💡┃suggestions-web

[undone bough]

i count a least 5 times, and its kind of ok in the end but not in the beginning and middle of the song

[sacred shard]

Is it something I imagine or does 3.5 occasionally have the tendency to speed up songs towards the end?

[torn cave]

try using a differnet ai for lyric generation

[undone bough]

well i do, i dont use suno lyric

[torn cave]

what ai do you use? I usually use copilot

[undone bough]

@torn cavemostly i write it myself but if i get stuck on something i just use bard or copilot

no need to pay for something thats free, those lyrics generators out there is just a way to make money on something people can get for free allready 😛

[torn cave]

i cant write lyrics for the life of me. i usually have quite long sitt-ins with copilot to explain it what i want

usually works out quite fine

[undone bough]

i start with a topic, then gets the info abpout it, paste the info in a text editor and start making verses with rhymes, when thats done i ask copilot to set it up and fix grammar etc and sometimes ask for suggestions for more verses. maby ask for a cachy chorus or intro/outro/brifge etc.. then its over to suno and magic happens 😛

[torn cave]

quite similar to my method: 1. ask copilot "do you know x" 2. Specify the topic and what its about 3. specify the the structure of the song 3. generate the lyrics 4. let suno do the rest

[left crescent]

Can you add it the option where I can duplicate a music clip I have to make it an instrumental version with the rythm going exactly the same as in the first clip

[short python]

The Ability to Crop a Generation

Sometimes, You get a Generation that's literally Perfect, except for a part that the AI Just Tacked on at the end without control, or if an instrumental goes on a bit too long
The ability to 'crop' these Parts off would be a Godsend

Closest thing we have to this currently is the ability to extend the generation from a certain point, and while this definitely helps, on occassion, v3.5 just will not allow a song to 'end', so a crop function would help

i can imagine this should be no different from Getting a full song from all the extensions, since it's not generating anything new

I'm perfectly aware i can crop the audio as a file, but i feel as if i get the best suno listening experience from actually listening on the website

[potent coyote]

Just wondering... Is there a way to organize the songs better? My "create" tab gets filled constantly and it makes longer songs pain in the ass to work with. I've already deleted all my old songs but that doesn't sound like a super good solution either.

[pseudo lake]

Not a perfect solution, but I've started picking a random keyword as a project name (THANK YOU DEVS for adding search through library)

• After making a bunch of generations for a song, I'll place them all in a playlist (just using it as a folder to organize, not interested in playback or sharing). for the purpose of bulk selecting songs to add to playlists or delete, ctrl + click adds additional songs to selection, selecting one song & shift + clicking another song selects all songs in between
• After grouping them, I'll rename the songs with ideas for extension points or TODOs for my external tools (stem separation, remove vocals, isolate drum track, etc)

[grave crest]

will there be a future paypal option to get a subscription? we dont all have credit cards

[pseudo lake]

#💡┃suggestions-web message

People have suggested using prepaid credit cards in the meantime, like vanilla or n26

[grave crest]

they usually wont work or be accepted for re-occuring payments, just ponder why it be so hard to add a paypal option 🙂 or direct banking 🙂

[cedar mulch]

you can use privacy cc, they link to your bank and you can generate a CC for every site you use, lock them to a single vendor, or make them one-time-only burn cards

[potent coyote]

I can see how this helps to organize the library tab. Alas, I don't know if it's any help for the crowding of the create tab, though

[dreamy sigil]

When the ai chooses a voice and style setting, it should generate some kind of seed that we can reuse to reobtain the same voice and style if we only want to rework lyrics

[potent coyote]

Maybe sounding stupid here, but why do you want to empty the trash? I just leave it. They'll self delete in 30 days anyways and it's not taking space in my hard drive anyways

This. SO MUCH THIS

[brittle crane]

Hmm.. sometimes you don't want to go for a subscription but still get something done.
Why can't free users purchase credits too? 😄

[delicate agate]

I read the FAQs and got concerned; I wrote my own original lyrics for some songs, but I am not a Pro or Premier subscriber. Do I retain copyright to my lyrics?

Also, if I download the track and remix it, remaster it, or recreate it from scratch, may I use my bootleg version of the track commercially?

[tepid trench]

Buying top-up credits is possible for subscribers but they are quite expensive compared to the subscription (1000 credits is about the same price as 1 month of 2500 credits.) So if they offered them to free users, there's not much point - in most cases it's more worthwhile to just subscribe for a month.

[pseudo lake]

You own the lyrics, but don't own the generated song.

In regards to creating your own copy, think of it as recreating a Beatles song -- have you made enough changes to the song that similar changes made to a Beatles song would pass copyright flagging? If so, I'd say you're safe, otherwise you're at risk of trying to skirt Suno's licensing by generating a free song, making (minor) adjustments, and not giving attribution to Suno for creation.

[pseudo lake]

In regards to retaining ownership of your original content (lyrics) after submitting them to Suno for non-commerically licensed generations.

Owning the content used for input doesn't impact ownership of the output, but it also doesn't impact the content you previously owned.

[steel fractal]

Allow to generate songs that flagged moderation but will remain unlisted

[delicate agate]

what about vaporwave/nightcore thanks!

[pseudo lake]

You automatically own the rights to anything that sounds like it's being played in an abandoned mall from the 80's, since it's been left untouched for roughly 40 years you're welcome! ❤️

[normal sable]

I would like to see the cover as a thumbnail in the mp3. Maybe even the style or the lyrics as the comment.

[hybrid crane]

Which list are you talking about exactly?

[austere flax]

The countries eligibility list for the Summer competition

https://docs.google.com/spreadsheets/d/1WF99cFAU22vbsT50CXdZEUSNtwGPBcdp-fAkiukwTfw/edit#gid=0

[hybrid crane]

Ah.. gotcha. Thanks for your suggestions. The developers look in this channel at least once a day (if not more) so I am sure they will look at it. 🌼

[austere flax]

Alright, thanks!

[soft meteor]

it's a bit confusing though that that list always opens with the restricted countries first... at least for me 🤣

[gaunt flame]

would like to see a negative prompt box/field. Specifically for lyrics. Sick of describing the song I want and the lyrics incorporate singing the genre I want and type of instruments I want lol.

[marble musk]

v4, will we get to change seeds for song generations?

also i hope they will add a small editor where we cut and trim our songs, since sometimes suno adds the outro then it continues it for some reason even though the song is finished lol

[violet kraken]

Hi is there a way to cut a generated song on the interface of Suno ? It will be great to have the ability to define the properly the end. In v3.5, it generates some noises at the end by reusing previous chorus.

[oblique pollen]

There should be more unique instruments added, like the medieval "hurdy-gurdy" - and greater ability to prompt for specific instruments
https://youtu.be/bvNZeh6f8vE

[wise scroll]

I'm not sure if this would be possible, but I'm wondering if future iterations could have the option for songs to split into their core components via multi-track. Separate vocals or individual instrument parts.

Or additional details like sheet music/chords/song key.

[pseudo lake]

Create an extension point a little bit before you'd want to wrap things up, then use lyrical tags like [outro], [end], [fade out], etc

[verbal scaffold]

Djent style metal doesn't seem to work well.

[gaunt flame]

i'd be very surprised if we don't see this within a yearish. From Suno or any other AI music company.

[violet kraken]

thanks for the method ! 🙂

[pseudo lake]

The tags aren't explicit instructions on what the AI should do, just suggestions. It's not exactly like writing code, so sometimes you need to experiment and work with it by piecemeal to get specifically what's in your mind's ear ❤️

Have fun creating!

[woven magnet]

A negate option to not include an instrument, like {no guitar}

[clever hinge]

Would it be possible to label the numbers in our profiles with some kind of text? My screenreader reads all the numbers, but as a visually impaired person I've no idea what they indicate, so I don't know how many views my songs have, how many likes if any at all, etc. Thank you!

[ocean musk]

I choose extend from 2:00 but why does it go back to 2:00 like the original or decrease to 1:50 or 1:30

[pseudo lake]

I would love to see if something similar to a LoRa model could be applied to Suno's generation? When trying to create AI images for my wife I can describe her as perfectly and precisely as I want, but without a created LoRA model to specify what I mean when I use her name as a keyword, it's just a gamble as to how it turns out.

Even if I'm generating a reasonably understood and predictable prompt (eg. "apples" for image gen, or "reggae" for music gen), there's still an astounding amount of interpretations an AI can make from those keywords (eg. "apple" can mean red, green, laptop, tree, patterned, archetypal, photorealistic, etc), so if you have a specific "apple" in mind you'd like seen consistently while applying different styles, using a LoRA can save a lot of headaches and wasted time/money by ensuring that there's a certain amount of "agreeance" between you and the AI as to what you're hoping to see, depending on the LoRA weights.

A small curated list should be all that's needed initially to help people understand the difference between results with or without the extra 'nudge' toward certain sounds. Some jumpstart ideas for "genre-plugins" could be; 90's grunge, 80's synthwave, classical baroque, British ska, Latin salsa, country Western, K-pop, dubstep, etc

{keywords for search: lora model, adaptation, low-rank approximation, genre plugin, addon, extension}

[prime stirrup]

I keep asking for folders in the create tab. Thank you devs you're the best

[viscid cloak]

2 things for Trending:

• Can we please activate that Trending has only one Trending Song at once? (So it gets more diverse and not the same people flood it all over) Would be also good for the competition
• Please deactivate the Downvotes considering the score, saw it multiple times that some users get Downvoted and some likes removed... (i bet you already saw that)

[glossy ginkgo]

I can't even see or scroll down to my song list on my album due to giving my playlist a decription. My suggestion is to fix this >.>

[drowsy dove]

its been bad 😭

[pearl mason]

I need that suno uncensores the moderation a bit, because death metal songs about violence is not getting through, and the whole point of death metal is to be brutal on the songs. Example: tear their guts and eat it" that is not passing through. Neither "grab my sword and fight" please make it better improve on that

[worn atlas]

Good question. Lot of countries are not restricted and not eligible. Ignored.

[leaden marlin]

I guess I was a bit naive in expecting 3.5. The key sound quality issue hasn't been resolved. Vocal noise is still noticeable, and the more sounds and instruments used in a song, the more noise is on parade. High notes in particular are painful to the ears. Even low tones are distorted. Also, it's not about whether the music can be output in WAV, but whether it was processed for composition in WAV in the first place. Nothing has been revealed about that. I downloaded several songs and compared the MP3 and WAV files on the spectrum, and I found that the compression of the treble that is common in MP3 was also found in the WAV file. It seems like it's just been converted to WAV and made available for download. That doesn't make sense to me. It's like a criminal who dresses up in expensive clothes to hide it. The service is good enough for casual use, but not for anything more serious. You can't release this level of sound quality, and you can't improve it in the mix, because STEM won't distribute it. I guess it's enough for someone with no musical knowledge to feel as if they're an artist.

[nova raft]

This time, many countries in Asia cannot participate in the event? I don't see China, Taiwan, Japan, Hong Kong, Singapore, and other countries on the list.

[delicate agate]

specify time signature (e.g., 3:4) and bpm

[thorny walrus]

Add options:

1. "Edit the lyrics" of the song after it has been created. Sometimes, the rhythm and lyrics are good but a word ends up causing strangeness in the song.
2. "Copy rhythm" to create a new song. Some rhythms are very good, it would be interesting to be able to use them in other songs.
3. "Cut/Copy/Paste parts of the song". Quick editing tools within the platform can bring agility, convenience and make music even better.

[tepid trench]

Devs were reluctant before to offer WAV because the model is what limits the quality at the moment. There is a little bit more detail in the WAVs so it's not just a converted MP3, but not much.

Annoying high pitched ringing mostly got fixed. That's been the main issue with quality over the past couple of months. Overall audio quality has never been perfect with it so far but it's at least bearable to listen to. It will no doubt improve in future but this is as good as it can be at the moment.

[elder gulch]

Hi do we have "render" function in the Suno? as for like I gave it a rough sketchy melody and they can sort of "render"it to make it sounds like a complete song?

[lusty blaze]

Melody 2 Song input. Allows a basic chord progression or even a sung melody to be used in the creation of the song.

[tepid trench]

Like this feature that's coming soon?

https://x.com/suno_ai_/status/1795878282631512512?t=_tI4U3nUx6RuRDUnRm-MOQ&s=19

Suno (@suno_ai_) on X

Make a song from any sound. Coming soon 🎧

VOL-2: Classical piano, and bring in some French accordion

Performed by Anessa (Pianist and Suno Software Engineer)

Twitter · May 29, 2024, 6:02 PM

[lusty blaze]

Nice. I don't support X so I don't get updates there. Hence, why I came here. 😉 Thanks for the update.

[tepid trench]

I'm surprised it's not in the announcements channel (there's a different example there though.) I originally spotted this on Reddit

[lusty blaze]

Yup, just found it myself. Pretty excited for this as it will be a valuable songwriting tool for me.

[elder gulch]

oh my goodness that's fantastic

so it is coming out? when shall I be expecting myself to use this function?

[hollow barn]

We need a shuffle playlist funkction, aswell as a replay playlist funktion

[crimson hound]

add a feature to separate the instrumental and the vocals maybe just have a toggle feature, would be nice but seems hard to implement

[mint sparrow]

I have an idea... is it possible the AI would recognize stuff like (solo) ?

basically some way to add some instrumental part like a bridge or something

[tepid trench]

You can put [instrumental bridge] in the lyrics. I don't make songs with lyrics much but when I have done, I always ended with this to make sure there was some instrumental padding at the end that I could decide after whether to keep or cut it.

[mint sparrow]

I'll give it a shot

thanks

[neat belfry]

I have been using suno for a bit now and I feel like the Intros and outros are a nightmare to get right. Not sure how to make it easier but maybe it would be an idea just [intro instrumental] [outro instrumental]. Something like that.

[minor gull]

i have used [Guitar Solo] that 100% works

[eager fulcrum]

Please add volume slider to /explore... so annoying

[mint sparrow]

that's basically the extension of my idea

[marsh orchid]

Guys I hope you improve the sounds and pronunciation of some languages like Arabic because currently it looks like an English person trying to speak Arabic

[cloud wadi]

So I just saw a video showing sooner would be able to make music based on sound you input. I'm curious as to how that works. Does it follow the tempo and the beat of the song? Does it follow the sound input or the pitch?

[tepid trench]

Looks like it'll be both. There are two demos I've seen - the first looks like percussion and tempo following the sound of someone hitting a watering can, and the other is of someone playing a piano whose melody is then turned into an accordion

[delicate shard]

Will there be an option to select and save multiple songs in one go as ZIP? It would be good if you add a short unique hash to the filename at the download, for example 'My Song 6a9e078d.mp3', to avoid overriding songs with the same name.

[latent ruin]

I also tried it and I agree with you. What I missed the traditional nepalese music, because it not sound like.

[visual brook]

what is the new URI format to download a large version of generated images from album art?

[leaden marlin]

Yes, it may be a masterpiece at the moment for just “listening”, but I'm one of those users who wants to use it for more than that. Unfortunately, the treble noise still hasn't been improved. It's especially noticeable on synths, acoustic guitar, piano, etc.

[neon moss]

Create a song #💡┃suggestions-web message

[formal kelp]

creat a song

[delicate agate]

Add a way to limit the number of jobs running simultaneously. I make edits often. Having 10 songs (50 credits) consumed simultaneously will consume all the credits waaay too quickly.

[leaden saddle]

Edit song on a timeline, with time stamps to edit the prompt will pay credits for this

[tawdry ocean]

Plz, I need statistics of usage... like last.fm

I need to be able to modify the lyrics after finishing the song 🙏🏻

Thx!

[nova thicket]

A Suggestion for discord side. Have Showcase Channels for Music Genres. Rock, Pop, Country, ect.

[safe ingot]

Please put some sort of marker to know if your song is blocked from appearing in New. Especially because of this contest now, very annoying having to keep checking and wonder if it somehow tripped the filter.

[proud dome]

On the library search can you also allow this to be used with liked or public??

[wooden igloo]

If dev's are using thumbs up and down to see what is liked and not, can we also get a 5 ⭐ rating system for ourselves and creators to rate the songs?

[upbeat mantle]

one easy thing to add into the Manage account area is a Delete account button due to some laws of country's needs it there and it makes it faster then sending in a removal ticket to the team

[wanton lodge]

Don't know if it was suggested already by someone else, but it would be great to have a multiple select function of the songs. It's very time-consuming when you need to delete bad songs, move them to playlists or download one by one.

[leaden marlin]

Ctrl + click will select multiple songs

[wanton lodge]

Yeah, I found out about this just right now, when decided to test with windows commands. Thanks anyway!

[wide rivet]

there needs to be a ticket channel

or a ticket bot

[hollow locust]

Just a heads up, I got a phishing email today claiming to be from Suno, and promised a reward of $1 million.

[upbeat mantle]

https://images-ext-1.discordapp.net/external/_wbX_RgvETlshu4to6oM-U5pp1yTij-ux-a3JuM5uDo/https/pbs.twimg.com/media/GO6iCQqXAAAu8Yp.jpg%3Alarge?format=webp&width=723&height=905

[hollow locust]

Okay, my bad.

[upbeat mantle]

no worries its in the Announcement section

[worn atlas]

To Devs. Is it possible to know the list of bug fixes between the early access version of V3.5 and the official release?

[keen marsh]

possibility to edit generated songs afterwards, for example the voices (I often have songs that sound good but the voices don't fit so well)

possibility to select a certain voice range as style (e.g. Female Voice 1, Male Voice 7 etc.) to have the same voice ranges for different songs.

[worn atlas]

Thanks for the display of total number of characters count over the total number allowed in the lyrics box as i suggested. Ie 782/3000 . Kudos to devs.👍

[brittle crane]

Though approaching the 3000 will still cause weirdness to happen. 😂
Bot hallucinates songtext that isn't there but finishes just fine..

[rustic anvil]

Can we get a check box or some indicator on the UI to clarify to SUNO that we intend to have the song end on that Extend and to wrap it up? [end] [silence] and anything I put for v3.5 is not working anymore. A final absolute indicator would take care of this issue and not have all the songs to continue beyond that extend.

