting to? Are devs actually reading these?

[digital crown]

Because MRR is important for a SaaS budget. It's better you subscribe

[digital crown]

Yes

[digital crown]

1,- not gonna happen

2 - I agree 🙏

[solar aurora]

Lovely
I've been generating like crazy, and trying to develop an efficient workflow. I wonder if other people would like this.:
I have a lot of half done songs I would like to go back to but they get lost, playlists dont help if you want to continue work.
As new generations get buried further down you main list.

As I might have multiple song ideas on the go, as ideas come and go. I want generations to go in which ever folder/playlist Im currently in.

i.e.
I would like to be able to have current generations go into the folder I have focus on ("current" folder/playlist.)
Then be able to switch (focus) to another existing or new folder and then have new generations automatic go there.
I hope that makes sense

I dont suppose there is any consolidated list of suggestions so far?

[willow storm]

Projected revenue arguments are invalid when you can get more stable projections from having larger plans available instead of having plans fluctuate in and out based on usage.

Suggestion stands: Bump the top ups to equal the usage limits on a $ to request basis.

Extra suggestion: Add two larger plan options with 20/40k credits.

[digital crown]

I've got 100 songs on YouTube wish I could teach my workflow to y'all

[worn atlas]

Introduce the concept of music production yield as the ratio between the number of clips kept and the number of clips deleted or "disliked". This will be a tool for measuring the effectiveness or the "efficiency" of our workflow . 🙂

[devout echo]

Totally agree. Folders needed.

I know this is a pipe dream but in time maybe. Instead of a song being created with all instruments mixed together, a song consisting of several parts mixed together to form a whole with the individual parts available.

[vale sphinx]

I don't know if it have been said before, It could be great to have BPM and Key used for generation, so it's easier to make it work with something else

[wind abyss]

added a proper lyrics generator to suno : like the free one to the interface. lyrics generators are all free and are easy to use.
https://playthetunes.com/song-generator/

[vapid isle]

I find it frustrating that you have no influence on what is generated, the extensions sometimes last 28 seconds, sometimes 1 minute, sometimes he sings high in the middle of the text then quietly again, that's an imposition.

[scarlet rivet]

Sorry I cant find a consolidated list of feature requests, so I will post here. For us it would be super helpful to be able to specify song duration(s). We typically require 30s, 60s, 90s and 120s.

The lengths dont need to be exact but within, say, 10% of the above listed times.

[royal venture]

Give us documentation that says we own a track we have created while under a plan using Suno. Even if its a generic document that is downloadable per whole song with the title inserted into the document. Currently there is no way to provide this documentation to a distributor.

[dusty spoke]

It would be great if a subscribed user could upload a certificate to use for timestamping downloaded MP3s (assuming this is technically possible). Of course, the user would still have to purchase the certificate separately, but it would probably (IANAL) be a legally valid proof of ownership.

[cinder epoch]

Hello SUNO. Thank you for being the pioneer in this adventure of Ai Sounds. Since Suno is a very powerfull algorythm for music and beats, and use already clip for text to audio, i am wondering how difficult or how close the technology would be for Suno to generate coherent music video. It would just have to generate an image from the lyrics, one per sentence let's say, and interpolate the image in regard of the rythm, pitch and style of music.
Are this technology already there ?
it would have in the UI some sliders for adjusting how closely or broadly the frames relates to the lyrics, to the rythm, and more specifically to the pitch, to the tone, etc... , the ability to customize the generated prompt and have artistic control over the output : like attach a character to a specific beat or pitch.

[delicate cedar]

Let me introduce you to Noisee: https://discord.gg/ZUcZVPsqwN

They take audio and use AI to seek the chorus, then generate a video based on that and some user inputs.

You can submit an image to help guide the style and give text instructions to determine the clip subjects.

[cinder epoch]

Thanks for the link. Why would Suno wouldn't build their own ?

[delicate cedar]

They may eventually, but this exists now and they work pretty decently with Suno so better to have than to want...

[cinder epoch]

Got it. Thank you !

[true umbra]

Set of Consistent Lyrical Tags

I've been exploring different AI music generation apps, and I've noticed that they tend to produce similar results. This is likely because they're based on existing music data, leading to outputs that often sound alike—although undeniably high-quality.

However, I'm considering using Suno as my AI co-songwriter. For this partnership to work, I need more control over how the lyrics are sung, which I believe can be achieved through consistent lyrical tags.

Additionally, I'd like to have the ability to define how different parts of the song—like the intro, verses, and bridge—are executed and performed. Again, this would rely on the use of steady lyrical tags.

So, my suggestion, as you might have gathered, is to establish a Set of Consistent Lyrical Tags. These should be versatile enough to integrate with any style of music. I'm confident this would significantly enhance Suno's utility as a creative songwriting tool.

[pliant shoal]

Suggestion: it would be great if we could cut off certain parts. I have a great part 1, 2 and 3 that form a great song together, but at the end of part 3 there are just some random sounds that I don't like. I would appreciate it if the "Get whole song"-option allowed to cut off X seconds of the last part.

[ocean anchor]

I'd suggest going back to the settings you had until mid April. The percentage of mind-blowing songs I got was astonishing, even with great verses AND choruses, and key modulation. Now it is uninspired at best. I'll keep chasing the dragon until my credits are gone in the hope things will go back to '''''normal'''''... sigh...

[potent rain]

Udio just got audio inpainting. nudge nudge!

[tidal wharf]

Do they already allow to “continue from anywhere”?

[jagged fable]

They've already had the option to "add section" from any point. The inpainting allows you to replace a selected section.

[tidal wharf]

hm, you can add section from any second in the generation? I thought only at the end or beginning (backwards)?

that changes everything, probably have to try udio for my next album as the quality seems to be up (more than) a notch

[jagged fable]

Yeah there's a toggle which allows you to highlight a portion and then choose to add a new section before or after it

haven't messed with it much, i'm still really back and forth between suno and udio because both have different strengths & weaknesses

[tidal wharf]

that's cool!

[tidal wharf]

what are udio's weaknesses so far?

(let us know if we're crossing the line here @hybrid crane )

[hybrid crane]

Why not discuss that in #🗣️┃chill-chat

[cloud zephyr]

Maybe the option to choose the art on a song from a database that generates images based on lyrics/genres. I have a heavy metal+ instrumental with the image of a bird dressed as a minstrel with a lute. Doesnt quite match the song

small example the first is perfect the metal version gets birb... whyy

[blissful wolf]

FYI UDIO May 8th:
" With Audio Inpainting, you can select a portion of a track to re-generate based on the surrounding context. This makes it easy to edit single vocal lines, correct errors, or smooth over transitions, so you can create the perfect track. Inpainting is available for subscribers starting today (only on desktop). "

[cedar mulch]

wow thats really cool for udio

[blissful wolf]

Ye - Many times I've wanted to change a line of lyrics without having to redo the whole song .

[cloud zephyr]

its a really good feature

[rigid rivet]

can anyone help me out with sumo and command prompts

[idle oak]

see i have been able to pop them out already have 9 songs from the last couple weeks. Getting my account set for my music is the next step

never actually had the kinda positive attention before its great.

Also the intro function doesnt work but i have found that if I just do very small pieces of the song at a time i get a better out come

like just the intro or just verse 1

i think my best songs have been done that way

[cloud zephyr]

How would you break it up? a lot of my verses have several parts, would you post the whole verse + interlude or each part of the verses seperately with a interlude or musical break between the 2nd or 3rd part of verse 1 im not even sure if any of what i said makes sense x-x

[gentle oxide]

@still parcel

a creation tree like the above to easily track the routes

and a search

[strong swan]

i have to 😏 a tiny bit... but just a tiny bit

[gentle oxide]

we should make some instruction guide, i have about 200 tracks on spotify, youtube, apple music tiktok snapchat etc, I write them then trial and error with the music, master them to to HD then upload them>
If any one is interested, i have developed a GPT to help with people uploading to streaming sites like soundcloud, just put in your artist name lyrics and the genre of music and it will produce seo description, hash tags and a album/single cover for the track

https://chatgpt.com/g/g-SbomTfARZ-music-promoter

it will also talk youthrough how to upload and guide on soundcloud

[cedar mulch]

what is your mastering process?

