💡┃suggestions-web

[jaunty moat]

I thought it was just me! Seems like every kind of generative ai model you're always trading creativeness for quality. If you turn one up, the other goes down.

[half lotus]

OMFG I did not read the fine print in the billing and just went from almost 4k credits to 11000 credits. I thought I was gonna have 15K credits today LOL... Damn it.

Suggestion: Please let us use what we paid for and let us keep our credits if we dont use them.

I would pay for this, yep.

[mortal zodiac]

Export as stems and export without final mastering compression. Would even pay for it.

[fluid narwhal]

Over the last couple days, I've made an album worth of songs with the following aim:
a. make them sound like they could plausibly have been made by the same artist.
b. make songs that have a decent structure, ideally with intro, verse, chorus, and bridge parts.

I used Suno for music and vocals, and used Emily Dickinson poetry for the lyrics (all public domain). Here are a few observations about features that would make it easier to do what I tried to do:

1. I was able to make intros by starting from an Instrumental and then continuing with lyrics. This only sort of worked. It's really difficult to get the AI to follow structure as denoted in the lyric box.
2. I would like to be able to add a part to the beginning of a song, in addition to continuing from the end
3. I would like to be able to trim generations, so that if I get a good ending I can cut it before extraneous bits pop up.
4. I would like more information about how to extend a song and have the lyrics continue
5. I would like the AI to adhere to instructions like [intro], [instrumental], [breakdown], [drop], [fade out] etc.
6. I would love to be able to organize my created list-- I did a lot of generations to find the right bits to stick together. It would have been easier to keep track of what extended what if the generations were threaded, for example. It WAS helpful to have the previous parts highlighted when playing an extension or a Full Song.

(here is the album if anyone cares to see how well I achieved my goal: https://suno.com/playlist/f2459ddc-cec2-4b20-bc1b-005500c492c4)

[tough prairie]

I will let a few quick suggestions to make Suno a better studio:
Over "style" make a separate one: create two fields, one for vocals; 2nd for instruments + Genre, subgenre
Tags autosuggestion when you're starting to write (predictable based on tags with usage on platform)
Selectable instruments from list

Plus service (for pro for example):
Adjust vocals and emotions as selectable tags (aggresive, clear, tenor, baritone, sad, narrative, etc)
Preview generation (a few seconds without charging credits)
Field increase max-lenght for style
Tag temperature colors based on sounds precision of fusion / quality of fusion
Studio quality export (wav, dB adjustments, stems isolation, etc)
Second try option (to fix some issues over lyrics or readable structures)

[crimson plank]

I want to request http get related resources.

[tough prairie]

For preview I mean after 1st generation part; about generations I would suggest to setup number of queue generations for specific part (1, 3, 4, etc), not only default queue generation

[leaden marlin]

would be super cool if all the MIDI from the songs generated could be downloaded to import into a music DAW so the trackss could re produced with higher fidelity. also a bump up in audio quality all around with how the songs sound, also paypal option,

[chilly fable]

Once again, I sincerely suggest it.
I would like to ask you to sanction the YouTube channel that creates music and posts it without modification, ignoring the current rule that the sound source itself should not be redistributed.
The owner of this channel flouts licensing regulations and goes around promoting his YouTube channel on other sites as if it were a cc0 public domain. ->@music-lover11934

Even though it stipulates that the music created without creating content is not distributed or sold, do you condone these channels? I feel that there is a need for a clear definition of the extent to which it is acceptable.
Can all users of suno share their ID with others after creating music or get sales or suno generation fees?

[waxen jewel]

Simple thing, but can we add option of selecting final picture for song out of already generated? Currently it's last one used, but some might be much better

[amber finch]

It would be cool to be able to add lyric to an instrumental after-the-fact, or seperate lyrics from a gen in-browser (i know there are tools do do the last part yourself but it would be easier to have it just there)

also, having each gen labeled in-title as 'A' and 'B' rahter than both having the same title (so it is easier to differentiate between all the gens)

[empty whale]

Is in your roadmap deploy a proper api for users? Im creating a product around suno, and there is not any official method to connect an application

[amber finch]

oh, separate style prompts for the instrumentals, vocals, and the song as a whole. so makers have greater control over output

[wooden yarrow]

I also would like to:

• I would like to be able to add a part to the beginning of a song, in addition to continuing from the end
• I would like more information about how to extend a song and have the lyrics continue
• I would like the AI to adhere to instructions like [intro], [instrumental], [breakdown], [drop], [fade out] etc.

[leaden marlin]

when it generates a song and you dont like it a refund of credit or a redo or something would be nice, because im only getting like 1 or 2 songs id keep out of 10 at best with the parameters im asking.

[amber finch]

i was going through my gens and found a 'credits refunded' one that was actually really good as a stand-alone soundbyte, perhaps having a model or setting specifically for these might be interesting?

[glacial tundra]

A feature request from a recent suno (web UI) subscriber... In Stable Diffusion, if a generation is almost perfect, you can re-use the same seed and modify the prompt or other settings to tweak the output. I have a suno generation that is very close to perfect but it slightly glitched the lyrics right at the main hook. I think if suno were able to import and re-use a seed value from an existing generated track, it's possible I could manually fix the track by changing a single word at the glitched lyric. Any chance that an "import / re-use seed" feature might be added?

[sly furnace]

The home area's excess of categories is overwhelming. Trending and new songs deserve immediate visibility, not a 'show more' hindrance.
It should be split into three pages:

1. main page with trending top 100
2. new songs (more of them!)
3. Archived songs and sub-categories

[slate radish]

Would be super great if you could:

👥 Have a profile connected to the music you make
So that when you share stuff you can know who prompted it and can maybe find some other cool stuff the person has done

💿 See other peoples shared library when going to a profile
So that you can view what other pieces the person has created (midjourney has a nice version of this)

[spice oar]

Using the "extend" function I thought it'd just use more credits to extend the song from the ending due to the lyrics being cut short. But it ended up going back to the beginning and starting the lyrics over, and there's a very noticeable cut in the music. I'm aware that the premium versions would do a full track, but I figured maybe with the free version I could use up more of my dailies to make a slightly longer song. Just didn't seem to do anything useful by extending the track.

Also, and I'm not sure how hard this would be, but there's oftentimes a chorus from one track that I really, really like (the melody it chooses or the particular inflection) but where the verses are completely jumbled up and sound pretty bad, and another version where it's the inverse - verses sound great but the chorus just doesn't do it for me. It'd be amazing if I could somehow tell it to take elements of both and make a new track. Almost like selecting the lyrics and saying "hey, take the temp of the lyrics from this part of this track, and the temp from these other lyrics from this other track, and then generate a new track with those composed lyrics". Sort of like audio in-painting.

Also, pulling the tracks into Ozone and running them through a slight spatial filter really adds some depth as most (not all) are very center panned. Not sure if you run the output through a processing engine at all, but I'm sure you could make some money by allowing people to enable a mastering chain on the output so that things are a bit more accentuated. Could even use another prompt field in Custom Mode to explain the sort of mastering parameters you'd want, though I'm aware that's just another model / training set you'd have to source.

Maybe this is already possible but I haven't seen it yet: being able to somehow notate how lyrics are to be sung note-wise, both in timing and note. Perhaps secondary lines underneath with a notation Suno itself has invented (I wouldn't mind learning the notation just to be able to do this). In my own compositions for vocals I often use dots, dashes and commas to represent quarter, half and eigth notes, and then another line underneath with the note values themselves (C, Db, etc.). Even being able to specify just a few notes and having the AI come up with the rest could be incredibly powerful.

Last suggestion: vocal-only mode. Sourcing actual vocals for a track is expensive, and oftentimes I don't know exactly how I want the acapella to sound without having track first, and in the other direction, I don't often know how I want the track to sound without having the acapella. Going back and forth with a vocalist is helpful when I'm working on a very important project such that they can do something new and interesting, and I can adjust the track for it, and vice versa - but as an indie creator that's often doubling or tripling the cost, which I don't often have out of pocket. A real usecase I would personally pay a fair amount of money for is generating acapellas from some lyrics, some note hints (from the above), tempo, etc. and getting a decent enough acapella to work with that I don't have to spend hours melodyne-ing, just as a prototype vocal track that I can send off to whomever for an actual session. I would imagine this would scratch an itch many indie and smaller artists have been trying to scratch for a very, very long time. I have a few immediate colleagues where this would be invaluable, where they are already impressed with Suno but don't pay for it because generating entire tracks isn't really useful for us for a number of reasons. However the acapella angle would be a game changer.

Bonus points if I can upload the track and have it use the track to generate the acapella somehow, or even two tracks - one for the track itself, and one with the melody line for singing. That'd be... wow, I'd pay a lot for that.

Actually on that note, individual instrument tracks in the same spirit - uploading the main track, telling it to play e.g. a talking drum over certain parts, and getting back a preview + the isolated track would be game changing. Percussion in orchestral work is something really hard to replicate in a DAW but booking a session for a particular, potentially exotic instrument is often cheaper than buying + learning it yourself and recording it but still abhorrently expensive for an indie artist. I often don't need something that is perfect since it blends in with the other bits of the track - Suno could be the answer to this.

[viscid cloak]

The "Liked" Songs on Home would be better if they arent there or atleast on the bottom. Since its a Playlist now, it kinda doesnt need to be at Home

[tawdry scroll]

the extra voices can ruin a perfectly good track more than often

[frank matrix]

a suggestion i have that make it so that if your song is privet it removes the trigger words like if i want to make a troll song i cant just put the limits on the public songs

and be able to direct the bot in lyrics like when you put () make it a adlip or when you put * it make it so that you can tell the bot to do what with the beat and other commands

[peak pasture]

Suggestion: Midi Files from Melodie & Chords, so we can use it in our DAW. 😉

[snow thicket]

I think I nice suggestion would also be editing existing songs. Like change the lyrics, extend and move parts/verses around. At the moment when you want to change the lyrics a little you get a full new song.

[sullen hawk]

If you have an account and dony use the credits today, it should be counted on the ones for tomorrow, so like if i dont use today then i have 100 credits tomorrow.

[peak prism]

Hurry up and improve the audio quality. The mp3 bitrate of the neighboring Udio is already at 320kbps!!!

[leaden marlin]

I'm trying to make game music, like Jak and Daxter style ambient music. Example: https://www.youtube.com/watch?v=KUjtjFiOlAg&ab_channel=sirberegond9

I have a lot of keywords, mostly for the known instruments in these tracks.

Some of them: Midi, Steel Handpan, Timpani, Precussion

• It would be nice to see a detailed breakdown of how each of my keywords actually impacted the final track.
• How was the keyword 'Midi' used? where? how? why?
• How was the keyword 'Steel Handpan' used. where? how? why?
• Let's say in the report, Suno is telling me 'Timpani' isn't being utilized.

I'm kinda just looking for a bar chart or graph showing me which ones are, and which ones aren't.

I guess, almost like a Programming Software like Visual Studio's IntelliSense? Telling me that 'Timpani' is an unknown, and therefore it's not being utilized - and to replace it. Maybe offering me suggestions of what to replace it with.

What if we can have Suno analyse this track? I can upload it, and Suno can make it's own keywords in an attempt to recreate the track?

What if it doesn't know what a timpani sounds like? and I could upload a sample of a timpani?
https://www.youtube.com/watch?v=wQx1R51haZM&ab_channel=FreeMusicSamples

Being able to communicate with Suno, taking an existing track and removing an instrument, or adding one, and not starting entirely from scratch with a fresh track. Eventually perhaps being able to have a slider per keyword.

If I'm not getting the results I want. I'll be able to know about it and resolve it via the reports and graphs, without resorting to generating 100 new songs to diagnose the issue.

[hollow locust]

For the chronological order of 'Create' and 'Library' to be in the same direction. At least give me an option!

[desert whale]

Is it possible to implement extreme gradient boosting to improve the fidelity / depth of the audio output? Not sure if this can be combined with bark.

[drifting basin]

A suggestion: being able to split the vocal and track and having seperate downloads

[steel fox]

I use a different AI that seperates vocals and instrumentals. It doesn't give the most clean result, but it does it's job well enough for now

By the way, is this channel only for suggestions about the song generation itself or are UI/UX suggestions also welcome?

[hollow locust]

Time & Date stamps on generations.

sub-folders in playlists.

Stamp of time-length on 'Full Song' so I don't have to play it to know how long it goes for.

[robust hedge]

You can absolutely create a system alike a semantic tagger which is trained specifically on input + tags that translate to certain notes, or frequencies, other nuances. Actually, they already implemented this: https://suno-ai.notion.site/Using-Metatags-34944efe09ec41d693e314f13af44695

[willow gulch]

Can you finally fix Intro thing? Other [ ] work but creating intro is just 🤡, for more than 20 songs (more than 100 tokens) there was 0 with intro despite the fact that i specified intro in many different ways

[solar prism]

I have a suggestion that you should let us wait longer before we can listen to a song that isnt done generating. Often it ends up with errors

[harsh lichen]

i lowkey didn't think they would actually listen to my suggestion. either that or they where already planing to add it. i finally found what i was looking for after a week.

[spice oar]

Doesn't appear to be what I mean. I'm talking about annotating specific syllables with length and note.

[soft meteor]

you can just start the song as instrumental? This should usually work

[glossy portal]

Hi ! Being able to choose the time of the song in the paid version. (being able to make very short pieces of 20/30 seconds for example)

[robust hedge]

That would be a fine-tuned version of the already existing system

[still halo]

Hi, good day. The songs generated by suno now don't include backing tracks, will there be a new version in the future that will be able to generate the backing tracks along with the songs?
For now, we can only get the backing track with some ai separator program, any ai program recommend? Thanks

[gaunt briar]

Would love to be able to upload audio that the song should be based on instead of or in addition to a prompt. (Reason: My taste is very particular, I pretty much don't like most music of any genre. So I need to be able to reference a very specific style.)

[fathom crystal]

I made a collection of songs and put them in a playlist, but it seems that playlists can only be viewed by users with registered accounts (unlike songs which anyone can listen to), so I can't really share them. Similarly, it doesn't look like you can view the playlists (if any) a song has been added to.

[fierce siren]

Continuations in the library would be nicer if instead of being appended to the list as its own entry continuations could show in a tree like structure as a child of its parent song.

[opaque lodge]

i agree

[ornate lake]

is light theme available?

[random fulcrum]

I'd love to have a way to see how many plays a song has. I have a few of my songs set to public, and would love to know if anyone other than me has been listening to them :). Would also be great to know how to get featured.

Would also love to be able to mix and match pieces of a song. So if I do 4 continuations, then go back and start over from the first segment, would love to be able to mix and match continuations that are born from the same root music segment to create a full song. Right now I jump into Adobe Premiere and cut it up, which is fine, but if there was a slick way to do that in Suno? That would be DOPE. 🙂

[vivid spoke]

1. Please add music styles: Kizomba (classic, modern), Tarraxa (Tarraxinha), Semba - it's all the one family. for now, it's quite a poor generation, maybe a small database.
2. Please add a very rare genre - fodencia. It is almost forgotten, only 2-3 composers supported, a few hundred tracks, but very simple in structure, can be revived and give new life to the genre.

If you need, I can give links, and authors.

[woeful rampart]

1. A way to regenerate Song Pictures

2. Rewrite lyrics after Getting a complete song with lyrics cut from pt1

[rotund breach]

1. You can add your own voice into SUNO and it can use your voice and create songs.

[vivid spoke]

The ability to regenerate a word/section of a song.
Sometimes a song is perfect, but one word/phrase sung incorrectly ruins the whole thing. It would be nice to be able to change the word/phrase, or average a few other variations of that section of the song. (For example, suggest highlighting words that need to be re-generated).

[warm solstice]

Unfortunately, people would abuse something like this by asking for refunds on nearly everything just to get their credits back.

[rigid eagle]

Would love it for the tool to be less restrictive in terms of what it would generate, it currently refuses "latino", "nacho" or "somo" because they are artists names, but they are also normal words that can have nothing to do with the artist at all.

[wise summit]

this is a known issue that's being looked into after we ran into the same problem with the word kalimba

[rigid eagle]

Good to know, thanks!

[wise summit]

a work around in the meantime you can spell words phoenetically and get the same result

[wooden fable]

@rotund breach tip: at least Suno seems to know both Soukous and Highlife ... now we're on the african genres. 🙂

[rigid eagle]

Hmm, how would "phoenetically" be? Using other types of characters?

[wise summit]

so latino might be latine-oh

or something to that effect, nacho -> nachohh

think about how reed and read, or write and right are spoken the same way by most people

[noble slate]

I used my own lyrics that I wrote in high school to generate a song. I loved the outcome. However, in my excitement I rushed and had 2 typos in the lyrics. It would be greate if I could edit the existing song. I tried to regenerate it but the result just isn't like the first one and now I'm sad.

[broken canopy]

why does subscribers do not receive free 50 credits a day? That's kind of absurd. It feels like if paying is for noobs...
Am I on the right place to ask this? 😄

50 x 30 = 1500 credits that they give free for non-subscribers. But if you subscribe for 2500 like me, you are really only getting extra 1000, as you don't receive the free ones

[sinful mulch]

Make the volume slider have its state persist through refreshes and reloads etc

[paper maple]

Yeah, $30 for a second account with 10k or $30 for 4k on the same account doesn't make a lot of sense

Arguably you'd think a top up would give more, not less, than the initial buy-in. Definitely not under half. The first pay in gets you other benefits, the top up only gives you credits.

[paper maple]

Depending on where it is in the song, you could do an Extend before the lines that had issues, cut the lyrics to start at that point, and then re-run them. If there's enough info to go off of, it'll keep the song going. Had to do this a few times when lyrical flow didn't match what I had in mind.

[leaden marlin]

??

[paper maple]

Paid users don't get the 50 free per day that the paid users do. If you pay for an $8 subscription, you miss out on 1500 free points. Meaning out of the 2500 they give you, you're only technically paying for 1000 points.

[noble slate]

I tried that but it pretty much kept the song more monotone whereas the original had better runs and flows. It was perfect. Just wanted the word pronounded correctly.

[paper maple]

Just hold onto it for now, inpainting is supposedly in the works

[worldly galleon]

Topping up credits doesn't make much sense. It seems the most optimal approach is to make a bunch of $30/mo accounts.