[worthy dust]

Since I really struggle with finishing a song that contains lyrics for 2 minutes (Suno always makes it 4:00 minutes), a function, a button when playing, for example, where I can set the end, would be fantastic. Something to force the end of the song when creating it 🙈

[potent folio]

Could we get a NSFW, or uncensored rather, trending list?

this isn't necessarily to produce intrinsically nsfw music, but it seems to me that even songs that are mildly "not kosher" are ending up off the list

which will end up impacting the outcome of the month

and ultimately, and this may be just me, I think there is an incentive for Suno to do this, as people will judge the service based on the winners

and like, variety would help imo, as there is a saturation limit to anime

in terms of music base

[keen marsh]

suggestion: in the library, check the box that marks all songs on the page for deletion

[stuck steppe]

Shift + mouse click you can select everything anyway (or ctrl + mouse click)

[keen marsh]

oh, thanks 🙂

[foggy cedar]

Hi, is it possible to extract only the instrumental of a generated song as a vocal remover ?

Please

[lone hawk]

should everyone had acces to an opensource Bark 2.0 soon?

[slender peak]

PLEASE READ WILL IMPROVE THE APP 10X: You have done a fantastic job with the upgrades mentioned. search was done immediately BUT I think having a search where each creator can search their own songs is equally as important as well as the ability to arrow to the next page there should be a Multiple page jump option as well with the ability to write it the desired page number. Right now, going from page 1 to 200 1 click at a time is drudging.

[wind crest]

hey devs, i know you're working on inpainting, when we can it can we have the option to just insert a new section at a specific point without erasing anything? as in, say i have this banger™️ acid house track now, and i think it could do with a nice dreamy interlude and drop, i'd like to be able to point to a timestamp and direct the bot to insert something there, without having to erase a chunk to inpaint with the new content

[tepid trench]

You can search for your own songs in the library.
Going directly to a page number would be nice though (workaround for now: Click once, hold down space bar or enter key to rapidly go through the pages)

[minor gull]

i would really like a way to adjust the volume balance of the music/vocals. as it is right now i have to run everything through external tools to lower the music volume by 40-50% to achieve the balance im looking for.

[wooden igloo]

A button to empty the trash please 😂

[dense thunder]

In the future, I hope to have the opportunity to edit a song I've made, to "fix" a part of it.
For example, Suno could offer the option to choose a 30-second block of the song, we can choose, in which we can modify the structure and/or the reproduced text.

Thank you

[inner rover]

When we going to get unlimited generations with a reasonable cost, for pro and paid users. 😓🥲😮‍💨🙍🏻‍♂️

🫨🤔🫡

[leaden marlin]

Have an option that when choosing extend to have the new generated lyrics be related to the previous lyrics and not just be random lyrics that have nothing to do with the previous lyrics

[hearty pawn]

hey

iam unable to use the chirp command

any suggestions?

[upbeat mantle]

make it so the songs get placed in their correct genre in your Profile of songs and that it shows the total amount of songs u have in number form

[viscid cloak]

There is no Chirp anymore. Use the website: https://suno.com/create/

[lethal flame]

It would be great if Suno could provide something to "clean up" audio that we've gotten from it being glitchy. There's stuff I that love aside from the glaring reverb or static. I don't want to have to just trash them.

[woeful burrow]

WAV export is super slow and apparently also skips files (!!) when you do it for many files during a short period of time (i.e. as if some files will just never even prompt download if you click "too many" before downloads start).

**Problem: ** if you first generate a bunch of stuff, then try to download all of them or e.g. "the best of them", and the server just wildly skips WAV exports, you lose track which files you already downloaded.

Solution:

1. Please place check mark or other indicator for "downloaded" (in Create section and Library section), that appears when server has sent the file to user client (not when user clicks on "download x")
2. Fix WAV export speed; and make sure all exports user has done arrive as downloads to user client.

UI dev team will probably think "nope, doesn't make sense, because there are WAV, MP3 and Video download options", but I can assure you that a graphic indicator for "any format downloaded" is still better than nothing. Would wildly argue that most users will primarily download same format anyway. Make a color coding for the icon (and relevant help page) if you want users to know which format they downloaded.

[woeful burrow]

Not to brag but come to think of it, I had to suggest the same thing (download-indicators, wtf) to letsenhance.io and runwayml.com before it was implemented in those services. It puzzles me a little bit that users doing batch-processing is not well thought through in UIs of services that particularly invite people to batch process stuff. The results are fantastic tho, not to sound like a complaining mofo.

[tawdry wave]

Output in midi files? When?
Output in multitracks like moises.ai? When?

[tepid trench]

Good suggestions. I generate a bunch of extensions and want to download them all but I have to remember which one I downloaded last and if I take a break or get interrupted/distracted I then have to leave my mouse pointer over a song as a reminder to myself as to where I got to.

[tepid trench]

The first part of this should probably be in #🪲┃bugs-web

[chilly junco]

What about bulk downloading all the songs in a playlist (mp3 or WAV option)?

[wide adder]

Please add:

• ability to modify part of a song
• seed - to regenerate song with the same style, music , voice, instruments etc. I got luck and generated a nice song, but i want to modify text, and get almost the same result, but now it is impossible.
• Better understanding of key words/tags such as [Guitar solo] etc, it is still random.

[woeful burrow]

Yeah, some "downloaded" icon would have been useful even before I came across the WAV export slowness. This just just made it seem even more useful!

[tender basin]

To be able to manually enter or at least copy the seed would be awesome!

[slow wave]

ability to generate acapellas, opposite of instrumental, which only generates vocals

[worn atlas]

Suggestions :
1)Display the total number of pages we have in our library. This can be done where the current page number is displayed. Aka 4/100
2) Add a tab for full songs.

[hardy falcon]

[ Search songs by model ] 3.5

[heavy quiver]

• negative prompt: being able to keep out certain unwanted elements would be a gamechanger
• intro generation: it really sucks when you generate a great song that starts halfway through.
• clean instruments: it's so hard to make instrumental tracks with the low quality instruments and mids/highs degrading over time.

[full tulip]

Hi there! I'd love if we could edit the text from a generated song and regenerate them with the new lyrics! That would be super helpful!

[tawdry wave]

We're waiting for output:

• midi files like Band in a Box 2024
  and
• multitracks like moises.ai?
  when comes?

[royal venture]

make a 3.5 with versatile song structure like 3.0, or make a 4 min 3.0 version

[potent crane]

The reason it is slow is it spends time converting the mp3 to a wav before it downloads

[south willow]

I have to agree with a lot of people in terms of being able to tweak text from songs you already generated whilst keeping the same beat and/or voice. It would be super helpful to be able to mix & match as well (for example, if i like the voice from one song and the instrumentals from the other, i could combine the two into one, new song).

[woeful burrow]

Is that a fact? It crossed my mind but then I thought that would make zero sense technically (what is the reason to even provide WAV in this scenario?)

[pastel kiln]

we should be able to creates tracks and assign them to playlists that contain different genres for organizational purposes. this will help people identify songs better and not have to spend hours going through thousands of tracks

[potent crane]

I always convert to wav anyway, I know that it doesn't improve the sound, but all my digital music archive including music I recorded as far back as 1972 is in the wav format, I never use mpr other than emailing demos etc, I have a consistent archive 100% wav. The training data is mp3 so unless and until they re-create the training data from scratch it will always be a cosmetic excersize outputting as wav.

[broken ferry]

Anyone know if the sound to song feature will include vocals?

[wooden igloo]

Similar to "get whole song", we can have another context menu item "edit".
Can show us a waveform similar to what Udio shows just to cut and get a new "whole song".

This can help on cutting the extra generated sounds and repeats after the end of the song, clicks appearing in the beginning.

[brazen prism]

Like this idea!!

[potent crane]

If they can do bulk delete they should be able to offer bulk download (multiple songs within a page) using the same code

[brazen prism]

agreed - although i have zero clue how the tech/code would work. 🤦🏼‍♂️

😂

[novel compass]

Would love more filter options for the library. For example, I would love to have a filter that REMOVES public songs from the view.

[timber drum]

UI on the website is already great, but i would suggest structuring creations with folders to unclutter the creation list.

my alternative workaround right now is doing naming conventions. essentially what i do is add tags to the title of a song

[novel compass]

same. That, and a bunch of playlists pretending to be folders.

[agile condor]

Add a feature where we can add our own background music/melody for Custom Mode so that the ai only sings provided lyrics overtop the provided melody (The ai generates a lyrical melody over the provided music file)

[novel compass]

I would love to see the videos the site creates, or that streams when a link is posted here, be updated to reflect the artwork and names of the sub-parts it is made from when we set custom ones. I would love to be able to use the images and titles as a kind of slideshow that would be cohesive and go with the music, instead of abrupt birds and whatnot. 😉

[twin bear]

VOICE PROFILES on SUNO 🙂

PLZ PLZ PLZ VOICE PROFILES on SUNO 🙂

[grand harness]

Can we get an edit tool, for stitching and terminating a song. Lets say we generate a song and it adds an extra chorus, we could say remove everything between those time stamps and stitch. Then when the ai has over generated past where whe wish to conlude we can say terminate song from this time stamp.

[modest frigate]

1. Place a symbol for downloading wav file directly there, in the line of the song. The pull down menu for downloading the file is waisting a lot of time and energy
2. Display the number of pages in the library and offer direct acces to a page. I have now 16 pages
3. Allow also partial string searchesin the search mask at the library like "tim" instead of "time"
4. Change MP3 rate to 256 or 330 kbit and place also a symbol beside the symbol for wav file downloading

5. Same with Splice: if you have only a like symbol is not enough. Some songs are better than others, so it would be better tobe able give one thumb for the good songs, but alsoto give a heart symbol for the best 5. best songs, so the users know which songs are the best and which songs are only good.

6. Make a tag to force the engine to use a certain instrument in the song and also one to specify intro, solo etc
7. Make a tag to force the engine to use a certain BPM. Also variable BPM in the same song would be great later (some produces accelerate the song slightly at the end or in certain areas)

[mint bridge]

So, I think this is being worked with, so I don't want to frustrate anyone, but if I draft up a song like the one I just did, I think it will fit within 4 minutes, but before I add style notes it's about 3100 chars, so with style notes it could easily hit 3500-4000. I just want to throw that out there and see what other people are experiencing.

[pliant tendon]

Adding that it could be pretty great if we provide our own vocals and Suno generates background music!

[tepid trench]

Likely going to be possible with a feature coming soon

[leaden marlin]

thank GOD!!!!! my game needs a placeholder OST and i have a few samples, but not even with a paragraph could suno get it down, maybe that'll help

[tepid trench]

Looks like you'll basically feed it some sound and it can figure out the tempo, melody etc. (there's some examples from the Suno team on Reddit and X)

[leaden marlin]

oh. i meant more like a 10-20 second sample of actual music

so i could make something 30 seconds long in my daw and suno does the rest

[tepid trench]

Yeah there's someone playing a piano in one of the demos and then Suno turns it into an accordion and adds other stuff

[leaden marlin]

mmm thats not really what i had in mind

[tepid trench]

I think you still get to give it prompts so it might still be useful. Hard to say. I expect there'll be some more examples to check out soon.

[spark river]

My suggestion is a "tip jar" option. Street performers can get tips from the crowd. If a listener wants to tip them they can do so with Credits. Tip your favorite creator.
What do you think?

[strong swan]

can confirm... it will do that. if you give it more ( longer and complex) input the extension will be more similar to the input. a 30 second piece should be plenty...

[ionic raft]

I would suggest, It would be great if you could provide a simple editing feature, like the one on a smartphone. Sometimes, certain songs may have a bad intro or outro, and we have to use other software to edit them ourselves. Having this feature would allow us to share perfect songs after editing in suno platform.

[hidden scaffold]

It would be great if there were an option to use my own voice instead of the ai generated vocals (by uploading my voice recordings, similar to what Kits Ai is doing). This is the best way Suno could help an independent artist.

[hearty pawn]

oh alr thanks

[safe ingot]

This guy has been doing this literally all day long, giving no one a chance to get views from the New tab. My suggestion is to figure a solution, I guess

[cloud wadi]

I had a feeling something like that was going to happen when to win the contest you have to have the 100 most popular songs

[safe ingot]

I don't think that is what the rules said

[glad steppe]

audio to style prompt would be very useful for those times you don’t quite know how to replicate a style of an existing song.

[rare agate]

It would be great to create a song with lyrics / voices / vocals and then have the option to download the song with and without vocals.

Another nice thing would be to use own vocals.

[tranquil sluice]

Are you sure? I tested phase invertion in reaper, and wav and mp3 song are difrent.

[feral aspen]

There was a bug right after the WAV feature where mp3s and wavs were the same file, but they should not be the same now

[leaden marlin]

oh my god my excitement is off the charts. do we have an estimated release date???

[potent crane]

That's good to know

[potent crane]

Encouraging news

[potent folio]

It occurred to me that this is likely already your direction buuuuuut

if it isn't: having the ability to create a station using genre tags would be great

for Suno moreso actually, bc you would get higher quality preference data

IFF it is in some way able to learn from users preference

and pull up increasingly focused songs to that user

but, presumably, this would lead to better matches between genre tags in user perception vs models in future fine tunes

while improving the listening experience

not trivial though to get right lol

and in all likelihood you would need a model that one's hots some set of preferences from the users to fast filter, almost like a lora but it's multipurpose and the tokens are previous likes and dislikes with compressed representations of whatever data the model believes is relevant for which songs it should produce. Idk, the more I consider it, the more it sounds like a user preference search algorithm

which is way way more than what my suggestion was

supposed to be

minimally though it would prefer songs other users with the same genre radio, and just things that are broadly liked

which is a lot simpler 😄

[sinful harness]

But Daily Timer On #🐣┃suno-showcase Show your song has bigger and longer time get peoples attention.

Timer could be like 6hours. So you could still post your song later on to the channel

[tranquil sluice]

By the way, just interested. In what format songs play on the site?

[feral aspen]

mp3 I think, the wav is only created if you ask for it, since it's very large file size

[bright topaz]

Has anybody found out how to make fast rap?

[stone steppe]

New Suggestion

• Negative Prompts. Pretty obvious in itself, but I will ellaborate.

Being able to break away from generic genres would be a big result from a feature like this. For example, when I attempt to make dubstep, I get nothing but the subgenre "riddim". Not that there is anything wrong with that subgenre, I really would like it to be more creative, sporadic or even old school (specifically artists like Liquid Stranger). If there were negative prompts, I could solve this issue by putting "stomping rhythm, riddim, modern".

https://tenor.com/view/vulcan-salute-spock-star-trek-the-original-series-live-long-and-prosper-gif-23575376

[wooden igloo]

As "spamming" Suno works well...
For 3.5 can you increase the input limit for style of the music?

[dim vault]

New Suggestions

• Being able to better dictate the type of vocals you get in a song. The AI currently dictates this by the style of musical instrumental way more than it should. IE: Giving Pop style vocals just because an instrumental is "Soft Sounding".

• "Delete All" Function for Trash.

• Have songs in the Trash be organized by when they were deleted.

[modest frigate]

Please make as in splice a check button for every song to enable multiple file download at once and a button to select all songs in a page or all songs. It takes such a long time to download all songs individually !

I could not find any manual/web page where the metatags/commands are showed... or the music genres the machine knows

[granite shore]

It would be amazing if you could also select multiple or all to delete. I have to individually delete all the disliked songs and since you can't sort by disliked to see only those songs, you have to sit and painstakingly scroll through your entire library. It takes forever.

[potent crane]

It seems to be the general view amongst AI experts that "negative prompts" are unreliable, and a concept that AI struggles with.

[hollow swan]

Put a BPM Option

Put DRY/WET Option

Put STEM SPLIT Option but Directly from the source , no f.e. (lalal or fadr or any other program type of stem mastering cuz of phasing and bad quality)
Same goes for MIDI

[tropic totem]

Hello. I am writing to you because I want to draw your attention to a very big problem in Suno v3 and v3 versions.5. Unfortunately, these versions in 80% of the compositions re-perform the verses and swap them. This greatly spoils the structure of the composition and its semantic message. I would really like you to fix this problem. There is also a suggestion in the window where the music style is indicated to create several buttons to select not only the style but also the dynamics (speed) of the music.

[visual brook]

when evaluating new generations, it sure would be nice to have the option to play back at 2x-5x speed (with proper pitch too)

[tawdry wave]

I would like, as a user, to have the option to pre-determine the Key, BPM and chord progression.
It is also essential that the result is download multitracks in .wav and .mid

[pale pelican]

Gonna assume that this was already suggested, but I'd like the option to see the seed and use it when I want to generate the same song type with different lyrics. Of course, this would mean that would also mean that instead of generating 2 versions, you'd generate one, cause I am assuming that it would function similarly to other AI gens that has seed use as an option.

[novel compass]

It would be very useful to be able to see which songs are linked together in some way. Meaning if I have an instrumental and then extend it into a song I would like to be able to see which tracks have been extended from it.

This would help keep me from making 2 different songs that have the same Intro section. It would also help people to see if someone else has extended one of their own tracks.

[tepid trench]

Unsure if this is what you're looking for exactly but on desktop, you can see the previous parts on the right of the page by expanding the "extended from..."

This doesn't happen when you have a "full song" though (it'd be nice if it did!) It'd also be nice if it showed the timestamp the extension was from.

[novel compass]

I would love the reverse of that to work as well. Like if you were to view the first entry that you could still see the 2nd one as a “child”.

The children see the parents but the parents don’t know who their kids are rofl.

[unkempt folio]

Please, Please, let us have more control on what parts of a clip we want to use. For example I generated a clip where the only thing I liked was the guitar solo in the middle. I would love to be able to use that some where in the song. So would like to tag that as such to get inserted when I see fit.

[hybrid crane]

4 to 5 week we get V4 with will have inpainting. And then you can redo some parts of the song.

[unkempt folio]

Also, can we have better organization with our songs. I have lots of songs I work on and currently it's a pain to go back in the library to find old song you would like to work on and finish.