[soft harness]

In the future, #💬┃general-chat would have been a more useful place to ask as this channel is suggestions for improvements related to Suno.

The chirp bot is retired, use the website https://suno.com

Some helpful starting links: https://suno.wiki
https://youtu.be/lj8utBd_IUM

[gentle oxide]

long winded if the quality isnt there as it hasnt been past couple of days, but my suggestion for people is to use a free mastering tool that can be found here:

https://www.bandlab.com/mastering

I however use a 10 way splitter at lalal.ai to breakdown the track then put in bandlab add effects, then put the track through the mastering and then upscale it to 96000bits in wave pad which is linked to bandlab and overlay the parts that are good and replace any that are curropt , making sure im getting the best i can out of those tracks that i really like.

I know long winded right....but in effect im not using pure A.I created music and really have a great sense of achievement when each one is then released, currently have 71,000 streams on soundclould for my tracks so far

[cedar mulch]

thaaaaaaanks

[sly blade]

Suggestion:

Add more control by allowing Music Theory influences. For example if I write

In A minor, starts with (B2, D3, G3) chords, (G1 bass). Chord-based with subtle melody in (fourths: 'B2, C3, B2, (long C3)'). Gentle dynamics, reflective mood

It should really stay in A minor for the composition.

Or maybe it should be possible to give a start melody (like image2image -> music2music) so I can record a little starting melody or track and it continues and improves it

[winter jungle]

that is a GREAT idea

[quaint valley]

Udio now has an inpainting feature. Allows you to edit parts of the song.

Suno needs to do this too.

[tepid trench]

I've heard this talked about as an upcoming feature in v4.
Be nice if it's available for free users too (Udio's requires subscription)

[winter jungle]

thats why ive never heard of it.

[untold shuttle]

A search function is really critical. Can’t understand why it’s not included from the start…

[sly blade]

if we have this + inpainting (which is already coming im quite certain), we will have A LOT more control.

[sly blade]

yeah lets hope the team sees that at all xD

[outer remnant]

The ability to add in random animal noises for example [cat meow] or [monkey howl] could add extra hilarity to the creations.

[winter jungle]

real

[sly blade]

do we need that for now? xD

[wind abyss]

i work on others too.

[sly blade]

[tepid trench]

Some of us (including me) try to do some wacky experimental things sometimes. I did manage to get some kind of ratchety duck quack but it's really hard to prompt for. So it'd be a nice addition xD

[drowsy dove]

please let us do Lyric video adjustments! sadging when it has [end] from the previous cut because i didnt think i was going to extend at first

[sly blade]

I see xD we all just gotta be patient.

AI Music for sure is gonna surpass everything humans can do not only in speed but also quality.

If AI Music gets combined with easy to handle voice creation / cloning, we can just choose a perfect voice in realtime, start with a melody and add instruments and sounds or other singers in the process. With outpainting, inpainting, music theory etc. this will be INSANE

it will be the easiest workflow for the most beautiful music

[winter jungle]

Mariah Carey sings Led Zepplin's Communication Breakdown with Micheal Jackson as backup, and Phil Collins on drums.

[odd tapir]

I want to create a voice I like and keep it for all my songs. Why can't we do this?

[soft harness]

Not currently a feature or focus for the team ATM AFAIK. There are other tools available that can help switch the voices such as kits.ai (see #🗣️┃chill-chat message )

[dusty spoke]

We can. Kinda, sorta, more or less.
You can create "your" voice and upload it to https://weights.gg
Then replace Suno's voice with yours using https://tryreplay.io
Details are still rather sketchy to me, but it looks viable. I'm still at a very early phase in my experiments with it, so I can't tell you more.

[winter jungle]

i just made a cover just using Weights, kinda mid but itll get better im sure

[strong swan]

there is enough space for 16 playlists on the profile page, but it always shows just 10. It would be nice to have 16 there. the names for playlists get shorted with ... , that makes it difficult to add extra infos like (MAXi) after the album name for example (sometimes the name is too long on its own). Maybe 2 rows of text or even the Text (with a shadow to make it visible on all backgrounds) inside the cover art ?

[wind abyss]

mastering the complete audio tracks, with a few filters ambient,thunder,universal,fire,clarity. to spruce the tunes up, and to get rid of some of the noise appearing on creating tracks gens one after another.

[kind bridge]

A solution to add music to the start of the song, i find that it often goes right to the lyrics and i would like the option to go back and say okay generate backwards so from 0 seconds to -3 seconds. So i can like add an intro that is just an instrumental. as all the music I generate goes straight to the lyrics and I would like to add music or generate in any direction forward or back

[wind abyss]

block music also, so we don't need to type [intro][verse][chorus][outro] makes it more manageable, with the implentations of adding the [] for special sections or a renamable block section=. [pre-chorus] etc etc.

[glossy cape]

Perhaps there could be a negative tag section. A big thing I often have is instruments being prioritized that I didn't add to tags, rather than instruments I did add. And if we have a negative tag to ensure certain tones or instruments don't appear, that could help.

I have noticed that adding certain genres or tones can add instruments, so like "somber" or "emotional", will add a piano and prioritize that, rather than the organ I added.

[ocean nacelle]

Yeah, like negative prompts in image generation such as "poorly drawn fingers" and it makes poorly drawn fingers anyway. Good luck, because I've already used stuff like "[ NO VOCALS ]" and it adds them anyway, even when only "instrumental" is selected.

[dark parcel]

the song always ends suddenly,even if from extend.I felt unhappy

[cold remnant]

A command to get the vocal to hold a tune for example ogeeeeeee

[obsidian gale]

how does that work? can we change parts of songs we already made

[sly blade]

yes we will be able to. inpainting means to change parts we already made and outpainting is to extend from the beginning or ending (e.g. add intro)

[obsidian gale]

that would be sick!!! some of my songs need a rework

[exotic python]

anybody know this?

[robust jewel]

https://huggingface.co/spaces/suno/bark

Chirp isn't on the Discord anymore. You have to use the website to make songs.

Bark is only on the huggingface page now. I don't actually know anything more about it than that it exists.

[exotic python]

bark training code is not open source too？

[feral aspen]

Bark isn't technically off-topic, but it's so low volume on this Discord you're probably better off looking in other Discords for specific Bark implementations, of which the Huggingface Transformers version is probably the most popular. You can also DM me Bark questions, as long you don't mind that it could take me a few days to reply at some periods.

[robust jewel]

Sounds like you probably know more about it than I do

[worn atlas]

Generation by sections (udio has already this) . Providing a mean to stitch all these sections in any order you want(udio does not have that). In this way you can only generate a few number of sections and manage them like in a DAW. And with the coming inpainting feature you will be able to modify any of your sections independently.

[trail fulcrum]

I would like to suggest that we are able to get vocals only, for example adding at the ... button Vocal Only (gets vocals for the song made and excludes instrumental) and then maybe at some point also add your own vocals to add to the song?

Currently to extract the vocals it doesn't come out well with other tools.

[untold zenith]

Improve the prompt with songs from the 80s, I was trying to make a song in the style of modern talking but the voices they make don't seem like songs from the 80s

[cold remnant]

If we had negative prompt option it may help sort some of the current issues out such as echoes, distorted, male / female voice, too fast , too slow reading ( instead of singing ) sudden ending etc

[solar steppe]

Darn. I prefer suno but these changes and prices are really good. The inpainting alone blows suno's current problems out of the water....