[cedar fulcrum]

Right? I've had some bangers that I just needed to change a verse or even word and it's basically gone forever 😦

[paper maple]

Or it costs TONS of points to fix it

[stray frigate]

Just wanted to come and say your website is incredibly well done. one suggestion is for the site to remember your sort order + and make the url change for that so i can share that page

[queen hill]

Suggestion: more control over phrasing in lyrical content - using some combo of existing punctuation rules

Also if there is mastering on your chain then please option of having it off.

Inpainting, multitrack, stems, acapella and instrumental. Obvs all been said but saying again heh. It’s still great btw!

[tranquil whale]

AN OPTION WOULD BE VERY GOOD, THAT WOULD SHOW YOU THE LETTER THAT IS SINGING IN REAL TIME AT THE SAME TIME IT IS PLAYING, SO I COULD INDICATE TO THE AI WHICH WORD DOES NOT FIT, OR WHICH LETTER IS EXTRA OR MISSING, IN MY LATEST SONG, THE CHORUS IT IS "THE ENERGY YOU TRANSMIT" THE AI EATS THE THE AND IT LOOKS VERY BAD. IT'S BEING IMPOSSIBLE FOR ME TO MAKE THE SAME SONG WITH THAT CHANGE, AND IT'S A SHAME THE SONG IS AMAZING.
THANKS FOR IMPROVEMENT, THE APP IS AWESOME

[craggy coral]

Suggestion:

As someone brand new to Suno (heard about it yesterday, started using it today and immediately subscribed to Pro), the settings within 'Custom Mode' are not entirely intuitive or immediately clear - **specifically, Custom Mode's 'Style of Music' section seems to be the exact same thing as non-Custom Mode's 'Song Description' section; however, it's not clear that these are the same. **

More details:

In the non-custom 'Create' menu, you're given the option to type a 'Song Description.' This seems pretty clear, like: "Describe your song." Say, for example, my description was: "A song for a military simulation game. Inspiring, slow, yet ominous and somber."

When I enter that description and hit 'Create,' the song generates. The system applies its own keywords to it, like "cinematic, epic, thrilling, orchestral." If for example I select the option to re-use my prompt, it forces me into Custom Mode and places the aforementioned keywords into the 'Style of Music' section, seemingly deleting and overriding my initial prompt.

So I guess I'm reaching out with a couple of things in mind:

• Is it truly overwriting my initial description, in favor of its own designated keywords? If so, why? I'm given the option to 'Re-Use Prompt' but it deletes my initial prompt if I re-use.
• UX Suggestion: Assuming they are indeed the same thing, align non-Custom's 'Song Description' with Custom's 'Style of Music'. Call it the same thing in both modes. The reason I believe they're the same is because the '?' buttons next to them show the same text when hovered over.

[junior surge]

picture AIs have developed to a point at which the prompting can be very technical, including focal length, specific lenses, color ranges, exposure and more. To the same end, Suno needs a way to prompt it rhythm, instrumentation and more technical details. As it stands, it simply ignores half of my lyrics half of the time, or meanders on the song until there is not time left to fit the lyrics. Other times, Suno speedruns the same lyrics in 80 seconds.

[pseudo brook]

I dont know if this has been suggested before but I will anyway.

Having specific vocal seeds or something that is given after a generation would be very nice to have, would be great to ensure vocals are as consistent across the entire song as well as keeping the same voice on multiple songs.

I have noticed that sometimes as the song gets longer the vocals slightly change based on how the last lyric's inflection was done.

Hopefully something we can get soon

[deft hamlet]

A lot of the songs i create would be really good for song covers. Can there be an option which can create the instrumental of the song i just created?
Like "Create Instrumental of this song".

[shell mirage]

It would be nice to have an Song Stucture Custom Mode where we can Specify exactly the lyrics we want for part of the song. Like there would be one box just for the Chorus, etc.

[sudden grail]

Would be awesome to have a way to refactor generated songs. Say you love the chorus but hate the intro and verses

[fallen osprey]

There should be a section just for completed songs in your library instead of mixed in with all the parts

[brazen ether]

The feathery voice and bad harmonization should really go. It's so annoying even when you describe it in the prompt and the song type it just keeps having it. Makes me wanna stop my 3 account subscriptions.

[jolly trail]

The character limit in the music style field is too small to fully express the instructions, which is frustrating.

[ember mountain]

Would be nice If you could make a live Version from your own creations. That would be awesome

[leaden marlin]

yeah, that feature would need to be implemented in a way so you can preveiw what it made in low quality and if you except what it generated i then allows you the download options etc,

[fallen osprey]

Can there a date created and tag of who made it.

[jolly trail]

It's absurd that while the lyrics field has no character limit, the style directive field is woefully inadequate, making it impossible for creators of instrumental music to fit a complete instruction.

[low helm]

Would be good on the web version to have the creationg zone in the main frame of the screen instead of a tiny column on the left actually. Some AI oriented website have a pretty good UI around that (images related but still) like Nightcafé or even Clipdrop

[reef marten]

can only see the top 21 of this week, what about past weeks ?

[low helm]

A future option to archive and better order our creations would be absolutely great as well

[dull pawn]

hi guys! why do the vocals in the third version suck so bad? the second version sounds much cleaner and more natural.

[quaint valley]

It should be possible to upload an audio sample that the Udio AI can reference off of to understand what it should sound like?

It should also be possible to upload a song with words and have AI do the same words but in a similar way or one with a higher variation style please.

[snow tapir]

I don't know if it's possible with AI right now, but imagine rather than just generating one track, imagine the AI having multiple tracks with different instruments assigned to each one all playing the same song. This with a track editor interface where users could select new instruments per track, make edits to individual track segments or delete specific sections and have the AI generate new data. You could even having a pricing model where you could sell access to more tracks/instruments based on customer needs. Example: User adds 3 additional tracks to their monthly sub, while also adding additional instruments to their sub as well all the while having a convenient editor at their disposal.

[low helm]

Another idea would be to be able to save some style prompts (when we see some we like across the ocmmunity) to have them in a small list for reuse

[rain ore]

Web Library Suggestions

• List total community likes for songs. Maybe an icon for those that are trending or top 10/100 or whatever.
  multi-select for adding to playlists, deleting, permanent deletion.
• Library sort by (date/alphanumeric/full song/parts/versions)
• attribute search with wildcards. (song name, genres, lyrics, and those listed above etc.)
• uncategorized songs are removed from the master list. This is needed to keep the "workspace" more clear while working on a song and reviewing segments/parts for extended songs.
• Alternatively from the above comment, create a "workspace" for song creation where songs and song parts can be isolated from the rest of the library.
• a "notes" section for songs that are "in progress" so creators can keep track of their ideas.
• An option to remove songs within playlists from the default library screen

[placid heron]

Hi there. The 'Explore new style of music with Suno' https://suno.com/explore/ is innovative and amazing. But how can I search a particular item by a name such as 'waltz'. It'll be perfect if there is a search by textbox.

[outer flare]

Please add the ability to generate a very short clip of your prompt for free and then allow to generate the full clip for credits if you like it?

[cerulean glade]

Allow users to search by a specific genre and pull up songs. Example: "Singer-songwriter".

[fallen osprey]

Allow users to search in general.

[cerulean glade]

Yes

[outer flare]

Would it be possible to generate a genre prompt from an existing song ?

[vivid matrix]

Negative Prompts on a roadmap? That would definitely be nice to have.

[pale pelican]

Here's an idea, have the custom lyrics be able to hold as much as the user wish, and have about 1000+ characters be registered when being created. This would allow as much customization too the lyrics that the user wants to do and putting everything down, while also keeping the input/output be limited so that Suno is able to keep working gradually.

[amber finch]

at the moment the songs are created in a single pass. perhaps a modular approach to song making? separating sections of songs and stitching them together? each with their own prompts and lyrics but own overarching prompt to guide the whole gen? Into part, verse part, chorus part, outro part, etc

[sudden valve]

Man the more I use suno the more I realize negative prompts are essential. Like, I need to lobotomize "pop" out of these generations sometimes.

[sudden valve]

Or like, if my life depends on generating smooth jazz WITHOUT saxophone solos. Without negative prompting I would probably be dead.

[vivid matrix]

with metal genres it really loves "shredding" sometimes lol

and not all metal needs to have shrediness

it honestly varies a lot

some genres ive had excellent quality, then others straight ass

[zenith cloud]

I agree, I generated a few good songs but 90% has been trash

[vivid matrix]

I do notice that some degreade quality wise over the time of the generation

This one starts out quite nice, as it gets towards the end its not terrible but seems like lower bitrate over time

i have a breakbeat thing im doing right now

yea it definitely needs more consistency for sure

is v2 worth using?

https://suno.com/song/d4d1690d-5a88-481b-8dd6-4a03c1dbf044

there does seem to be this white noise ish thing over top

ive new to using suno but out of all the options out there its really hard to beat, most others are way worse

i can see that for sure, ive been making music for over 15 years though and so far its been a fun "toy" as most of these ai tools have been

[vagrant crow]

Needed a feature that generates the instrumental version of song based on a same song with vocals. Tried to do "instrumental version" in Music Style, [Refrain] and [Reprise] and even a "set vocals = false" (Python subroutine). It, fu...king FAILED.

[vivid matrix]

i actually did have a few almost comically bad loud bad generations

it was like everything was at 25% and bass weas 75%

what they should implement is a better point refund system actually, if you downvote, get point back since it removes that gen anyways

downvote/report = point refund

with the added thing if you download that gen, you cant refund (to help prevent abuse)_

[latent ruin]

Hey, does anyone know how to make a good music ends. Because most of the time it just stop.

[vivid matrix]

manually edit

thus why i added, if they download they cant get refund

no worries

i can see though where people would still just then record the output to bypass downloading but my answer to that would be to then implement preview audio quality being 128kbps and download minimum of 320kbps (currently downloaded audio is 192kbps which is still lossy)

im actually more annoyed that i just noticed downloads are 192

paid should definitely be 320

my guess would be cost/storage reasons

even though it would be only like 1-2mb difference

oh for sure

BUT then you can argue, there should be a paid option for 320kbps or a slightly lower option for 192, etc....

id agree there for sure

same, i only did 1 month sort of just for a test period

1 month and being out of credits in 4 days lol

[vagrant crow]

@everyone, for those who couldn't finalize the song, please do "Extend" and choose "Instrumental", then in style of the song please type three letters: E. N. D. Simultaneously. No spaces or dots. Just "END". E-N-D!!!!!!!!!!11ONEONE 🤬🤬

I hope it will help those who are suffering from the song interrupted at the middle.

[leaden marlin]

no

😎

[hybrid crane]

Luckily @everyone doesn't work here.. but things like that are so not done... We have over 340k in the server.. and you wanted to ping them all? You think they would have been happy with your ping?

[vivid matrix]

i do not like being pinged like that

[vagrant crow]

Finally you came. I have a problem, so please go to #🪲┃bugs-web message where I have pinged You. Really, it's so stupid if somebody don't know how to finalize the song. I had to do that.

[hybrid crane]

I am sorry.. I am volunteer and it's 1.30 am for me.. this is not the way to draw my attention. I will come back tomorrow on it.

[vagrant crow]

Sorry. Really have a problem, I need to get a solution. If not, I will also send a report to [e-mail]. Well, havanice night.

[hybrid crane]

Again.. I am a volunteer. I have no knowledge about errors in the software, that is something the Devs wil take care off. Please understand my role here.

[vagrant crow]

Oh. Sorry. Though You have helped me twice in a row. Sent a message to [e-mail], they should get a solution.

[vagrant crow]

Really. Sorry.

[hybrid crane]

And you are welcome.. don't mention it.. Just don't ever try to force us to help you please. I understand your frustration, but some things are not within my reach.

[vivid matrix]

i would add this is channel is #💡┃suggestions-web none of that was a suggestion :p

[twilit wolf]

Is there any way to see about outputting a song ending to not stop mid vocal or song note? Maybe instate an option when extending to similar or new Song Style of some sort.

[viscid seal]

Please add daily credit reset time to the FAQ.

[rain pike]

Would like the ability to select a previously generated song as a musical or vocal reference for a new song.
I'll often find a tune or voice i would like to reuse with new lyrics.
You can technically extend a song and write new lyrics, but you're left without a proper intro, unless you do some post processing surgery.

[outer flare]

This 100% sometime I get the perfect sound early on but needed to fix a typo or tweak a lyric

[rain pike]

oh yeah, that too. Forgot about typos. 😄

Also, there's been a couple of instances where i generated a song with no vocals, and would like to add vocals to it.

[teal linden]

Not sure where to post this but is there an API to use or does it have to be through the website?

[tepid trench]

No API yet.

[teal linden]

Is there an ETA

[vivid matrix]

https://studio-api.suno.ai/api/session/
https://studio-api.suno.ai/api/feed/

it's probably not that big of a deal but curious why the clerk "publishableKey" was not put in an environment variable

[floral tide]

• it would be great if the "Get Full Song" allowed you to specify cue in and cue out of the segments. Sometimes there are a few seconds at the start or at the end that you really do not want.
• The ability to make a generation to be prepended or appended and marked as a normal segment or an intro/outro
• Ability to remove the lyrics when there are no lyrics (only tags) / add text.
• It sounds impossible but an option to replace a specific time or segment would be incredible
• Having an easy way to play all the parts, in order, up to the current part to hear the whole song without having to "get full song" and waiting for that
• Ability to re-generate the album art or at least pick any one of the album arts from the segments that make up the song.

[pseudo jasper]

AND: Correct the Text at and from the extending timestamp. The first text lyrics always saved, doesnt matter if you put in a new part a whole verse earlier

[floral tide]

Indeed, which is why adding text or modifying the text would be great! sometimes you have to get phonetically creative so the AI pronounces the words as you want it to, and it looks ugly in the lyrics when you go read it.

[pseudo jasper]

AND: Is it possible to get something like a seed? That we can reproduce an earlier generated track with new lyrics? Right now we will allways get a complete new music track

[floral tide]

this was something that not_riley was requesting a few lines above. yes. I agree.

[pseudo jasper]

i would love to get a seed of this arrangement. i tryed it over 50 times... I never received another song that was even similar.

[floral tide]

someone was saying that you could extend it and add lyrics to it, but then again you would need to cut it in post once the song is ensambled.

[pseudo jasper]

i will give it a try with Adobe Audition

[floral tide]

Darn, I just got 3 segments of 10s of less and 1 of 15s as the end of a song that I was not trying to end. That seems like a waste of credits 😦

[rain pike]

Reduce excessive word moderation on the platform, please. Was able to use "motherfuckin" last month, and rerolling the same song will flag it for moderaiton now. However the word "fuck" works just fine. Doesn't make much sense.

[ionic skiff]

add a separate fade out functionality - i'm tired of losing credits

[foggy gale]

music request channel where you can just say "looking for anime opening vibe and hip hop tracks, DM open" and trade music in

[floral tide]

Maybe it would be a good idea to add to the FAQ what happens to the licensing rights if you take a song marked as public and extend it. Can you then use it?

[terse pumice]

how to use musical instrument for chinese "erhu"

[flat spruce]

I said something about this a long while back but only one person ever talked about it and just said what Suno's doing and never said if my suggestion would be viable to the team at Suno or not. So here's the suggestion in question (not copy pasted tho):

Implementing a feature that allows both free and premium users (both tiers) to purchase rights to songs created using the free version would greatly enhance the user experience. Presently, even upon upgrading to a paid subscription, users are unable to obtain rights to their previously created content. It seems logical that if users are willing to invest in securing rights to their future creations, there should also be an option to acquire rights to songs made with the free version. This adjustment not only empowers users to fully own their musical creations but also presents a win-win scenario by generating additional revenue for Suno. This could potentially lead to increased user satisfaction and loyalty.

[feral aspen]

It's come up and I've also mentioned a related idea myself. At the very least, I think a reasonable grace period would smooth this out a lot. So a paid subscription is retroactive for just an extra 24h. That may cover most of the edge cases.

[flat spruce]

Well, it depends on how far back the coverage is because I don't want anyone coming after my ass for a song I made on free version when I get one of the subscriptions. Please bring that up and see if we can get anything definitive. Your comment is much appreciated, mate.

[flat spruce]

And I wanna know for sure before I do anything with spotify etc

Yeah, AI gen content is a grey area. But ownership could be to some extent

[summer lark]

hear me out..... a negative prompt line

[dreamy ember]

If i can make a suggestion for a "Copy" button so we can copy the exact Chorus etc to the extension of a track ..generating a new one doesn't line up with the first chorus that was generated. Thank you

[flat spruce]

Yeah, your comments have weight. Suno should defo change their ToS here and there.

Yeah, now I'm just waiting for answers

Still, my suggestion has a lot of weight in my opinion

[lofty perch]

have a clear way to separate the song's genre from the song's content in the simple UI. right now everything i generate has lyrics about whatever genre i put in the description in addition to what i told it to make teh song about.

[whole orchid]

Suggestion: The ability to make both a lyrical and instrumental version of a song either at generation or after a song is generated you can turn a song with lyrics into a version without.

[vagrant saddle]

suggestion: for pro users (premier plans or lower but paid) to add the melodic line like: C D E F G A H minor or major and to change it based on the preferences, like to change the accords and stuffs (maybe put their voice and just autotune it)

[odd talon]

Hello. Can I create long audio tracks for 2-4 hours? And how can I do this?

[mystic plaza]

I love the "Explore" tab. May I suggest adding a button to like the song that's playing and maybe also add it to a playlist???

[dreamy rock]

We really really need a feature that can edit or regenerate just a part of a song more precisely
Just burnt 8000 credit to get 1 song right and still failed

[ember mountain]

Maybe in the explore Tab a search bar for specific Genre Like Rock than it shows every Rock Genre

[dreamy rock]

I'd use it even if it costs more credit than generating a full length song

[naive quartz]

Minor suggestion but a lot of effort, would love a UI change when creating a song from the current song list UI to more of an Adobe Audition wave form editor UI. Parts generate sideways and attach to each other as the song forms to get a better feel that auto plays through the tracks you have created for that song only.

Minus the sliders so we don't accidentally delete parts of a song we wanted. Using the current "From "this" time" format we already use.

With that as a base line UI, making changes to be able to capture a middle section only of a track like people are asking for would be more easily done.

[wise moat]

Idea: What if there was an option to compile different verses into one another for more consistency?

IE: You prompt the first verse, pre-chorus, chorus, bridge, etc. Then you can piece them together in any order you like.