[unkempt folio]

does that mean we can fix vocals where they might have created an artifact. like stuttering? or repeating a word when it should have been said once?

[hybrid crane]

Yes.. but only for songs made with V4

[unkempt folio]

It would be great if we can also edit the lyrics in songs after they are made to fix artifacts or change a lyric post creation.

It would be awesome if Suno could incorporate edutainment into there song creation to help budding musicians and composers actually become and understand the process of creating music. Currently, Suno is a musical Genie, but would love to see this grow into a tool for musical education and growing more musicians.

[brazen prism]

THIS!!!

[frosty python]

is there any gpt prompt about suno

[hybrid crane]

Plenty.. just look on chatGPT

[frosty python]

hmm most of them just prompt writers
I mean a lyric writer with all the tags

[hybrid crane]

Nobody knows all what is possible yet, so none of them will be complete.

[frosty python]

great I will share mine when I do it 😄

[gilded thunder]

please Add: custom instrumental exempel :[intro]

• [pads] (Warm atmospheric pads)
• [melody] (Gentle piano melody)

[verse]

• [drums] (Light drum pattern with snare and hi-hat)
• [bass] (Smooth bass line)
• [synths] (Soft melodic synths)

[pre-chorus]

• [drums] (Build-up with additional percussions)
• [synths] (Rising arpeggio synth)
• [effects] (Subtle risers and sweeps)

[chorus]

• [drums] (Full drum pattern with kick, snare, and cymbals)
• [bass] (Deep and punchy bass line)
• [synths] (Layered synths with catchy melodies)

[verse]

• [drums] (Light drum pattern with snare and hi-hat)
• [bass] (Smooth bass line)
• [synths] (Soft melodic synths)

[pre-chorus]

• [drums] (Build-up with additional percussions)
• [synths] (Rising arpeggio synth)
• [effects] (Subtle risers and sweeps)

[chorus]

• [drums] (Full drum pattern with kick, snare, and cymbals)
• [bass] (Deep and punchy bass line)
• [synths] (Layered synths with catchy melodies)

[bridge]

• [pads] (Atmospheric pads)
• [melody] (Reflective piano melody)
• [drums] (Minimal percussions)

[chorus]

• [drums] (Full drum pattern with kick, snare, and cymbals)
• [bass] (Deep and punchy bass line)
• [synths] (Layered synths with catchy melodies)

[outro]

• [elements] (Gradually removing elements)
• [pads] (Ending with warm atmospheric pads)

[effects]

• [reverb] (on pads and melodic synths)
• [delay] (on pads and melodic synths)
• [automation] (Filter sweeps and volume changes for transitions)

[tepid trench]

The full-size videos are nice for posting on other platforms but it'd be nice if Discord embeds weren't so big.

Very rough MS-Paint mockups of layout idea with crude copy/pasting from the Library page 😄 - the idea here is to show the style tags first and then switch to displaying lyrics as the song progresses. I think it'd be good to have it close to how it appears on the website in terms of styling and layout.

Top half of this example mockup would be for before playing a song, and it'd switch to look like the bottom half during playback

As it is at the moment, #💬┃general-chat is often very busy with people posting their songs which can make it difficult to follow the conversation.

[stone steppe]

unreliable, but they work.

as long as it works

it extends the chance of ridding genericity

[sharp arch]

when u go in libary thatu can see how much views and likes your songs have on suno

[quartz steppe]

I just tried to extend a track, after several thousands of generations, but ran into a problem.
I can set a "new" list of tags for the next part in song style section, it looks ok, I am "fixing" tags this way.
But when I click "Get Whole Song", it restores original tags in the result, throwing away new ones. I can't control this process in any way and there is no possibility to edit tags at the moment.

This is a huge blocker. It doesn't allow me to extend any track, and potentially publish it, without placing wrong tags and ruining all training progress for AI.

[forest mauve]

Hey Suno team,

How about a front-page category just for custom lyrics? Imagine a lyrical playground where every wordsmith can shine. Plus, it’s a perfect hangout for those who adore clever rhymes and catchy phrases. Let's give our lyricists a stage to rock their words!

[potent bough]

man if any devs are here, i would love a "pay extra credits to re-run a song at high quality" option, im ultra bummed out suno generated a fantastic song that is almost entirely unintelligible by halfway through

[leaden marlin]

special account for registered .edu emails (Student pricing).

Would give you the same permissions as the basic package for a lower rate. Verify student email .edu to gain access. Other AI startups have begun to do this, and OpenAI just announced a university package for their GPT suite. Spotify, too, has a student rate to get their Premium service.

[raven bluff]

I’d love the option to generate an instrumental first, right, then add vocals onto that using AI. Alternatively importing a beat you made, for example, and using suno to generate Vocals for that. I think a lot of people would like that.

[subtle latch]

Suggestion: Reposition pop-up notifications to elsewhere (like the top). Current notifications cover player controls and sometimes gets in my way when I'm needing to play or pause a selection.

[amber violet]

when generating, generate into a 'folder/playlist'. sometimes I want to work in 2+ different ideas... and can't track it that easily.

[upbeat mantle]

add it so there is a free song tag on the ones u made during the time u dint have a subscription easier sort them out

[tepid trench]

This, plus a way to "buy" songs made on the free plan (using credits) without having to contact support to ask about getting ownership of the songs

[leaden marlin]

Please sound effects

Suno ai is getting boring

[vague prairie]

Have you tried extending the song from the point before you found you stopped liking it?

[soft harness]

There are sound effects, you need to find out what terms to get the AI to use it though.

E.g. *gunshot* for gunshots as demonstrated in https://suno.com/song/2ad4ff91-a1ae-4bda-a05c-3e103d37488b (source: #📚┃suno-school message)

[lofty hatch]

The option to split gen between models: instead of 2 v3.5 do 1 v3 and 1 v3.5 . Or have it gen 4 (2 of each) Helps get a feel for the difference between models and also stops me from having to toggle back and forth. currently I end up using v3 instrumental beginings as the sound and gen better, and exntending it with v3.5

[scarlet shell]

it would be cool if you could have like an ai generated producer tag sort of that plays at the begining of your songs when you post them

[potent bough]

i have but what I've found is that once there's "noise" in your song, no matter what from that point forward every re-generation or extension keeps adding more noise, its like exponential

i generally instantly delete songs that begin noisy and not clear

[dreamy sigil]

The show case section in discord should be broken into forms stipped into 10-20 ish core main music styles

[low crane]

Add FanArts disc guides channel. Simple Guide of a productor and youtuber @hidden vault (and more)
https://silver-writer-ed8.notion.site/SUNO-AI-de-0-a-PRO-f3578a0a567b41bb8302a69e29313906
I'ts very util chance to create profesional lyrics prompt, too. 😄 (Expemple because this ytber is Spansih producer) jejeje ❤️ Onle Love One Kiss ❤️‍🔥

[jolly inlet]

Let us regenerate the videos with new album art

[raven nexus]

Please figure out a way in your UI to give more control to users about intros and endings.

I can't tell you how many credits are wasted by trying to extend instrumentals and music only to have the algo continuosly want to just end the song - 50 times in a row.

Annoying is an understatement.

Or allow us to force a length... If you say it can generate up to 2 minute extensions, allow us to force the model to do that by putting in how long of an extension we want

[woven pilot]

It would be nice if we can select the instruments used in a song

Or atleast the main instruments

Also select no drums

[amber finch]

anyone know the max length expected for v4?

[urban sinew]

I see some other questions about not using certain instruments. Is there a way to do this? I've been trying to get some cinematic soundscapes, ambient but I don't want piano. I tried doing custom with [no piano] but doesn't obey. New to this so sorry for repeat question if it's been asked. On a side note I've been extending songs and it gives me like 10-19 second songs with low sound quality and I lose credits. Reported a few but feels like I am just losing credits.

[urban sinew]

was just gonna post this. Do you notice that some of the extended songs are very short like 10-19 seconds and they still take credits?

[earnest imp]

I don't want Instrumental to add the "woowoo~" of the human vocals.

[jovial shoal]

I would be perfect to cut section of a song in my case i just make a great song but the end is a little nuts it would be perfect to able to cut it

[urban sinew]

Instead of the reuse prompt it would be a good option to reuse a song and give variations of it.

[amber finch]

suggestion: vocal controls for lyrics? adding ^lyric^ if you want the word to be sung higher or accented? ~lyric~ if you want it softened? things like that?

[pliant tendon]

Using negation in a positive prompt doesn't work. If you tell Dall-E 3 "and no trees," it will generate trees. This is like telling yourself not to think about pink elephants - you will likely think about pink elephants.

But Stable Diffusion for example has two prompts (SDXL has 4) - a positive prompt and a negative prompt. The positive prompt describes what you want to see, and the negative prompt describes what you don't want to see. From my experience, negative prompting is incredibly effective at removing what I don't want in an image.

[leaden marlin]

Fail, it's still music and that's a drum, it doesn't count because it go chaotic

[barren loom]

Can we please get negative prompts?

[gloomy python]

NEED Suno to save split tracks for editing.

NEED to replace short phrases of lyrics that come out electric like Max Headroom, or are slurred and indiscernible without redoing the whole otherwise perfect song, and to remove odd tics from otherwise perfect audio.

[uneven belfry]

Let us regenerate the videos with new album art & title

[gloomy python]

You can click the tripple dots, chose Edit Details, and replace the album art. You can right click on the song and chose rename.

[uneven belfry]

I got that part I do appreciate it though but the lyric videos

[gloomy python]

NEED to keep the awesome voice and music, but redo odd lyrical timing in small sections.

[soft sky]

What should I do to generate a song through lyrics

[amber finch]

preemptive suggestion. for the inpainting. being able to highlt multiple sections at the same time for revision would be useful (like if you were trying to replace the chorus throughout a song)

[frigid pivot]

1. Style Negatives field.
2. Ability to either cut existing songs, or stop generation in progress.

[limber turret]

Will Suno ever entertain entering the world of crypto & digital assets. A suno token airdrop will provide value to the largest stakeholders (the users) and will surely become a powerhouse

[sinful harness]

Make Version that allows to make 3minute songs too. Currently l had been having the issue that l my writing lyrics that can fit to 3minute song for no issue, ( what is the referred length l use. But everybody has their own taste. )

But so far if the AI run's out Text in V3.5, it will try fill out the song to meet the 4minute mark sometimes. Even though l have added in addition to the Ending like [Bass Solo] Or [Dramatic Ending] and still it keeps going and trying to fill out the rest 30s or whatever to try make it 4minute long song.

And the annoyance part is l have to either cut it away with Extend or some other way to make sound more fitting.. 😅

[keen marsh]

we really really really really need an editing option for created songs. example: suno gives me a song that sounds good but the vocals don't quite fit and i can create a new version based on this song. sound, hook etc. remain identical

[sinful harness]

That would be nice. Like have the same Vocal the AI created, but resetting and "re-recording" so to speak. So it will try use the same Vocal again, but try get it more correctly the other time. Would be really nice!

[keen marsh]

in my case: unbelievably brilliant chorus, catchy af! but suno put too much white noise in the vocals for most of the song and doubled some words that shouldn't be there, experiments to keep only the chorus and change the rest didn't work. i really miss an edit function...

[sinful harness]

if you put words into (like this.) it will use Echoing type of singing. So void using those if you can help it. But otherwise, yeah it likes to do that. Sometimes it feels really good, and sometimes unnecessary in some lyric that do not make any sense 😅

If you want example. l can sent you Song where l use it. So you can see what l mean by Echoing effect

[keen marsh]

sure, i can send you my song too where you can hear the white noise what destroys it 😦

[sinful harness]

Sure thing! :3 l would love to hear it, and maybe l can help with the overall song construction and see what might cause it!

[keen marsh]

i have sent you a pn

[sinful harness]

👍

[late skiff]

Improve the distinction between different variants of the same language in your training materials. Maybe tag them with the corresponding language iso code, like en-GB, en-US, pt-PT, pt-BR , etc . This will hopefully help ground the song to a specific language and accent. You can see that easily in Portuguese where the singer frequently alternates between Brazilian, Portuguese, African Portuguese accents

[fierce sky]

Hello, I would like to know when the midi version of the music will be available. A bit like AVAI AI. This allows the composer to retouch it more easily

[raven nexus]

Yes, a lot more now

[waxen creek]

I have noticed that the latest Version, when you input Custom Lyrics, the AI seems to ignore verses marked as (Bridge), and skips strsight to the next verse or chorus.

[sinful harness]

It does that sometimes

[sinful harness]

Also Use [] Marks rather Brackets

[fair cipher]

Given that we now have longer context, and the AI likes to try to fill up all the space even if it has to repeat a bunch of stuff to do it, maybe you could give us a slider for how long a song we want to generate? Not every song has to be 4 minutes. And if you extend from the 4 minute mark, it's always gonna try to give you another full 2 minutes even if you just need 30 seconds to do an outro. Also if you let us generate shorter parts, it uses less compute.

[signal depot]

Hello, Can we please get an update on where we can download the stems and trackouts of the song? that would be just perfect for remix purposes!

[lethal trail]

ohohoh, big suggestion here. Seeing how both the AI and website has been updated and upgraded over the past... insert time here, I think it'd be super duper cool if the id:guide could get updated to reflect this

[proud dome]

Can a Folk or Singer-Songwriter genre channel be added?

[proud dome]

I would love a way to trim songs. There are so many times the song ends then after a moments silence it kicks in again with a random section that then over runs the 4 min mark

[dusty river]

((( i suggest WE ALL show our love to the devs and ALL type WE LOVE SUNO in general to show our gratitude for this amazing sight!!!!!!!)))))))))

only half joking. 😄

[oblique patio]

Yeah, [Bridge] works strangely in 3.5. Also, annoyingly [Verse 2] or [Verse 3] starts to repeat the verse based on the number...

[oblique patio]

Extend the song where you want it to cut. Empty the prompt and put [End]. If you started with V3.5, switch to V3. For me, it creates like 3-10 sec ending.

[proud dome]

[thanks i will try that]

[oblique patio]

Then just get whole song from the second part.

For some reason V3.5 loves to just keep going. 😄

[dusty river]

it tries to gen a full song of 4 minutes

typing [end,short] 3 times and then [end] seems to help some of the time.

[hidden vault]

omg thanks, thats me haha. There´s a full youtube video (in spanish) about all the little things 6 months of experimentation taugth me in suno :). I maybe should record the english translation

[lyric quail]

Add vocal to an instrumental song?

[hybrid crane]

Here a sneak peek of what is coming soon: uploading your own samples and let Suno make music with it:
https://www.youtube.com/watch?v=6rzt1WNGUSc

[scenic oracle]

It would be great if you guys made a feature to separate the stems of our generated songs

[tardy lark]

I would love to have a feature to regenerate a track with one prompted tweak. For example, "remove reverb". I have many times when a track extension is spot on to what I'm looking for, but then there's a reverb dialed up to 11 on one of the instruments that just ruins it.

[gaunt briar]

there honestly should be a setting that lets u choose how high quality the song should be but would take longer

[delicate cedar]

They hope to make it easier somehow, I know this has been mentioned before. But as it stands now and possibly will always be the case, the method they use to make the music has no stems. It is created whole cloth as one piece.

[umbral flower]

Suggestion: reverse music: feed the music to the suno get the prompt (will be so cool to steal other songs make more songs you like)

[tropic prawn]

If a clip comes out to less than 10 seconds and it would refund the credits have an option to click and opt for the refund, sometimes literally just need the instrumental to outro in under 10 seconds but it drags the outro out, like the song comes right to the end but then picks up for no reason

[zinc void]

I think we should need a button like :
+The Remix Button [this can help you with the same sounds but other beats]
+Instrumental [Remove all the voices with the music]
+Lyrics Changing [When there is a wrong words or any words that have mistake , you can change it back]
Hope this helps!

[indigo ginkgo]

The main two things I would love are:

1. the option to add a negative prompt, since sometimes there is a sound you really want to avoid. I feel that the newest model is very good for pop, but it's too pop in other genres

2. I know this might be a pipe-dream, but I would really love to train lora's for suno, to nail down a certain direction of sound or certain combi of sounds even. it would also free suno of any legal questions, while allowing people to train on whatever they want. Lora's are awesome and it would really help and could spark a creative community.

The same goes for "control-net" like features. I come from the stable-diffussion community, but img2img => song2song would be pretty cool, keeping the temp, structure or sound specifically.

Basically adding things to the latent space would be cool.

[jolly inlet]

Man I wish more AI services allowed stuff like Loras and Control-Net. It just gives you so much more control, but so far SD seems to be the only thing something like that has been implemented.

[fluid saffron]

For training something similar to midjourney would be cool. Where you have two songs (let's say similar prompts) and you vote for one of them based on some different parameters (like which one has higher compliance and which one sounds more pleasing etc)

[robust agate]

select all/delete all for trash please.

[vague moss]

Can you make the explore searchable so we can look and see if Suno knows a type of music. While the wheel looks cool its very hard to find what you want because things are not properly grouped

[dawn olive]

I assume it's that it's training database doesn't include too much of this, but some genres are just not that well implemented with Suno. Post punk electronic late 70s to early 80s for example. 90s synthpop dark electronic is another. I know they may be limited in accessing data sources for these sub genres due to copyright but it does feel that the prompts just don't get it, as there's little in their dataset. I hope they can sort.this out, otherwise I can see Suno being limited to more modern dance pop.

[cosmic terrace]

i would love (if not been suggested before) more for convince 1 : ability to sort my own playlist\albums.
2: pack several playlist into a folder (sub playlists\folder under a main playlist \folder)

[scenic oracle]

I see. Thanks for the info!

[undone bough]

so the next update makes it possible to edit the lyrics and genre plz! it's a mess if u extend and when there are repeated verses the text is missing for that too. I know millions of people have repeated this, but it can't be said enough. It ruins the whole export video thing totally! Text and comments and commands in the exported video just don't work well. Anyhow thanks for a super app and your effort to keep up with suggestions and requests 😛

[strange ginkgo]

An AI tool idea came to my mind last night. With AI, things that were once impossible can now be achieved. There is a creative enhancer tool for images now. Why not an audio version of this? In other words, it is an AI tool that analyzes a simple demo recorded at home in a short time and improves the production quality. Suno.ai specializes in audio in the field of AI. Can it provide such an opportunity to people in the future? It will also present 2 or 3 variations of the improved song in the output section. How much the original song will be affected by the creative enhancer can be determined with the percentage bar. Something like this came to my mind and I wanted to share it. Regards.