[placid marsh]

Allow downloading in lossless.

[strong swan]

then you would have the same audio quality just in a lossless format. its not the mp3 compression, its the sound of the AI.

[placid marsh]

You sure about that?

Try loading a file into spek

[tepid trench]

What would you see in spek with regard to audio quality?

[placid marsh]

Frequency cutoffs at specific mp3 encoding levels

And the tell tale artifacts of a compressed file

Unless the AI is compressing it while it performs the prompt it should be able to output as lossless

[tepid trench]

Frequency cutoff I'm aware of / have seen but I doubt that would be noticeable by ear.
What do the telltale artifacts look like?

[placid marsh]

The jagged bits along the top are where the encoder tries to detect even more quiet parts it assumes it can remove without affecting the audio

[placid marsh]

And the 16khz cutoff you see in nearly all the songs is an archaic form of mp3 compression. Like 1998 VBR LAME encoder.

It should output as flac

The file size will be reduced according to what is on the spectrogram

And nothing removed by bad encoding

Wav is gonna take too much bandwith as it takes space for emptiness.. Flac is true lossless without taking space for the empty parts of the spectragram

If it makes sense

George even said it there. They are reducing the quality. The output might be bad already but no need to compress it further.

Should be doing all they can to make it sound better.

[tepid trench]

FLAC is a good format. I'd like that when possible.

[placid marsh]

Yeah. I'm probably sounding like a bit of an audiophile snob. Maybe I am. I'm a sound engineer. But I do have ears like a wolf and I'm fussy with quality.

[tepid trench]

I'm kinda just happy this exists at all 😆 and expect they'll bump up the model and file quality at some point. As long as it's worthwhile and they don't give us 320kbps MP3 with 128kbps quality music in it as a placebo.

[placid marsh]

Yeah true. I guess it's still in its' infancy.

Will get better with time.

[rough tartan]

Hello, can you make an option to modify the lyrics of the music we made without it changing the music too much ?

.
I created music that I love but there are two or three words that I want to modify except that I can't, what I can do is continue the music with extension but that's not it What I'm looking for

.
Would it be possible to do this ???

[mortal anvil]

Lyric Editing: Add a feature for users to edit lyrics on full songs to prevent repetitions and errors.
Remix Functionality: Replace "reuse prompt" with a "Remix" button, linking to a page where users can create and listen to remixes.
Copyright Monitoring: Implement monitoring of prompts to avoid generating music that mimics copyrighted material closely. for example people using "beminem" to generate music similar to eminem
Flexible Extend: allow users to extend the previous sections, use case: replacing a songs intro
Custom Images: Enable uploading custom images for songs and albums for better branding. ☑️

Monetization:
**Tiered Access: **Offer listening of full-length tracks for subscribers and limit non-subscribers to short clips. Create a pool from subscription fees distributed based on song popularity.
Competitions: Host contests with cash prizes or subscriptions to boost creativity and engagement.
Licensing Marketplace: Allow creators to license their music for commercial use, sharing revenue with Suno.
Remix Revenue: Share revenue with original creators when their songs are remixed.

[sly blade]

^^^^^
up vote this boys we need it

[novel gazelle]

I feel like you guys at suno discontinued TTs alltogether except hugging face

[dim hearth]

i would downvote bad songs a lot more often if the songs would go into trash right away. instead they disappear from the create window and litter my library instead...

[pastel lynx]

We also need trashed songs to disappear after 3 or so days.

[obsidian gale]

let use change the parts that are problematic

[loud nimbus]

Sheet music to accompany tracks would be super useful 👌

[woven valley]

I agree. Also the ability to select all in trash, and then being able to delete all of them with a click.

Please put back the function of remix that allows us to remix with the current music, but change the voice/lyrics. The earlier version had that and you took it away. Why???

[soft harness]

They never took it away. Remix was the previous name for Reuse Prompt, the new name accurately describes what it does

[woven valley]

Nope. It doesn't work the way it did before. In the past, it would keep the same music, but now it totally changes it...at least this has been my experience since V3 came out.

[rare ermine]

I'm an aspiring song maker xD I love to sing but the problem is am not really good with instruments or anything like that hahahah so I was wondering if it's possible to sing a tune that I imagine and AI builds a track with that and I could play with it to make a song that I wanted to make :))

[tepid trench]

I remember people used to ask why remix changed the song so much, which is what led to them renaming it.

[soft meteor]

it never kept the music though

[sly blade]

why is there no information from the team? Udio is releasing so usually Suno should do so too

I don't like udio I want Suno only but I need more control

[placid marsh]

• Strictly no lyrics/vocal cuts when "instrumental" is selected. I keep getting lyrics every once in a while in songs. Sometimes it's not even words.

[leaden marlin]

Brazil FUNK! How can we add more genre and foreign language

[soft harness]

For foreign languages, you just need to write in that particular language (which the AI will interpret). However, you will need luck getting the right accents or pronunciation (e.g. there have been a lot of complaints for a long time from Chinese users that the AI sings the Mandarin characters as if they were Cantonese)

The AI actually does know a lot of different language styling, you just gotta find out the correct genre or keywords. I find adding the language or culture as a prefix helps.

If you want to make fusions of different styles, try "Fusion of X Y Z" where it will usually give interesting new melody or rhythm

-- A few examples --
Different languages:
World Hello!!
That's Life (short)
A cappella/Rap fusion:
Track 1
Track 2
+ German Club (DE/EN mix)
Latin/Mozart (kicks in at 40s due to [Drop]):
https://suno.com/song/46531a71-c9de-45c7-bf6f-52e72f984b34

[young basin]

Addition to:
➡️ #💡┃suggestions-web message
➡️ #💡┃suggestions-web message
➡️ #💡┃suggestions-web message
➡️ ‼️ #💡┃suggestions-web message
➡️ ... and other Addition ...

Yes ! We need a CONSTRUCTOR of the final song 🙏 ☺️ , for example a graphic one, so that from other generations we can insert (replace) individual fragments into the finished version (!)
(see screenshot)

And also in each such fragment - you need to write PROMT - what needs to be done?! :

• replace the music,
• change a piece of music to (add an instrument, remove DRUMS, change drums...)
• replace words,
• replace music and create outro ...

[shut iris]

Hi Everyone. I'm struggling with some of the songs' output quality because of background noises, artifacts. I have tried to clean it up with Audition, but it's not much better. Anyone find a solution around this? suggestion?

is RipX the only solution?

[lyric cedar]

Remove the option to be able to flag your own songs, as in the original creator should not be able to do that. That doesn’t make any sense. Only listeners should be allowed to flag a user’s songs

[hybrid crane]

no it's not.. If you get a song that has some hallucinations curse words, you want to be able to flag it so the team can use those outputs to train the AI not to do that

[pastel lynx]

The censorships has gone too far, it's pretty sad👎

[lyric cedar]

Ohh, okay I understand now. That makes sense. Sorry, I was confused

Perhaps I’m not understanding something😕

[strong swan]

If suno creates music it can sometimes happen that the Voice is activated without text in the Lyrics box. These are called hallucinations. depending on the style suno tries to make up vocals that fit to the music. and sometimes there are bad words in it. you have to be abled to flag these generations. is that the part you didnt understand? if not please let us know.

[lyric cedar]

Yes, I think I now understand, thank you. And yeah I’ve noticed that.

[lyric cedar]

I fully understand now. I’m sorry for taking things out of context😔

[hybrid crane]

It's fine 😉
Another case what could be helpful, is when the created image is more adult.. it's rare but it's AI and it can sometimes come up with a bit nudity 😄

But then you can flag it

[lyric cedar]

Haven’t come across that yet but I’ll definitely keep a lookout lol

[acoustic pike]

Please add some Ghazal, Kawwaali style music.