[feral aspen]

Suggestion: Add Extend and Reprompt buttons to the fancy new Explore song. So many cool starting points there.

[serene eagle]

Bring back and entire tab for new songs. Now almost nobody listen to my songs.

[thorny arrow]

It would be great if one could provide a previous gen and get a new song that had the same style ... or could get the seed used to generate a song and then regenerate with the same seed, but different lyrics and cues? Some way to make several songs with consistency across vocals, tone or style...?

[serene eagle]

Having a lyric video would be amazing if possible.

[feral aspen]

In the meantime, you can still generate AI lyrics in custom mode. Type your song topic in the lyrics box then click "Make Random Lyrics". Then type the style in the separate box.

[peak oyster]

It would be nice if there was an easy way to get instrumental versions of the songs we created. I wasn't sure if this was already possible as I couldn't find it

[obsidian snow]

Suggestion: When you run out of credits on a paid plan, it gives you 50 credits per day (capped at 50), like with the free plan, so we can finish some songs or start with new ones while we wait for the renewal date.

[delicate comet]

SUGGESTION: I produce music professionally and getting the right vocalist is always a major challenge. If we could upload an instrumental and have vocals generated for it, that can be downloaded in stems by themselves, I’d pay quite a lot of money for it. It would have to be in WAV, not compressed though so it can be used

[shut iris]

Is there a way to integrate a theremin?

[delicate comet]

Should be a paid option for wav or it can’t really be used properly. Apparently it’s something to do with the AI though. It can’t render to high bitrates yet

[vivid matrix]

Started using udio earlier today, they provide 320kbps

Quality is actually pretty nuts with that one

[rocky oak]

hello

[lime moss]

Suggestion: I'd love a way to edit the lyrics of a song after it's been made. Because when you run out of time but still have lyrics left I haven't been sung yet it gets repeated twice when the whole song is strung together so there's a bunch of repeated sections of lyrics that need to be edited out

[thorny arrow]

It would be very cool to have a "stream mode" where "Extend" would just keep playing until the user runs out of credits (1 credit/15sec?) or presses an "end" button where the song would wrap up with an ending.

[feral aspen]

let's think bigger: you can type like mad while this happening and the music style is whatever is in the style box at that second and changes every time you change it

keyboard dj-ing

[thorny arrow]

Yes, dynamic changes to the parameters would be cool... and the vocals / lyrics.

[toxic tundra]

Suggestion: Don't generate an image of a bird just because a song is set to instrumental. Figuring out what song is what gets really confusing when they all have basically the same cover.

[ember drift]

Any chance of light mode?

I really struggle to see stuff with dark mode.

[tidal wharf]

I hate how “multiple add to playlist” works right now. I go from generation to generation by clicking and add each to a respective playlist, but guess what, Suno considers it a multi-selection unless I manually “unclick” the previous one and adds all the clicked so far to the playlist

[tidal wharf]

There might exist a browser extension for that. It definitely does in the other direction

[ember drift]

I currently use a browser extension indeed.

[azure needle]

just add jersey club kick support 💔

[covert spindle]

there are some difficulties that should be addressed in the future:

1. More accurate response to prompta. In about half of the cases, the AI gives me a female vocal, despite the fact that the prompta clearly indicates a male one.
2. There are obvious difficulties with the extension. For example, most often, when trying to extend a composition, the melody in the next part changes a lot, and sometimes the rhythm. In addition, the AI stops following the text and starts talking nonsense, splitting and gluing the lyrics in any order. Maybe there are some special text hints to minimize this?

[dusky charm]

a "send all (thumbs down) clips to the trashcan" button would be grand

[lapis osprey]

Like inpainting in image AI

[lapis osprey]

For endings, I always use [End]. Mostly I use [Outro] followed by [End]. Seems to work pretty good. You can also add lyrics or Ooooo or Aaaa to either of those for good effect.

[leaden marlin]

No

[worn atlas]

In the Library window we have now "songs" and "playlists" tabs
In fact you should add à third Tab called "clips"
Full song will go to the "Songs" tab and clip to the "clips" tab
Add subfolders In the song tab to link with related clips

[feral aspen]

Problem: A lot of users keep trying to Extend the same clip with the same prompt, and the lyrics keep not quite lining up smoothly. Experienced users will then try changing the Extend Time or their lyrics prompt, but this isn't obvious to many users.

Suggestion: If the user does not specify a Extend From time, instead of defaulting to exact end of the current clip, randomly back up a tiny bit (up to 1s?) from the end instead each time. Only do this after the first normal end-of-clip generation, and only when no part of the prompt (lyrics or style) was changed and the user has not specified a time. (they didn't type in the end of clip time, they left it on default.)

Theory: Jittering the time even just up to 1s may be enough to fix many not-quite-there extended clips with brute force, and at the very least, users retrying will get a little more range of possible outputs. (Test this theory first, of course, since I can only speculate.)

[shell quest]

**I need a feature to search through my library ! **

[sudden grail]

Anyone found a reliable way to extend a song with a guitar solo?

[rigid fable]

could you add a search feature to the home page, so you can search for specific songs

[wet pond]

Suggestion: Add the option to bulk delete tracks
**Addition information:**Many times I need to generate over and over again to get the sound I'm looking for and have many tracks that I no longer wish to keep once. The ability to delete in bulk would allow me to keep my library cleaned and more usable.

[tulip pumice]

Need affiliate program for us to spread the words.

[wet pond]

Suggestion: Add the ability to use the 'sound' (seed) of a song with different or updated lyrics.

**Additional information: ** There have been multiple times where I generate a song and the style, vocals, etc are really what love, but then i notice that my lyrics have a mistake, or something doesn't quite flow as well. Unfortunately today, as I hit generate again, I'll likely never get that same seed from the generation and will need to regenerate many more times to find the right 'sound' i want again.

[timid veldt]

Suggestion: Option to regenerate song artwork
Additional Information: Everything is birb.

[native forge]

It would be nice if you could:
-Choose song bpm
-Choose song instruments
-Choose song structure (Example: A B C B)

[upbeat rock]

It takes 5-10 rerolls to get a proper solo in my experience

[sudden grail]

And do you specify the solo in the style or the lyrics prompt?

[uncut tundra]

id love an option to save a voice/singer type for other songs.. or is this possible already somehow?

[upbeat rock]

I do it as the only thing in the style

(instrument)Solo solo solo solo

U can sometimes set solo mood. Like happy, fast, sad, slow

[wet pond]

**Suggestion: ** Provide detailed documentation formatting and how it may influence the song generation.

Additional information:
I'd love to see what words and formatting are taken into consideration, and potentially how to add weight levels to what I'm prompting or writing in lyrics. It would be great to understand at a high level how interprets what I write in genre and custom lyrics.

Understanding what sort of options I could type in to genre to further influence to get the right sound would be great, for example to know if I write "Underground rap with heavy bass hits" would actually do anything different than just "underground rap".

Also, Formatting (line breaks, punctuation, caps versus lowercase) for custom lyrics and how it influences the generated song, would be great to get closer to the right sound we are looking for.

[upbeat rock]

Example. Skip ahead a min if ur in a rush: #💬┃general-chat message

[wet pond]

**Suggestion: ** Add the option to specify what the primary and secondary sounds or vocals should be in a section.

additional information:
If I'm writing a barbershop quartet, I may want the vocal lead to start, then in the verse 2 only add the tenor, verse 3 add the baritone, chorus have all singing. A way to also indicate that this should be a strict vocal solo for the lead, or whomever, would be awesome. This would also be great for writing duets between a man and a woman.

The option to also indicate what sort of sound you'd like in the background such as ooohs and ahhs, long sustained vocal notes, vocal beats, specific instrumentals such as having the instrumentals gradually or quickly reduce from the full band to a single guitar.

[dim seal]

Not sure if this is the correct place, but is there an API for Suno? Would 100% pay for API access...

[small beacon]

I would like to have the possibility to work with the same song. E.g. select part of the song and ask to add guitar or change bass with a special prompt. Many times there is a very good part in the generated song, but remix/extend changes the whole thing.

[heady pier]

Suggestion: It would be nice if you could give indications of the structure of the song and perform it in its entirety and not just a few minutes. I have also seen that they have removed the remix, that would be good if they put it back in so you can make variations of songs that you liked. In the future they could put some boxes to place the structure of the song and you can design it as you want and within them you can indicate what you want it to do, for example... you put an intro box and inside it you can tell it what you want an intro with violins, then you place another box that is the verse and inside it you tell it to continue with the violins but now add some drums and a bass, and so on with the entire structure, once you have all that structure finished you will create it complete.

Another point that would be good is that you could add example audios so that it makes music similar to the example audio, so at least we can know that it can make something that we might like or the style we want, because it is good that they are built randomly, But many times we spend points with many songs that we did not like and the same thing happens with extensions, many times we create extensions by spending points so that later they come out almost the same or bad or that they do not pay attention to the instructions.

[oak escarp]

Not work /chirp

[livid ruin]

Suggestion:
Don't stop what you're doing. Your service is fantastic and sounds like something coming from an actual studio.

You could make an option for “live” sounding music like Udio but please I prefer the actual studio-like sound that Suno makes. I think people will tire of the style Udio creates until they can make proper clear songs that don't sound like they are in a tunnel.

[spiral marsh]

A button to "remaster" a track, even if it costs more than a normal gen?

[frank jay]

Suggestion:

Multidownload button for playlist
add BPM and KEY under track name
Stem file export or atleast MP3 320kbps or 16bit WAV quality for the highest tier sub

[reef grotto]

I can second that. Same story here. I'm a producer, but I don't sing + getting right vocalist is a struggle + expensive. I'm impressed what Suno offers, so I'd love to get STEM files, especially vocals. I'd be willing to pay around 40-50€ per month just for a service like that

[vale trench]

guys maybe add a thing in the UI that says Commercial or Personal Use next to tracks

since FAQ says you dont get commercial rights to songs you made under free even after upgrade

[real frost]

I wish I could do a prompt + style so that it would generate different lyrics with a consistent style

[rich pecan]

if Consider looking into the iterated versions and their corresponding prompts (for personal review only), similar to LangSmith or Git Branches?

Because I found that multiple versions of the Segments in the generated Video already had Prompts and Titles, if they could be displayed in a Tree-like structure, it would be clearer, and users wouldn't have to record them themselves like what suno.wiki.

[cosmic geyser]

MIDI & STEMS PLEASE !!! ❤️

[native grove]

• Have the option to control how many songs are outputted.
• A length slider that can do shorter or longer than whats available now, but also allows you to generate a certain length instead of random.
• The length of an output would change credit costs. Example: under 30s would cost 3 credits instead of 5. Over 30seconds to 2minutes continues to cost 5 credits, over 2minutes will raise by a credit every 10 seconds or something.
• Length slider can work with extending in the same way.

[mint thorn]

In view of the announcement that has just been made, I assume that the developers have stopped listening to the community? the turnover being sufficient I imagine...
It's still sad to see that we've been crying for weeks for audio quality and what are they giving us? the possibility of having a nickname! AHAH yes! thank you so much ! 🤡 🫶
Crying for stems about mounth.. nope nickname ! eheh :3

Sorry maybe i gonna be banned... but i need to say this.

[formal spear]

Suggestion regarding subscriptions:
Comparing the free plan with the Pro Plan, you "only" get 1000 credits more.
My suggestion would be to still have 50daily credits added (counted separately so with the same terms as the free plan, they only renew if they were used up) for all paid plans.

[lusty nexus]

Tag our profiles by musical genres, and allow people to search public songs / profiles for certain genres. Improve discovery!

Getting on Trending requires luck in the New list 😦

[plush scroll]

Suggestion: able to generate an intro or extend at the beginning of a song.

[fickle hedge]

Be able to import the Instrumental you made and mix them with the lyrics

[glossy wasp]

suno insane: https://suno.com/song/c15f0251-fbac-4a30-a3e1-002dbc78cb79

[rigid aspen]

I would really love to have an option to chose the voice. The best would be, to synth my own voices as the base of the songs.

[hallow tide]

Suggestion: Search for Explore

[sudden vessel]

Suggestion: if you have predefined prompts that guarantee an outcome, it would help tremendously when creating songs. As an example "continue the beat and guitar rhythm" or "continue the melodic pattern add more technical drums". It seems some prompts always work with little randomness, especially related to vocal or ayth/keys.

Ps. If there is a prompt cheat sheet and I missed it, my bad

[latent ruin]

Thanks. I did that and finally I made my first music.

[sturdy niche]

Make Suno extend longer tracks not just minute or shorter

[sudden vessel]

Shorter ones seem to be bugged in most cases, reporting them or marking a bug will at least grant the credits back xd

[sturdy niche]

ok and?

i prefer to make it longer tracks not just bugged 1 minute at most

[sudden vessel]

All I'm saying that a short one that lasts seems a bug in all cases as typically 1 minute track Is made

[sturdy niche]

still too short

it should create another two minutes at most

and you can just continue from any moment you want

[sudden vessel]

I both agree and don't agree. Sometimes I want it to make 20/39 seconds that matches what I want to achieve in a song. Otherwise I don't mind ( or actually appreciate) a longer generated song. I'm on board with that idea

[sudden vessel]

This

[sturdy niche]

it should depend on how long the lyrics are

[ivory condor]

I made more than 2000 songs/clips with Suno , now we need a SEARCH thing in the Library, I also think the Library could do with better display as in larger thumps , half of the songs deleted already , I suggest adding more futures , I have been a paying memmber for few months and shall continue supporting Sono . thanks guys .

[sudden vessel]

Suno does not 100% follow the input lyrics, I've had plenty songs where it cut them short or picked random lines. It's not 100 % gyaranteed

[sturdy niche]

yes that's why i want them to make it better and depend on lyrics

[sudden vessel]

Yeah, it's a pain when you get the desired music but lyrics are a mismatch

[wooden ridge]

Suggestion: Descending garbage instead of ascending so you see your most recent deletions first. I don't want to have to scroll dozens (or hundreds) of pages to see my most recent deletion

[sturdy niche]

thats right

[sudden vessel]

Thnx for a good chat, off for now

[shell mirage]

If it's feasible for Suno to create a "Refiner" capability where you just select the Clip or Full Song, and comb over it to Master it/Remove Audio Distortion/Reduce or Eliminate Echo, that would be dope. It would be better if it actually just took the melody and everything and copied it, but made it better, so that it regenerates the song with improvements.

[sturdy niche]

and also always have an option to correct your extension of song

[vale trench]

suggestion: update server icon to the the new suno logoing

for more consistent brand or something

[viscid cloak]

Would be awesome to have some Social Links at the profile and put your Playlists at the Bottom

And maybe a Showcase with the Trending Songs?

[polar tapir]

suggestion: add a seperate option for making a new thumbnail for your song.

I made a couple of italo disco songs and they had great thumbnails, but when I made the last clip (which wasn't instrumental unlike the others, because i had to signal the AI to end the song via "[End]") it chose a bird with a guitar as a cover, instead of the previous city neon scene. I'd like to be able to change that.

[surreal steppe]

Suno:

• Prompt weights, where
• 1 indicates no preference (default, same as not applying weight)
• >1 is higher preference
• <1 is lower preference
• 0 avoid at all cost
• Example: pop, (uplifting:1.2), (retrosynth:1.4), (autotune:0)
• Negative prompts. Allow users to input negative prompts. This would be a blessing for those of us who don't want FFDP-style autotuned vocals in our thrash metal tracks, or T-Pain vocals in our 80s chillwave.
• Optional cover image generation. It just wastes bandwidth and GPU-hours in 90% of tracks (you know, the tracks that are throwaways since you're trying to nail down the genre-prompt since there is no way to preview, say 10 seconds of a track...)
• Previews. Allow users to have a 10 second preview of a track before generating the whole shabang. I'd gladly use 10 credits on 10 x 10 second previews rather than 2 x 120 second throwaway tracks...
• Seed control It would be sweet to have this. *Implicitly (now explicitly) assuming that the model is static and deterministic.

[shell mirage]

Would seed control be like being able to replicate a song verbatim?

If so, I agree, and that would help with many things

[visual brook]

i request the profiles all have a simple RSS feed on them too, so that i can use an android RSS feed app to see when community members post new songs on their profiles as public

a simple RSS feed that goes from my android notifications and consequently to my garmin watch while i work

[wooden ridge]

Suggestion: ability to follow a profile, aditionally be able to see all your followers, and people you follow.

[delicate cedar]

Do you think it is at least possible that these new items were longstanding requests along a road map that were simply released as they were completed?

Or maybe that a group of developers can work on more than one item at a time?

Maybe even that to create even the possibility of stems is something structural that will take a while to complete because it requires fundamental changes in how everything else works?

Are you really truly upset that major structural changes were not completed in mere months?

[simple crow]

Allow for no lyrics (for instrumental tracks and stuff like EDM dance music.) Too often suno uses the prompt as lyrics even if you ask it not to.

[marsh isle]

The ability to highlight your tracks to display at top of profile

[tame tundra]

are you checking the "instrumental" toggle?

[visual brook]

seed control could get into "banned seeds" and DMCA issues if temp is also allowed to be set to 0, so this feature would need a robust regurgitation regression test

[simple crow]

Wait there's an instrumental toggle? 💀

[tall latch]

they should add suno ai in app directory
, it is one of the best bot and funnest bot to use it with friends
the bot is not in app directory , thats why most person dont use the bot
it can be better than mid journey

[viscid cloak]

And i hope you can Follow some Profiles in the Future

[shell mirage]

Well, they will need seed control on the backend either way. If they want to Master or Refine clips/songs. They will need a way to copy a songs generation verbatim and make it better or paint-in

[feral aspen]

I also appreciate the social potential of Suno bot. I hope some day Suno Chirp Bot returns better than ever 👍 🐣🐦

[jolly inlet]

I think the only thing I'd truly want from Suno is to be able to use a previous gen as a reference.

[frank jay]

Add BPM / Tempo range for render

[shell mirage]

That would be good for both the user and for Suno

[jolly inlet]

Like I have this gen that sounds amazing... but it screwed up and missed ONE lyric within the middle of the song so it doesn't make sense 🤷

[shell mirage]

paint in feature is already coming

[jolly inlet]

👀

[shell mirage]

but I would like for it to not only paint in

but to improve the sound and remove audio distortions/echo etc.