[vale flax]

Letting other people add songs to my playlist (with permission) This way we can add songs to eachothers playlist and don't we have to both keep sending each other cool songs.

[novel compass]

Please let us enable shuffle mode for playlist playback. 🙏

[thin palm]

Please add a commands channel in this server

[hybrid crane]

what do you mean with that?

[thin palm]

To play with the Suno bot

[hybrid crane]

That is the old way. They removed that because just 1% was using it. So it is now fully on the website https://suno.com

Suno discord bot is not coming back

[thin palm]

I understand

[hybrid crane]

website is better version anyway. Bot here was just V1. 😉 🌼

[thin palm]

Ok, I will test it

[hybrid crane]

Have fun.. and if you need help, here a great video I made about the current version: https://youtu.be/hsecBccWETM

[thin palm]

Thanks you so much!

[hybrid crane]

Your welcome 🌼

[worn atlas]

Great idea for having translated in several langages.
Thanks @hybrid crane

[hybrid crane]

Thanks.. 😊 🌼

[astral isle]

Making songs public by default in the account settings

[grim geyser]

for us musicians I would like it so that we can give it a midi file to follow, so we crank out the melody, and it gets supplied along with the lyrics, and the vocals follow the midi track.

[harsh laurel]

Can AI create Cantonese songs and pronounce them accurately?😩

[tepid trench]

Hmm... I wonder if this could be a possible future expansion of the upcoming feature that lets you upload audio into Suno (to then generate something based on it.)

Initially you might be able to e.g. render a WAV file locally from a MIDI file using SoundFonts and then upload that, but eventually that could be something Suno does automatically to handle MIDI input.

[grim geyser]

I was suggesting melodies for only the vocal line so as to avoid copyright issues where people submit melodies from instrumental parts, and I don't generally use soundfonts, I can make generic MIDI files that I use for controlling VST plugins, and outboard gear.

[tepid trench]

Ahh right. I understand what you're asking for now.

[grim geyser]

for whatever is used to send the melody data to the AI it would need to try and maintain a certain level of human personality, so it should only be used as aguideline anyway.. 🙂

some other neat feature woudl be to be able to tag [call] and [response] in the lyrics box so as to make more interesting music using antiphony. those are all the ideas I've got today

[tepid trench]

I think it might be possible with the new feature to just sing the lyrics as you want them to sound and have it transform them somehow. No idea if it'll do a complete vocal swap or keep the melody intact. Might have to give that a try when it's available.

[grim geyser]

I like that one, Rifffusion has that, but it can only generate 11 seconds of music.. lol

I like how Riffusion handles it though, they time stamp every word, and so you don't have to sing it, or be talented. lol

[tepid trench]

I don't think Riffusion takes pitch into consideration, just the word timings (at least, when I tried it.) I ended up here as 12 seconds is not enough 😄

[amber finch]

suggestion for inpainting: being able to change how much time in in the inpaint, using the time time vs less or more, something you need an extra sec to balance the song

someone else suggested it a couple of days ago but there needs to be (some) way to stop gens. lots of gens with typos or defaulting to the wrong model. wasting a lot of credits.

[quiet lark]

if the devs want another interviewee pm me would be interested money not needed

[rotund sand]

I wished they allowed a comment section on the songs we listen to. That way, people could say they liked it or even offer advise or tips. That's just me.

[wispy tree]

Heavily agreed.

[fallen lintel]

Just curious, will Suno be allowing us to generate songs with a specific end time/duration or we are able to set one in the future? To prevent hallucinations as the music gets longer.

[warm aurora]

https://tenor.com/bCjEm.gif

[formal bear]

yes probably, or at least give you an editing option, so you can trim it yourself

[fallen lintel]

That's good to hear, really looking forward to it as I have been enjoying myself with Suno, cheers Team! 🥳

[formal bear]

oh yeah great idea

[formal bear]

@spring silo

[formal bear]

inpainting will have a mind of its own lol

we will have some duration control but probably not guranteed

[delicate agate]

It seems there's a certain pattern of image generation. It would be nice to be able to select the general style of cover art to help distinguish the tracks.
• Color Splash
• Bird with a Guitar
• Winding River Valley
• People
there's probably more, but these are the ones I've seen recurringly

[brittle loom]

Not sure whether it was already suggested, but i have thoughts:

• A feature to regenerate just a part of a song (like you choose start and finish timecodes and Suno generates something that will fit there).

• • In addition to this: a feature to tell Suno whether a new part's length should be same as removed part or it remains up to Suno

• Extend a song back, for example, to add an intro

[heady pier]

Hello, the subscription credits should be accumulated each month, not all people have all the time in the world to spend them, I find it very bad that you pay for those credits and if the month passes they will take them away for not having used them, in addition to that buying credits are much more expensive than what a subscription is worth, I would also implement more subscriptions such as... another one of $16 with 6000 credits, I would even implement another one of $32 with unlimited credits, that would be good, in addition to the prices of of the lowest credits if you have a subscription, I see it logical that they are expensive for someone who does not have a subscription, but for someone who has a subscription it is logical that they cost more expensive when taking out another subscription would be more profitable.

[wind marsh]

I would like to create songs similar to the ones Type O Negative created. The lead singer's passing was unfortunate back in 2010. Such an iconic voice.

[thorn monolith]

There needs to be a style that uses acapella harmonic/polyphonic vocals like 90's quartet group Boyz II Men or more modern groups like The Wellermen

Suno doesn't appear to really be able to do harmonies very well/often

[full ridge]

I'd go one step further and suggest being able to collaboratively create songs together. So you create Part 1, and then you can invite editors (like Google Docs) and someone else can create Part 2. I think this would be incredibly fun and cool.

[viscid cloak]

Can we deactivate dislikes already?
Right now the Trending is a mess and some songs get downvoted and people targeted.

My Pride Song was downvoted into oblivion and people are being targeted

[winged sable]

It would be nice if I could choose which music the AI ​​would base the generated songs on. It's very difficult to avoid music ruined by forced choruses everywhere. For the same reason, it is very difficult to avoid anything being synthpop or arena rock. Maybe with a meta tag like [nochorus].

[hardy scroll]

I want a feature where the ai can copy other music from yt (ofc no copyright) and take the music without lyrics.

[thorn monolith]

If you're looking for that feature right now, you can do that in Audacity as long as you can get a recording of the song.

[forest mauve]

Suggestion for the Dischord:

How about a dedicated feedback and criticism channel? That would be really helpful! Showcase and General are different! You don't wanna tell someone how they can improve when they are showcasing their latest masterpiece after all, and be a buzzkill... that's why we need a dedicated channel where people can come together and pick songs apart, to give constructive criticism.

[inner rover]

We definitely need an unlimited song generation subscription for a lower cost for paid users.

[safe ingot]

Having a dislike feature that nukes the songs make zero sense if they are going to have a contest like this.

[forest mauve]

Dislike butten doesn't make sense if it removes likes

If they simply ahow the numbers thats better

Show*

[inner rover]

And Suno, when we going to see a discovery page not just a trending list, would be nice to have a more accessible song discovery method so other creators could find, follow, like and engage with more users. 🔥👌

[leaden marlin]

I will leave it here as well:

my suggestion: like/dislike only in Create mode for the creator, and for everyone else a heart/favorite button instead

[viscid cloak]

Also its used for hate, example Pride Songs.

Remove the Dislike Button

[forest mauve]

The songs that blow up often have a very polarizing ratio of likes to dislikes

Don't remove the dislike button!

Just make it not matter how many dislikes a song has

[viscid cloak]

The users dont see the dislikes anyway

[forest mauve]

Thats fair! But its good for the creator to see it

[pliant tendon]

That's probably a better idea because dislikes based on biases aren't productive

[safe ingot]

The creator doesn't see it either

[forest mauve]

Yes! This!

[viscid cloak]

The song creator doesnt see the dislikes either

[safe ingot]

It literally does nothing useful

[forest mauve]

Oh. Ok, so here's how to fix. Keep like and dislike button. Show creator the numbers. (Maybe show everyone)

[pliant tendon]

I gathered it must affect something on the backend?

(From this chat)

[forest mauve]

And most importantly many dislikes and many likes at the same time means your song is awesome!

All good art is polarizing

[lethal trail]

The only use it has is to affect the AI's ability to generate something similar down the line

[forest mauve]

But DON'T make dislikes mean anything to the algorithm

[viscid cloak]

Honestly, Dislike is just used in bad faith. Dont see why it should exist in the first place. 🤷‍♂️

(Not the AI Training one)

[forest mauve]

Because when you encounter enimies you know you are going in the right direction. Like in a videogame

Just the data is useful. It could only be visible to the creator, if that is preffered

[frozen geyser]

Nah. Just remove it.

Why would it be useful to know that ppl did not like my song. What Am I to do with it? I do not know why they don't like it. Is ist the genre? The text? Some in between stuff? The only good feature is to tell the AI what was good and bad so it can learn better.

[vagrant elk]

Hello everyone! 👋

I’d like to suggest adding a "Stop Generating" button. Currently, even when I **delete **a song and move it to trash, I get the message, "Please wait for your generations to complete," when I try to **start **a new song.

It would be great to have a stop button because if I listen to a song in progress and don’t like the style, I want to be able to stop it immediately and create a new one without waiting for the current generation to finish. The new v3.5 takes some time to generate a full song, and this feature would **significantly **improve the user experience.

Thank you for considering this suggestion 💗 🎧

[winged sable]

Great idea!

[shell valley]

After i read a little bit about the dislike button...it could be helpful to have more Dislike buttons. Is the Voice god or bad? The music/instrumental? Did the AI use my keywords good? Because after 3.5 the music style means nothing if i use it.

[tepid trench]

I use the dislike button if it simply sounds bad (especially poor audio quality). On the "..." menu there is a "Report" submenu where you can report a bug, which is useful for highlighting specific problems:

[vagrant elk]

I have another suggestion 🤩

it would be really helpful if we could **see **and **copy **the ***seed ***for each generated song. Currently, every new song is created with a new random seed, which changes the voice, tone and music style, even when using the same lyrics and music style.

This is problematic because if I make a *mistake *in the lyrics or want to make minor adjustments, I can't replicate the exact same music style and voice. I end up with a completely different generation.

Having the option to use a specific seed for **consistency **would be fantastic and greatly improve the user experience.

Thank you for considering this feature as well! 💗 🙏

[winged sable]

Have you tried to extend that song from almost the beginning? The style remains the original, and you can fix the mistyped lyrics insert bridges etc...

[keen epoch]

How do I set up SunoAI to create a Korean Trot song?

https://tenor.com/view/hi-gif-27584362

[hallow dagger]

After the many times, i want to give a chance to Suno for v3.5, Music structure is good and 4 mins is really good but Vocals are still robotic, not natural. First, Suno has to solve this problem. Then it will skyrocket.

[sand scroll]

I'd like to see something where you can copy the seed one generation used, in case it had the feel you were going for but like you wanted to make some edits to the lyrics or something.
It would make it a lot easier to be consistent with a song's feel.
Also, maybe something where you can crop a generated song to deliver a more finished result in case the AI goes off track and starts spouting nonsense after its already over.

[wispy laurel]

Just noticed the add of images for playlists and individual songs. what would be great is a video version option to download the lyrics with a green screen so that you could overlay you own video images. Ideally without the suno brand even if it cost tokens. When things like open AI Sora is released this would be a huge add, as creators will be able to create music videos with the overlay, just a thought. Also I own a software company so I know how challenging these requests can be and yes it's a "Could you just"😆

[wispy laurel]

That would be amazing if you could create a voice or seen to use as a library👍

[thick charm]

What would be really useful that I need would be a way to preserve the essence of a song while making tweaks to it. For example I have a song where I love it so much but the man is singing the woman's part and vice versa, and there some small lyrics changes, and if i generate a whole new song i'll lose it. So maybe if the AI could copy melodies, even if I can hum it the melody or sing it and it extrapolates that something like @sand scroll said

[marble bough]

i'd really really like a timeline that shows stuff from people i follow, and also maybe (optionally ideally?) recommendations for me

i was looking for where stuff from people you follow shows up and i have no idea >___<

[nova raft]

The V4 version hopes to have shamisen, taiko drums, and pipa instruments more obvious, as well as a random refresh button for style labels.

[vagrant elk]

Omg that's cool idea i didn't thought about this i just tried it now and it worked :D 🔥 ✅

But it only preserve like 60% of the style not 100% the voices and intro are not the same for example the original song was female voice when i extended it it was male voice and it had similar intro and structure but it was not the same as the first one

also i can't enter the time as 00:00 I had to put it as 00:01 to start from the first second but I think if the seed is transparent and you can copy it and give it to others so they can build on top of it would be more cool and flexible and convenient than using the extend feature

Thank you so much for the idea I really appreciate it and I like your pfp man 💗 🙏

[glad sorrel]

Sometimes you created a great piece of let's say... 1:44 but there is a small error in the middle, just a word error... If you regenerate from the previous second, you lose the great parts previously generated. Including support to regenerate a time interval in the middle of a previous one would be so great. I'd love it to fix 99% right songs I don't want to recreate... Is it possible?

[cosmic terrace]

this prob been answered already but if not you can generate you're own cover art on each part now all up to u tbh or add another own image 🙂 (thanks devs for that addon ❤️ )

[delicate agate]

yeah that's not what I was talking about but it do be facts

[cosmic terrace]

an intro inpainting would be wonderful im sure it will get there eventualy 🙂

[cosmic terrace]

then i appoligies for misunderstanding it 🙂

[flat comet]

Hi, It would be an awesome feature if we could tag our songs for languages or country and people could filter for that on the main feed. This way songs on other languages could reach people with the same native language. It may be even an opt out for suggestions (so people in Italy by default see the top Italian songs). English songs obviously dominate the main feed and it's hard to organically reach listeners in another language 🙂

[fathom ivy]

1

[rustic heron]

I wasn't aware it emptied itself. It's more an OCD thing that functional lol

[winged sable]

Your welcome! By the way I never tried to extend at the very first second, just adding new verses to earlier unfinished songs. But it also flattens the end of the song a bit. The finer details disappear from it. But that's not so bad, because in the end we'll record the ideas with a live band anyway :)

[timid wasp]

Discord channels for sharing specfiic genres would help with discoverablility

[worn atlas]

Already exists

[soft harness]

#1154790975240155247 - If you do not see this channel, go to Channel & Roles (below the server banner, above Info category) > All Channels > Scroll to the bottom and enable Genres

[spare orchid]

Add a feature where you can ALSO get an instrumental version of your song!

[timid wasp]

OMG - thank you

[potent crane]

A way to store your favourite prompts somewhere, so if I want a punk song in a very simliar style to a previous song I can just select it from my library of prompts, at the moment I have a word file with a list I have to cut and paste.

[upbeat mantle]

increase the prompts in style of music from 120 to 200 gives more creativity that way

[velvet vector]

It would be great if you could regenerate the video with the new cover image. Also if you could edit the lyrics that are seen in the video in case you have a typo or something.

[grizzled coral]

Exactly what I was coming here to say about regenerating the video. 100%!!

[vagrant elk]

That's cool man <3 I really like your ideas 💗

[prime widget]

Is there a way to rearrange the songs in playlists? I have been doing this the long way by dropping them in the order I think I want them, but there must be a better way?

[umbral grove]

custom images would be nice for free users just give it a try

[light hawk]

I would love it if after a song as been finalised be it from first-generation or with extensions that you can edit the lyrics so you can have them reflected properly in the finished product. 😄

[light hawk]

Oh you already said this bahaah, nice! I am glad people agree with this. I also like the first suggestion you have about being able to edit in new images for the song as well 😄

[leaden marlin]

remove "kill" from the lyrical blacklist

[toxic otter]

Prompts to finish a song, [end] [stop] [fade out] version 3.5 just wants to keek going and going and going and going ahhhhhh

[light hawk]

I think prompts like: [Intro] [Verse] [Bridge] [Pre-Chorus] [Chorus] [Outro] [Instrumental] should be recognised if it isn't already hehe. I think these can help people format their songs better as well. That and they are terms used in the music industry that people can look up like for example: Here

[tepid trench]

(Reminded of this by recent post in bugs channel):
Replace the "Page #" in the library with a text box that behaves similar to the search:

1. Show "Page 1" in it
2. When the user interacts with it allow number input
3. Then, when there's no input for a second or so, load that page of songs
4. When the text box loses focus, replace its text with "Page #" (with # being the new page number)

[vestal cargo]

I leave my suggestion, maybe someone has already asked... extracting just the beat of the song without the voices, like generating only the instrumental with the same "seed", something like that, being able to use the same beat with other lyrics and other voices

[subtle latch]

Add an area for a description that's user taggable. For example: "In collaboration, lyrics by @user, styles by @user. Created together in partnership for 'event'"

I also think allowing descriptions can allow for more communications from the creators.

[warm mirage]

Hello to you! First of all, many thanks for making this tool for assisted recreation or supervised imagination accessible 😉 After a good use of songs with original lyrics I must point out certain problems encountered: the first is the total lack of control the length of the timing of the pieces. Added to this is the fact that the [End ] command does not allow you to definitively stop the song. Most often the AI ​​fills the remaining time by repeating a random loop of choruses or verses! Would it be possible to force it to respect the chronology of the piece without making random additions? Thank you all for your work!

[fair turret]

Would love to see us be able to weight the styles in our prompts. So we can control the infuence of each style also please quadruple number of prompt characters

[fair turret]

Please give us basic presets separated from stype prompt. Like Voice Classification, which instruments to include. I feel these should be seperated out from styles. I should be able to say I want a Male Voice with Ominous tone and that be the setting for the song along with what instruments must be included before even looking at styles which I think would be more helpful if totally separate with regard for analysis. Combined with the above suggestion would be absolutely awesome and a huge step forward.