[true yarrow]

Sorry for the fieldback. This may be an issue that has already been raised, but as a Japanese person, I would like to see a guessing function. I'm not very good at pronouncing Kanji. It might be useful to have this even when making vocaloid style songs!

[dusty dove]

Please, checkbox option in Style Music, I think this is will be very good update.

[tough needle]

Please add some Rigodon music

[limpid moat]

The function to Extend but instead create something leading into a timer, so recreating the beginning of a now second part.

I am trying out some lyrics with melancholy styles, and one of the generations came out absolutely beautiful only the first line read is shlubbed everything after that is MINT and trying to get a recreation of the generation will be heavily credit consuming.

So I want the Functionality to recreate the beginning.

The functionality to also save the music and the vocals separately would also be great, one song I did had mint vocals but the music track for it seemed to not know when to come in or continue.

Maybe even a clean up function, where you can select a time line 00:01 - 00:06 and it will try and redo/cleanup just that portion

[hazy rock]

suggestion:

stop censoring autistic experiences in lyrics you ablist 🤬

[soft harness]

Already mentioned it is not Suno censoring the lyrics, it is ChatGPT, and likely objectiving due to some descriptive phrasing around a sensitive material/issue #📚┃suno-school message

[rustic anvil]

Would be really nice to have an in-house EQ for our songs and ability to fade in and fade out the first/last few seconds

[toxic otter]

Better vocals as it starts out good but the turns into a robot half way through t

Not to cut off words at the beginning of songs

Option to have an intro

Option to choose a key and tempo to song

Edit words after song is done without altering full song

Option for voice ie deep, low gravely, rough, falsetto,

As a musician I’d like more control to be able to hone into my idea of a song rather than a lottery where you press the button and hope for the best, wasting lots and lots of credits

[fading heart]

[background] [end background] and [repeat] [end repeat] to add background vocals and repetition

[hazy rock]

I think this is not possible with Suno. they may have painted themselves into a corner. likely (im guessing) they trained off of datasets of fully mixed music, so you will never be able to separate out and control individual instrument aspects of the generation. We had to work very very hard in the training when we were developing models to reconstruct speech pathology to have this level of control. its an audio version of MOE. I know how it could be done, and I might just do it as an opensource model after I'm done with my commitments on my current project, but its a start from the ground up rebuild of the model.

[granite summit]

They should have a feature that will automatically link Derivatives to the Source Song , or even have it in a specific category such as Remixes / Attributions

[pseudo swallow]

Suggestion: Lawsuit Muzo who is infridgeing onto the Suno name

[desert condor]

Adding lyrics to our instrumentals

[pseudo swallow]

That would ruin instrumental themselves

[graceful carbon]

make a section to choose male vocal and female vocal so that the ai understands better because right now its like 50/50 if the ai understands it. and also easier way to tell it if you want a deep voiced singer or light vocal and that stuff. (repost so it gets seen)

[wooden fable]

my guess: that might not be possible if this info wasn't available in the training meta data.

[sick shale]

When will Vocal + Instrumental download come back?

[hybrid crane]

Come back? You can still download? Click on the 3 dotted menu and choose download.. there you can choose between an mp3 or mp4.

[frosty breach]

There is no method, atleast on my handheld to inspect parts and the previous parts within the chain. Unlike PC where you can open a part in create and click the dropdown for the entire chain.

[true umbra]

Extend gives you some control over developing your song, but it leaves you with a problem when you wish to create a song.

I've been enjoying the Explore options, and this does produce some nice snippets. You can Extend or 'Continue' with this, but you can not add an intro to it.

I tried using the Style of Music I liked from Explore, reused the prompt and created an Instrumental with the intention of making my Intro. However, this does not work. As you'd expect with Ai, it didn't replicate the style of the song I liked. In fact, in both versions, it changed the rhythm, BPM and Style.

I would like to see two options added to Suno:

1. The option to, 'Add Intro.' this being based on the track you select, as it does with Extend.

2. Replicate the style and BPM from a given track. This would extend the options to Extend, Set BPM, and Keep Style. I suppose the Style option would be down to some sort of, Seed, option, as with graphical Ai apps.

Any news of Updates?

[frosty breach]

You know, you could extend from the song you like, cut in to around the instrumental you like and just continue on?

[nova dawn]

I tick box option for male or female vocals.

[frosty breach]

Maybe a crafts channel to share outside of suno creations? Such as third party album cover generations, remixes in audio software, video edits etc that all use Suno, but might be cluttering #💬┃general-chat a bit too much.
Perhaps it being under an opt-in role?

Just spitballing here, might as well share whims than letting them be forgotten.

[true umbra]

Thanks for the reply. Yes, I'm working on your idea at the moment. It works well enough if you like the style and tempo and there is no singing at the start! 👍 I did try extending from 0 seconds, but you have to have at least 1 second, but it does create the Extend section in the same style etc.

I'm just checking out your pieces.... very nice!

A Clean Slate. When you start a new song, it would be nice if only the tracks applicable to the song you're working on were in the Create Section. Hope that makes sense.

[soft meteor]

Suggestion:
Stop overwriting the prompt when I click extend (so many wasted gens 🥲)

[toxic otter]

Better quality rather that just mp3 a bit like mid journey where when you get something you like you could generate a better quality version

[leaden marlin]

I think it would be cool that, when making a song, the song generates like normally, but, the generated audio, is generated separately, I mean, that, you can download only vocals, or only instrumental, and, it syncing

[feral zephyr]

Sub-second Parameters Please
Let me Continue from
1:56:20
and not locked to either
1:56 or 1:57 for example.

I've had so many generatons where music goes quiet at the end of a second but the start of the second doesn't include the word so cutting before loses a possible word and cutting after generates to a quiet section which isn't what I want.

[blissful wolf]

There hasn't been a Generate+Extend ANNOUNCEMENT or a MINI-UPDATE for over 1.5 months ! Surely the DEVs have something to say about progress and take us somewhat out of the darkness ?

[true umbra]

Simplify Delete, Please! I've been going around in circles.

The playlist is okay, but it has very poor management. I don't know why it is not just like every other simple filing system: drag and drop, right click and folders.

When working from a playlist, it would be beneficial if I could create and develop a song within the playlist itself. Currently, I'm not finding a practical use for playlists.

[stray raptor]

Maybe a better "Extend" Option. Late i noticed, that i can adapt the time, that the second part joins in. Maybe a slider or something, so i can hear it joining together. Or a option to adapt it later. Always need to delete the song and start over...

[subtle tendon]

Really loving Suno, here are a few suggestions to improve it:

• ability to add before an existing track (eg like an intro)
• ability to change specific parts of an existing track which are slightly wrong
• output a musical score of the track
• change lyrics of an existing track

[sharp sundial]

A BPM input for the next version would be highly appreciated. 🙏

I am encountering difficulties achieving faster BPM speeds. The highest BPM I have been able to reach is only 210 (even with speedcore tag, it requires lottery), whereas I have expectations of reaching 300+ BPM.

[shadow torrent]

/chirp

[strong swan]

Chirp is retired. Please use https://www.suno.com to make music.
Also have a look at the community Wiki for a guide on how to use SUNO. https://www.suno.wiki/ ✌️

[compact juniper]

#language-suggestions

Not strictly a suggestion but, the current support for lyrics in the "Sinhala" language (Sri Lanka) is already near perfect! 🔥🔥🔥

This language is currently not in the supported languages list as well. It definitely can be added.

Pronunciation is almost perfect.

If this is accidental then please keep it as is! Thank you!

[urban sleet]

I made a song in 100 year old saxon english, that does not really resemble english and uses a lot of those crazy phonetic characters - and it was near perfect too. Although it might have been made in V2, I'm not sure...