[tame tundra]

I think seeds and parameters (temperature etc) would help with that

let advanced users keep the seed and prompt, but tweak the specifics etc

[shell mirage]

Right, yeah, we are all talking about the same thing

[surreal steppe]

IMU: seed != denoise. Latent noise generation is the only thing the seed should influence. Assuming a diffusion-type model, a seed of 0 generates a equally denoised output as any other seed, only the result is deterministic. This allows for minor tweaks to a known result. ||temp, denoise factor, or whatever you designate it, was not mentioned, as it is obviously detrimental to introduce that... :P||

[tame tundra]

Correct, temperature, scales (think cfg in stable diffusion), or whatever parameters are sent to the model for inference

[shell mirage]

I'm not familiar with a lot of these advanced terms, but as long as they are explained in a wiki somewhere, sounds good to me

[tall latch]

i am sure it gonna take place after midjourney

[elder badge]

it'd be great to have multiple general chats - it's too fast to chat about much of anything

[tame tundra]

I'm not super familiar with chirp, but in other models that use the same base, there are numbers and strengths you can tweak while keeping the same "random" result. So for stable diffusion you could get the same face, but with less intensity of the prompt. So if you put in "blue eyes" with a CFG (think of it as strength) of 9, the eyes will glow blue. If you use a CFG of 3 the eyes may be normal, but the image may look less stylized overall

[tall latch]

only if it comes

[marble halo]

Hey! Probably already suggested but a search box in the library would be amazing

[visual brook]

good reply! just for convenience, suno previous work is not diffusion-type, but called "bark" and i think it's a audio encoder based GPT (transformer)

[shell mirage]

Yeah, I used Midjourney a bit, so I am familiar with weighting. But it's been a while

[mint thorn]

No idea, we don't know the road map :/ so it's just speculation

I may be a little bitter, it's true, but I'm sorry having a nickname is not a useful feature :/ in my opinion...
or perhaps not a feature missing to date which has significant repercussions on the way of creating music.

[shell mirage]

As long as we can paint in, and improve songs either automatically or via minor tweaks via advanced command (OR BOTH), that sounds good to me

[surreal steppe]

With the prompt, and seed, another user could make exactly the same track as you (assuming that the model is deterministic).
This could also be useful for doing minor tweaking on a track.

[tame tundra]

temperature and a lot of those parameters still exist in a token prediction/non-diffusion arch, so it's still a valid suggestion to expose them once they've got stuff polished up and stable

[shell mirage]

Thanks for explaining, that's what I figured was the case. This would also be good for Mastering/Improving Audio/Painting In, etc.

[tame tundra]

yep, hopefully they get to that point, it'd increase the value of the product to have it produce consistent and customizeable results

[delicate cedar]

That's valid. You don't know the roadmap. (We moderators don't either.) The difference is perhaps that we've heard them discussing these things as QoL improvements for a while now so we expected it to come sooner or later.

I suppose some sort of general roadmap would be a nice feature. But would people get upset if there were delays and the team was working on fixing various things or speedwalking this feature over that feature? I'm sure there are reasons why there isn't a public roadmap.

[mint thorn]

I didn't know, sorry if I seemed too haughty.

Yes, maybe there will be unhappy people, but I think many people wouldn't say the same things every day because we would have an approximation of the release date.
I think that psychologically for the community it is better to have it, even if there is a delay, at least we know the progress in progress, there will only be some stupid people who will complain about the delays because they do not will not be aware of the difficulty (like me sometimes 🤐 )

[brittle hatch]

how to use speicfic vocal like man/women ?

[surreal steppe]

In case of a transformer, it's a bit trickier, but still doable. Randomness in the initialization, data shuffling and dropout can be controlled. Deterministic mode is also an option, but it disables some optimizations, leading to higher processing load. Some GPU-accelerated convolutional NNs are nondeterministic, in which case I'd be happy with just a preview function...

[tame tundra]

put "female vocalist" or "female vocals" or "female singer" in the style prompt

[quartz bronze]

Make it possible to search songs by name please

[delicate cedar]

Usually it helps to list it in the style prompt.

[haughty valve]

For the profile can we have a hide profile for songs?

some of us have some brand schizophrenia lol

[haughty valve]

Hide From Profile would be the name of the feature

[feral ether]

maybe a channel to ask questions if someone has something they cant find the answer to?

[haughty valve]

that's general chat

[feral ether]

yesh but its way too cluttered at times for that imo

and a lot if things get missed

[haughty valve]

we try

it doesn't help that a lot of questions are FAQ questions

[feral ether]

would just be convenient to have a direct channel for it for everyones sake imo

keeps it out of general too so it can maintain its own flow

[feral ether]

thats fair

just a help channel would be neat is all

[mossy belfry]

I wish I could save or riff on or share the songs in the new Explore 😦 sucks that they just disappear into the aether

[visual brook]

doesn't "public" have the same function ?

[haughty valve]

no

public or not public

it's still up there

[visual brook]

oh. my.

[haughty valve]

That was the first thing I tried

[visual brook]

i will not look at your profile until that feature drops, will only keep the hymncore main playlist open

[feral aspen]

I tried it quick and it did add and then remove a song from profile page when I toggled public off and on. Maybe it's cached and takes awhile though (or buggy)

[haughty valve]

Yeah, cached, you're right

[marsh isle]

A way to highlight a track to display at top of profile

[bright talon]

i wish i could add in a male character into a verse to switch it up a bit kinda like a feature. Or vice versa

[visual brook]

hi suno engineering, limit profile name changes to once per 6 months to lower your cache/redis usage lol - i believe in you, you can scale!

[zinc harness]

Make the thumbnails generated by the user or you can choose any one based on how many parts were made instead of keeping only the last one that was generated as the cover.

[nova prawn]

the program doesn t use violin or viola in a song

[viral juniper]

Accepting payments in Bitcoin and Stablecoins for people who do not have credit cards to make international purchases 🙏

[sweet ibex]

Top of profile, a showcase bar, allows user to put their favorite top 5 songs right there. Along with multiple pages to see all public songs from that profile.

[jagged fable]

100% agree - The server is way too busy for all discussion to get shoved into the #💬┃general-chat channel. Especially when most people use it to share their songs which takes up a chunk of the page due to the embed.

A #prompt-crafting channel is really needed for people to discuss techniques, strategies, keywords, metatags, and generally how to best utilize Suno

[sweet ibex]

I also feel like there should be some contextualization over gen's with paid access versus unpaid. On this profile I have some stuff I made during my sub, and now I have stuff made without the sub active, and outside of memorizing which was which, theres no indicator to the user to help them go back and find something they might want to publish. Maybe a user-end only system that marks the paid gens with some sort of tiny icon in the corner of their thumbnail.

[feral ether]

aye, I was thinking that too. i was gonna mention a group collab channel before remembering thats not actually a function, but the sentiment still remains that it'd be better to have dedicated places for questions and song generations

[wheat quartz]

In the home page, where you can listen to other people's songs, there should be a opportunity to filter on categories, so you can listen to categories that you are interested in.

[strong swan]

i think the discord needs a room where people can put the profile pages. just as a idea. i see them scrolling by so fast.

[last jacinth]

Please, if you are part of suno’s team, please let paid users use double credits to generate a 4 minutes clip at once instead of blocks. Second and third blocks suffer a severe downgrade that sometimes makes the songs unusable

I have wonderful first 2 minutes block that get completely broken in blocks 2 and 3

If your songs are dreamy with echoes it gets completely killed in extensions

I would gladly pay 2x or 3x the credits for a 4 minutes generation

[pallid horizon]

Hey, my suggestion is that, it would be totally cool if theres a feature to just take the instrumental cleaned out of a song we make, since lets supposed i created a song with bot vocals, but i wanna make it my own, and i wanna be the vocals, so it would be so so great if theres any way to get the instrumental cleaned

[last jacinth]

It sounds completely crap

Let me show an example

https://suno.com/song/daaab139-35da-4cb3-b143-d798f79f871b listen to the crisp quality

Now listen to the continuation https://suno.com/song/98360fa0-0bd6-4abe-9da5-7861e003b5fc

Sounds completely compressed and crappy

It happened to me DOZENS of times. I’ve literally lost half of the credits in crappy blocks like these

This should be prevented if you let@the generation make an entire song at once. I don’t mind paying 3x the credits in exchange of a superb quality all of the song

I am thinking on stop using the tool because of this despite I’m loving it more than any other recent discovery. I’m enjoying it so much I don’t mind recharging it once and again… but this quality issue is starting to making me feel frustrated and bored because I can’t finish the best songs

[feral ether]

you are able to make just instrumentals by clicking the slider for it

[pallid horizon]

i get that, but like i love the structure the bot vocals gave to the song and thats where my suggestion goes, thank u for responding 🙂

[feral ether]

oooh, then yeah, I get that

ive suggested the same thing before but idk if it's possible or not

the ability to separate the vocals and instrumental in case you only actually like one of them and wanna keep working with it

[subtle barn]

Prompt editing of a compiled songs/pieces would be great so it won't look like a mess 🤔

[feral ether]

aye

[pallid horizon]

yess, that would be so dope, it sounds really hard but i feel that if they already created this incredible ai, its not so impossible haha, they're genius

[ocean inlet]

When can we alter the lyrics of an already generated sound, I’ve been trying for hours to try and semi replicate one 💔

[subtle barn]

Only with "Extend" from particular time stamp (don't forget to shove proper part of lyrics in the Lyrics field).
You may probably also need a billion credits too.

[paper seal]

when you press the extend button, type other words from the second the words start and continue

[subtle barn]

Like seriously, I wasted like 150-200 credits because it keeps making extensions that sounds nothing like existing parts of the song 😢

[jagged fable]

One of the most useful potential features I can think of would be one that is able to tell you which words in your prompt are actually affecting the output. And furthermore, how much "weight" each word has in the prompt.

For example, my prompt could simply be 3 genres: Genre A, Genre B, Genre C and the outputs will clearly show that Genre B is consistently having the most prominent effect on the output, implying it has more weight than the other 2.

On the flipside, when using less common terms (such as some music theory terms), it is really hard to tell if those are making any difference at all. Then it just becomes a waste of credits trying to figure out what works and doesn't work. Having no reliable guide (besides a user made wiki) just adds to this constant game of trial and error.

[deft wraith]

Is suno ever gonna makes stems and midi available? would be insane if it got to that!!

[hybrid crane]

It's on the roadmap

[jagged fable]

There's a roadmap?

[hybrid crane]

They have a roadmap... it's not public

[jagged fable]

oh... Well it'd go a long way if they make it public I don't see why they wouldn't

[subtle barn]

Even with just general goals. Without any details.

[jagged fable]

They should take note of how midjourney does it with weekly updates on what's in the works, general timeframes, collecting user feedback on what features are most important, etc.

[deft wraith]

you domt go public with dates in case they move and you dissapoint

[jagged fable]

the #🗳┃mini-polls is a great start to this

[jagged fable]

then simply add *subject to change or mention that it's not guaranteed

[deft wraith]

altho, giving us an idea of stuff that is in the pipeline would be AMAZING 🙂

[jagged fable]

Yeah I'm much more likely to keep my subscription and continue to support the service if I have an idea of what's to come in the future.

[deft wraith]

I'll subscribe for a month for now (i subscribe and instantly cancel) to get a vibe for it. ill prob come back and forth to it. i see amazing potential

[deft wraith]

Probably the most requested thing which is why i 100% had to request it again. take my money! give me stems!

[hybrid crane]

I love to take your money.. but I am just a volunteer moderator 😉

[inner sonnet]

Would love to see the ability to add vocals to your own inputted music in the future

[calm heath]

Happy Friday. How can we create full songs on the first try without extending them?

[ionic skiff]

an option to generate further lyrics for the song

and also, to get the AI to hold the same style for the next track - we need such an option urgently!

we're losing credits otherwise!

[inner sonnet]

This one is so true, need this feature ASAP

[lapis osprey]

Fantastic! Another thing to try is get AI to describe the style of your favorite song/artist, then summarize it into the style field. Usually works very well.

[last jacinth]

Aldo please add a clear button and the possibility to stretch I out fields in cellphone, specially the lyrics

[cunning yacht]

I LOVE THIS TOOL. About every 7th or 8th song is a BANGER. Suno is amazing. I wonder if the devs would consider (for people like me who write their own lyrics) A simple Clickable "Song Structure Template"? I realize there are 300 + different structures, but when you follow the structure more closly the AI usually EXCEEDS expectations. Just Click on the (for instance) "Pop Template" and it shows [intro] (your lyrics) [verse] (your lyrics) [Chorus} (yours) etc. IDK is that a dumb idea? Anyway, conteinue the fine work.

[dusty aspen]

Would love to sample a verse from parts of the song that fit exactly what I'm looking for with the ability to work around it.

[woeful rampart]

please let the songs continue to play while i browse the site

[remote rock]

Feel like these channels should be Discord threads as well so I could see what was already suggested and talked about

[vapid isle]

BRO ! i accidently clicked a song while i was in " create" TAB and now its all gone...the song i clicked take the whole screen and boom, i clicked back but my text is goine...can you make this better ?

[charred basin]

Can we please.... for the love of AI.... please can we have 2 volume buttons, 1 for music and 1 for voice? I hear an amazing piece but can't use it because I can't separate the 2... can I download voice separate from music for further edits?

[teal perch]

How realistic is the scenario in the future. So that you can download an audio file in its likeness, could suno do something similar?

[deft wraith]

is it just me or does it make 2 versions every time i generate

[charred basin]

You are correct. Its 2 variables

[zealous parrot]

hello please, is there an option that allow me to have one type of voice like one style one personn singing all the songs that i create ?

[wind abyss]

it always 2 ( because the power of 2 : is strongest)

[deft wraith]

1 winner and 1 loser

[remote rock]

Also I want to follow other profiles if possible

[torn lantern]

When writing lyrics is there a faq on special text commands to insert and how they work? And if not then that would be really nice to have. In fact if the lyrics were treated more like a scripting language then could really control things that way, maybe?

[remote rock]

https://www.suno.wiki/faq/metatags/

[pseudo venture]

Major suggestion, because I can’t stand this.

When it says can’t make the song, have it at least point out what needs to be changed or removed.

Am doing a fucking guessing game with what needs to be changed.

Like there was once it wouldn’t let me make the song, unless I gave it an extra line of blank space.

[remote rock]

From what I've heard gpt does the automod so putting the lyrics in and asking whats wrong with them should work

[desert whale]

That’s really exciting!

[twin silo]

Love suno, it's one of my favourite things to do. The default video has a random picture that isn't always the best. I wish it had an "srt" export so we could make our own videos more easily

[acoustic pike]

Please add music style - Indian Devotional and Indian Classical

[surreal current]

I thought, can I add some instrument tags to the composition page to make it easier to add the desired content in the songwriting, for example, I plan to add some instruments I want to this song, I click on the instrument I want to add, for example, I want to add bass or other instruments in minutes and seconds, I will add the bass tag to the time paragraph I want

[last jacinth]

There are ai to separate them. I use one and then mix and master everything in logic audio

[rapid edge]

Would be nice to have a way to edit lyric text after merging songs from "extend". Appears to not properly merge if you extend from say half way into a two minute song.

[tulip fox]

A channel to showcase your favourite webui generated songs (bark and chirp dont mean much to me - always use webui).

[tulip fox]

Ability to edit the lyrics after the song is created to match the lyrics to the song (as we all know the AI doesn't always get it right, but still can create some great songs that stray from the lyrics).

[charred basin]

Please give me this wisdom!!!

[deft wraith]

https://www.lalal.aI is one

[pseudo venture]

Something also to suggest.

Allow a sort of editor to make cuts maybe between or after the song

There are times when at the start it'd take too long for some instrumental

Or at the end where vocals unnecessarily repeat some lyrics and just cuts of inbetween.

[opal storm]

should be able to generate your own track names

with the dice

[leaden marlin]

New suggestions for Suno V4:
• be able to import a reference song with a link
• be able to create voices in Suno and save them to make songs and make collaborations between these voices
• be able to write much longer song lyrics
• Suno mobile app
• Having the option to make songs that sound like a concert, as happened to me in this song by chance:

https://suno.com/song/ce5dc85d-3a39-47d9-bd57-f4f3352f269d

[deft hare]

Please add the ability to add audio that you need to focus on when creating a cover

[leaden marlin]

don't know if it's been mentioned, but could we get some folder structure for the "Create" section? not sure if I missed something, but the page gets messy fast with all the generations in just one list view

[rose elm]

What would be truly Amazing!? Output MIDI for each part/instrument!

[timid vapor]

trying to get specific instruments and genre is hard, like shinto prayer music and gamelan

[twilit hedge]

I'm seeing more success in creating concert style by specifying "live" in the Style box... it works in many genres and styles too..

[real lynx]

Two suggestions:
Seeing the waveform of each song would be nice. visuals can help keep tracks organized

Building off of seeing the wave form, being able to edit specific sections would be nice (Think 'inpainting' from generative images, but applied to music. In-composing? haha)

[warped herald]

It might be a little bit to advanced, but it would be nice to have some sort of editor. For example, it you like how verse 1 came out, to be able to cut it and paste it further into the song and let the AI smooth the transition.

[reef basalt]

Not sure if it's been mentioned before but having a way to know what styles of music Suno decided to use when you just input lyrics into the custom field and leave the style blank for it to generate something would be nice. If its possible. Had plenty of good music be generated from leaving the styles section empty while inputting my own custom lyrics but can't tell what styles were used after it generates cause the section that normally tells you says "No Style".

[reef basalt]

Also the ability to change your music art.

That "bird" haunts us all. lol

[barren osprey]

when it gets flagged by moderation, it should tell you what part is flagging it 😵‍💫

[soft geode]

Song editor would be nice, so that you can cut out the song or something.

[stiff hollow]

So I'd like to throw out a suggestion simply to help with continuity. If there was a parameter similar to [chorus] that allowed you to specifically say "Repeat the chorus that we already created earlier in the song" I think that would be really helpful. Too many times when I've been extending a song, the chorus gets redone with a different melody. It could literally be as simple as [Repeat Chorus]. Not sure what the backend programming required for this would be.

Second, if the downloaded audio quality could be better (320Kbps), that would be very helpful.

[dusky mountain]

Organizational suggestions:

• In the list of created tracks, have extended tracks (part 2, part3, etc) be shifted to the right, which will make them visually set apart and be distinct from the other tracks.
• Extended tracks are put right below their parent / original track, thus creating a parent/tree organization system.