[thorn monolith]

Hmm... saw something about disabling dislikes, and I guess I'll have to be the voice that disagrees. I'd actually like more ways to filter music out that I'm not interested in. For example, I'm never in the mood for Rap or Heavy Metal—it's just not my taste, but I recognize that there are others that may want to listen to it. Could you create a filter that allows us to not only find the type of music we're interested in, but allows us to filter out the music we don't want to hear?

On a side note, it'd be nice to have negative metatags to filter out styles, sounds, etc that we don't want in our creations.

[tiny breach]

It would be great if you add Portuguese language in Suno (I mean Portuguese from Portugal. The Portuguese you already have are from Brasil. This way people would be able to try out making Fado songs)

[fair turret]

I dont think you should dislike someones music if you dislike a genre that doesnt seem fair to the artist. I agree there should be a way to filter genres though.

[warm mirage]

I hope that a "split part" system to dissociate tracks and layers will soon see the light of day, this would REALLY be a major option for being able to evolve a specific song and use it as a basis to modify certain tracks as desired and create close variations!

[thorn monolith]

Not sure why downvoting a genre of music isn't fair to an artist as it's an attribute of the song just as much as the lyrics and composition are. Suno doesn't restrict features or access if a song gets downvoted into oblivion, does it? The risk of making a song public is that it exposes that song to be judged by the public. Not everyone is a rockstar because some people are afraid of rejection... or create music others don't like.

[subtle copper]

Add Bahasa Indonesia at #1154772025940922438 @here

[feral aspen]

Hey a suggestion I can actually just do: https://discord.com/channels/1069381916492562582/1247772106872193034

[feral aspen]

Please don't ping with here or everyone though!

[fresh wind]

is there any way to state the length of instrumental breaks

cause I don't want to go to lyrics right away but do some instrumentals for a bit

[late skiff]

Add a Description text box to the song details, so we can add a short text description explaining the song or providing a translation, etc

[mystic smelt]

I would like to see two things on the service:

1. The ability to recreate your own/someone else’s generated track in the same style. There are situations when the generated track suffers from bad vocals with bad accents, and the instrumental is perfect, while another track is the opposite. I would like to see a function for re-creating a track by vocals (that is, only the vocals will change, the instrumental will be the same) and by instrumentals (the vocals will change), well, or generate a similar track in both aspects.
2. I think this will be difficult to implement, but I would like to see the opportunity to create my own Suno submodels for specific genres/styles (for ex, add phonk tracks to the list and create a model that only does phonk)

[vital bay]

midi generator would be the most sick thing ever since it would give the power of diting the songs and perfect them

[lucid wadi]

1. suno confused Mandarin with Cantonese, please fix this. for example "看" sounds like "kan"(mandarin) not "hong"(cantonese)

2. hope we can control voice/instrument/music Key by tag in Lyrics

[languid beacon]

or if we could export stems from the generated music (separate tracks with bass, guitar, vocal, drums, ...). That would be a killer... (for paying subscribers of course)

I'd love to have an option to upload the recording of me humming, whistling or even singing some melody and having Sumo work with that. Probably not possible with that model, but sharing this wish anyway 😉

[vital bay]

there are many ai's that can pretty well do that. But i didn't find an sinle ai that can make good midi out of a mp3. would be really nice for making music box songs and making sheet music and karoke versions

https://youtube.com/watch?v=pqw3RwRgVXc&si=jepTfouD1fhDWANG

https://youtube.com/watch?v=8urpmkr0_3Y&si=cuYgT00TtoNXEYvW

https://youtube.com/watch?v=nDHA35x6JK0&si=5TmSZCNI2p0HAueL

https://youtube.com/watch?v=ymTiGzZJg4g&si=WsjJlrxKmFYRH3sa

[vital bay]

this ai is really good at seperating tracks but still really bad at creating midi's https://github.com/rakuri255/UltraSinger

[misty trellis]

Hi there. AN AMAZING PRODUCT - I just wanted that of my mind. But, do you have an API that can integrate this AMAZING PRODUCT into 3rd party website?

[low crane]

A button to put random play on playlists🔀

[languid beacon]

A way to generate only one song a time instead of two.

[narrow vector]

I would love to choose when the song ends. Most times the AI just does not know when to stop. So many times there was the perfect ending and then it just ruins it by not stopping. Also, maybe [End] should also mean to stop the song.

[sullen rain]

I would like to see an download option for a SRT file synchronized songtext with timing. So I can upload it into my video editor

Is there a command to make an end to the song ? When I extend with [Ending] it still continues for 2 minutes with a lot of extra unwanted stuff. Which has to be clipped out in my DAW. I like very much the extended time for the songs, but maybe a time limit would be nice to set up for the extension

[signal depot]

if only we could export stems from the generated music (separate tracks with bass, guitar, vocal, drums, ...) please work on it team.

[teal quail]

I would like to upload my own songs that I have started, and have Suno complete or provide variations on

[hybrid crane]

That is coming: https://youtube.com/playlist?list=PLi8taL4TbG_9x9FH7KZvTaQAD65mSq8t2&si=Qw5FkaU7mwD3iWhG

[teal quail]

Can I get beta access to upload feature?

[teal quail]

Can I get beta access to the upload feature?

[hybrid crane]

I have no say in that.. sorry

[tepid trench]

You could try taking part in Suno's contest where 3 winners will get early access to the feature: https://x.com/suno_ai_/status/1798036388329472380

[rugged light]

I don't know if anyone has already given this suggestion, but it would be fantastic to have CONSISTENT VOICES, just as it is already possible to achieve with images. This would be very useful for forming a band, for example, and distributing the music on various platforms.

[desert charm]

Hello everyone,

I am a singer and I would like to use my voice to create tracks on Suno. Is this a feature you are considering adding?

[fair jay]

Did anyone suggested to add note\chords tokens into lyrics already? If so, sorry. If not
Wouldn't it be funny to be able to write

{C} London Bridge is falling down
{G} Falling down {C} falling down
{C} London Bridge is falling down
{G} My fair {C} lady

and get that one song? I bet it would!

[thin ermine]

We'll start to see some of that with v4

[copper siren]

I usually generate my songs and use RVCv2 to replace the voice with my own, it can work pretty well sometimes

[plush lily]

I think it would be very fun if you would be able to change the seed of the song to get the same kind of song every time in case you wanna change a certain lyric.

[desert charm]

I use Kits instead

[dreamy sigil]

Suggestions after about 2 weeks of playing with Suno... almost **5500 ** credits used so far.
You guys got me addicted to making music haha

I got a programmer / customer service background. Im from québec so im half French.

1) Set a Playlist order (not based on name or what ever its based off now)

2) Add a filters [suno version used], Add a on/off params to songs to declaire it is Final, So you can add a filter to the playlist search that would allow selecting unfinished songs or finished songs.

3) Add an option on a song "hide" from profile overview", my old songs with alot of views can still be accessed in a Playlist I called Old songs. I don't want them to show in the overview of my profile.

4) Trim a song, cut the start or the end like youtube lets you do.

5) Discord needs more showcase channels.. by major styles "Pop, Metal, Electro, Rock, Rap, etc" maybe establish 10-20 core styles that can regroup all styles. so new songs are easier to find by style.

6) Possibility to edit the final lyrics (without it affecting the song just edit the final text). After a song was extended the lyrics don't match up anymore. Maybe in the lyrics section of the page have tabs. it shows by default the manually fixed lyrics if you have some set and you can click the 2nd tab to see the suno prompt lyrics that we have now.

7) The suggestion channel should be in the Forum format and you should have a mod who will add frequently asked suggestions (linking to the forum topics about them)

8) Self Boost your song for credits, on the suno home page add a section for self boosted songs. You pay let says... 1000 credits you get placed in the boosted section for 1 hour. You have a line up... first come first serve maybe show a position in que so you know when your turn to shine comes up. Maybe show an estimated time when it will be posted when you come to boost. So you know approximately when to expect it to show up.

9) Add a section on a song page for song notes. Suggested by Mr.Tree

10) Add a clear list of supported statement tags in your help. As a side note to this, the [End] tag doesn't seem to work very well that's listed on the wuno wiki... 

11) Add track layers? Like photoshop. You can add a layer on top of the generated track and record or upload the track for that layer. (To add extra vocal or instruments or what ever) once your layer is ready you cook it togetther.

12) Ability to gift credits/month sub or trade credits.

[modest frigate]

It would be great to be able to let a comment to each song....I could write e.g. "nice verses but bad refrain..." to be able to redo it later..." or "nice intro, but the rest has to be redone",...

[delicate cedar]

I'd like to second that and build on it.

I'd appreciate a way to include a note on my songs. Something that can give me a place to mention the inspiration, give credit if I'm using a public domain lyric. It may not be required legally but I like doing it so I'd appreciate a way to do that.

[modest frigate]

For a server based application small changes and bugs removing in the User Interface (UI) could be done very fast 2-3 days at the most, including testing. Is there any proof yet for any changes done in the UI based on customer suggestions ????? Is here any customer who noticed any changes in the UI in the last weeks? Or do we all talk in vain... 🙂

[modest frigate]

there is already an option for the instrumental version right there in the creator, under the mask where you feed chatgpt

[toxic otter]

I’d like a way to identify vocals on songs already generated so the style/tone/gender could be reused on other songs 😀

[empty rover]

please
please make bpm changes possible by simply mentioning it in the style or 'song description'

• for more specifications, I'm frustrated not being able to allow suno to do bpm changes or bpm overtime. I've tried multiple phrasings and combinations to either trick it or something, and it just never worked... (and no, saying "600 bpm breakcore" did not make it sound faster than any 200 bpm track, that I've ever heard)
  it should be obvious to allow it, unless I sound stupid then forgive my uselessness >_>

also offtopic to this suggestion: it didn't understand creative items like "replace drums with machine gun sounds" or similar equations, but I'm not bragging about that, at least not yet.

[marble bough]

the site looks great but i wish it wouldn’t obscure the album art so much sometimes

just accidentally flagged my own song for review because i couldn’t figure out how you’re supposed to close the options overlay on a song 😵‍💫

if it took up 2/3rds as a panel it’d be more intuitive i think

shrugs

[thin ermine]

lol! Hopefully they give it a positive review? 😉

[marble bough]

i hope sooo ty lol

i’d love to be able to do a simple fade or a trim 🥵

[leaden marlin]

I would love to see an "albums" feature. E.g. if you produce more songs, some may fit a certain theme, others may be very different. It would be great if songs could be organised like classical music albums / folders. Makes it easier to organise.

[tidal isle]

Something that suno end a song if I just write [End] after 5 tries at the latest. And that suno dont sing a verse a second time after 5 tries at latest if I dont have this Verse in the lyrics at the extend-point.

[cosmic shard]

Make a danish chat

[stuck steppe]

Give the possibility to get a "full song" tagged without the auto-mastering

[thorny barn]

I would like to see a MIDI download for the instrument tracks. 🙂 THAT would be a real producer feature 🙂

[hallow dagger]

We all are need ONE thing: Natural Vocals 🙂

that's my suggest. Have a good day all !

[leaden marlin]

Hey, I have made a song and I want to replicate it

Especially the vocals

Can I link it here?

Also should you be very specific when creating?

[copper siren]

The ability to pay for beta features (An extra Tier)

[leaden marlin]

https://suno.com/song/3fd00e86-d7b9-44e8-9eaa-dda766968141

Inspired by SCP-096

(Obviously)

[copper siren]

You guys should add a feature to ‘Use Voice’ from already generated songs

[leaden marlin]

No way! Thanks!

[dusty river]

i was wundering.
( @wanton compass )

would it be achievable or financielly viable to add a generate demo option for suno create screen? due to my issues i have to gen 1 song anyware between 5-10 times due to sentance structuring, grammer, spelling, lyrics not sounding good with music type etc etc etc,)

i realise these are "a me problem" however. a little assistance would be hugely appreceated. ...the demo would be non publishable, free, and maybe only generates 20-30 secands. please get back to me with wether this would be something you would be willing to look into. as i feel it would benafit a lot of peaple with issues like mine. am willing to descuss issues further should that help.

[leaden marlin]

Idk if anyone even notices these suggestions but it would be a total game changer if we can edit lyrics while keeping vocals and sound. I know we can put custom lyrics but there have been instances where I leave everything to the AI and I love it but want to make some small adjustments to the lyrics

[atomic bloom]

Ok... So.. Suno knows Vocoders. But... Why not Talk Box. I have tried everything. Can't get Talk Box effects. Any DEV. suggestions?

@leaden marlin you can try to extend a track with same vocals then change the lyrics then use an audio editor to snip what you want. So far I think that's the only way

[leaden marlin]

Tried it and managed to turn a song into two but doesn’t really work for changing an existing one. It changes everything too much

[atomic bloom]

True. I have run into the same thing lol

[robust frost]

I would like to create a voice that i can make sing again and again in different styles. it should also be possible to define recognizable elements of the voice. just like i can define a stylereference (--sref) in midjourney

[woeful hound]

Is it possible to use the same seed for the song but with other lyrics? sometimes I just got a banger rythm/music and I want to replace some lyrics to better fit the overall flow, but the only thing that I can do I am aware of is to reroll, as "extend" still puts "old" portions of the song into the new one.

[dreamy sigil]

extend the song but start the extending before the lyrics you wana change you can already do that somewhat

[woeful hound]

you mean to extend from 00:01 for example?

[dreamy sigil]

yeah more or less

[woeful hound]

this does not work, it replaces the "original" vocals and overall flow of the song...

I want to preserve the vocals, tempo and flow of the "original" but just replace the words that are in lyrics

[dreamy sigil]

you gotta try a few times you wont get the exact same thing every time

read in here people complain alot about the voices not being consistent

you get the same voice but the pronoucing and speed can vari from what i saw

[woeful hound]

Thats why I am asking a question here as usually for any random stuff you get a "seed" that you can re-use, I would expect to have the same thing here, as lyrics are the only variable that could change by purpose.

it would be a nice feature to get as sometimes I made a mistake or just a bad wording but the bit and flow was awesome 😦

[dreamy sigil]

yeah its been suggested alot alrady just telling you the work around for now

best you got atm

[marble bough]

playlists should show how long they are

[woeful hound]

so best would be to experiment with the cut off of a existing song just before the lyrycis?

or just after the first Verse?

[dreamy sigil]

yeah i found using a diffrent starting point makes a diff

[rare grove]

It would be good that the use of forbidden names (e.g. artists) is better checked, especially for artists who have the same name like a normal thing. E.g. I run into that "Richter" is forbidden as there seems to be an artist with this name. But "Richter" in German means "judge" so I can't use this word for German songs. The AI should be more accurate because it is rather clear that I don't violate a personal right of an artist in these cases.

[marble bough]

and not lock you out of skipping to the next track in the playlist because you mistakenly entered the wrong mode due to clicking the song to play and not the playlist’s play button despite viewing the playlist

at least i think that’s what’s been happening

i’d like the option to call a playlist an ep or album or whatever too (✿´ ꒳ ` )

i love the core thing and find it awfully addictive to make new songs but the site kind of sorely needs more social stuff, but i’m sure that’s planned

as it stands i think it’d be pretty lonely using the site with no discord

i love it overall tho 10/10

[marble bough]

imagining a dropdown when you edit a playlist lol. lp, ep, album, release, single, mixtape…

but like overall i really like the ui, and im picky about ui because i find it interesting

i think it’s a nice looking site as well

maybe kind of better looking than spotify in ways

i’d love being able to customize things on releases like bg gradient and button colors

if there was social shit like profile customization and polls and stuff i think i’d find it even more addicting 🥵

even if it’s just colors

if any of this stuff happens i’ll do pointing leonard dicaprio

[leaden marlin]

Yeah, but that is way too much because I think the AI would likely have to recreate the entire song if you don’t want those removed parts to sound like crap

[leaden marlin]

I hope I’m not being mean, but are you learning English or is it your second language?

[dusty river]

niether im 39 and english. i suffur from several learning dysabilitys.

[leaden marlin]

Gotcha, sorry

[dusty river]

no need to for sorry. you didnt do anything rong

[leaden marlin]

It’s weird, like I know what it means even though it isn’t exactly what I’m used to

I mean it’s interesting

Can anyone give me a tip on how to stop spending so many tokens so fast?

[dusty river]

due to how i had to teach myself. i spell things how they sound wich leads to it atleast being "mostly" readable

[leaden marlin]

I have to wait 20 more days until I get my next shipment

[leaden marlin]

I’d say it’s all readable, and I think yours is actually how they should be spelt, like, who made “wondering”? It should be “wundering”

Where do I put the prompt ‘use voice’ for my songs? And does it work on creating new songs?

[feral aspen]

It's a good suggestion but please don't ping Suno team members. They're checking the channel.

On to your suggestion, I would recommend using ChatGPT totally separately, on the official website, and explaining your problems and your request to check and fix for as many as it can predict. This should catch grammar and spelling issues. I would literally tell ChatGPT about any disabilities, it helps a lot. When I was trying to get ChatGPT to make less obvious mistakes in chess, I told it I was dyslexic and have terrible eyesight, and that made a huge difference in it's ability to find blunders, very interesting! Also reminds me of people telling ChatGPT they have no fingers to get it to not skip over computer code in the outputs.

I can't find the text prompt but I did find an old screen, there is an example at the bottom.

[dusty river]

think we should take this to #💬┃general-chat

[leaden marlin]

Thanks

[dusty river]

my oppologise for the ping. i just thought as it was due to disabilites it may have been worth the ping. i meant no disrespect or disturbance. and as for the gpt. i thank you ever so much for your responce and the screenshot helps too. im sorry again for the diturbance.

[dusty river]

semi connected side note suggestion. if you dont want peaple being able to mention some roles. you can change that in the discord server settings if you go to roles and then you can go to i think"guest rol" and turn it off there. or it may be on the team role i cant remember exactly witch role is edited. but it is easaly done.

[copper siren]

The ability to do duets

They happen automatically sometimes and its amazing

[hushed pulsar]

Please make it easy to save my vocals separate from the instrumentals or let us create Vocals over a instrumental track we already made.