[upbeat lily]

Please add european Portuguese to voices. They have great songs and very diferent from the brasil ones

[tall sedge]

Would love to have a statistics page that shows how many people have listened to the songs i've generated where you can change from 24 hrs, 7 days, 1 month, 1 year and all time for example. Maybe what country or region as well if that is possible. This could also show if people visited your account through a link or through browsing the showcase/trending page.

Would also be really cool to have a comment section for the songs so that people can comment on songs they like :)

[leaden marlin]

Would love to be able too get the vocals and insturment separated.

[elfin pelican]

🧠

you dont have one of these dont you?

[tepid trench]

Bit rude.

[storm hatch]

we should have the ability in suno to pick the voice/gender of the ai singer

[strong swan]

you have some control over it by adding something like "male singer " " female voice" or other indicators in the style. its not guranteed to work... like basically everything... but it raises the chance to make it happen.

[storm hatch]

ok i tried that once and it didn't work so i thought they didn't have that feature thank you good to know

[strong swan]

once is quite a small sample base 😄 for the thousands of variations possible (wich style is ignored, wich is used, wich voice is generated ...) some more generations are needed. every new variable we introduce has to be compensated by more generations to basically brutforce the randomness

[storm hatch]

yea

[spare juniper]

Just better audio quality and a search option in the Library would be great. I have so many pages and finding old songs is tough. Loving Suno.

[timid light]

image

i

rap

i

[devout echo]

Suno is still putting the genre into the lyrics ( "I'm singing this ballad") or the requested instrument (" as I play my guitar"). Sounds ultra cheesy and screams AI.

[leaden marlin]

the moderation prompt is too sensitive i dont know how else to put it.

[dark willow]

Hi there,
i'm clicked the "inappropriate" button accidentally report my own creation, please ignore my report, my account is "SmoothQuarterNotes694"

[shadow holly]

Hey Guys - can we get more vocal ranges?
I have tried playing around but can't get bass tones. Unless my prompting os off?

Female Voices:

Soprano: Typically C4 (middle C) to C6 or higher.

Mezzo-soprano: Typically A3 to A5.

Contralto/Alto: Typically F3 to F5.

Male Voices:

Countertenor: Typically E3 to E5 (falsetto).

Tenor: Typically C3 to C5.

Baritone: Typically G2 to G4.

Bass: Typically E2 to E4.

I would love the option to control the mix or have a button to remove distortion. Or perhaps improve mix quality. Especially when trying hybrid genres there are too many distorted elements usually vocals.

[robust jewel]

You may be able to prompt for bass-baritone to get a bit deeper, but that is likely to end up with more of a musical/operatic sound

[shadow holly]

@robust jewel thanks. I have tried about 50+ variations but it usually ignores. I have also tried simple prompts like baritone vocals or bass vocals... also Basso Cantante, Basso Buffo and Basso Robusto along with the obvious like deep voice, Bass voice which all didn't produce any noticeable changes in the voclas.

[robust jewel]

Vocals are better than they use to be (it use to just ignore even male/female most of the time) but I'm not surprised you can't get a specific vocal range.

[glossy stump]

anybody up for some basketball? 🏀 https://suno.com/song/814b20a7-734b-40ee-af78-d634f269dec0

[feral aspen]

Suggestion: allowing trading maximum generation length for extra long musical history

[royal cipher]

This company put an app called suno.ai on the APP store to cheat money, this is their company name and address, it is a United Kingdom company
VIVID RED

This company called VIVID RED has launched an APP called suno.ai in China to cheat money. This is not an official product of Suno. Go to Apple and report them

This is a scam, report it quickly, don't be deceived by them

[tepid trench]

Thanks for letting people know.

Just one thing: the company you mentioned and put screenshots of was "dissolved" in 2013 which means it is not active any more. Someone probably just used the same name. You might want to remove just the last 2 pictures as they can't be related (since the UK company is gone)

[worn atlas]

Move 'disliked' clips to the trash after a given period. Let each user choose the period in the profile settings. The user can put 0. It means move it immediately to the trash.

[hollow locust]

'modular' generations. Sometimes some generations get one verse perfect, or a chorus, or something, but not the rest of it... imagine if you could pick and choose the best generations of each and stack them like LEGO.

[dusty spoke]

If I don't like the first 30 seconds (sometimes even less) of a generation, I usually stop playing it and move it to the trash can at once. It'd break my heart to know there was an awesome chorus starting at 0:47.
If we could chat with Suno's AI and tell it "great chorus, but the verse is somewhat off" instead of basically rolling the dice again... but I guess we will get there, sooner or later.

[hollow locust]

Seeing song duration without having to play it.

[strong swan]

i just wanted to say thanks... you really changed it. now the playlists look super nice.

[still parcel]

of course, thanks for the suggestion

[frosty breach]

Top or Popular always seemed to be based around Views, but apperantly it's likes on Suno? Would be neat to be able to filter by likes or listens.

[glad citrus]

suggestion: let us extend custom mp3 tracks, so we can upload our own mp3 and suno will extend it

[spark pawn]

I don’t think the context window that Suno uses in order to ensure that the next section matches the song is long enough. If I get a song that hasn’t had a chorus in a segment, and I try to bring that chorus in in the next segment, the chorus isn’t anywhere nearly the same.there are good cases when an entire segment might not have a chorus, such as when you want to have an instrumental solo

Also, it would be nice to be able to start in the middle of a song and work your way in either direction.

[worn atlas]

Copyright issues with this suggestion

[hearty star]

What would be ideal is if there were different forms of brackets for different directive styles. For instance [] can describe style and flow, () would describe soundeffects, <> music theory, {} beatboxing, etc. It is possible on the back-end this can be accomplished with the correct prompt templates and prepending a directive to each bracket type on the agent side to achieve the desired effect. If we take Ollama with Llama3 as an example we can create agents with specific roles tailored to specific tasks so in the case of Suno it could be a mixture of agents approach to prepare the final text output prior to audio gen (i.e. each bracket would be preprocessed by its own style of agent to produce the correct style text input resulting in the intended output, even without changing or retraining the current model).

[sly blade]

yeah awesome idea wow

[opal gull]

It would be great if we could change a part of the lyrics or reuse a musical base and singing style.

[worn atlas]

I am not adding an extra pressure on devs for the inpainting feature but now that udio has released it and put it only its Paid plan, suno devs should release this asap. Even in v4 bêta.

In my last experience, Suno generated a beautiful music and melody for me. Unfortunately, the lyrics at a certain point need to be changed to fit the melody created by Suno. And that's where the nightmare begins. It's right at the beginning of the song, and one simple modification causes Suno to change everything else. I am still burning crédits to try to catch the initial melody.

[quartz steppe]

I would not use that. If I "dislike" a track, it doesn't always mean that I don't like it, it might be the opposite. Atm I evaluate tracks not by quality, but based on how close it to the prompt. I have a lot of "likes" on crap that is messy, noisy, hallucinating, just because it contains some elements that I want.

[worn atlas]

Than you will put "never" for the périod to specify. 0 means move immédiately. Other people can put 1,2 ... days or never.

[quartz steppe]

Yep, such things should definitely be optional

[worn atlas]

Sure. That is why i suggest to add it somewhere in user's settings.

[last jacinth]

Yes, the interface needs to be improved for mobile phone, some exampleS:
· Fixing a weird ghost click now in the library, that takes you to the top and doesn't click the song you expected
· Making the custom input LONGER or extendible in mobile. Now it's really a pain
· Make buttons to CLEAR the input fields in the song mode
· BUlk select and ERASE directly songs in the library
· Nested view in library
· Search field in your library
· Enable custom input FOR INSTRUMENTAL, since you could ask certain things.

[tidal isle]

Make credits from abo last if they don't get used, so it is much more attractiv to buy the yearly abo.