[worn atlas]

Adding "live concert" in the style often works. At least for me.

[humble grail]

Do you guys think that tokens could be available somehow? Based on number of Likes, Streamings, or Token Royalties for Remixings and Extensions?

If we could also create in a flowbuilder would maybe be a good idea to not lose parts that we like lost in the middle of so many generations.

[feral aspen]

Try "Liverpool" too

[worn atlas]

I would prefer get an error from suno when a style is not recognized instead of generating a random style that while not usefull to me, burns my credits.
As we dont have a reliable list of styles that work, we try several. But this has a cost. So giving a clear error message (if possible) should save our money. I guess that maybe devs dont know all what shirp is able to do so this request may not be easy to implement

[brazen ether]

the only thing I'd be excited about for an announcement is them announcing clean vocals. until then it's bye subscription for me.

[thorn hazel]

Can you make an interaction with Spotify - to upload the songs?
Some of these songs can be absolute hits! 🙂

[warped finch]

probably adding voices like boyfriend and girlfriend from fnf (like you did with vocaloid)

[rich pecan]

When I intended to click on Get Whole Song, I accidentally clicked on Move To Trash instead, even though there's a 🗑️ (but when my mind is tuned into the melody, I still spaced out). Maybe it could be changed to a different color, like orange.

(I remember Delete Permanently used to be red before, now it's a 💀, I think the icon lacks visibility without color)

[quasi merlin]

I would really like the ability to produce a proper outro. Songs shouldn't just "end".

[humble grail]

Try creating a free account to make those tests first 🙂

[dapper heath]

I'd like the ability to for the first section have one voice and on the extended portion be able to change to a new voice. (Example: Male voice first part, Female voice second part)

[gusty plinth]

this is already possible if you start the style as Female - Male Duet, and you use tags for which part to be sung by who, but you still need a huge amount of luck

[dapper heath]

That is good to know, thanks for the tip!

[gusty plinth]

oh yes, for quality of life enhancements,, I hope you guys add a button to clear the canvas on the create page, removing previously generated songs from view, so we can start generating new songs without the clutters

[dapper heath]

I'd love to see the amount of times my songs are liked in my library

[worn atlas]

Well! i have done this already and when i got finally a good clip i didnt know how to continue it using my paid account . As Suno does not ouputs the exact prompt it is internally using you have no chance to get the same style.

[limber turret]

the extend feature doesn't work great is it possible to max the original output to 3 mins instead of 2 mins

[tidal wharf]

would be great to be able to un-publish all previously published songs (I just did them because it didn't mean anything, so it was like "double thumbs up" for me)

I'm trying to do this by hacking the browser console, but not quite nailing it

[toxic scroll]

Hi people! Have I missed something? It seems like Suno AI gives much more free points for song generation. What happened?

Is it a trick or something? 😄

[worn atlas]

Link Only

[arctic coyote]

"We added 500 credits to your account as a small token of our thanks for your patience and so you can explore the new improvements now live at our new-and-improved site"

[tidal wharf]

I know, but I have 25 pages of them

imagine clicking them all manually

[toxic scroll]

Highly appreciated.

[tidal wharf]

Anywawy, I got a hack in case anyone else needs it: Go to the Library => pick "Public" => open browser console => type in:

Array.from(document.getElementsByClassName('chakra-switch__thumb')).forEach(element => element.click())

This will click all the toggles, setting them to Link Only (the toggles should be visible, so shouldn't be mobile view)

spring cleaning

nice, clean as a newborn's tear: https://suno.com/@vova

now gotta start looking back and adding what I really want to make public...

[tepid trench]

If access to chirp v1 is going to be completely removed, it might be a good time to consider open sourcing it. (Only v1)

[narrow talon]

Hi, I would like to have an option to pre-select the singer's voice, or maybe an option to create a voice and use it in different songs.
Also, we should be able to use the chorus multiple times in the same song. When you regenerate it in another part of the song, it often sounds different. This often makes no sense, as it is the centrepiece of the song and should have a recognisable value.

[sharp robin]

Hey, I would like to suggest any form of partnership with content creators that want to show Suno to their audience with a referal link

I cant find anything like that here, I dont know if I didn't check it correctly but, if not, I think Suno is missing a great opportunity to get known with this method

[latent ruin]

Thanks again. I tried that, it gave me a quiet good Rammstein style music.

[heady viper]

Hey there. I'd love to suggest a feature, which at least I would love to see.

Is there a way to get the lyrics shown under the song correct? Mine are rather messed up in order and even showing text that is not sung.

Example here: https://suno.com/song/b489e395-4e86-4c07-87c1-207194c811cb

[wind crest]

really need this feature, as well as being able to reroll the cover image of a song. every time a song is finished, the lyrics to the side are a garbled mess compared to what is actually sung. this is especially crucial since one of the download options is to download as a video, which puts the incorrect lyrics on it

[broken elk]

The new https://suno.com/explore is great but it's very tedious on mobile to copy the genre string. SUGGESTION: when a genre clip is playing, the little "bouncing levels" icon next to the highlighted genre name should be clickable and it should copy the genre to the clipboard

[wooden ridge]

Suggestions:

``UI:

• Change the position of "Get Whole Song" to the top with a divider between, similar to how the 'move to trash' is at the bottom

• Change the order of the list in the Trash Bin, so newest deleted files are on the first page, not the last

• After deletions of songs the list should refresh, rather then me having to refresh the whole web page

• Allow me to use the Delete key to delete a song after highlighting it, rather then needing to select it from the UI.

• Add an additional filter view, Currently you have 'Liked' and 'Public', please add "Disliked" so i can filter it by dislikes when i wish to bulk remove songs to the trash.
  ``

• If i downvote something, please don't auto play it. It's downvoted for a reason.

[quartz steppe]

doesn't really work for me, since I am trying to get something, that is quite far from mainstream genres
I think the problem is that my feedback is limited to like/dislike system
model won't learn much this way

[vagrant rock]

Bring back V3a

please

[rich pecan]

If there is one instance of a blurry/unclear audio quality in the generated segments, it will lead to the subsequent generated audio also having a continued blurry/unclear quality. => need filter

By the way, here's a suggestion for the user experience:
can bind the <Space> key as the sole button for the player? Because every time after clicking the toggle button (Custom Mode/Instrumental),
forget to lose focus. When casually press <Space> to play audio, it triggers 'Custom Mode/Instrumental' instead.

[toxic venture]

Probably already have been asked, but a search bar in the library could be nice, kinda annoying to go back to page 270 to get lyrics.

[cedar umbra]

Would be nice to implement a feature to specify chord progression for verse, chorus...etc e.g. [Verse (C, A, F, G)] The same goes for time signatures e.g. ([Signature 4/4]). Also woul be cool to be able to specify BPMs e.g. [BPM = 90]

[twilit hedge]

It's been asked a lot... but I think clips management is REALLY needed, it's very difficult and time consuming to go back and search over hundreds of pages... also I don't even know how many pages I have in my Library.

[rotund orbit]

I miss the remix button

a lot of the recent ui changes are really messing with my muscle memory

[toxic venture]

Same here Lynn 😅

[past briar]

It would be really cool to have all my Public songs visible in my profile, at least in multiple pages. It would be easier to share my profile to friends than keeping up with all playlists

[copper geyser]

Let us favorite and follow users!!! 😄 amazing stuff btw!

[ember dock]

Hey hello, If we could get a way to modify only some words and keep everything else the same that would be amazing!

[deep onyx]

That requires distribution channels that are not available to suno. You can how ever upload your finished versions to Soundcloud, their paid tiers allow you to upload to Spotify

[pseudo swallow]

ALBUMS. WE NEED THEM. Imagine this becomes AI Spotify

Like seriously, I have a ready album named Multilayered Planning and I can’t make it actual CAUSE OF THE DAMN ALBUMS NOT EXISTING

[delicate cedar]

Do playlists not work?

[stray sparrow]

Free users have became a problem: They overload the servers to the point paid users can't use suno. I know giving a free trial is the best promotion for atracting new users, but you could modify the conditions for the free trial, I think you can give free users 250-350 credits weekly and let them spend the credits as they wish (i.e.: all in one day) with only two restrictions, free credits must be spent on weekdays and not in rush hours.
That way paid users could enjoy suno without constant downtimes caused by the free users.

Also, the credit renovation for each free user has to be randomized between monday and friday to avoid a massive affluence the credit renovation day.

[pseudo swallow]

They don’t show up in the scrolling area

[vapid isle]

Need: "Use this Vocal for Next Creation" Button so we can keep this Tone for better Beat to become

[delicate cedar]

I see.

[pseudo swallow]

No, like you can’t see other users albums / playlists

[delicate cedar]

Oh!

I'd show you something but the website seems to be slow at the moment. IIRC there is a way to share playlists.

[past briar]

Actually, ranking own profile songs by (newest, most likes, most listens) would be better

[quartz basalt]

May I get help extending a song i created and liked ?

I get a error

And tried other song to test extending and the circle dont stop spinning and already passed a long time

[leaden marlin]

add a feature enabling you to upload a sound sample to add to generation process.

[delicate cedar]

I'd share an example of a playlist in a DM to show you it's possible but you apparently have DMs closed.

[pseudo swallow]

I don’t I think, I’ll turn them on now

Done

[trail kiln]

where can I ask quetions?

[jolly swan]

Would be great to be able to rate each generation with '1 - 5' or even '1 - 10 stars'. The current option of 'thumbs up' or 'thumbs down' feels too binary.

[jagged fable]

As of now #💬┃general-chat is where you can go but you just gotta hope that your question is seen amongst all the spam and link embeds. I heard they’ll finally get around to adding a #support or #questions channel in 2025

[karmic kernel]

hi any moderator of developer available????

i have a feedback i discovered suno like 2-3 days ago and i am facing a problem i have a suggestion if that can be implimented that will be greattttt

[cursive robin]

A spectrogram or similar visualization, zoomable as well, of the song timeline would be helpful when we get get inpainting/(inplaying?) support; would likely speed up the splicing process.

Would also be a huge fan of being able to download separate stems; but I don't know if the model produces audio this way, or uses a diffusion model or similar on the full composed output instead.

[jagged fable]

Stem separation is (supposedly) on the roadmap

[karmic kernel]

i have written some lyrics and wanted to see how they might sound so i created like 3-4 variation of the same on suno and i loved one of those though i have updated the lyrics since then 3-4 times i have no way of producing the song on the same music with updated lyrics and i am hooked to the previous generation that i am not able to like any new generation there should be an option to update the lyrics in previous generation or to recreate a song with previous songs @delicate cedar @hybrid crane @void prairie

[karmic kernel]

@jagged fable do you have any views or updates on this?

[jagged fable]

I’ve seen this feature commonly requested, you’re not the only one. Whether or not it’s in the works, I don’t know. The roadmap is not publicly available. Any information about what’s in the roadmap comes from moderators giving random bits of info about it

[karmic kernel]

oh ok i am trying so hard to recreate my original generation music but it will ofc not happen and it's bothering me so muchhh 😅

[vale rose]

just a random thought I had...would it be possible to expand Suno to accommodate for the creation of audiobooks with sound effects? I mean, I don't know if that is possible, but I think that would be really sick

[vernal quail]

add Frenchcore and Uptempo Hardcore to ai, I'm trying to make one of them but it just comes up with rock and metal and that.

[teal girder]

Hi! An ability to change the cover of a song would be great!

[royal belfry]

1. if v4 is going to be higher-fidelity, it'd be nice if we could run our songs through a quality upgrade filter, so that they're just the same but they get that fidelity boost.

2. I'd heard that v4 will get inpainting to fix mistakes - I don't know if this is an inherent part of that, but it'd be nice if we could also insert lines, not just paint over but insert new stuff and push the remainder of the song forwards, so that it becomes longer, if that makes sense.

Also please add a phonetic mode or some way to know what it will sound like through the TTS before you hit generate.

[daring kettle]

Expected length when extending. So that you don't get a bunch of unwanted 00:11s extentions that ignore your input and just wrap up the track asap.

[pliant creek]

it would be nice to swap an instrument not typically used in a genre. for example maybe i want to hear nu metal with a sitar instead of a guitar

[stray echo]

We can edit the title of a song after it's done - can you allow changing of the lyrics (like the actual text, just let us edit it; especially after combining parts)

[hasty rose]

I'd love to be able to use that to update V3 songs to V4

[tame jewel]

i wish we could choose and customize the voice, as i dont like how you get a random singer (even if you have a prompt for it, the chances of getting the same voice is near 0) every time you generate a new song. would be nice to have a library for it too.

[hasty rose]

Ability to cut and paste elements of a song - Currently once it's done it's done, I'd love to be able to go to a specifc time (e.g. 2:15-2:40) and adjust that part, re-generate it, or copy-paste it to a later moment and use that part (e.g. re-using the same instrumental passage again, now getting that is a 1 in a 100 chance)

[viscid cloak]

That the Liked songs are not on top of the Home Page and the New Page was better when it was an extra site, now it kinda goes under. Also some New ones get skipped, since it just moves a whole chunk on one doing.

[true thicket]

ok, here we go. i was thrilled at first, but i think the results are getting worse rather than better over time, which is a shame. no track endings, text repetition. pure waste of credit. other AIs, like udio, have achieved better results. i don't know what the problem is, but it can't be that the track is 20 minutes long at the end just to get an end in the track.

[surreal mantle]

Are we gonna be able to clean up the lyrics on the finished songs in the future? Having verses and choruses repeated because I generated from a timestamp in the middle of a clip makes the final songs lyric look way too messy and unsightly

[sudden ledge]

Allow the ability to change the lyrics of a song but keep the same melody,

Allow the ability to put a input audio to copy the melody of another song

[west galleon]

Request for Suno v3: Really need to be able to create song project folders on the 'Create' screen. This should probably happen automatically when you generate a new song clip. The generated song clip plus all 'remixes' and extensions for the song clip would all be contained within the song folder for each song you create. So, on the main 'Create' screen you will only see individual song folders for each song you have created (and those folders should be sorted in alphabetical order or optionally by creation date). When you click on a song folder, the folder is opened and all you see are the song clips and extension clips for that song only. This would make the Create functionality much much easier to manage! 🙂

[surreal mantle]

Have you tried using [Outro] or [End] at the end of your last generation? I get plenty good song endings that way

[pearl pond]

yeah and close the roads and replace them with a bike lane

[stray sparrow]

You may like It or not, but is not fair that paid users can't access the service because free users are overusing It. I didn't ask for get rid of the free users (I was a free user and I will be again in the future), I ask for let them use suno but with some restrictions that could guarantee the paid users can use the tool properly.

[pearl pond]

while I sympathise you were unable to use something you paid for perhaps it would be easier if they just put free users on a separate server.

[stray sparrow]

That would be fantastic, but I think that wouldn't be that simple, I can't say for sure, but seems suno is growing faster than It should, to a point the admins can't keep with the massive ammount of new users; so I guess managing and mantaining two separated server farms for free and paid users It's out of the print right now.

[jagged fable]

I think a fair middle ground would be something like this or having a "slow" vs "fast" generation mode, with the fast mode being prioritized to paying members if too much CPU/GPU is being taken up from free users

[wraith escarp]

bring back the "remix" button

[jagged fable]

it still exists, it is just "reuse prompt" now. Which is a more accurate term for what it does anyway.

[pearl pond]

well they did give priority to paid users

but im sure they are working on something

[proud jolt]

one request: to include [blergh] for breakdowns prompt

[wraith escarp]

'reuse prompt' doesn't work the same way as a remix. I can just copy/paste my prompt if I want to reuse it

[stray sparrow]

If paid users can't use the service, It means the measure didn't work.

[leaden marlin]

Good evening! Please tell me how to add more time to my song, Sun generated 1:58 and the song stops halfway, how can I make it play until the end?

[jagged fable]

I am 90% sure they do the exact same thing. What did remix do that reuse prompt doesn't do? All remix used to do was allow you to autofill everything from a previous generation (lyrics, style, title).

Someone correct me if I'm wrong though

[pearl pond]

possibly. perhaps some paid users had used their credits and suddenly had 500 credits and went wild.

im sure they know

[wraith escarp]

reuse prompt produces random crap and doesn't feel like a remix, which used to allow me to look for melody variations

[stray sparrow]

Well, that's true but I would bet my balls the proportion of free/paid users leans to the free users side.

And I love my balls, thay are made of porcelain and I use them to decorate the christmas tree every year

[pearl pond]

well, you are a braver man than me. All I know is I was unable to create anything either. The problem was down to 500 extra credits most likely

[jagged fable]

It's likely placebo, here is a guide stating to another user that it was just change of wording

[pearl pond]

well we have three solutions now. Just have to see if any of them are actioned

[lapis wyvern]

yes i tried it but i confused about extend from

[true thicket]

I will try that out. thanks

[wraith escarp]

Then this feature never worked as expected. Useless button

[jagged fable]

Yeah that's why they changed it from the word "remix", it was a misleading term for what it does. All it ever did was autofill a previous generation for people to use as a starting point. It confused me as well

[leaden marlin]

Is there a way to get the same singer and song? When I use the extend button, it often gives me a totally random singer. It'll go from female to male. Or sometimes will even change the style of the song. I would like to just use the same song/singer but add different verses and stuff

[leaden marlin]

so how can this be fixed?

[woven pilot]

wish suno would tell us the bpm and key of the music

[edgy aspen]

Pls devs get rid of the white noise, it literally destroys 80% of the songs.

[dry robin]

ive paid my subscription today for the next month unless its sorted i suspect ill still have 9000 creds at the end of the month because i can finish a song, having to utilise udio to get songs, that free at the moment, is it as good in some cases yes in a lot of cases no but they are on v1. 13th of may is my deadline to generate a good song or thats it im out. cant wate 30 a month on something that does not deliver anymore.

two weeks or more and have not been able to deliver a good song, had 2 albums of awesome songs before now nothing, still had 6000 creds left before remewal today

excuse the grammer and spelling, i have had a beer #

[teal perch]

When will Suno learn how to create such a sound and style?
https://www.youtube.com/watch?v=5vgkgXt9zFg

[autumn hinge]

Let the ai say bad words pleaseeeee. 🥺😔

[earnest storm]

I know a lot of people probably already suggested this, but if not, cropping generations when extending

[crude widget]

Hi, devs! After generated a song, later when I used continue function edit the lyrics. The old lyrics would still be displayed along with the new lyrics. If I edit few more time the lyrics became chaos. Could you guys find a solution for that or at least let user erase /edit their lyrics. Please and thank you.