[marble bough]

i feel like one of you should tell investors this at some point lol @devs

[hushed pulsar]

Pleas fix mastering of the tracks, they need worked on over half of my track are distorted at the peaks of the song and some are the whole way through.

[timber drum]

I would like to get higher audio quality in general. i noticed it is not really HQ, but i compared it today against some .mp3s i bought a couple years ago and the different is very noticable. I would even say that no serious music artist should use the .mp3s Suno outputs as they are. Music is more like my secondary hobby, so i'm not a real artist and despite that i noticed this.

Should have at least a bitrate of 250kbps. Music i bought is at 320kbps max. .mp3s from Suno are at 192kbps which is very noticable.

i just compared the .wav and .mp3 from Suno. Apart from a very slight volume difference (.mp3 sounded like it's slightly louder), i didn't notice any difference in quality

[leaden marlin]

I wish Amazon Pay, it´s powered by stripe!

[pearl ice]

Buy White Robes: I was just sitting here basking in this beautiful, instantly created song that v3.5 made for me. Soaking up the moment of actually hearing something that only a second ago, was just a thought. Now a beautiful symphony I can play over and over on the way to work tomorrow. The freedom and joy it gives my life. What a blessed time to be alive. What such amazing people you are developing something like this. So my suggestion is for yall to buy white robes, because if heaven does exists, I bet you get to sit in the nice seats! 🙂

[bronze granite]

Let paid users download a wav version of created songs without main voices (a sort of karaoke version) to re-record a song with our real voice.

And give us paid users the possibility to shorten a song if too long, without realaborating the style

[fair jay]

Dream big. Song is generated with lyrics. Use something like https://huggingface.co/audo/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/README.en.md to separate them. Works like a charm.

[verbal remnant]

How are we not getting credited for generations that are 12-15 seconds when the lyrics/ prompt have many sentences of text?

[bronze granite]

Another useful thing would be the possibilty to listen to lyrics pronunciation BEFORE creating the song. So we could correct some words to sound better. This would be very helpful for foreign languages. In italian I must often delete a good song because some words are pronounced very bad

[sage mirage]

With v4 we will get inpainting so it would work, instead, as you have your song but you would 'paint' over the mispronunciation and could regenerate that section

Essentially rerolling just those sections where the pronunciation isn't what you want

[verbal remnant]

Until recently, misgenerations like this were always credited... I've probably spent 2000 credits tonight just trying to get something to properly extend and generate.

[modest frigate]

In the last 2 days there are a lot of distrorsions and clicks! I barely had the 2 days before...

[marble bough]

i’d love if paid users could have the 4 minute limit raised to like 6 or 8 minutes. certain genres get frequently cut off at 4:00 and extending it doesn’t always mesh with the captured lightning in a bottle

[fair jay]

It's model's limit. It's literally impossible.

[marble bough]

aah i see

ty!

[feral aspen]

It's totally cool to ask for features like 8 minutes though, that's what this thread is for. Even though 8 minutes isn't possible today, if Suno knows people want it, they can take that into account when deciding what improvements to make in the future. So all good to ask.

[frigid pivot]

PLEASE give us the ability to cut off a song. I've lost some insanely good songs to trash endings

[slate belfry]

You can just crop the video

[marble bough]

what if there was a creative commons license selector built in for songs

[fair jay]

And it will most likely be done in v4. Still, there will always be limit. So, cropping song and im- and outpainting it is the best solution. As well as smart cropping, to mark best places to chop it.

[fair jay]

But money!

Suno is made to make money. And it's not a bad thing. You want CC, you buy premium and release your songs under CC.

[marble bough]

fair

i just like cc _(⑅；ω； 」∠)_

[fair jay]

CC is mostly used by big companies for advertisement. I doubt this is what Suno needs now.

Say, Google would never make Chrome free if they wouldn't profit from it by ads and personal data collection.

[marble bough]

download playlist as wav would really help me lately 😵‍💫

maybe even zipped???? 🤔

i feel like the awards should be separated by genre and maybe have some voting and editorializing like irl music awards so they don’t become the $700,000 electro swing suno fund (nothing against the genre as a vocaloid fan ofc!!!!!!)

could make a fun youtube or twitch or podcast thing too for promotion and awareness purposes too, it’s basically free advertising

like there are plenty of relatively underrepresented genres that aren’t really surfaced on the current summaries

it could stand to be more like twitter or deviant art etc where it’s fun to just scroll essentially

i get recommended some lofi but it’s just the top 3 most popular things

[tawdry nebula]

there's gotta be a way to get some negatives working in the prompts. Thought saying "No Pop" would work, but alas.

[sage mirage]

If we get negative prompting it will be a total separate section. If you put 'no pop' into Suno now it will go "Ah you want 'pop'!"

Since you mentioned it at all

[tawdry nebula]

that would be smart

thought i'd give it a shot lol

[sage mirage]

It's been talked about months ago

[tawdry nebula]

word. i'm not the biggest discord user so i wouldn't have seen it

[sage mirage]

Yeah no worries

[feral aspen]

For the general problem of "3.5 is too mainstream or generic" some things to try here though: #💬┃general-chat message The short list of easy things is:

1. SPAM phonk -> phonk, phonk, phonk, phonk, phonk, phonk
2. Change spaces to dashes -> acid house -> acid-house
3. Avoid big genre words like POP, ROCK, and METAL. If you can't replace, make sure always a dash like nu-metal instead or be creative tape-metal whatever.

[tawdry nebula]

i'll try that lad thanks

[sharp meadow]

Big suggestion here, Custom song lengths notated by the user before generation.

I've noticed the AI has trouble when it comes to stopping a song over the past few versions, even after user input of end notation through "(end)" or anything similar within the lyrics. Most of the time, the AI will generate entirely to 4:00 minutes and then abruptly end in V3.5, but the user might want to instead keep the song shorter at about 2:45 while not needing to switch down to v3 to do so, and proceed to lose out on the benefits of the new model. Even after a second generation to end the song, the AI will still insist on generating at least a 1:59 minute long generation before finally listening to user input and ending the song, usually offering mostly 6 minute tracks for one that should have been shorter and ended sooner.

[merry osprey]

I appreciate the ability to use the service for free, and understand the limited amount of credits I have each day. Server time is expensive after all. But I would really like to have a way to spend the credits well. When I create a new song, suno will run for 30 seconds and generate two 4 minute tunes for 10 credits = 20% of daily credits. I'd rather have an option to make it create eight 30-second bits for 5 credits, as I can usually tell from the first line of lyrics whether the style and voice are what I want or not. This would be cool to have more leeway to experiment with the prompts and then expand on what works rather than spending your time and money on generating additional 3:30 on a song that will move to trash immediately.

[marble bough]

fade in + fade out points and lengths, and start/end point selection pwease 🥺

that's the only thing i'm needing to touch audio editing to do

[rustic coyote]

Changing lyrics - once you stich your songs together, lyrics get messed up and sharing them in suno or by video is messy

In general it would be cool if Suno could separate lyrics from the instrumental part. Sometimes I get an awesome background but a typo in the lyrics messes up the entire result. The only solution I´ve got is expanding from the very beginning, hoping the result will be close to what I had before

[forest mauve]

Hi. This is for the Suno team.

Have you considered creating an online equivalent to Billboard Magazine? Where you interview your artists/creators on their creative process, how they market themselves, how they think, etc?

AI music is only popular because of the people making it, and I think it will remain that way. No one watches AI play chess even though it is the unchallenged grandmaster for as long as I can remember.

A top 100 list would also be great. Everything the regular music industry is doing will work in the AI music industry

[marble bough]

trackmania has an in-game radio feature where you can queue and vote on songs etc. i think it uses an m3u stream winamp style. food for thought

[raw mist]

Please add ability to: download song part stems (with vocal separated from music at minimum) and MIDI song parts (chords at minimum). Thank you!

[hallow frost]

If you do not make the popular genre's on Suno you just get left in the dust - you guys should do more to bump people making stuff outside of the mainstream, not only to show that suno is more than k-pop and poop memes, but also to encourage people to explore more and showcase strange and eccentric music that is possible with suno.

[modest frigate]

is there any meta tag to tell the AI NOT to use an instrument ? Like "no drums" ?

[upbeat rock]

not yet

[opal spoke]

Premium have exclusive song layout with shine lyrics or smth

[raw mist]

Please add ability to create vocals that are not distorted. Its hit or miss right now. Thank you.

[floral basalt]

Hello

Uh im new here

[hollow locust]

I'm sorry, but I did not type "Lights are bry" nor "every ny". Let me guess, It'd pronounce a word which means 'to consume food' "Ee". 🙄 Please fix this. Thank you.

It's doubling every verse and chorus.
It's doubling every verse and chorus.

[stiff orchid]

When adding an extension, we need an option to add before the track. Often a track starts without an intro even if intro is put in in brackets. If I like the track and want an intro before it right now that is not possible. Please create this option. I have seen this option in another platform online. I think it would be an important add on

[hollow locust]

The problem seemed to be in "Theatrical" and "Opening number"

[lethal saffron]

Consistent, EXACT song flow would be neat for lyrical. I'll type verse 1, chorus, v 2, chorus, v 3, chorus, but it'll jump around and take chunks of each in no order.

[twilit hedge]

Can we keep a playlist on repeat? curently, only a song can be on repeat... but I mostly listen to my favorite playlist and I don't want it to stop 🙂

Please add that feature.

[cold wigeon]

I was wondering if we could use extend feature to work around this? I haven't tried it myself but this seems to be a possibility?

[frigid pivot]

With v3.5, it creates songs up to 4 minutes, but most of the time, for me at least, it will end the song at like 3:12, go silent, and then bam we're back ladies and gentleman, singing another verse, wild encore, getting grea- and cut off

[cold wigeon]

How can I create a soundtrack without a voice from any particular composition?
Ideally, I would like to remove the voice completely or make it sound much softer
This is important to make a karaoke version.

How can I regenerate a voice, while keeping the song the same but just regenerating the voice to make it sound proper?
Most of the audio issues are related to voice distortion but music sound ok.
Is there such feature request already - to regenerate voice only and keep the music the same?

Some images generated by Suno are not relevant for the music or very disturbing and ugly.
I can replace the image and upload my own or click to regenerate it with prompt.
However, the video keeps the original image.
How can I regenerate the video with new image?
Is there such an option today or is there a feature request for it?

[twilit hedge]

I do this all the time!

[cold wigeon]

Very often music and voice sound great but at the end of the song I hear strange voices, laughs, coughs etc.
How can I cut the music at specific time code and remove the rest?
Can I use "extend" feature for this purpose?
Is there a feature request for this to be able to cut song at the end?

I noticed an issue with video.
When song goes into repeating part, video stops scrolling the lyrics and waits for the repetition and then shows the rest of the lyrics. This is not good as people can not read the lyrics.
This looks like a bug. Is there a fix for it or feature request?

[twilit hedge]

been a most requested feature since day 1 🙂

[spiral acorn]

Would it be possible to have an option where when you extend from a "Whole Song", it always keeps the original audio intact. So whenever it applies the mastering on a "Whole Song" it's only using the original audio of all the parts so they retain their full quality.

In the past I have tracks where the start of the song becomes muddy because of extensions at the end of the song. Maybe this has been improved more recently?

Like is it not possible to have all the tracks be stitched together, and only master those for each Whole Song? So if you extend from a Whole Song it will use the original audio underneath, without any mastering applied, and then when you make the next Whole Song it does the mastering again (while also keeping the original audio intact). So that way you are never losing the quality of the audio.

Or is there a sound levelling issue that makes this less ideal? There are definitely times where I'd prefer to have the song degrade gradually rather than losing quality throughout.

Edit: Or is this already being done, and the later extensions are somehow causing the mastering to use all of the wrong levels and therefore it messes up the start of the track.

[thin ermine]

It's probably not a simple mastering issue - more of a rendering thing. I'm sure they hate it as much as we do.

[spiral acorn]

yea, i do get that. I'm more just saying, since it's possible for me to download the individual parts in their original quality, stitch those together, and apply mastering, why can't that be automated on the server side? I haven't tested that yet, so maybe there will be a "gotcha" i'm not aware of

I know there would be a strong reason for why it is the way it is currently!

Totally seperate to the above thing:

• Would it be possible to have a "sort by date" when searching the Library?

It uses date order when you're using the library generally, however when searching I'm guessing it's sorting by relevance. That's useful a lot of the time, though sorting by date is useful is you're trying to find the "oldest" version of a song for example

[sage mirage]

Would be a nice feature, yeah

I expected that's how it would but then it just mixed all sorts of stuff including v2 gens

[spiral acorn]

yea there might be some logic to it, because in certain cases the relevance might be more important, but it depends what you're doing

or unless its not sorting by relevance and its just random order 😆
(i havent bothered to look into this much)

[haughty halo]

New to suno. Interested to know if there is any plan for integrating sheet music, or direct play into the ai. Similar to how you can feed Midjourney an image, can I sing a tune and let the ai run with it?

[spiral acorn]

the sing a tune part is coming sometime soon!!

you'll be able to upload any clips, compositions you have made... random sounds, singing etc... and it will generate the song out from this

[haughty halo]

Especially helpful for an in painting situation. I'm sitting here thinking, this is great but I wish it would go up here instead of down. "La La La" boom 💥 magic

[sage mirage]

As mentioned yeah an upload feature is in the works.
This one has the first 17 seconds I uploaded, Suno generated the rest

You prompt it like normal past the initial clip

@hybrid crane has done her own voice to make that the vocalist

[spiral acorn]

ooh awesome, didnt realise you were a guide and already get to test it!! it sounds magic 🔥

[sage mirage]

It's phenomenal, imho

But then again I always like to mess with things like this and uploading allows you to get real weird

[spiral acorn]

i can't wait!! im curious to upload some of my own non-AI music and see where it goes if i just upload the intro

[sage mirage]

There was someone here last night that said they had hundreds of intros but always ran out of steam to complete them

So this, specifically, is directly targeted at that

You can take other Suno clips too from other songs and things why not

@strong swan did the Windows 95 boot up chime

[spiral acorn]

oh yea, i can see that being useful too since you could take a part that normally appears further into the song

[sage mirage]

Yep

[wooden igloo]

I think we need an "ultra quality" switch as well. It's ok for me for this to burn more credits.
As long as the vocal + instrument queality is not suffering...

[bronze oriole]

Library Folders

[fair turret]

PLease turn off autoplay on an individual song page when trying to evaluate multiple clips for possible use it interupts the process. Thanks

Not sure if they are getting plays for that but if so that would make views a little inaccurate

[modest frigate]

The 1256 suggestion for the User Interface (UI) which will never be implemented probably...: make user defined templates for the lyrics window... so we dont have to to add manually intros, solos, interlude metatags to all songs... usually is the same job everytime... we have to do it manually in the time of computer and AI! Think about this "disacord": the SUNO AI makes better music than humans but it's UI is from Microsoft DOS time!
When do youfinally hire a good UI programmer???? 🙂

[stone steppe]

obviously not a big issue, but would be convenient when I click "extend" it selects the timestamp I am currently paused on rather than me typing it in manually. just a simple feature for the lazy people out there (aka me)

[cursive breach]

Please make the AI not repeat lyrics unless told, and end the song when the lyrics are over unless told otherwise

[winged sable]

This is one of the most useful requests :)

[cursive breach]

I've wasted countess credits with [end] and [fade to end] but it just doesn't care. Today I have 0 credits left 😦

[winged sable]

[Finish] or [Big Finish] usually works for me

howewer a simple [end] with imemdiately shutdown even teh power grid effect would be really good

[winged sable]

By the way, the [break] works too, with a little hack. Put the [break] at the intended end point. This usulaly put a short silence and other stuff,. The main point is the short silence. Download it and simply cut the song with any DAW.

[cursive breach]

Next month I'll try this for sure. I've tried [break] but only mid song. Also sorry but probably shouldn't be talking on #💡┃suggestions-web 😛

[leaden saddle]

If we can get specific genres channels in discord

[primal loom]

The songs are amazing and humanly realistic. However, the most important thing for any service is to be useful, and the lack of stems in .midi and/or HQ .wav makes this impossible... 😦
I know a lot of people have talked about this, but this feature is really essential.

[hushed pulsar]

Distortion is a huge problem!!!

[marble bough]

trim start end end time!

[wintry gazelle]

If you make a song with good vocals that sound good, we should be able to extract that sound or voice and put in a new song

If someone does that it changes everything

Fake musical artist on the come up

[stiff dome]

I guess this has been requested, but could not find anything if is something that is coming, in the future (maybe v4) would be possible to from a song that we generated get the song without main vocals? like a karaoke song? I have tried several web-based tools to remove vocals, but is far from perfect, it lowers the quality and introduces some noise

[signal sleet]

Cut and paste options would be nice 3.5 repeats alot to fill out the 4mins of play but sometimes it repeat a verse or chours that sounds/works better would be good if this could be a thing and also end stop and finish should work where you put them in the song eg. [END] [FIN]

[cold plume]

a drop down list for tags to use in custom lyrics would be a sick idea for those new or inexperienced to musical terms.

For example:
If i have a song that reads

[Verse 1]
[Insert musical cue here]

in this section, if i wanted to add a musical cue of some kind, perhaps a drop down menu where it shows a vast selection of tags that suno recognizes to where, similar to the style of music section, can be as easy as a click to add or copy for a simple Ctrl + V command wherever the user would like to place it. So if i wanted a sick Kazoo solo, i'd know if it's possible with this dropdown menu.

It can even be something where it's structured like

"Are you looking for an instrument or a musical term" (I select instrument) > Which instrument would you like to use? *(Clicks kazoo) *> What would you like to have the (insert instrument here) do? (Clicks sick kazoo solo)

[lunar spire]

lets get some sponsorships going on for small and large content creators so we can tell are community about suno! give cods for a one time price drop, more for a year sub

[wintry gazelle]

Send credits to someone you want to support from your credits, like a gift

[glossy ginkgo]

Button to give creators a 5 credit tip

[sinful grove]

@dire frost Per your request here

Other apps on my mobile device and laptop often have the icon that I have attached. This icon typically enables a person to stream the services from the site/app to a list of compatible devices. The devices listed are typically ones that exist already on the network of the home wifi or LAN such as televisions, computers, mobile devices, etc.