[spiral pulsar]

when will it be possible to submit one's chord progression for Suno to compose based on it?

also can you make it technically possible to submit my singing mp3 or a midi to Suno for using it as a guide in generation?

Also, Suno could definitely generate midi files that one can reuse/edit in a DAW later.

[hidden ibex]

There should be a search for created songs and whatnot

[digital owl]

it defenitely cant generate midi files.

[cinder violet]

Beat competition, hip hop, sumo ai beats only , cash prize or something idk

I want smoke lol

[edgy elk]

When we extend a track from a specif time (i.e., from 1:10), make sure to record this time in the extended part description, so users know from which part the original track has been extended. Better yet, you can create an extended section which also includes the original part, as a single unified section just like Udio does.

[zenith zephyr]

• Please allow us to edit the posted lyrics of the song. Currently, if you generate a song and you want suno to recreate the end of the song you can use the Extend From feature to have it remake the end of the song, but when you combine the parts the lyrics that you cut off by changing the Extend From time will still be present on the songs page and in the generated video. Also sometimes the AI will sing parts or leave out parts of the song and I would like to correct the posted lyrics of the song to match what is actually said.

  This will cause the video to need to be regenerated as well.

• Another suggestion would be to use previous parts (when extending songs) to generate the ai art. I have found that when you want it to extend the song and just add an outro by putting [Outro] for the custom lyrics, the ai seems to always generate black and white pictures, usually of birds holding a guitar, instead of anything to do with the previous parts of the song.

[rigid eagle]

• I wish the "Extend" and "Get Whole Song" were an atomic operation (since Extend sometimes doesn't perfectly match what was already there, and the best way to check is to listen to the whole thing)

[rigid eagle]

• Assuming it's actually possible to extend from any point in time in the song (not only a full second), it would be incredible to see the sound wave of the song, and be able to select the exact time at which we want to start generating (I have a few cases where e.g. extending from 01:32 is too early, but from 01:33 is too late).
• If we see the sound wave, selecting a range of the song and being able to re-create just that part, would be a game-changer for the music industry.

[woven cradle]

You can "Get Whole Song" at any point and then use that as the new base. Using "Get Whole Song" after finding an "Extend" that you like is a workflow shown here:
https://www.youtube.com/watch?v=5uJw5kb05as
I wonder if it would be too costly to happen automatically since many extensions are never used.

[tepid trench]

Make "Monthly Billing" the default.
It seems odd to have the right-most option the default.

There is already "save with yearly billing (20% off)" text on the monthly billing subs to encourage considering the yearly plans.

It could help avoid people accidentally taking out a yearly subscription when they didn't intend to, e.g.:

#🗣️┃chill-chat message
#💬┃general-chat message
#💬┃general-chat message
#💬┃general-chat message

[vivid anchor]

I find that the suno software has a difficulty making instrumental music with erhu. Is there anyway to implement erhu

[unkempt yoke]

Here is a suggestion for the Explore page. Please implement a search box that will either give us the results, or highlight the names in the list. Until then I will continue to use the web browser search on the page. 🙂

[vocal portal]

we need a timer on this page:

I am about to lose 1300 credits I paid for, but no idea at what time...so no idea if i have time to use them lol

[leaden marlin]

I think it'd be neat to have "Trending" sections or playlists for different genres and styles, that way more songs would be discovered

[cinder violet]

@urban sleet why the face?

[tidal wharf]

here's a super competition-defeating idea: Let users remix/continue tracks they created with Udio. I understand that the main reason for not letting users upload their wavs is because of copyright infringements, but in the case of Udio-generated tracks you can be sure there's none.

(And yeah, I'm also submitting this idea to Udio discord lol.)

[lilac schooner]

Let us chose the Ai voice we prefer. Also let us get just the instrumental off of a song we finish(this is probably what I want the most).

[vocal portal]

add the ability to search our own music to find specific songs. please 🙂

[hidden ibex]

There should be a Suno.Ai Google Chrome footer extension.

[rigid eagle]

You can "Get Whole Song" at any point and then use that as the new base
Yes! I do that. What I'm saying is that I almost always want to do that (i.e. I rarely want to have only the "extension" part of the song, I need the whole song to see if the new part combines correctly with the older part).
I wonder if it would be too costly to happen automatically since many extensions are never used.
But yes, fair point, I thought it should be a very simple append operation but that takes time, may be to re-encode everything or something. Thanks!

[rigid eagle]

Is the restriction on not letting users upload their wavs only for copyright infringements? What a pity, I thought it was a tecnological problem. I thought maybe Suno saves more data on the generated songs, to be able to continue them with the same vibe and stuff, but if it's indeed possible to extend from any point of any song, that'd be crazy good (one could extend personal songs in the making)! 🤯

[soft harness]

Most likely one part of it, the other part is security since there WILL be people that will try to exploit the system.

IIRC, the ToS of Udio made it sound like they may potentially do what you want in the future (based on the wording that Top Music Attorney mentioned in her Udio ToS coverage video)

[velvet hill]

Hey just curious, I'd like to make a suggestion where there could be a separate description to describe what the singer could sound like. Not sure if this has been suggested but this would be great for hopefully more consistent results 🙂 .

[tidal wharf]

Jukebox could do it very easily, and if Suno's anything like (it which it most probably is), it (transforming raw audio into internal "tokens") is actually the easiest and the fastest part of the model.

[cloud crest]

The one I wish Suno adds in the version is being able to to change the length song

[soft harness]

In case you were not aware:
https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/
Make sure you check out the other entries

[ionic skiff]

allow to upload EDITED downloads from Suno BACK into Suno

[royal venture]

Needs a search bar, playlist pages on profile that side scroll like netflix catalogue, and a delete all button in trash can. Also a select all button on create screen. Also a trim button. Oh yeah an option to select any thumbnail for song you have generated. Oh yeah let me rewrite the lyrics on song page. Cherry on top, let us customize our profile pages like Steam profiles. Background themes, profile picture borders, animations, ect, and tie customization to the monthly subscription to encourage more sells of that tier. Give yearly subscription unlimited credits which will encourage sells at highest tier.

[velvet locust]

Uh...

[potent rain]

There badly needs to be an UNDO remove from playlist. i just lost track of half an album while trying to add them to another playlist. Suno's interface is unusable for serious work ffs

Why is it that Udio is several feature releases into inpainting and their general UI while Suno still doesnt have search, yet is releasing useless social features like "Following" and "Followers"??? who the hell is managing the product roadmap , a bizdev intern that time traveled from 2009?

[fair cipher]

Are there any plans to make an app version? It'd be easier for me to subscribe if I could use Google Play.

[sturdy viper]

It would be nice if you gave us the opportunity to independently choose which genre will be in a certain place of the song)

[proven wharf]

I suggest ‘able to set up Start section and End section of song or use that for next generated song

[main stump]

You should allow people to have retroactive rights to the songs created with the free plan when they finally buy the paid one. You can add some terms and conditions for a win win situation (ie, they must remain on the paid plan for x months etc.)

[long folio]

I agree, or the ability to use credits towards converting them to rights

[amber finch]

please groups extensions of tracks together with their original track. finding them in the library is hard if you have a lot of them.

[drifting root]

Hello is there any way to change the thumbnail/generated art that is produced with each new creation? Not a huge deal but would be nice to use the art from other creations and use it for my final song edit as the cover art, thanks for any answers. Definetly a suggestion if its not possible at the moment 🙂

[wooden fable]

i don't think the cover art is a high priority for suno. it's probably more like a little funny extra feature. personally i'd rather see them work on the sound quality than the cover art, because you can always generate your own art using something like midjourney or stable diffusion, where you have far more control over things.

[robust jewel]

I wouldn't be surprised if they just included a small change for the cover art stuff at some point without ever hinting at it in advance. Just like "Hey, someone had time between other stuff."