[earnest storm]

Wait what how do you get that

[leaden marlin]

what did I get? I didn't understand the question?

[earnest storm]

“Extend from (minute)”

Nvm I see it

[leaden marlin]

and how should it be?

I don't have a continue button

[earnest storm]

Scroll down?

Idk

[surreal mantle]

If youre just wanting to continue a song then you press extend, which is located between the thumbs up/down and public switch

[cyan umbra]

I know it's been asked and answered. But I didn't like the answer, so I'm going to ask again. Bring back v3a as an option, please.

For my use case, it makes music I enjoy more. I was on Spotify today and some of my v3a creations came on - right after I was trying to make some v3.0 music and it really brought it home how much more in tune it was for me.

I'm willing to be abused for my weird fetish. Under resource the servers for v3a. Put it behind several extra clicks. Just bring it back as an option until v3.1 or v3.5 or v4.0 or whatever is coming. Please. I've been a good boy using dislike to train v3.0, give me a reward!

[leaden marlin]

I clicked there below and they give you a choice of music time but I can only do less

[surreal mantle]

The time they give is where you want the continuation to start from so if you set it to 30 seconds for example then the next clip you generate will continue from the 30 seconds mark of the first clip

[craggy echo]

If a person could upload their own guitar track (or any instrument, I just play the guitar) and have Suno fill in the other instruments to complete the song that would be so cool. I haven't seen any other AI music site doing that. Suno is already the best I've found, but that would put you miles ahead of the competition.

[visual brook]

make a turbo version of v4 that releases along side v4 that costs less credits than v3 but is of similar or slightly better quality than v3

[edgy vessel]

• Continue option that simply adds clean end to song

• Like playlists

• Tell us if a certain modifier or genre isnt understood or if its doesnt have much training data.

• Ability to trim songs or add fade outs.

• Allow us to regenerate the AI art based on our desired inputs. (useful for instrumentals)

• Tell you what parts of the genre tag were most and least used in the output

• Allow you to save/pin certain genres or modifiers/vibes you like
  Genre ex. Dubstep, Trap
  Modifier/vibe ex. Melodic, sad

[calm sonnet]

I suggest that the continuations are hideable in a drop-down tree with the original song as the root

[cinder orbit]

https://suno.com/song/3c527da8-c642-45c7-91b0-78c72a1a5bec

[leaden marlin]

does subscription points expire if you dont renew your subscription the next month?

[calm sonnet]

Allow us to buy song generation with Paypal or with gift cards while maintaining the terms of service of the basic plan. I believe that many of us do not want to subscribe because we only do it for fun or we only use it occasionally. However, at some point we might want to enjoy generating songs without the daily restriction.

[frozen sonnet]

I would absolutely love an option to delete all parts to a song after making the full version (have it in the 3 dots) This is mainly so I don't have a million parts sitting around that I'm not doing anything with anymore. (conviency of course, I know I can already delete them individually. I just want a catch all button.)

[sharp hornet]

I would like an option to download a track in it's component parts, unmixed essentially, so that I could mess with the mix. I could use the AI generation as a jumping off point, scrap what I don't like. It would be even better if there were tools built in to let me add and adjust component parts of the mix.

[serene tusk]

I would like the simple change of sorting out created songs by ones I haven't rated instead of just the ones I've liked.

[edgy vessel]

Shuffle in playlists

[oblique elk]

Discord channel suggestion for sharing Artist profiles. Looking at #General people want to do that so maybe a specific channel for self promotion would be good.

[edgy vessel]

Option to choose the style of vocals in the lyrics section. IE, Soft Female Vocals, Angry Male Vocals.

[hollow locust]

Ability to 👍 👎 per verse/chorus/bridge... Sometimes it friggin' nails one part of the song in particular and the rest is not so good... and vice verse.

[azure bluff]

Would love an option to download Stems for the songs to get the individual instruments and vocal tracks to edit.

[subtle barn]

Funnily enough, creating a song would probably take the same amount of time as the "real" one due to generation cap per day limitations. Probably even with subscription, cause this thing tends to unnecessary improvise too much when it's not needed.

[tacit idol]

When extending a track, have the "Extend from" timer sync up with the play time of the track preview on the side bar.
We can play the track from the thumbnail, pause when it gets to where we want to continue from and the time is already set.

[dusky charm]

often asked feature. The engine right now builds the clip 'as one', not instrument by instrument, so there are no stems to speak of that could be exported.

[sleek aurora]

pls, take "remix" comeback !!!

[floral tide]

how can you end and instrumental song? is there any way to try to tell it that you want it to come to a closure?

[floral tide]

isn't "reuse prompt" the same as remix was?

[dusky charm]

continue the song, remove everything from the style prompt and just type in END

[tacit idol]

"remix" is still there. It was just renamd to "Reuse prompt"

[copper mason]

Are there any resources or guidelines for what is and isn't censored? My prompt is very vanilla power metal shit but I keep getting "flagged for moderation." Worst things in it are references to war and battle? THe fact that it gives no indication of what exactly is causing the issue is very frustrating.

[tacit cloud]

Text prompt is too short for 190 bpm

[tidal wharf]

just curious, how are bpms and prompt size related?

[tacit cloud]

Lyrics "fits too good" in prompt. So last 20 seconds are empty because no lyrics left

[severe glen]

Ability to import midi chord progressions to have greater control of what chords should play in order and not just a random generation...

[tacit cloud]

Just access to seed is more than enough for 50% cases. And there are lot's of messages with seed request

[severe glen]

NOPE, advanced users who work with DAWs need precise control.... we work with audio and midi offline...

[feral aspen]

Can I ask what you are using to prompt 190 bpm music reliably? I have trouble getting it that fast.

[tacit cloud]

genre: "nightcore, happy hardcore, pop, female vocal, 190bpm"
Sometimes it makes 170 bpm, but at least it's correct direction
lyrics should be really good aligned. Each line 6-8 vovels

[feral aspen]

I have not had much success with things like 190bpm in prompt. But if some bpms are very common, maybe it does work for very specfic set of bpm numbers? Hmn

[tacit cloud]

only modulo 5 bpms (120 125, but not 124) are working as far as I experimented today

[merry fern]

Wouldnt it be cool to get the option to download the song in its seperated layers ? like Bass, Synth, Percussions, Vocal, Pads etc ?

[floral tide]

tried that, and tried [end] and it just kept going... segment after segment 😦

[icy cargo]

Hi Team,
**Feedback: **

1. Song extension is having some issues, like consistency of the song differs, my last one came out faster in pace and more iterations make it completely different. I know it is challenging but that will make it much better in experience.
2. Mostly the starting music from the songs I have been trying is missing and AI automatically starts singing lyrics from the begining. It would be nice to give this feature to composer to control starting and ending music (before and after lyrics start).

**Suggestion: **To avoid above problem:

1. You can increase the length of first iteration of song from 2 mins to max 5 mins. That can be also optional input from user. (If not for everyone atleast for Pro user).
2. A audio editor screen might also help and give more power to user to do minor edits ourself.

@tacit idol @feral aspen

[feral aspen]

A few quick tips for workflows that could minimize some of these problems:
1.) If you are not doing "Get Whole Song" on clips before extending them, that will improve music consistency.
2.) There are two main ways to deal with this.
a. The most reliable way is to start from an instrumental (no lyrics) and then use the Extend From feature to extend the song from an early time so you use the instrumental clip as a song intro.
b. The less reliable way is more abstract: I find the more unique and diverse your style prompt, the higher the chance that the singer does not start signing immediately.
For example **country **will sing immediately with no intro usually. But try something quite strange like azzzzzzzzzz, zaaj?, banjo harmonics, stereo-scale, experimental remix, country, indie folk, emo scat funk, swing lyric and it often has a nice long intro (this example is very extreme, but I like this song 😄 )
https://suno.com/song/b6b37b20-b0d0-44e6-aaa4-64eefaf70a82

I love how these wacky style prompts give you such unique singing voices. There I am just mashing my hand against the keyboard a bit and adding it in the style, sounds great

[dusky charm]

weird ... This was the first part of the clip https://suno.com/song/fa9156ef-fd95-4363-bc8e-aea627794748 then I just removed everything and typed 'END' in the style box and those were the two gens https://suno.com/song/95f09584-0b2f-473d-9c80-d080b167a7d3 and https://suno.com/song/6744f0fd-1157-4dd7-a1df-757055cd6f2c

so it worked for me...might be a genre problem though. V3 has really trouble with some genres, and the problems seem to be different depending on the genre (like the impossibility to have a constant solo singer on rock and jazz, or the over-dramatic chorus/overlaying effect that is apparently plaguing continations of Electronic songs)

[winged horizon]

First: I love SUNO, amazing work! Right now I mainly use it for inspiration, I feed it with my own lyrics of songs that I wrote and just see what it does with it. The most amazing feature for me would be, the possibility to train an AI model of my own voice which Suno then uses to create new songs. I have a studio and am able to create hours of training data with different pitches from my voice. Right now, if I like a song, I have to extract the vocal of Suno to get the instrumental (which usually then does not have great quality) and re-record everything by myself. It is a functionality not for the masses I guess but it would be amazing 🙂

[hollow locust]

Make is so adding 'female vocals' to 'Style of Music' doesn't make the vocals start at 00:00

[vast stream]

Please please please can we get batch select for Trash in iOS. Right now it is minimum 7 clicks (down flag, select in library, moved to trash, scroll to the end of trash, select clip in trash, click delete permanently, then click to confirm you really want to delete permanently) to get a clip from create into trash permanently. Multiply that by 100 and it gets crazy fast.

[dry robin]

fellow metal heads, has the guitars got any better for you, or are we still rolling in a washing machine

[proud jolt]

i am more concerned about the mechanical voice and choruses sound like reverbed folkmusic. Guitars are a bit muted as well.

[broken elk]

I've been using Suno for a long time, and now I've also been testing Udio. I actually prefer Suno. It has better compositions, and I feel the vocals are more interesting and integrated.

Udio has a more hi-fi sound but it sounds more like covers. There is one thing that's great about Udio though: the ability to ADD AN INTRO.

Quite often I wish I'd be able to add a short intro in Suno. I wonder if the Suno model could be forced to do it with some trickery (such as reversing the latent clip).

Even if Suno trained a small dedicated model for that, and if this only produced instrumental-only up to 15 secs intros, it would be great.

[rich pecan]

This snowman emoji always generates the same song. Is this intentionally designed, or is it a problem with the dataset?
I've tried many, but there aren't many special ones like the snowman, so interesting.
https://suno.com/song/4dff3bd5-fb8d-442f-91b2-ef2309e3119f

[waxen glacier]

Add a button where you can deleted all old music on the trash place!

[feral aspen]

Replying to remind myself to check this out later, that is wacky

[woeful rampart]

playlist shuffle button please 🥺

[subtle quail]

how about way to move songs around in a playlist

[wooden ridge]

• a plan between Pro and Premier for 5000 Credits.

[unreal tree]

Best idea ever - cat voice

https://www.youtube.com/watch?v=lwLLFbC1H0c

I'm ready to pay for a premium membership with such kind of voice

[hearty prairie]

A great feature for us who want to use SUNO professionally could be a way to get just the vocals and the instrumentals seperate.

[sturdy viper]

Give us the opportunity to choose the gender of the vocal voice! you are welcome! This is very much missing! Credits are stupidly wasted when you want to make a female rap. stupidly it doesn't work out from the first attempts. And sometimes it even happens that the neural network does not care about 1-2 lines of a song and he sings only 3-4 in the chorus and repeats them later!

It also bothers me the moment when you indicated what the chorus is and would like the chorus not to change the style, but the neural network always changes the style of the chorus when you start a new part of the song with the chorus! It's really annoying. Because of this, you're stupidly wasting your loans. I feel very good about the neural network, but it really annoys me when I can't achieve the desired result, but stupidly MAINTAIN one STYLE in the whole song!

[autumn blaze]

It'd be pretty cool to give paid (topped up) credits more versatility by creating a tipping system to donate some to song creators that you like.

[cyan umbra]

I don't oppose this at all, but since I've altered my workflow to do this, I thought I'd provide it just in case it's helpful. I always generate an instrumental segment in the genre I'm interested in until I find one I like, then I continue from that usually around 0:07-0:10ish with the vocals.

But, yes if Suno could add an intro or would respect the intro tag in a song better, I'd love that.

[patent sonnet]

An option to change the thumbnail of a full song to the thumbnail of any parts its uses

[brave willow]

A suggestion for the Style and Genre List: Shanty
There are some options that come quite close, but I don't quite get the right Shanty vibe with them.

[broken elk]

That's true, but if you start with the instrumental intro to get the first 10s, you're flying a bit blind. I've been extensively using the "continue from specified time" on Suno, and indeed it's very useful.

[wooden fable]

Okay, I have a suggestion: Better documentation on the wiki how this all works behind the scenes. More in-depth, preferably written by devs instead of users trying to bruteforce and guess their way through.

Current documentation seems to be written by users mostly.

It even says "Suno doesn’t reveal details about how Chirp creates it’s voices.". And I think that should change.

[woven pilot]

please fix the issue with the chorus changing when it comes back in

[worn atlas]

Like udio give the possibility to generate intro and outro separetely

[hidden vector]

I would kill for an option to, aside from "Extend", select the parts I don't like for regeneration. "Fill", if you will. I keep generating songs in the verse 1, chorus, verse 2, chorus, outro format, and too often verse 1—chorus end up perfect, but verse 2 sucks, or verse 2 and outro suck, but verse 2—chorus—outro are too long for Extend.

Also, it'd be nice to have the possibility to edit lyrics after Extending a few times, since it doesn't overwrite lyrics, only append to them, even if their audio was overwritten.

[swift parcel]

When scrolling by touch within a playlist on a tablet, the order of the songs is reversed, which is very frustrating. We hope this issue will be resolved.

[subtle barn]

The problem that it starts to improvise instead of sticking to existing style and it's just maddening.

[thin talon]

In addition to creating new music, I'd like to see the program have the ability to turn existing music into a different genre, but the results of this feature are not publicly available and are for my listening pleasure only.
The lofi version of never gonna give you up is an example of this.

[torn beacon]

I would love to connect my Suno tag to my discord tag. 🙇‍♂️

[vale rose]

Another suggestion I have, is if Suno could create like special sound effects (either simply the sound effect as a stand alone, or explicitly in a song - eg. thunderstorm, electricity or something like that)

[visual brook]

i request an "Extend --- as a new song on the same album" for those seeds that we love but just want to use the same embedded context as a new song

[severe glen]

All songs start right away from lyrics? How to get it to start with instrument intro?

[visual brook]

make part 1 an instrumental generation, then extend that into lyrics at XX:XX timestamp - shame because you lose 2min gen

[iron echo]

can I generate 1 generation at a time instead of 2 generation.

[jagged fable]

The cynical side of me thinks that they’ve intentionally held off on releasing documentation/guides because users end up burning through their credits faster through guessing/trial & error.

This is one of the most commonly requested things and I haven’t seen any acknowledgement or plans from the devs afaik. But I hope I’m proven wrong.

[strange timber]

They just want more money, classic with such AI things. Look at GPT-4, it's the same.

[tight hamlet]

suggestion for more revenue: make a plan(could be added onto pro/premiere) that allows users to accumulate daily credits. it could also allow basic users to accumulate credits past the max cap of 50

[calm sonnet]

If I subscribe and extend a song generated with the basic plan, can I use that song commercially?

[worn atlas]

Right. Suno sometime will sing something even if it is supposed to be in instrumental mode.

Make an unlimeted generation plan.

[wooden fable]

Subscription models are the new goldrush.

So, it's really hard not to suspect what "Tiredness" mentions just to some degree.

[wary hazel]

I'm angry! I lost over 2000 credits because it didn't do it right! In the first verse I added rap and other sounds, it worked, but in the second verse, where I wanted to continue the song, I wrote that I wanted a woman to sing and there was a choir. Still nothing!

😠

[unreal glade]

Hello, I am writing this using translation. Can I make a song using my own lyrics with Suno and publish the song on applications where we earn money, such as Spotify? I will buy a paid plan.

Please tag me and reply. I may not see your messages.

[wary hazel]

I am asking for options so that I can choose who should sing, a choir, a woman or a man.

[wary hazel]

good question, I'll add to the question

[worn atlas]

Yes if you have a paid souscription

[unreal glade]

@worn atlas So, is there a song publishing company you would recommend?

Tunecore distrokid and other ?

[wary hazel]

so how can I create a song in Suno, can I download it and share it on Spotify? Yes? I'm asking again to make everything clear! ps: I buy credits on Suno.

[unreal glade]

@wary hazel look the message bro

[worn atlas]

You should have a look on the FAQ item How to monetize your song
https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448

[winged ivy]

When are we going to see an api? There are some other solutions like Udio that are out there but still, there is no API for SUNO. If Suno does not develop one, the other competitors are going to take over with this

[worn atlas]

Do you have an API with udio?

[worn atlas]

Use the download audio and take the mp3.

[winged ivy]

No, but they just started a few days ago, I am just saying that this is not going to be a monopoly forever and that other brands will take over if they do not develop the api

[storm crag]

Is there an option to convert/regenerate a song to instrumental after it’s already generated?

[worn atlas]

There are some features like the possibilty to add an intro in a post generation phase that udio is already doing. I think suno people are aware of what the first serious competitor is doing. Yes they have to move faster.

[twilit hedge]

Not currently.

[worn atlas]

I have several times talked about the lack of an official documentation updated for v3
@hybrid crane answered Devs are working on a new version of the FAQ.
Still waiting for this update.
We have with suno an unusual use-model in which the users are paying to write the documentation of a product because they need to try and retry to learn how the product works. An unlimeted plan would be fine for this purpose but there is no unlimeted plan for Suno

[worn atlas]

Maybe devs really do not know all what the model is able to do.
Someone has asked for the training data set and another one for suno's algorithms. We should not espect to have these informations one day. This is normal.

[wooden fable]

i'm sure they know where the training data came from

[hybrid crane]

As a volunteer moderator, my role is to relay information as accurately as possible based on what I learn from the developers and other moderators. It's important to understand that the developers are also learning how to effectively communicate with the model, much like the users. The community often provides insights that can sometimes be more immediate and practical than the official channels. Your patience is greatly appreciated as the team works on updating the official documentation, and your suggestion regarding an unlimited plan is an interesting point that I can pass along to the developers. But it's unlikely due to the high cost of GPUs, which are crucial for running the models.