It would be great to have such a feature on SUNO where the player that plays the music has this button as a function and it will petition to find other compatible audio devices on the network that the host device is connected to and stream the sound to that device.

The end result would be something that a user might see as pressing this button, selecting a TV or smart assistant, and now the sound is coming out of there rather than the speakers of the host device.

[wintry gazelle]

Or click on profile - donate and then u type how much u want to give

[pearl mason]

I have a question, if I buy premium from sonu ai, can I still sell my music that wasn't from premium? Does it become part of premium?

[glossy ginkgo]

Idea: song from a URL where it would read a webpage and use its contents as a song prompt

[glossy ginkgo]

No, but email support and it may be possible to convert

[cold plume]

My last suggestion is a longer character limit for the style of music section.

Often times, I'd like to add specifics like instruments and vocal styles but run out of a character limit soooooo fast.

[rugged light]

Exactly. I had suggested the use of consistent voices just as is already done with images. It would be fantastic to put together a band like that 😅

[wintry gazelle]

Yeah i have a song with amazing female vocals and it fits jazz perfectly I want to use it all the time, I'm sure they will add it

[glossy ginkgo]

Idea: chaos slider for how creative the song is. Low chaos adheres to prompt better, high chaos is more randomness

[pearl mason]

I have a idea, sonu should let people buy the songs they make to copyright it and sell it if they don't have premium, this could be good both for the sonu ai business and the creator of music

[glossy ginkgo]

I like this, non premium songs get converted to premium/user owned for a five dollar 1 time fee or something via easy checkout process

Rights issue solved.

[pearl mason]

Yeah it should be like that, it could benefit sonu and the creators

[rugged light]

It depends, if you create more than 10 songs per month, you'll see that it's more advantageous to pay for the premium, unless you only create 1 to 3 songs per month.

[glossy ginkgo]

Idea: regenerate song videos for $1, uses user uploaded thumbs and the current song title

[dusky plinth]

A trim feature would be nice to cut out parts at the end that you don't want. Since it often adds random stuff at the end or even starts a second song.

[pearl mason]

What I meant is convert non premium songs to premium to sell it

I have lots of good songs I wanna sell

[rugged light]

Oh yes, I support it as well, I also have good songs, but in the V2 version. Maybe if I master them they'll turn out well, I don't know.

[sinful grove]

ADDITIONAL SUGGESTONS

1. In addition to "reuse prompt" it would be great to have other options for reuse specific parts that you want such as "Reuse Vocals Only" or "Reuse Instruments Only" and then rename "reuse prompt" as "reuse lyrics and tags only" or something like this. This way if I love the vocals but not the instruments I can reroll only what I want to change, etc.

2. The ability to select a Time Signature and/or Key along with the prompt but, if left blank, the AI will operate as it currently does. Modes would also be pretty cool

3. The ability to pay more credits to generate more layers to the same song. (i.e. maybe I think one segment would benefit from something more - it would be cool to be able to select that section and then pay credits to generate another "layer" of sounds on top of that which work with the original generation. )

4. An ability to toggle some basic features of the track such as volume, pan, envelope, etc.

[pearl mason]

When people buy premium their old non premium songs should immediately turn premium

[glossy ginkgo]

No, then they would skip out on premium for 2 years then convert it all lol

Then cancel after 2 hours

[pearl mason]

Yeah or maybe people should choose specific songs they want to become premium

[sinful grove]

If anything it makes more sense to have more of the advanced features behind the paywall. I.e. trimming, loading samples, any of my other suggestions - those all make more sense to be a luxury feature you would pay for.

[pearl mason]

Look at this

I wanna make this popular

[lunar spire]

any way we can get voice seeds in an update?

[pearl mason]

Yeah whatever can be done to let sonu creators sell their ai songs would be good

Sonu ai has potential too, sometimes I wish it sold stocks, to become even better, I love this company. But I understand if they wanna remain private currently

[cold plume]

Suggestion:

Export files as a DAW file so aspiring musical engineers can learn from any songs they make and like by bringing it into their DAW of choice and disecting it. 😄

[glossy rose]

I appreciate the existing feature for creating instrumentals. I would like to suggest adding an option to generate both the vocal and instrumental versions of a song simultaneously. This would streamline the process for artists and enhance the platform's functionality.

[marble bough]

radio station generation like pandora............................

[wanton pivot]

A comments section for songs where users can give feedback and or discuss to each other about the song?

[wanton pivot]

As cool as that sounds unfortunately it could very easily be abused

[wintry gazelle]

How

[wanton pivot]

People could easily multi account and just spam give themselves points

Ooh they could make it so only premium accounts can gift credits that would work

[marble bough]

download playlist as zip

also: gift points to ppl in exchange for wavs, zips of full albums, etc etc…

id be ok with the only giftable points being prepaid ones

[wanton pivot]

A middle kinda plan
Maybe 5k credits $14-$15/month?

[lean zodiac]

can you add a comments section behind musics ? or a PM system ? i want to dialogue with some A[I]rtists :/ (when they don't have a SN link)

[warped sierra]

Alguém sabe como faço música Brasil funk

[marble bough]

i wish the function to autoplay related tracks after one finishes and seamlessly play an entire specific playlist were separated

afaik they’re the same toggle..?

[wanton pivot]

Is it better with pro?

[tepid trench]

There is no quality difference between free/subscribed. The available functionality for generating music is the same.

[marble bough]

plz implement a show romaji button for my in car umi singalong seshes

and for ppl who can’t read japanese

[robust jewel]

The only times where that isn't true is when an alpha test is happening (v3a, v3.5a). Pro has been given access to that earlier.

[tepid trench]

Oh yes, there is that.

[fringe reef]

It would be great to be able to have the vocal and music tracks separated in case we like the music but think the singing that the AI generate isn't good and want to cover it ourselves

[dreamy sigil]

Suggestions after about 2 weeks of playing with Suno... almost **5500 ** credits used so far.
You guys got me addicted to making music haha

I got a programmer / customer service background. Im from québec so im half French.

Updating my post with new suggetions...

1) Set playlist order not based on current criteria.

2) Add filters for Suno version used. Include an on/off parameter for songs to declare if they are finalized, allowing filtering of finished or unfinished songs in playlist searches.

3) Include an option to "hide" songs from profile overview. Older songs with high views can still be accessed in a playlist named "Old Songs" but won't appear in the profile overview.

4) Enable song trimming, allowing users to cut the start or end like YouTube's feature.
   4.1) Introduce the ability to adjust voice and music volume using key points on two tracks, enabling fine-tuning of voice and music balance.

5) Expand showcase channels on Discord, categorized by major music styles such as Pop, Metal, Electro, Rock, Rap, etc., to facilitate easier discovery of new songs by genre.

6) Allow editing of final lyrics without affecting the song's structure. Implement tabs in the lyrics section to display manually adjusted lyrics separately from the original Suno prompt lyrics.

7) Transform the suggestion channel into a forum format and assign a moderator to compile frequently asked suggestions, linking to relevant forum topics.

8) Introduce the option to self-boost songs for credits. Include a section on the Suno homepage for boosted songs, where users can pay credits to have their song highlighted for a set period, with a queue system and estimated posting time.

9) Add a section for song notes on the song page, as suggested by Mr. Tree.

10) Provide a clear list of supported statement tags in the help section. Investigate issues with the [End] tag listed on the Suno wiki.

11) Implement track layers similar to Photoshop, allowing users to add additional vocal or instrumental layers on top of the generated track.

12) Enable gifting of credits or monthly subscriptions, as well as credit trading between users.

13) Ensure Electro is included as a main category on the Suno homepage.

14) Allow for switching between male and female voices, possibly using designated tags such as [Voice Signer1], [Voice Signer2], [Male Voice], and [Female Voice], with support for multiple signers. Investigate and address any inconsistencies in voice switching functionality.

[marble bough]

the video embeds could like ken burns across the art and spike the brightness in time with the music or something, in addition to the timed lyrics (which are amazing)

[somber dove]

Idea: if you are a person who doesn’t understand a language’s script that ChatGPT gave you (I.e. Arabic script, Burmese script) you should have a pronounciation guide window where you can paste in the real pronounciation and how it should be pronounced

[forest mauve]

Guys! We need something like a "New Music Friday" list, where people can submit songs and someone with a backgroud in the music industrys full time job is to curate it. I am sure you can afford hiering someone to do that, and also we need an online blog where you interview the top creators etc. I mean... this community is going to become incredibly large. Play your cards right here, and you will be the next spotify/tidal/apple music AND a new social medium that never existed before all at the same time. You don't only need to have the best AI. This community you are building here, and also the possibiity of songs as a social medium are also very important focus areas! Especially now that competitors are entering the market.

Thanks for reading my idea rant😘

[stuck vessel]

Could be cool to have suno/chirp on hugging face. I know the old model is on there but it’s quite outdated considering the advancements this year alone

[dense swift]

I'd like a way to ask suno to describe an artist. Kinda like /describe that Midjourney has. But in other words a way to categorize a sound clip or artist into styles that could be used to make something similar without infringing on any intellectual properties. But rather describing the intellectual properties in a way that should be legal in theory.

[marble bough]

i would love this with romaji for my songs for karaoke purposes

[verbal remnant]

After failing to do what I used to do easily on your platform for a few days ...

Suggestion: Make a much higher priced, premium version that gives way more compute horsepower for $$$ and follows better simple English commands.

I'd pay.

[marble bough]

ssc would be a great suno m3u station theme 💦

also per-artist or per-group radio that plays stuff on shuffle......

[reef grotto]

Negative keywords, please, add them for the 4.0

piss off, scammer

I'm sick of tracks that Suno generates, when I've used 'Deep male voice', but I get a female high pitch voice instead.

so negative keywords, please

[raw mist]

Please add ability for us to specify: key, chords and BPM.

[lunar thunder]

Please fix the video generated. The lyrics are a total mess. If you extend a song and then combine it some of the verses will not have lyrics at all and it looks very very ugly and over busy. The title and the prompt have no business to be in the video. Also the [prompts] do not belong there. And it looks weird and kinda spoilers the song that you see 5 lines at once.

Best would be to be able to set your own lyric/video settings with an optional picture, text position and size, line count etc. etc.

[drowsy dove]

I am once again asking for a poll bot for discord that (if abuse is the concern, mods can use it only) can be used in the event of community events.

[indigo cypress]

I must say, everything you do is amazing, huge 🙌 . But: v3.5 is still a 8-bit-ish sound quality.
Plus I came to the conclusion that the fact that this new model is now able to generate an entire song, does not benefit the quality, because as long as we were building from sections (v3), we always had the opportunity to choose the better version of that section (or regenerate) and build the next section on it, and choose again the better version from that. Masterpieces were born! Now, however, there is no choice, so there is no room and tools for refinement. The song is ready immediately and is what it is. Generally worse than what we were able to create using v3.

[thin ermine]

I don't disagree with any of that, but you CAN manually do 3.5 in sections... Just don't paste the entire song in at once - do it in small bits. Pick what you like, merge it back in and start on the next extension. I started doing that with a piece last night and am quite happy with the results. It ate up a crapton more credits, but that extra level of control is great.

[forest mauve]

3.5 is much better for me. I agree that Suno must become better at mixing, or removing the 8bit'ness to the sound, but having the opportunity to do full songs have been nothing short of miracolous for song structure. Much easier to write a pop smash now for example. If you don't want a whole song at once just write like this:

"[Instrumental/intro]

[Verse]
These are lyrics,
Sing them out,
Suno, Suno make me proud,"

*generate

*click extend

*write more lyrics

"Suno, suno
mimic sound,
Sing my lyrics loud!"

*click get full song

*generate

*click extend

*write more lyrics

"[Prechorus]

You give my heart a voice
Both harmonies and noise
Fuck Udio you are the only choice"

*click get full song

*generate

*click extend

*write more lyrics

"[Chorus]
Ai, Ai, Ai music
Make me famous
Get me groupies!

Ai, Ai, Ai music
Do it, do it, now!

Ai, Ai, Ai, Ai Ooooooh!
Give me groupies, go! (Go!) (Go!)"

*etc, etc, etc...

[bold holly]

I would personnally like a fader option where we can choose a time in the song like say the last 5 seconds and have it fade out. This would alleviate the issue of the song springing up a new line without enough time to play it through. Right now all you can do is extend the song which may have undesired effects on the quality of the song. If I can fade it, I can keep the song I like without having to mess with it further.

[sinful grove]

IMPORTANT SUGGESTION

I want my playlists to have two more buttons. I need a "shuffle" button to shuffle my own playlists. I also need a "repeat" button that repeats the whole list and not just one song.

[indigo cypress]

Please, seeding! 🙏 To regenerate the song if something should be refined in the lyrics but the song itself is good enough and we don't want to get something else after the refinement.

[vital hollow]

Can we please get the bot trained on Jeremy Soule😭😭😭

[long hawk]

Quality should be Job 1....Sound Quality

[karmic steppe]

SUGGESTION

To improve the usability of Suno AI for sound engineers like me, please allow the export of stems for every finished track, including separate files for the instrumentals and vocals. This feature will greatly assist in the mixing and mastering process.

[modest frigate]

another suggestion that would probably be never implemented due to laziness of the UI guy: the different columns in the application could be scalable in the width. For a wide monitor it is very annoying to have the song logo and name on the left side and anything else on the right side.

[tepid trench]

Calling the UI developer(s) lazy is not constructive. Similar to what I said in the bugs channel yesterday, this kind of thing is likely low priority compared to other work.

I have a very wide monitor. I find it impractical to use most websites maximised so I just have it resized to about a third of the screen and placed in the centre (Windows 11 makes it easy to "maximize" windows in certain areas of the screen.)

[sage mirage]

21:9 crew

Unless you are one of those freaks with the ultra ultra wide

In that case I'm p jealous

[tepid trench]

I don't know the ratio of mine is or what it's classed as but it's 120cm wide. It relieved my itch for wanting dual screen.

[sage mirage]

Usually it's 16:9 then 21:9 then ??? for the super super wides

But yeah. Wide is better

[north jasper]

Would love to see 1) Ability to only change the vocal track after a generation. 2) Ability to fix mispronounced words or certain sections of a song after generation 3) Cleaner vocals that don’t sound robotic 4) More FAQ coming from the devs themselves as to how to best prompt seems like we are doing a lot of guessing.

[north jasper]

5. A full clear button. When i extend a song sometimes the extension doesnt extend the song i picked and the clear button doesnt resolve the issue

[cosmic terrace]

i would like to see a note addon to own songs , were i could write things i must fix instead of having that on a txt saved on my pc or in my cloud. (again more for handy small things) 🙂

[hazy hatch]

probably was suggested hundreds of times already so just joining the chorus of
asking for suggestions on the homepage to look more like youtube's where the suggested tracks would be recommended following the subscribed artists and the likes for each user

[urban grail]

Sorry new here, has the suggestion been made to be able to take a generated song and then be able to create an instrumental version of it?

Or be able to write lyrics into the instrumental creation platform so it will generate based on your expected lyrics?

[tepid trench]

This is "stem splitting". It's something that is requested a lot. Until it's possible to generate these separately, there are tools to split the audio to get separate parts but it's not the cleanest result.

[urban grail]

Understood, so since it generates the song as a whole instead of layers separation isn't an option.

[twilit hedge]

It's ok to make a request or a report on certain feature but it doesn't seem nice and considerate to call them lazy.

[thin ermine]

It's a good thing that I'm not in tech support. I'd file ignorant and presumptuous "requests" like that right into the trash bin.

[pure gazelle]

Would love to have a specific toggle to force a fade out or an ending. I'm throwing credits in the trash generating extensions over and over trying to get a song to end, and seeing no results.

[thin ermine]

'til they fix it, do an extension with v3 from the point where you want it to end (or a bit ahead) and include only [End] in the lyrics box. You may still get some longer ones, but you may also get some very short ones.

[pure gazelle]

thank you, will attempt this

[thin ermine]

Sometimes, the extension clips will come in so short that you'll get a credit refund

[wooden fable]

Okay, so if this 𝇟 has a special meaning, what other non-ascii characters should we know about? I think this is exactly the kind of stuff that the devs should reveal to whoever is struggling with writing documentation for this thing.

[pure gazelle]

that's what I've been banking on, it's the only thing that's worked for me so far. It's not really a huge deal, but when I'm throwing 100 credits out to try and get a half second of silence it starts to feel like I'm not getting my moneys worth.

[thin ermine]

Honestly, I think it's much more of a crapshoot with symbols like that than it is with the v3[End] method.

[pure gazelle]

That symbol didn't work for me, I feel like most solutions are a crap shoot at this point.

[wooden fable]

i haven't tested if this 𝇟 character even has any meaning or not. to be honest, i hope not.

the documentation is sparse enough already. having to guess ourselves through the whole unicode table would be a nightmare.

i'm going to assume that it's not a thing for now.

[thin ermine]

It's used to indicate the end of a repeating section in a song.

[wooden fable]

so it does have a meaning!

what other secret symbols are there?

[thin ermine]

technically, it does. But in Suno's "mind"? It's probably noise.

[wooden fable]

if suno was trained on this character meaning "end" then it has a real meaning

but that would mean that some oddball actually went and wrote that weird character into the meta data

which i hope wasn't the case

it suno doesn't know this character, then it's just gonna be weird noise (which of course might have some effect)

so... i'm gonna assume that noone thought of filling up the meta data with obscure unicode characters that aren't even on anyone's keyboards. 😄

[twilit hedge]

https://en.wikipedia.org/wiki/List_of_musical_symbols#Repetition_and_codas

maybe some info there

[wooden fable]

Sadly not. 𝇟 is not one of those. (i'm a musician btw)

wait, it does say it's a music symbol. 🤔 but i gotta admit i've never seen it before.