The weirdest cover image I've gotten. Not like any of the other images generated for the song. Was kind of thankful I wasn't seriously looking at using that clip.

[torn cave]

id be fine with selecting a cover from one of the generated tid-bits that i used to create a song with instead of just the last one thats been generated

[wooden fable]

i think the question here is: should suno be a platform where listeners come and find the music creators they like, or should suno just be a tool for creators to make audio, and then they can go to the normal distribution channels to release their work. i have no idea what the long term goals of suno is. the website says nothing about this.

if suno is supposed to be a complete solution where end-users (listeners) also come, then the artwork becomes important too. but my guess so far has been that audio generation was the main topic, and that the graphics was just a fun little afterthought.

but without more info from the suno team, we'll never know what the long term vision is.

[worn atlas]

Sure. I have already said that and someone replied that these social features have been asked . IMHO devs workload should be dispatched in better way focussing on the features with high added values like search, inpainting, stems séparation ... , realeasing incrementally these features and better communicate. Suno is not a toy.

[worn atlas]

There is at least a third category : professionnals using suno to have creation ideas. They will modify, or use in a DAW for a bigger project what was produced by suno. But effectively if suno creators were thinking only to social networks interactions than, yeah, suno is not the rigth tool for them.

[torn cave]

i do think suno is more akin to a toy rather than a professional tool (whilst still being one) it's just that i don't se myself or anyone being able to make any notable amount of money with suno, except maybe suno itself... there is just going to be to many "artists/creators" making alot of music and and people are not gonns have enough money for everyone of their favorite "artists/creators"

[tidal isle]

Fixed promps that always work for some usual things for example [#Short Instrumental] [#Constant Chorus] [#Exactly Match My Lyrics] [#Quick End]

[wooden fable]

@worn atlas i totally agree. those exist too. but if they are the target group, the sound quality needs to go up to v3alpha levels, and they need to offer .flac or .wav export.

take it from a sound engineer: studios don't work with mp3 files as a part of their internal workflow if they can avoid it.

gotta be lossless for that target audience

[upbeat mantle]

make it so u can control the Male or female vocal pitch with more complicated promps

[potent rain]

By who? Hundreds of people asking for inpainting all the fucking time, ive never once seen anyone ask for social features. Suno is a toy so far.

In that time Udio has added tons of features to the UI and improved generation quality significantly

[zenith zephyr]

bbl drizzy

[wooden fable]

i wonder why you would even need inpainting with the current sound quality. it would have made sense with v3alpha imho.

[tepid trench]

V4 should have in-painting so I hope the audio quality is improved too. They have incentive to, given that there's competition around.

[wooden fable]

indeed

[cold remnant]

A remix option for ' full songs ' using extra credits being 50 or 100 credits and obviously taking more time. Sometimes the song all together does sound right so this could be used as an option to change from say blue to bluegrass or rock etc

[torn cave]

@zenith zephyr what?

[worn atlas]

#💡┃suggestions-web message

[full ridge]

This be a game-changer. It's made such a difference with generating AI images. Before, you could have an image that's 98% pefect but there's that one bit that's really bad and noticable amking the whole image a write-off. But now you can edit those bts out. Fixing weird bits where Suno forgets a word in the chorus or there's a bunch of background noise at certain points would be one of the best things they could add.

[zenith zephyr]

google it

[wind crest]

got an odd generation where a song ended well, there was silence, and then a whole different song started in the last 10 seconds. hopefully along with with inpainting in the next version they'll also allow us some very basic editing, just the ability to cut a chunk off of the end/beginning would be great (had other generations with 30 seconds of silence tacked on to the end, currently i take tracks like this to audacity to cut a bit off but it'd be really convenient to be able to do it in-app)

[potent rain]

Ive had the same problem last couple of weeks

[tepid trench]

I have that happen a lot
Basic editing tools would be nice.

[torn cave]

@zenith zephyr can't be that important then

[visual brook]

i would love a field when making a song of "influenced by" that allows a few links to other Suno entities - like a Suno song that helped inspire a prompt for a new song gets a link

[wind crest]

this is such a crucial suggestion, with inpainting on the horizon. i'd requested more modular control for inpainting than the current by-the-second one, being able to see the soundwave would be so infinitely useful for clipping and replacing a single garbled word (or for selecting an entire chorus chunk to duplicate it after verse 2 and not having to worry about it singing the chorus the same, just telling it to continue verse 2 into the copied chunk)

[frosty breach]

I do think it is a matter of perspective. A painting set full with brushes and some paints can be seen as a toy, yet in capable hands it can still produce a masterpiece.

Tools and Toys are very much alike, depending on their use. I have a bunch of sculptors clay, and often I give it to kids to pass the time as a toy. But that evening I will sit down with the same clay to sculpt a new miniature.

[zenith zephyr]

renaming songs should regenerate the mp4

[worn atlas]

Possibility to add Fade-in and and Fade-out effect in post-generation. As adding this in the lyrics does not (mostly) work as expected.

[frosty breach]

00:00 extend time.

For us experimental folks, 00:00 might be very interesting to play with. Unsure if it's able to tho.

[sly blade]

will Suno even be updated? I don't hear any news

[tepid trench]

It will be. They're working on the v4 model since v3 was released (toward end of March). I don't know how long it takes to train it but probably quite a while (plus other development work, like getting inpainting working.) Until v4 is ready we're probably just going to see general website updates only.

[sly blade]

I see. well they better hurry. If elevenlabs releases their model there's gonna be competitors

[zenith zephyr]

wouldnt that just be the same as creating a brand new song? Thered be no context to go off of

it would be like using img2img with denoising set to 1 in stable diffusion

[frosty breach]

Well that's what I want, an extend with no prior context.

[zenith zephyr]

thats just plain generate

[frosty breach]

No it is entirely different.

[zenith zephyr]

i understand you think it would be different

i guess the difference would be that you could join the two parts later into a full song

[frosty breach]

There you go

[zenith zephyr]

ok

[frosty breach]

You can then trick the AI into generating more like it.

[daring kiln]

I have a suggestion. It would be awesome to look up the genre for some music in the same category.

[zenith zephyr]

category = genre?... not sure what you mean. Have you looked at the explore tab?

[daring kiln]

I have but it would be awesome if they add a search bar to it.

[zenith zephyr]

what would you search for?

[daring kiln]

me mostly rock style songs.

[zenith zephyr]

no i meant literally what would you put in a search bar?

i guess having a way to search through and find a sub genre might be useful

[strong swan]

in the best case the music is analysed when switching it public or so, and a new genre is added that is searchable

[delicate cedar]

I would like to suggest a channel that is just music videos. By which I mean something with some motion in it, edited together in some form, using Noisee or some other video generator to create the clips. Not trying to take away from the works that people make that are audio only or which are only lyrics. But it'd be nice to have something to showcase AI music videos in the classic sense.

[torn cave]

@frosty breach that was (kinda) what I was trying to get at but more like the ability of being able to make a notable amount of money with suno, from a "creator" point of view and while I think a small number of "creators" will be, or is already capable to make enough money from monetizing AI generated music that will be just a negligible percentage. I'm not saying that it isn't a tool, I'm Saying that I can't see it as a tool. And in the end it's just my opinion. I don't want to downplay anyone's efforts of making money with it, but we gotta look at it realistically...

Btw I really enjoy a good conversation 🙂

[frosty breach]

I think I can see your POV. Mine is entirely different, give me anything with customisation and my head starts spinning. Whether it's the old symbol edittor for Call of Duty, Sims, AI generation... Hell even in Excel every colour needs to be just right. So I see almost everything as a tool to use my creativity with. For context I am a traditional artists with a ton of different interests in other creative fields as well.