[worn atlas]

Yes but they will never expose them

[tepid trench]

Quite a few people seem to be frustrated with their lyrics being flagged for moderation. The common suggestion is to paste the lyrics into ChatGPT to ask what could be causing this, which does give some good insight into why. But it might be nice if the Suno website did this kind of thing automatically instead, to suggest to the user why their lyrics might not be accepted - rather than giving a vague message.

[wooden fable]

i think you're right, farafina

they won't

so guessing what's in the meta data - and thus, what is good words to use - will be some tough bruteforcing for us users.

my guess is that they somehow got hold of a database similar to spotify or such, which would mean they have stuff like genre, but probably no information on which key or types of chords each song uses.

[trail dust]

Suggestion, add a prompt for the cover art generation?

[jagged fable]

For me it's not even just about the training data, I personally don't expect any AI services to reveal all training data. It's about just them being able to give some guidance on the functionality of their service rather than having hundreds of people waste their generations wondering if something will work or not.

One big example are [metatags]. Find me one place on the official Suno website where metatags are mentioned. Knowing how to use them (and knowing which ones will work) is an important part of having more user control with the output. But instead, any public knowledge of that system has been compiled by the users and put into a 3rd party wiki. I don't expect hand holding, but giving users tips on how to best utilize the service is 10x better than just silence with an implied "you figure it out"

[vestal nymph]

I've been referencing this Wiki about metatags: https://www.suno.wiki/faq/metatags/

[jagged fable]

Yes, this is the one I mentioned. It's just weird that it's up to the users to compile these fragments of data through trial and error instead of being given some concrete info on what does/does not work.

[trail dust]

Another suggestion for payment methods specifically : Add a PayPal option, and if at all possible, iDeal ( Which is Netherlands specific so maybe not that interesting for you guys right now ). Taht would be great!

[worn atlas]

There is an official page for the metatags. But as you can see the metatags are for chirp v1 and obviously completely outdated.
https://suno-ai.notion.site/Using-Metatags-34944efe09ec41d693e314f13af44695

[jagged fable]

Yeah I'm aware of that one but I figure it is mostly irrelevant to V3 given how much the model has changed. Though I am sure there are still things that carry over. I've tried a few of those tags and some have never worked. Some work part of the time. Some work more consistently. That page implies that they will just always work.

Like, it'd just be nice to know the best overall practices for increasing the likelihood of something like that working. And only the devs would really have that knowledge. I have faith they'll eventually have an improved FAQ like the mod mentioned, it just should've already been prepped prior to release imo

Hell, I make infographics and would love to make some for the community on utilizing Suno if I knew for sure the information I was putting on them was concrete and coming from the devs

[worn atlas]

This unlimeted access can be granted to a few beta -testers like you. They will help devs to write reliable informations .

[hybrid crane]

I actually am the proud owner of 2 premium accounts which I fully paid for a whole year by myself

[worn atlas]

I share at 100% your thoughs
Suno is a complex software, offering a multitude of advanced features to meet various professional and technical needs. However, as you mentionned, the provided documentation may seem inadequate for unlocking its full potential. This can leave users feeling as though they are navigating uncharted territory without a complete map. Ultimately, using Suno today is like trying to pilot a spaceship with a bicycle manual.

[dusky fog]

(I don't know if this is a question or a suggestion. So feel free to move this where it belongs.)

Is there anyway with v3 for me to upload one of my human made tracks and have the model add vocals on top of it?

[jagged fable]

As of now that's not a feature. Not sure if that's something they're considering in the future though

[viscid cloak]

The top 3 songs always get the most attention, and the other 15 kinda go under. I think it ticks every 10 minutes trough and some songs seems to be skipped.
Can we get the extra Site back for new Stuff? I really liked it more and i could listen and explore better.

[quartz barn]

maybe make it possible to download instrumental and vocal tracks separated from each other? and if each instrument was successful separately, it would be absolutely crazy

[robust jewel]

There aren't separate tracks, it generates a single output. You can download it and use programs to rip it apart, if you want, but the AI isn't generating each layer and then merging them.

[quartz barn]

I write suggestion man, to future options

[robust jewel]

And I'm explaining why that isn't actually an option with how the AI works

[quartz barn]

It is possible of course, few AI sites have this option 🙂

[soft meteor]

if you can provide a large scale dataset with raw stem files a couple changes to the model and a lot of training time it might be possible... until then what @robust jewel said holds true not possible with current model architecture

[solar bolt]

Selecting the lyric voice

[sour echo]

uploading our own modled voices to use as vocals would be pretty darn cool

[granite hedge]

As Suno evolves, introduce a feature to 'remaster' songs that you made with earlier versions (V3-V...), to make the audio/vocals clearer.

[upbeat rock]

As someone using the site daily… 10k plan is not enough. Would love a model that allows infinite slow generations on top of 10k priority gens, like slow and fast hours in MJ. Please

[sterile shore]

Greetings, I don't know if a MOD will truly see this but i'm enjoying Suno v3 on the web interface. I can't seem to find an email for suggestions so here I am.

1. Allow export [ download ] of WAV or AIFF.
2. An option to adjust or remove instruments or lyrics.
3. Allow some post production [ siimilar to mubert ]

As mentioned I enjoying myself and I would eventually like to see these things or possibly better.

[upbeat rock]

Devs read here. What post does Mubert have? I used to use it alot but on a free version

[sterile shore]

I was watching a video and there was a feature to edit the beat as if it was in post production. After playing with mubert [ free ] could there be an option to manually input the time limit within a spectrum.

I also noticed that sometimes it will read tags and somethings in parenthesis. @upbeat rock

It's fun trying to figure out how to make something from literally words. The details get lost but still the result is awesome. I'm enjoying your product.

[granite willow]

please make posible to remaster the song

[sterile shore]

Agreed. It's the file type.

[upbeat rock]

Use Brackets not parentheses

[upbeat rock]

People use bandlab for mastering. But more control is coming dw

[sterile shore]

I thought brackets envoke tags.

Oh I see maybe I worded that wrong. I apologize. The brackets would be read out. In some cases. I will have words on instrumentals.

[prime sundial]

How hard would it be to have suno output 24 bit files?

[dusty fiber]

Can we get a separate tab when we want to extend a song? I have multiple songs that I'm extending at the same time, even when I extend the one by one I sometimes mix them up. For clarity would it be possible to have a separate tab or be able to select the song you are extending and only see the extended/related parts?

[spiral gulch]

Suno is amazing, but there are some basic things I feel should be added.

We must be able to add an intro to a clip song without repeating the song.

The ability to generate the same song (music), but with different vocals.

I personally would like to see a feature that allows the user to control vocals. Perhaps a sort of slider bar system.

[dusty fiber]

A select option would be great so we can select multiple songs to add to playlist, download, etc

[severe glen]

For now, post production in a DAW you can use a tool named UVM 5, https://ultimatevocalremover.com/ ... a does a decent job of removing voice..

[unkempt folio]

I would like to suggest the ability to go into a song already created with Suno and add a beginning. Say the song started off with vocals, but you want it to start with instrumental first but like how it turned out regardless. Also retaining the sound of the song to be used in other songs. Like a band template.

[clear sentinel]

Indicators like how V2 had, stuff like [Drop building] or [Enter synth and drums], I miss that so much from V2

[unkempt folio]

also clear separation of songs as project folders. The organization is lacking.

But Suno is by far one of the greatest applications I have ever seen and grateful and blessed that I could take my own musical ideas and see them come to life and even better than I could have imagined! A year from now, soon. People will be creating so much music that it will be a deeply personal and tailored expereince. I pretty much only listen to what I make or I am creating. Thank you because of Suno I could write songs that my family will live on beyond me.

[leaden marlin]

im not shure how to ask this but when you render a song and like the volcals it would be nice to use the same vocals for a different song somehow? and also an acappellah re render option? also some better defined instructions on the list of commands and how to use them maybe some youtube video tutorials in the future?

[floral geode]

Another suggestion: Either let us clean up the shown prompt/lyrics of a fully generated song ourselves, or let Suno analyze which lyrics have actually been used during generation.

Example:
You create a song with [Verse 1] and [Verse 2] - but you only use [Verse 1] and make an extension to create another [Verse 2] generation.
Now if you're happy with the result and generate the full song, Suno however shows all the full prompts in the final lyrics: [Verse 1], [Verse 2] (from the first prompt) and [Verse 2] again (from the extension).

I wanna have it either notice that in this example, [Verse 2] from the first generation wasn't used in the final version, or that the website lets us edit the final lyrics ourselves to "clean up".

I know it's more a visual than a technical problem, but it peeves me to have the final "lyrics" be extremely cluttered and wrong if you needed to puzzle together a song with several new breaks, extensions, and attampts in between.

[vestal nymph]

It'd be nice if you could have it generate the lyrics/structure and let you preview/amend it before generating the song.

[arctic sedge]

custom mode, write subject, "generate lyric in the bottom"

[severe glen]

Advanced Users: You can post-produce final touches in DAW by removing vocals with UVM5 and add intro by cut/copy/paste from resulting intrumental... I use Mixcraft Pro... others prefer Reaper

[vestal nymph]

Oh thanks! I didn't realize you could put a prompt in the lyrics section. I kept trying in the title and style.

[arctic sedge]

haha! must have been some mixed results 😄

[craggy echo]

Do the programmers or any employees from Suno actually read this, or are we making suggestions to ourselves?

[arctic sedge]

was thinking the same

[severe glen]

Advanced Users: Female solo vocals can be converted to midi with Melodyne VST in your DAW..

[arctic sedge]

they do (y)

[severe glen]

Advanced Users: Replace vocals with yours or anybody elses vocals with https://www.tryreplay.io/

[hollow locust]

Ability to upload our own art for each song. Choose which piece of art we went for Playlist cover.

[severe glen]

SUGGESTION: For advanced user subscribers allow wav/aif downloads for post-production in DAW

[craggy echo]

Question: Is there a way to fade out instrumentals at the end? They tend to end abruptly. If not...

Suggestion: We need a way to end instrumentals so the ending sounds intentional rather than like someone turn the power off in mid song.

[woven cradle]

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

[summer oriole]

I have some recommendations...

So let me give you a little bit of perspective on what it is that Musicians in particular are looking for. It's great that Suno caters to people who don't necessarily know a lot about Music, it's great to bring down the barriers to entry...

But this could also act as a beacon of light in a world where Artists and Musicians are growing ever more apprehensive towards AI in general, which could have bad results in the future if not addressed.

This could be a very exciting thing for Musicians, IF we are able to get more control over the musical creation process, and instead of running over the work of Musicians by reducing it to a text prompt, there are a few features which could be integrated to improve the experience for Musicians.

• Audio to Audio: Firstly, this could ensure that Musicians are able to work more collaboratively with Suno and provide their own input audio similarly to image2image. You'd obviously want to scan the input to check for copyright infringement.
• Voice to Instrument: I have seen musicians have a lot of fun with that. You could use your microphone to record some sounds with your voice, and turn that into an instrument and build a track from there. You would want to be able to quantize the input and layer in some tracks of maybe have Suno pass over it to add some additional instruments using the transformed voice as the scaffolding.
• Track Splitting: This should not be too hard to impliment, you can literally do that in Audacity. You could split the tracks down into their individual instruments, and maybe apply different filters or omit instruments from the track. It's not nice when you get a great output from suno but there is something wrong with the sound and it's unlikely that exact output will come up again. Or it may just need a few tweaks. Sure you could just do that in a DAW but isn't it better to do it with Suno?

^ I have more strategies than that, if you're interested we could discuss this more in depth.

Overall though, the point I'm making, is integrating more audio editing features and making the User Experience nicer for Musicians could be make or break for Suno in this highly competitive AI startup landscape. You need to offer something unique that others are not in order to keep your place.

Getting Musicians on your side will be powerful, you could break the trend that everybody is following right now, where everyone is running over artists and musicians and taking away their agency.

What if you could empower them instead? Standing strong as an Ally of Musicians in a world that threatens to reduce them to nothing. What a powerful way to rally a supportive community who will in turn rally behind you, as you would represent their empowerment and competitive advantage.

There are some powerful opportunities ahead for sure. If you want to leverage those opportunities to your advantage, I can offer much deeper insight, so let's discuss that in depth.

[edgy vessel]

Credit Bonus when songs hit certain "like" milestones. "high enough that the value isnt worth the cost to bot likes, but low enough that advertising your best hits on the site is worth"

Incentive to like others tracks, I feel some may not be using the feature to like others songs so they dont inadvertently create competition for exposure on the site.

[desert whale]

Hi guys, the ability to add consistent verses and choruses would be amazing. Right now there’s still so much variation when extending songs, even if the same lyrics are used within the same song. Don’t get me wrong it’s still amazing but currently it takes over 1000 prompts to try and get enough pieces of audio to compile together a perfect structure.

[rustic garnet]

Endless mode - on some instrumentals, having a mode that literally never stops would be an absolute blast (pro version for sure)

[tiny terrace]

maybe key changes. also, maybe work better with instrumentals at the start of songs.

[unreal moth]

when the possibility to blend tune ?
Uploade a melody and ask the AI to hack it with new vibe ?

[neat vortex]

Hi All, i have a suggestion is it possible to remove the character limits on the lyrics field and style of music as this really limits on what one can do in my humble opinion

[spiral vessel]

My suggestion: cowbells. For some reason Suno doesn't seem to know what a cowbell is. https://suno.com/song/baccf786-a792-4b2f-a650-3802c573715e

[summer oriole]

That's one of the things I was suggesting earlier

[proper sundial]

I don’t know if I’m the only one but trying to get it to just to sing in with one voice has been a challenge. It keeps wanting to harmonize constantly. I tried playing with the prompts to see if that would work but I’m getting the same thing. And with it trying to harmonize all the time it makes it sound more robotic. Anyone have any tips?

[rustic cargo]

Dear Suno AI Team, I have a suggestion
please teach AI about throat singing
AI has absolutely no idea how it sounds like
and makes something like this
https://suno.com/song/f685dbb1-db81-4ce3-940b-3f3683ed50fd

[torn cedar]

On the web page in create custom mode, can we get a clear button to wipe out the lyrics.

[thorn ginkgo]

when it generates a good song:

1. I would like to be able to export an instrumental version. so that i can re fine it and have an instrumental copy of the song
2. I would like to be able to take that instrumental version and generate different lyrics or try different things. we need the ability to save songs that work and build on them. try them again with slightly different lyrics. try with different voices.
3. would like to build on a song that is working in more specific ways. sometimes when we continue a song the new one is too different or there is silence in the middle.

[weary burrow]

Hello, I'd like to create a male-female duet on Suno.
However, I found it's not very controllable.
I can't specify [Male Vocal], [Female Vocal], or [Male and Female Vocals] for each section separately.
If I generate each section separately, it's too cumbersome, and some sections are too close together to separate smoothly.
I'd like to confirm with you whether my approach is incorrect or if this feature isn't yet available.

[silk flare]

Suggestion..sub genre discussions so people can post up for those that are interested in a genre of music

[silk flare]

Could we add alternative/indie?

[green parcel]

Please add a referral system where I can earn credits for referrals

[thorny arrow]

Please keep this up too date and expand the tags that can be used. https://www.suno.wiki/faq/metatags/

I would really like control over

• Beat [BPM]
• Time [1/2][3/4][4/4] etc
• Keys [C-sharp major] [D-flat minor] etc
• Vocals tags: e.g. Male, Female, Tenor, Falsetto, etc
• And a way to keep the same voice, e.g. refer to them as [Voice 1], [Voice 2], [Voice 3], etc

So a song might look like this

[110BPM]
[4/4]

[Verse]
[D-flat minor]
[Voice 1 Male Tenor]
La la la la
[Voice 2 Female Falsetto]
BLa bla bla

[Chorus]
[C-sharp major fifths]
[Voice 3 Choir]
Ha ha ha
Ha ha ha

It would be very nice to be able to recall a voice used on a previous song with some kind of identitifer so that the same voice generated would be used again, so that we could make more than a 1 hit wonder.

[serene pendant]

Hi, May I have ability to search style on the Explore feature?

[ionic skiff]

can we import a downloaded soundtrack back into suno?

[kind spoke]

• to previous question. Very important to import soundtrack. Maybe a short audio, midi or gtp-tab to show the melody I want

[idle hill]

So the playlist feature was a nice addition. But it would be helpful if we had some sort of workspace feature. Say I start with a single generation. It would be cool if I could send it to a workspace, similar to the playlist. But every 'Continue from' within this workspace will automatically populate in the workspace. Within the workspace you could add a Branches feature as well.

So basically just a playlist folder where continued generations automatically align to from within, with branch folders that you can setup within that could host whatever path you're working on. This would be a lot easier to navigate when working on something with dozens or hundreds of continue states at various branching pathway points

[sinful sphinx]

When creating Brazilian funk, it only generates the variants from São Paulo and to a minor degree, Belo Horizonte.
It would be cool if we could add tags like "funk_RJ_150bpm", "funk_cuiabá" to differentiate between subgenres because they do sound very different one from the other

[prisma aurora]

The perfect product for me would be to upload my original song and then use the text prompts to describe how I want the finished track to be and then Suno is my producer. I would pay very good money for that. Even better would be individual stems as it produces then I can’t start prompting for different take on the guitar or drums, etc. get a vst plug in for Logic Pro and it would be even more amazing

[marsh lion]

API API API

[snow elk]

Hello, could we get a detailed resume when our prompt is not accepted?

[peak gull]

has anyone been able to generate (mostly hiphop) beats based on chopped/looped samples with vocals? I have not yet been able to generate something like this https://www.youtube.com/watch?v=qP7-SrP70ZU

[icy chasm]

Greetings. Can you tell me Latin rhythms? From South America. Especially from Peru or the Andean Region: Huayno. Yaraví. Skirt. Tondero. Sailor. Celebration. Landau. Thank you!

[west galleon]

Feature request for Suno: The ability to upload a short music clip file to our Create list, and then have the ability for Suno to create song and extensions clips closely based on the uploaded music clip. This would be awesome!

[visual brook]

rhyming is over-rated. auto generated lyrics should have an option specifically "without rhymes".