[thin ermine]

I've never seen it in use. Doesn't seem necessary since there are other methods in place to handle how many times to do a repeat

[wooden fable]

i've tried using musician expressions before with suno, and most of the time they don't have much effect. like if suno simply doesn't have that sort of info.

i've always had the impression that the meta data of suno was probably similar to what Spotify has about a song and the album it's on.

and that's stuff like "genre", "year" and such, but nothing about whether it's major or minor chords, or even the sex of the person singing. and this seems to match that it can be tricky to get a female voice. but of course, i'm only guessing. there's too little documentation on this thing.

So that's my suggestion really: More documentation from the devs or whoever actually knows something about the training data.

it would save everyone a lot of time

... to know what works, and also what doesn't

with the amount of paying users that suno has, i don't think it's a unreasonable request

[twilit hedge]

yeah I really wish they put out some guides to help us to construct the prompt and instruct the AI to read lyric... we know it does read but what are the things that it considers as "meta tags" for instructions... so far, we rely on users experience to share... but it'd be nice to have some real official doc... especially new model comes out.. so we don't have spend so many credits to experiment.. unless that's the intention!

because if everything is clear and works with instructions... then no one will buy more credits 🙂 daily and even lowest level of pro.. won't have enough credits to experiment

[dawn olive]

I'd really like a "kits.ai" feature of being able to choose from an extensive menu of vocalists with ability to blend them , for no other reason than to stop getting that mid Atlantic drawl speak sing bloke all the damn time! 🤣

[tribal sentinel]

Please, please make it possible to keep the vocals or other stems, and specify what you want to replace and with what. There's this beautiful song, and the bass/music part sounds really awful. I just want to redo that and keep the rest. That would be soooooo cool.

[deft pecan]

Not sure if this has been suggested or not i apologize in advanced if it has been please add better song prompts and the ability to edit the songs that has been already created.

[prime stirrup]

Please add Folders in the create tab. Thank you.

[twilit hedge]

It's a good thing if we want to showcase the song on suno

[modest frigate]

Are the devs from the UI still sleeping, along with their bosses ? There is so much work to do! Wake up finally! I wonder how many months they need to do a minor change others makes in 1-2 days. To have such a great AI and in the same time one of the most primitive UI in the world, which is not even capable of banal things is such a big contradiction I have no other words for it....

[gritty sinew]

A suggestion for developers is that we can choose vocals from other songs, and add them to new songs, to maintain a type of "voice" characteristic of each user!

[tepid trench]

Again, this is not constructive feedback (and it's not even a suggestion, it's just complaining at this point.) In case you are not aware, there is an audio upload feature being added - something that has been suggested a lot in this channel and is something which will be appreciated by many users. This would involve UI changes, and there are likely other UI changes apart from this which won't be visible until they have been fully developed/tested.

Please, if you are going to make suggestions or report bugs, be civil about it and don't berate the developers.

[ember beacon]

I Hope, Dev can make a new UI.

For Male, we can adjust pitch slider (young to old), (Soft to rough)
And How many lyrics male can sing, and make a Male color text let say a (blue text), so all blue text in lyrics box is male will sing.

Same as Female, it will be color soft pink..

And for voice sound, we can separate, with music genre.

Like sample, A music is Japan, with Mexican accent. or Yodelling voice style.

Edit : They just need to take reference from any website Image AI. so refer the idea and apply for music version.
For Music, what they need is.

A Negative Box (Remove any music instrument we don't want)
A tick box default : This one, need a dropdown slider (list out all instrument, and we can tick any instrument we want it to play in music later)
A BPM Box : default number or (we can edit any BPM we want)

Suggestion : I learn all genre from this website https://www.chosic.com/list-of-music-genres/

[modest frigate]

Suno got enough constructive feedback and suggestions, reports about missing important and even essential features but NOTHING has been done so far. NOTHING! And we are talking about UI things that can be done in a couple of hours and tested in one day. I am a developer, I know how it goes. The management is even more guilty about this situation. So all these pages with suggestions are in vain ....more time loss for all the users actually. If suggestions are ignored for weeks or there is nobody to implement this baby things any fivver developer can do in one day, then what remains except for complainging ???? How can we call a developer who implements a search feature in the songs only AFTER user complains? Or one who forgets to think of a direct access to the 1000 pages of songs??? Come on! This is actually incredible!

[twilit hedge]

...and you're not even helping to make it better

[modest frigate]

yes, if you all would complain like me, you would change something! But you give infinite suggestiosn without any real changes !!! Is like talking to a wall...I hope you notice some day ... 🙂

[twilit hedge]

screaming at them.. calling them lazy isn't helpful... not like they will hear that and said... oh I need to work more... I'd ignore ppl like you

no one said you shouldn't complain

[modest frigate]

Is useless to defend such amazing developers and management, the facts speak for themselves.. look how often they have small updates in the UI ! Only one ! And this in a server based software which is so easy up todate...

[twilit hedge]

I'm not defending anyone

and I'm not saying the UI is perfect

just that no one know what's going on ...they might just still be a small team...

as the company grows... I'm sure they'll have a bigger team and do more

[modest frigate]

If you have no idea of software you' better stop commenting...

[twilit hedge]

everyone all needs a good UI... good software... it'll come

[thin ermine]

They're just trolling at this point. Stop feeding them.

[twilit hedge]

speak for yourself

ppl like you are not helpful

[tepid trench]

People had previously suggested playlists, search, profiles, to be able to change the song artwork, and many other things that have since been implemented.

Repeatedly being disrespectful to the developers is not going to change how they prioritise their work. We all want things to be improved, but coming here to rant in an unconstructive manner is not helpful.

I am issuing you with a 1-day timeout as all you are doing at this point is ranting and it is not constructive or helpful at all.

[twilit hedge]

good idea

[dry crow]

Have you considered changing suggestions from a text channel to a forum? That might help with that kind of situation (not a single thread to rant, more visibility on achieved work, less repeated suggestions, easier prioritising)

[tepid trench]

I think this was talked about a while back, but I guess this arrangement is "good enough" for now, though personally I don't think it's ideal given that amount of repeat suggestions/bugs.

The devs seem to check the two feedback channels fairly often - I guess they probably have their own internal way of tracking things. I don't think I've ever seen a dev confirm that a suggestion is being worked on - we find out about changes when they appear in #📢┃announcements or #🚀┃updates afterwards.

[dry crow]

Ok, well as long as it's been discussed 😉

[opal spoke]

Suggest a option for translate your own songs to other languages,without just reprompt and the song sounds different from the other one.

[somber wasp]

lol it's called google translate

[twilit hedge]

I think he meant.. kept the background music but switch the language on the singer

regen the voice on the generated clip... with a different language

would be a cool feature but I don't think it's going to happen.. not when they all are generated together in one stem..

[inner rover]

Dear suno, we gotta see a feature soon for how long you want songs to generate, they seem to always try to push out to the furthest limit, always being at 4:00 for 3.5, love this version but as a song that don’t have the lyrical fill for a full four minutes, the program just starts scrambling you lyrics where it wants them randomly to fill in for the 4:00, but great work ❤️

[karmic steppe]

bump

[quiet warren]

Hey I went to my friends house and wanted to show them my songs and searching my username didn’t find anything, we had to search the song title then click my username on the song. Just a suggestion to allow usernames to be searched.

[nova basin]

What I miss is the possibility to select/mark several songs in order to delete them or copy them to a playlist.

[wooden igloo]

You can select multiple items like a file in a folder. If you are using mac, use command key, probably it's ctrl or shift in windows.

[drowsy dove]

Make playlists include:

Total Playlist Length

Total Number of Songs

[wooden igloo]

Hey devs,
Is it possible to disable any audio mastering happening on the Suno export?
It feels like there is a compression happening on your side, and some eq and gain changes. I do not know how your backend works, this is a complete guess on my side.
If you do, I'd really like to see what the original audio is, maybe it will be easier to recover from the noise.

An additional wave download without output going through the mastering maybe?

[rough oar]

a slider would be niiice

[thin ermine]

Please give us some sorting options for the trash... or at least sort by last in, first out. It's pretty easy to accidentally delete a songs right now ( ::cough:: confirmation dialog ::cough:: ). Trying to find that song in the trash can be difficult, even when the art stands out.

[visual brook]

it would be nice to be able to drag and drop playlist songs locally, instead of the playlist being locked into whatever order songs are arbitrarily added to it

[thin ermine]

I'm able to do that in the web interface

[visual brook]

omg thank you @thin ermine !!

[thin ermine]

lol

[opal spoke]

But when you reprompt in other language the song is not the same

[pallid gust]

Hello. I have some accessibility feedback, I hope you guys improve the accessibility in Suno AI.
First: The like and dislike buttons don't give any indications to the screen reader that they are activated or not, so I hope this gets fixed.
Second: there is currently no native context window for suno.
If I want to download a song, for example, I'd have to use focus mode on the list of songs, select the one I want, play it by pressing nvda+numpad_enter, pause it, press right arrow to exit the list and focus on the current song being 'played', press b thrice, then open the more options menu and either up arrow three times to find download, or down arrow a bunch of times. Yeah, it doesn't actually wrap the menu so it's either one of the two options.
I hope this feedback gets considered and fixes to the problems above be implemented. Thanks and regards

[balmy cairn]

how do you make the singer come out with a deep voice, what do you recommend?

[frail sequoia]

I would just like to be able to edit sections of songs or get each track within the song as midi

[wintry gazelle]

Please I beg you please when this happens it should be a question mark near the thing and when u click the question mark then it will tell you what lyric is bad, or multiple of them. I had to rewrite everything and it's still bad. Please add this. Or get rid of censorship.

[soft harness]

It got censored from ChatGPT (which moderates the lyrics in Suno) not liking something.

==> Ask OpenAI's ChatGPT itself about "List what censors the following: [content]" and it should tell you. <==

(If the reply gets censored, ask "what changes to make to avoid censorship" after and it should work).

Usual causes are swear words or interpreting harmful situations. Even if it worked before, it could be a small change caused its interpretation to become different.

I remember people getting censored for using bones, ribcage, and skeletons in benign situations but they made ChatGPT assume some sort of horrific tale.

[wintry gazelle]

I tried gpt but they seriously should add the question mark I think it's a good idea

[soft harness]

Yeah, I agree

[real lynx]

I'd really like the ability to check-box select items on my library and delete them at once.

Less storage space for you, better organization for me aha.

[wintry gazelle]

To avoid censorship, you could consider altering the language and content to make it more appropriate for a wider audience. For example:

1. Replace explicit language with milder alternatives.
2. Avoid violent or aggressive imagery and themes.
3. Focus on positive or neutral topics rather than controversial ones.
4. Use metaphorical or symbolic language to convey emotions or ideas.
5. Consider the impact of your words on others and aim for respectful communication.

Haha it's a classic

[high trail]

Could you please add the option to edit songs in the form of text so that you can actually annotate an already created song or composition? Additionally, it might be interesting to have a feature that allows you to download only the music or only the vocals from a song where there is singing.

[strong swan]

i need a way to follow artists from the radio playlists (in the best case on each song) instead of going to the song page.

[long hawk]

Intro? Got a hard start song.

[potent folio]

may have been a bug

But before it was mostly instrumental

and/or 4x4

Ok, it was a bad "roll"

Removing

[leaden marlin]

for future updates, hopefully its possible to make it so when you create a track it also shows the key its in and tempo as well. and further down the line... allow separate channels for one track. this tool is such a cool and unique way to get inspired and also would be amazing to download one section or just one channel from the track.

[tidal ridge]

hey Dev team, please, put a way to generate the whole song with a specific timestamp to end/stop music.

[visual brook]

Not sure if I need a larger sample size.... Song Radio does not seem to take into account Lyrical similarity - or at least maybe some of my songs are too lyrically unique to get similar songs back in the Radio and Radio prioritizes musical similarity above lyrical content awareness

[stuck swift]

做歌曲

[blissful karma]

Another suggestion: co-creation - I should be able to share a link that allows others to edit/extend. A different link than just the share link. Something that puts a song in both my and the co-creator's create feed.

Another suggestion (50-gazillionth time) - ability to specify voices in chorus. Choirs frequently work on the basis of having different voice sections do different things - timing, etc. Right now there's no way to prompt that.

[potent folio]

OK, so I can say: if there is a way to somehow have gpt gen songs, especially pure gpt gens, separated, that would be nice. Bc it's the primary reason you hear people saying "soulless" and "AI" imo. Now, it can be improved but it will require some randomness I the internal gpt prompting bc basically the issue is, as long as it remains the same, then you get these samey songs

in any case, to be more specific

it generates a particular rhyme scheme

and it forces things in that are just awkward in many cases ESPECIALLY in rap

but that may just be bc it's a genre that can be done really wrong ie it's much more sensitive to error

bc the subject matter is basically extreme tragedy, and you're telling it by bragging, and yea, that's tough

To pull off; but the gpt tracks stand out, much more than someone even just writing stream of consciousness

and I think the overall quality would be improved for the majority of people if starting a station based on a custom written song only included custom songs

On a secondary note, you can probably get better LLM outputs at this point on a llama 70b fine tune

for this particular purpose. although I'm guessing you may have some contractual discount on gpt via msft. In any case, having some randomized factor in the gpt prompt would help alongside internal testing to make sure it uses a variety of approached to songwriting, as opposed to a static. Ideally, you don't want the requests to be anywhere near converging.

if two people request a pop song on the same topic, ideally, they should have different structure, different approaches, etc

and then likely it would be less noticeable, and an improved win rate

[hallow seal]

We need a seed number to get the same singer 😉

[harsh musk]

Is itvpossible to bulk download all songs we made?

[sinful harness]

When option that we can choose to only make 1 song, instead of 2? 🤔
As l think the AI would make better work only to having making 1 song instead 2 at the same time. Giving better results in future.

Or heck make an option that how long you want the AI take its time to pre-plan the song. To get better accurate songs.

With of course costing 10 instead 5 credit. So basically if "Free users" want more power to make better complex songs, they can. But now instead been able to make 10 songs, they can make only 5 now.

But still you have the freedom to choose making either 2 songs prices of 10 credits or 1 song prices of 10 credits, but better and more complex.

[daring slate]

For the Music Ai App can you please do like on the Website that will have to go on the App or no?

[strong kernel]

Don't know if this has been suggested yet, but control over approximate length of the song or length of generation, based on the style would be an amazing addition.

as well as being able to lock in a specific seed and being able to create multiple songs with the "same style" eg vocalist and general sound of another song that we have already generated.

keep it up, keen to play with the upload feature 🙂

[tepid trench]

I'm not sure what you are asking but there are currently no Suno apps. If you are using an app that you have downloaded, it is a fake/scam.

[raw mist]

Please add ability to create vocals that are not distorted. Its hit or miss right now. Thank you.

[elfin reef]

Hi, I'm blind and so I'm a screenreader user. I have some feedbacks regarding Suno accessibility, especially on mobile. Where can I transmit my feedbacks and maybe works with devs to improve this aspect of this geeat project? Thanks

[light ruin]

a little nsfw kinda tag while creating songs with separate home page so people can choose to go with PG-13 or not and still have a change to go trending (there can be still filters on some words)

[still parcel]

DM and I’ll give you an email contact that will get it to the dev team. thanks!

[old quiver]

is there an option to edit the lyrics in for songs you extended so you don't have things repeated?

[plucky sequoia]

Would love to be able to use Suno as a collaborator. Have the ability to create just a bass part or drums, vocals, etc., in whatever key, bpm, and time signature. Or the ability to feed it a song and have it create a missing part. Sure it’ll get there eventually

[tepid trench]

There's an audio upload feature coming soon where Suno will continue on from what you upload. It's pretty versatile and whilst it will be as unpredictable as AI usually tends to be, it will pick up elements of it such as the tempo and chord progression and try to apply your prompts to it. So it may be useful for this.

[marble bough]

per-artist or per-song comment wall might be cute and build a sense of community. maybe dms too

[marble bough]

it would be super cool to be able to automatically display foreign lyrics like this

[spark coyote]

It would be great if Suno's new upcoming upload feature can be used by free users with limits or increased credits usage and unlimited usage (as normal) for subscribers. So that the free users can also show their talents too

[hallow talon]

i wish it would bd free for poeple that may not have alot of money like if your a youtuber and wanted the song for some think would of make me much happier

[steep canopy]

Could the term "progressive" not default to the metal crap of today? Please let it learn 70's prog rock in its various forms. Thank you.

[hallow talon]

i would love to use the song i make and put it on youtube or others

[twilit hedge]

You can, just can't monetize with it if you're not a user with pro plan.

[pliant tendon]

This has been brought up a few times, but extending and stitching all parts together completely jumbles up lyrics because they are also stitched together. Could we receive the option to simply edit lyrics manually akin to the Edit Details section?

[pliant tendon]

One more suggestion I have is the ability to do audio2audio. Now, I'm not sure if this is what's being worked on currently, but it seems we will be able to upload original audio and extend it. Could we also get an audio2audio feature with "denoising strength" for the lack of a better term? Let's say I'm using some really bad-quality midi piano samples, and I want to run them through audio2audio & perhaps also extend afterwards to get a nice piano solo that adheres to my original audio but sounds cleaner.

[undone stream]

It would be amazing if Suno could export lyrics file (I don't know what its name is. something like .srt or so) so making Lyrics video can be made easy

[royal gust]

Loving the new version, especially with the time extension. However, if the custom lyrics end before the limit, the AI will try filling in extra things vs just ending. It would be nice to cut off the "extra" at the tail end

[tardy anvil]

tracks with 1 play and 1 like shouldn't be in a trending list.

like, hello? no one has even played the track, but its somehow "getting lots of activity"?

[deft pecan]

I got one more suggestion please the Western Metal Genre.

[light ruin]

oh yeah lets not give people any chances to show themselfs and only the people are lucky that they are well know can be on the trending page sounds awesome

[robust jewel]

That would be what search, playlists, categories, or radio would be for. Trending implies that it's something getting attention already.

Also, what the showcase section of this Discord is for.

[tardy anvil]

thats what NEW section is for :P, among what @robust jewel said as well. How can your track be trending, if you have zero activity on the track?