[tepid trench]

There's clearly an appetite for v4, with some impatience growing. I have no idea of the processes and timescales for training/testing etc. but considering the fairly short "alpha" period of v3 and the common issues people encounter with the final v3, I'd like to suggest a lengthened period for testing and feedback, along with getting something out there sooner (even if it's far from complete):

1. When v4 is at a "minimally viable product" stage, make it available as "v4 alpha" to subscribers. Include obvious warnings that it can and will change (banner on site, announcement in Discord.) Warn of updates a couple of days in advance so there's an opportunity to finish songs before potentially drastic changes. Refine the model based on user feedback (use some mechanism other than thumbs-up/down though!) This is a time of chaos. There'll be excitement and frustration.

2. Make "v4 beta" available at a point where users are generally happy with how the alpha is. This should be of similar length to "v3 alpha" - basically a short phase to iron out final bugs. If possible the model should only receive minor tweaks (I don't know if this is possible though) to avoid users wishing that they could go back to it after final release.

3. Release "v4". Celebrate. Most users enjoy the new model. Parties are held in the streets, with music about cats, capybaras and neon lights. Some guy stands at the side of the street yelling about how there's going to be a better service somewhere else but nobody cares because they're all having a good time. Even gran's bopping to the bangers. Life's good.

Some of this is probably obvious or has already been considered. Still, just throwing it out there as a suggestion of something that could keep users engaged with this exciting technology.

[dusty river]

(((adding more roles to enlarge the staff and share work load for moderators creating a tree like staff structure. having 1 mod in charge of for example server tidiness and maintenance and another mod in charge of community feedback ect ect. this would allow more control and better over all management of the server. adding a channel that allows the masses to post staff specific contact requests to remove the need to spam staff with direct messages. allowing any staff to reach out when there free. rather then getting randomly picked. would also go far with creating a clear line of communication between masses and staff)))

i just hope staff are reading thses and this channels not just here to make the masses "feel" like there feedbacks inportant. 😄

[strong swan]

no. its actually read. no worries ✌️

[rain pike]

1. clearer crisper voices
2. better adjustment tools (like the ability to change/fix a 3 second piece in the middle of the song)
3. ability to receive instrumental and vocals as separate files.

[sleek abyss]

As things go forward, features to remix full songs into new versions or just "remaster" things to be clearer and cleaner would be very helpful and encouraging to long-term use. I've got several bangers I'd like to make more perfect as time goes on.

[wind crest]

really hoping later versions will allow us to remaster old songs that have the fuzzy vocals problem

[granite star]

Hello! I'm not sure if this has been mentioned but I recently joined the paid monthly plan to try out the program and I have to say that I love it and I'm really enjoying making music. However, I have been paid for less than a week I think and I've already used a large portion of my credits mainly on trying to get a song right. I've rerun many times because the lyrics were either off, the song didn't end or sound right, and I was thinking, there is a thumbs down option and when you press it, the creation goes away. What if, you can down vote a creation and earn your credits back for that creation? For one song I probably made more than 20 generations just because the program wouldn't get the song or lyrics right. I feel like that is a lot of money wasted on just working to get one finished song. I would love to see something like this as an option so that it doesn't become a waste of money to use the program. Other than that, I am excited to see where the program goes in the future!

[fair cipher]

This would be awesome from a user standpoint, but not awesome from a Suno company standpoint. I'm not sure how much it costs for them to run the AI, but it costs them the same whether the output is good or bad. So if they tried to refund us for generations that weren't what we wanted, they'd probably go broke. Also people would abuse it to make them refund perfectly good stuff. Refund for generation not working properly (you only get a few seconds of crap) is probably the best that can be expected.

[granite star]

I suppose what might make more sense is an editing feature. If we can edit part of a generation where the lyrics aren't clear, etc. then we wouldn't need to continuously generate new clips and waste credits.

[fair cipher]

I'm sure editing would still require a certain amount of credits. Anything that uses the AI would still cost the devs money. Unless you're talking about adding in conventional editing software, rather than making the AI regenerate part of a track. Which would probably still cost money because like, most existing editing programs cost money.

Devs don't work for free, they gotta eat too. And this thing has a whole hell of a lot of users. Running something at this big a scale costs major moolah.

I use a text AI regularly and one of the devs there said he once got a bill for $200k for something. So yeah, this stuff is expensive.

(which is why I try not to whine too much about free users not getting many generations per day... Like, at least I CAN use it for free)

[rigid crest]

My suggestions....

1. Be able to add in a setting where there are no vocals until the desired bar.. (so the song could have some instrumental space for an intro.. then the vocals come in. Instead of vocals randomly coming in too quick.)
   Also just to be able to do collaborations with the songs. FOR EXAMPLE- Ai generated a verse then chorus, then leaves 16 bars of instrumental only, then Ai generated chorus comes back in after the 16.

2. Allow the stems to be downloaded by PREMIUM users so we can (if desired) go and rearrange the song as we'd like.

3. Add a randomize option to the playlists. I have to listen from the top down in the same order every time when I am working and having a more randomized playlist would spice it up and make it not so predictable

[brittle geyser]

I only use udio now. Your melodies are better but the quality of the vocals are so bad they are unusable.

[worn atlas]

Read ...certainely ! But with no feedback . Expected feedback may look like : "on track", "too difficult to implement", "works as designed", "will not fix", "lower priority", "highest priority" etc ..

[frigid pivot]

I know I'm just repeating someone, but

1. Negatives
2. The ability to stop generation before it's finished (sometimes Suno makes a song keep going when it should have finished)

so many credits have been thrown into the void trying to generate a specific style, and it could be avoided simply by stopping generation, or adding negatives.

[soft harness]

For point #2, you can use ending tags to tell it to stop (though uncommonly the AI ignores it): https://www.suno.wiki/faq/making-music/how-do-i-end-the-song/

[verbal remnant]

Your software is awesome. I have a ton of fun with it.

As songs get longer, the quality diminishes. Noticeably. Vocals no longer match the beginning of the song. They have crackle and are of a noticeably worse quality.

As songs get longer, the predictions/ suggestions worsen. Newly typed lyrics are lost. The AI reverts to prior lyrics and often truncates or refuses to go on.

If you can figure out how to keep "extend" capabilities in the 2nd or 3rd "level" of song extension as good as the first round creation ... you will really be getting closer to a world class product.

That said, there is nothing out there as good as this right now, so rock on. Can't wait to see this type of music ability in 5 years.

It would also be amazing to "add before" on a song instead of just adding "at the end" or "add after" 😉

[glossy stump]

i don't think you need to wait 5 years for that. i think we're all looking forward to V4, as rumor has it, that it will come already with significant improvements. there's an ongoing competition between the software providers and timeframes seem to get shorter and shorter...

[devout echo]

The original prompt should be kept in full, not just the genre prompt. "A rock song about finding oneself", minimalist, acoustic guitar. This should not be whittled down to alternative, acoustic, male.

Rap song, rap the verse, sing the chorus. This prompt worked in 3 alpha. No longer works.

[verse], [Chorus] no longer works. 3 alpha towards the end gave verse, chorus as the default song generation.

when creating, the library is inverted and the generating songs are at the bottom. Click on playlist and they are at the top.

[robust jewel]

Based on how the previous couple of versions have released, Suno seems to go several months between updates but release big changes when they do, putting out a new model along with notable feature updates rather than doing lots of little feature updates spread out. Makes it a very noticeable upgrade each time but without the feeling of it constantly improving. It will make it a much bigger deal when V4 releases, though.

[nova raft]

"Regarding the trending list, we can suggest that it's exclusively available to members with paid plans. This ensures that only paying members can benefit from having their content showcased on the list for a wider audience to see."

"It seems that many lists are dominated by free users, making it challenging for paying members to get listed or have their content featured prominently. This creates a significant disparity in the visibility of paid and unpaid content."

[tepid trench]

🤔 how have you found out which users are paying and which ones are not?