[crystal fossil]

I would love to see a feature like Photoshop blender tool. A music editor where you can trim sections, and have Suno reblend the two sections

[open vessel]

Please, please add vertical space for the text box for song creation. The experience is a pain on mobile (Android Pixel 7) when I need to edit text and insert or delete something in the middle. It's unscrollable with up and down drag (whole page scrolls).

An even better idea would creating a button to expand the box to full screen, like a pop up, and an X in the corner to close.

Mobile is painful right now. But I love to make little songs while I'm sitting and waiting somewhere and just have my phone.

[lofty kettle]

[Download feature] I suppose this has been asked a million times before but, IMHO all published songs, not only ours, should be downloadable. Hidding the option does not provide any extra security for the song (we still can download it, but we need to bother using 3rd party programs, or "tinkering" with the websites to do it). Plus your direct competitor Udio, allow downloads of all songs. So please, allow us to download ALL songs.

[frail abyss]

The ablity to download the songs as a stem would be amazing! Alllow the author to load itup in thier fav audio editor and adjusting parts of the song or pitch/tone of synths etc.

[icy cargo]

Is anyone else facing this problem ?
When I extend the song, it limits within 1 minute. around 58 secs.
Not sure if this is by design or just a bug or happening randomly.
Suno AI is awesome in general and hope it will be the best in future.

[wind bridge]

I think the discord really needs to enable thread posts. We would be able to have a more collaborative interaction with the community.

[wind bridge]

This is something I noticed after the main release version if V3. I think it's intentional.

[frail abyss]

I think that is by design, that seems to be my experiance as well.

[icy cargo]

Is there any trick or [filler] for intentionally creating additional music between two versus ?

[rich pecan]

When Extend , hope to keep the audio source in the instrumental toggle state of the previous level.
Instead of having to manually switch to Instrumental after I Extend.

[frail abyss]

I extend using commas or periods. Some times commas will grab some of the previous lyrics, but not all the time.

[icy cargo]

Let me try this trick too

[rich pecan]

You might have misunderstood me. What I meant to convey is:

Version1: Instrumental Music 1 (where Instrumental is ✅ toggled on).
Then, I Extend.

Version2: (But at this point, in the WebUI, Instrumental is ❌ toggled off).

I want the Instrumental toggle state in Version2 to automatically match that of Version1(last version).

[frail abyss]

again it is not always 100%, that is why I wish we could download the stems.

then you could just loop that 16-32 beats over again.

[molten orchid]

I think the way you devs tweaked the AI made it procude extreme metal genres worse (deathcore, death metal, black metal...). V2 is still more fun to play with when it comes to these genres. Is there a possibility to provide 2 versions of the same model? A "More accurate" and a "More creative" options like Bing Chat has? Since I don't think you can please everyone with a 1-for-everything solution.

[rich pecan]

Possibility: Time Index(Location any second)

If LLM could review any point in the entire song, then the strength of the prompt might increase directly by one dimension. In that case, markers similar to the following could be used:

- [At <second>] # can be used for referencing prediction generation
or
- [From <start_second>][To <end_second>] # can be used for "Ritornello" or more 

[potent crane]

Many days into V3 and we are STILL getting choral voices all the time, and no intros, and jumping verses, and missing lines. Can I suggest we roll back to V3beta so we can actually get some songs produced?

[dense grotto]

2 things: I just had a song with lyrics come up as only instrumental. Also, we really need a good fade at the end. My songs are getting cut off in abrupt ways that don't sound good. Other than that, I think you're building an amazing product!!!

It would also be great, when regenerating or using existing lyrics, to be able to ask to use the exact same sound, voice, etc., but to go at it again for a new version or mix.

[untold lagoon]

Will there be an API for developers? I'd love to be able to create music as part of automated workflows

[rotund turret]

Please move the Undo Button for deleted songs a little bit. I dont see a reason why those two Panels Overlap eachother

[azure mulch]

If suno could strictly follow the lyrics when singing? Sometimes there is no need for suno to improve my lyrics, especially when I'm trying to produce a song in a minor language.

[fluid rune]

please change it so we can use our own song cover

[summer forum]

This is seriously good. This is a great product. I can see myself using this more than Spotify for some cases. Seriously this could compete with Spotify.

I want to be able to make collaborative playlists. I want a 'coding music' playlist I can add to (generate more songs and save to) with other users. ◀️

Make it happen

[tiny terrace]

Will there be a day will we can generate songs longer than 2 minutes?

[frosty ferry]

It'd definitely would be nice, since it'd save on credits and generations.

[low plover]

Would be awesome if we could edit songs from our libraries such as being able to edit certain parts from a song we would want to edit or completely cut out/replace. I've generated many songs where I felt like certain parts from one would fit better in another song. Another thing would also being able to alter a singular instrument too, ive gotten a few generated "death metal" songs that come out with power metal or doom metal sounding riffs but the drums are spot on. Some of these generated songs would be perfect if certain portions and instruments could be altered overall

Along with this maybe allow edited songs to be saved or overwritten along the way to either save progress or revert mistakes

[jolly inlet]

So what are odds we'll ever get the ability to reference other audio, even JUST previous gens would be amazing.

[marble bluff]

Suno can't figure out how to make Swancore. Maybe train it on songs from Dance Gavin Dance, etc.

[verbal remnant]

It would be amazing to upload your own song idea or base track and have vocals able to be generated. Or to upload a thematic riff or melody that the song would consider + use as the basis for its main melody etc.

[trail temple]

I wish they would add a timer thing to the website to let users know when their time is up and they can do more tunes.

[tepid trench]

It's not a timer. It's based on how many songs you generate. When you run out of credits you can't make any more songs until the next day (free users) or month (paid subscribers)

[trail temple]

So in the free after 10 tunes ya done and no more?

I was on yesterday and today it is all greyed out

[tepid trench]

How many hours ago? I think credits should reset for free users in 5 or 6 hours from now (I'm not sure). If you're a free user it's meant to reset daily around the same time.

[trail temple]

sometime yesterday afternoon

[tepid trench]

Hmm that's strange. How many credits does it show you have? (Maybe send a screenshot of the website)

[trail temple]

says 0

but it saud that yesterday too

Shows the tune for April 14

[tepid trench]

Unrelated but: Wow is that... Windows XP?
I assume there are no songs from today in your library, or in the trash? (Just wondering if someone else has somehow used your account - unlikely but I can't see how else credits would be used)

[trail temple]

Devuan Linux XP theme

I have some in trash from yesterday.

I guess I can just keep waitin and see what happens.

I was just experimentin with takin some poetry and see what happens tryin it as a tune.

[tepid trench]

Yeah check tomorrow. You could try logging in with another account (e.g. different Google/Discord etc.)

[trail temple]

My own poetry

I thought of that but didn't wanna get in trouble for it

[tepid trench]

It should be fine - it's only a problem if you're using multiple accounts to get extra free credits. Since you can't even use the 50 you're meant to have, I don't see a problem.

[vestal nymph]

This has probably been mentioned before but now that v3 is emitting about 2 minutes worth of music I think the lyrics/prompt should be made a little bigger. Depending on the genre you can exhaust 2 verses and a chorus with a lot of time left to kill. (checking custom looks like you can squeeze maybe another two/three verses in there, maybe it would be helpful if it generated more verses)

[trail temple]

I notice the tunes like to just stop abruptly. I tried [Fade to End] but it doesn't works.

[pine kiln]

It's probably been said already! But stem separation would be impressive!

[plucky violet]

Adding mock-up lyrics to an instrumental helps me to define song structure, especially when working with unknown music. A function that separates lyrics from instrumentals would set Suno apart from its competitors.

[limpid fjord]

Please add the sounds of a chainsaw as an instrumental

[winged horizon]

Don't know if that is already possible (it didn't work forme yet) or if that was mentioned already, but it would be cool we coul set the Key in which an instrumental or song is created (F-sharp-minor, C-Major etc.)

[sturdy magnet]

Is there any kind of markdown syntax to do things like correct pronunciations, change a rhythm here and there, etc?

[chilly kelp]

Some updates that would be extremely appreciated in Suno would be;
Add the ability to choose what instruments are played
Ability to set exactly the order the AI should play
Be able to export a sheet of how the song is played

The AI makes some banger melodies & riffs from time to time, and I'd love to extract it to play around with it, but I'm horrible at hearing what is being played

[chilly kelp]

Even better if you could create riffs or import tracks you've recorded and let the AI play with it

[frigid oyster]

The pricing "structure" is ridiculous - $10/month for virtually no improvement over the free account (AND it doesn't let you roll-over those credits) and then the next tier is three times the cost? They need to learn some fundamentals of (or better yet, talk to someone about) things like anchor-pricing and so on.

I'd like to give them some money for this, as it has a lot of potential, but feel like I'm being played for a fool by this pricing, which my brain doesn't cope well with.

[regal citrus]

How come I can generate in a web browser, yet not here in discord...what am I doing wrong?

[edgy vessel]

Try [Outro]

Suggestion:
When adding a song to playlist it should have an icon for playlists the song is already in to help prevent accidental duplication.

[trail temple]

Thank you

[jagged fable]

If you are trying to use Chirp, it is because they depreciated it a few days ago. The only way for now is Suno through web browser.

[regal citrus]

Thank you for explaining that. Guess its just a timing thing...So is there an alertnate / new way to create songs?

[jagged fable]

Everything is done through https://suno.com/

[regal citrus]

Gotcha, thank you so much!

[jagged fable]

No problem, enjoy!

[shell bobcat]

A tool for placing accents and logical (semantic) stresses. Perhaps - broader tools to explain to AI what exactly and how it should sound. For non-Romance languages ​​:)

[dark raft]

Can we get a category on the site or a channel on the discord for songs that don't use ai generated lyrics? Some way to filter them out, don't care how.

While I'm here suggesting, the ability to upload sound files and extend from them would also be amazing

Oh and a text file that comes with the lyrics and prompt attached to audio downloads

[prime rock]

Many people get tired of not having anything special in the way of covers for a longer song, just a bird. It’s obvious why you don’t want to deal with the moderation for letting people add their own art, why not having a “regenerate cover art” button with maybe a way to input a few words?

[eternal chasm]

It'd be great to be able to organize songs into "projects". When I'm working on a piece, all the generated pieces get pushed out of the list and sometimes I cannot access older ones to experiment. It's also difficult to work on multiple songs at the same time.

[cyan wigeon]

Question about copyright, right now I've signed up for payed account and dowloading generated music, but I don't see any copyright proof. If somebody claims this music been stolen how I can get a proof that I have a copyright for it?

[devout sparrow]

Is anyone else giving the prompt when forming the song?
????????
The songs are not generating.

[jagged fable]

You retain ownership but I am not sure what can be done with copyright. Here is Suno's statement regarding copyright. It's a tricky situation given the novelty of AI and lack of existing laws

[cyan wigeon]

@jagged fable thanks, sounds like at least I can embed this songs to my game

[woeful coral]

My suggestion is for [guitar solo], I find numerous times it feels impossible to pull out a guitar solo when you want one, or you end up spending near 400 credits where eventually the [breakdown] finally has one included

[torn lantern]

Hi I have an idea on the "voice training"... to train a voice, obv you need recordings of that voice. Well, to submit, auto generate some text the person must read and submit (as audio) along with any other number of voice samples. Can then do analysis on that, to confirm the text matches what was supplied. Can probably also do analysis on all other voice samples to ensure all the records are the same person, perhaps could even detect if the recordings themselves were potentially AI... if all tests pass then the voice can be trained, if not then reject it (perhaps have method to appeal the rejection, etc...)

[west galleon]

Suggestion for Suno: Need an official Suno prompt list from the Suno team on what official text prompts Suno should recognize along with any required syntax details, and the expected response from Suno for each prompt. Seems right now it's all pretty loosey goosey and very hit and miss in how Suno might respond.

[swift parcel]

I am interested in a model specialized for underground music, similar to something like V3 Alpha. I don't believe the model needs to be a unified one. The issue with a unified model is that it can result in a loss of creativity and diversity.

[serene pendant]

It would be great to be allowed updating the final extended (full song) lyric. Then, the public sharing more efficiency.

[cobalt ibex]

Need a way to mark songs already added to a playlist.

[edgy vessel]

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

[bitter tapir]

I'm sure it's been said 1000x but having a seed function would be awesome, especially if there was a seed for the instruments and a separate seed for the vocals. If you get good music but bad vocal alignment, change the seed for the vocals and rerun the prompt. Would allow for so much more fine tuned control over the output and increase consistency between sections of your songs. Please!

[formal bear]

It is not impossible but technically a bit challenging to fix seed in production environments

but it sounds like, the feature you want is just "modify vocals" or "modify instrumentals" controabillity in editing

rather than a random seed

[blissful mantle]

New user here and loving this so much.

Could you please add a random toggle for the player? So the library/playlists don't play in the same order every time and don't just stop after the last song.

Also, maybe add an instrumental toggle to Explore, or add instrumental electro swing to the list of styles there, please. 🥺

[grim cedar]

I think one of the most frustrating things when really digging down to get that song you're happy with, especially when you're invested in the lyrics and want to do them justice, is the lack of consistency sometimes offered by the [Chorus] and [Bridge] tags, not within themselves, that is actually really useful, but with the song as a whole. There are effects applied to the vocals on both of these that can almost change the voice of the singer, then the backing track can sometimes get too loud during these too, you can go from quite a mellow chill song with front and centre vocals, to the vocals being drowned out by the backing track in the chorus or bridge. It's a huge fight to avoid that, lots of generations needed to get something that works. Probably comes under the currently known vocal layering or choir effects, but it feels like a slighter broader issue from my experience

[torpid storm]

please let us choose the picture from one of the parts it generated instead of just using the final

[noble frost]

Exactly this - thats what OpenAI realised very early on, the API is their most powerful product yet. Its become a crucial infrastructure for alot of devs and that gives OpenAI a moat. Model companies who dont realise this will be eaten alive by those who do

[low cosmos]

I don't want to edit music extensively with Suno, but I would like to have the ability to cut out unnecessary parts of the song in order to create a nice ending to the song.

If the cadence(end form) is repeated, it will feel strange.

It would be great to have song titles/prompts search and tags.

[late elm]

It would be nice to have a filter by language so i could only see songs in english, spanish, french, etc...

[unreal moth]

can we have ui with dnd tokens part ?

[shy zephyr]

Include editing the voice to be male or female, as well as duet within the same song.

[jolly trail]

Anyone else think the drums are too loud and mismatched? Despite asking for no drums or light beats, I get loud, metallic sounds that ruin the chill vibe. Is it just me, or is this a common issue?

[dry willow]

The volume on the vocals on most generations is very low and oftentimes hard to understand the lyrics. Would be great if you could increase the volume a bit or do some sort of compression. Cheers!

[river pasture]

Hello,
Is it planned to be able to edit the lyrics from the video in case they don't exactly match with the song ?
Especially when we try to keep the same "seed" by using an extension at 00.01 (it keeps the lyrincs from the first part even if we don't hear them anymore).
Thanks 🙂

[buoyant dawn]

Hi I actually have a question, not a suggestion, but the general chat was too busy, and I wasn't able to get a sufficient answer (hope it's okay to do that)

So, the number displayed next to the 'play' button of a song when I go to my profile is the number of times that that song has been listened to right? Is that the number of times I listened to it, or other people? The reason I am asking this question is because when I listened to the song, nothing changed so I assumed that it was the count of the number of times that other people listened to it. But then, just as I made a song public, it somehow said '23'.

@river pasture umm.. why exactly did u react that

[river pasture]

That's a good question, that's why 😉

[true coral]

All that copyright blurb means is that the US court system has been fairly consistently been ruling that AI art cannot be copyrighted. I’m not sure there have been many lawsuits regarding AI music other than artists suing because they feel like their work was used without permission to build the AI models. The courts did not rule in favor of the recording artists.

[loud nimbus]

Creating sheet music for all instruments in a track would be an awsome addition.

[river pasture]

Another one that might alreadydy have been suggested, but being able to export/download in a better quality thank "just" MP3 would be great

[severe glen]

Can DJ drops be made with this?

[wet gyro]

**When SUNO creates a song and there is No Style of Music Chosen by the user, can it still display what style of music SUNO chose to assign to the song?

That way if the User likes it, they can use it for future creations?**

[dusty swallow]

Download vocal acapella to producers use and artwork bigger

[fading iris]

I wish I can splice tracks from one extension into a different extension. I like how one track sounds but it's extended from a different track.

[real osprey]

Is there any way to be able to make a 3 min song?

[delicate cedar]

Yes, by pressing the Extend button.

[wet gyro]

**To be able to type in a style of Music and then have it generate a sample so that we can preview it before spending credits on a song to see if we like it.

OR, you could create a search feature on the EXPLORE page. But the dice are pretty cool too.**

[real osprey]

But it's not actually extending the song is it? It's creating a new track "overwriting" the lyrics earlier?

[delicate cedar]

The way the process works is this:

You roll the dice for a first segment you like. You get the music and the first verse\chorus the way you like it. (Whether you do this by continuing from time to build a franken-sound and vocals or whatever. You get that first part down.)

You hit extend on that part that you liked. Now you add the next verse and your chorus and try to match the previous part in sound and feel. When you like that you extend again.

You continue doing this until you like the song and are ready to end it. Then you close it with a [fade out] or something and select get full song.

Now you have a song longer than three minutes.

If you're getting a new track over-writing the lyrics from earlier you're doing something wrong or there is an error somewhere.

It should be a chain.

one part linking to the next part.

[real osprey]

Thanks a million mate, i'll check it out

[delicate cedar]

Glad to help out. 🙂

[jolly elk]

Hope no one has suggested this. I was listening to a song yesterday and the singer extended the word 'I' almost a whole bar. It would be great if when you type lyrics you could type something like / next to it denoting how long the word should extend. Like : I////I can see clearly now , the rain is gone// ---- so like the # of slashes tells how long to hold the note. Great stuff you guys are doing here. Million thanks.

[finite ether]

even I am facing the same issue, unable to extend any song since today morning

[delicate cedar]

Might need to put that in code blocks so it shows up?

[jolly elk]

Got the same yesterday evening

[jolly elk]

I don't know what code blocks is. If it's apart of the forum I'm not privy to it. Spread it around and maybe the forces behinf will hear.

[delicate cedar]

Like : I////I can see clearly now , the rain is gone// ---- so like the # of slashes tells how long to hold the note.