💡┃suggestions-web

[hybrid crane]

lol

[jagged grove]

Please voice cloning

[robust jewel]

Which, of course, NOBODY is doing because that would be directly against the terms of service and would be a bannable offense.

[topaz dagger]

It would be awesome to be able to upload a piece of music, provide Suno with lyrics and have it generate the voice as a separate stem

[zinc gyro]

A personal appeal... Please Suno... forget about trending. It's hard to imagine that whoever is at the top didn't use an artificial boosting bot to do it. The top 1 is ridiculous (I'm embarrassed because it's in Brazilian Portuguese). Is that what you want to use as a benchmark for something serious? It doesn't represent me and it doesn't represent Brazil.

[robust jewel]

The trending page isn't even really active yet. It's currently just a few songs they've selected to show off what can be done, that way visitors can see samples.

[zinc gyro]

It is... but apparently they've been contaminated by booster bots.

Worse... driving garbage.

[modern vessel]

we are wolf

[toxic wharf]

is there anything confirmed to come?

for future updates

[feral aspen]

1. A new model, V3. (This may not be very near, but devs already posting teaser clips in chat here.)
2. Custom continue times. (This is huge, but may take time for people to understand why.)
3. Misc UI and organization improvements (song title editing, some kind of better organization with folders or playlists).
4. Multiselect, multiaction.
5. Better accessibility, especially for screen readers.
6. Maybe longer duration? Hinted at in public chat by sunomc, but no details...

[robust jewel]

I think I can explain why #2 is big with one clip. 29 seconds of this is perfect and I want more of exactly whatever it is. The rest of it can be voided. https://app.suno.ai/song/101e274b-2e65-494c-aeb5-f930c501052a/

[cold vine]

7. Backwards-music-creator (to create intros)
8. "impating" parts editor

those would be game changer

[leaden marlin]

please remake Suggestions for new updates:

1. please do something with the voices for the songs. Here I am referring to the voices for Japanese songs and fantasy/epic style songs. I would like clearer voices, something more organic.
2. Do something with the anime style of the songs. I would like more types of anime genres: happy electro rock anime, fantastic anime, emotional vocal anime, etc.
3. To add and fix the current anime style, because the rhythm and tone of the genre that you have is not very good, not evenanime voices for japan music. Please fix voices

is the control panel available?

[woven merlin]

Suno is telling Twitter app, that on the 6th of Feb, they will publish new experiences. probably the next version. Hopefully, it includes a huge improvement in sound quality.

[winter zephyr]

General question. Not sure if this is already possible? Is there a way to become an affiliate for suno.ai?

[formal bear]

No artist affliate yet. But working for suno ? 😄 yes.

[winter zephyr]

Lol.. off topic, you looking for product managers? 😄

[safe sand]

Did someone mention that the PRO Plan (that I'm on) deletes unused credits every month ?????

[steep jackal]

I logged out of the site and logged back in ... all my library of songs are GONE. seriously, wtf? As a paying member, I figured the songs would remain there for later use..... !!

I had several finished songs -- all fuckig gone!

[steep jackal]

are you kidding ... what a rip-off.

[hybrid crane]

It's not different than midjourney in that regard, or telephone plans... @safe sand

[hybrid crane]

mail them.. it shouldn't be gone [e-mail]

[karmic fiber]

guys what i can do if suno give me this: Suno is for generating original music. Try different lyrics, and use genres and vibes rather than specific artist names.

[robust jewel]

That is a generic error message. It typically shows up in a few situations. Pasting in a whole bunch of text, lyrics it decides are "too violent" or which could "sound like a slur", if it thinks you are asking for a specific artist's sound, if it thinks it sees lyrics for a song it knows, or sometimes with seemingly random, perfectly innocent, words. Oh, or if you don't put anything at all as input.

[fast ocean]

It can even happen if you write a person's name in the song's title. I wrote lyrics for my friends with their names in them and no problem, but if the song's title included my friends' name it showed me the same message.

is there a way of choosing between male and female voice? all the songs i've generated so far are sung by male voices.

[robust jewel]

You can try putting male/female vocals in the style or [male/female vocals] in the lyrics section

[marsh thistle]

Suno should make a search feature where you can search for songs that are made public!! I never really ever got how suno has the "Top trending" when no one really is in control!! Suno could also make the song tell you who made it with their discord user or whatever they use! so the to trending is what there cummuinity likes!

[robust jewel]

The Explore page isn't fully implemented yet. It's currently a curated list just to have something to show off as samples of songs, so new users can see something.

[leaden marlin]

Pleas Pleas make an option for if you want music or a song because every time I try to make a song I end up wasting my credits because it never generated a song and only gave me music

This has been a problem for a while

[flint fossil]

Go to Library and then select the "Liked" filter, that might bring them up. Also check on phone, and desktop browser.

[torn adder]

https://app.suno.ai/song/a4f2c680-3f53-48ce-a9ce-e0547e8cc936

hey everyone, how do you feel about this song?

what would you call this genre and what bands sound similar to this? I wanna continue this song in this style. I tried 7 other generations and they don't sound as good as this

[glass cliff]

Does anyone from Suno actually come through here?

I've gotten to the point that I can usually generate a full song (4 or 5 parts) in just 4 to 5 generations in multiple genres by crafting my prompts well, but the current fee structure for credits disincentivises those that are actively paying customers from using them. I work in Finance, and would be happy to outline some alternatives that would help increase revenue by rewarding the paying users rather than penalizing them for having subscriptions. Win-win all around, unless the goal is to reduce server traffic; if that's the case, the current plan will certainly help curtail usage quite a bit.

[torn adder]

I'm from suno, the website

[robust jewel]

The team read stuff here but rarely respond

[glass cliff]

A website user, or someone that actually works at Suno?

[glass cliff]

I kinda got that impression. I've sent a couple of emails to dispute some of the egregious generation issues and try to get some credits back, but have heard nothing bac from them. ☹️

[delicate cedar]

There has been an unusually large influx of email in the past few weeks, this has caused email responses to lag back further than anyone would hope. There have been steps taken to improve the response times and those should be taking effect soon. I apologize for the lack of response you have had in this time period.

[glass cliff]

I don't mean what I said in a negative way, just stating my experience. This is a beast of an idea, and I have the feeling it's blowing up at an accelerated pace. I'm sure there may be some growing pains, for the most part I can navigate around what I've needed, save for the billing structure and penalization of subscribers.

[delicate cedar]

Sure, I'm just explaining what was going on and (hopefully) giving the good news that the situation will be improving soon.

Oh and apologizing for the growing pains you have experienced.

[steep jackal]

With Midjourney, errors show up as errors, or the bot hangs when generating images. With Suno, do ongoing "credits refunded" signify high traffic or bot problems? I've gotten 20 in a row.

[toxic wharf]

wow can you show me some V3 Clips?

[feral aspen]

I can't, but if you search V3 the devs teased a few early tests in here somewhere

[fallen pecan]

v3 sample here => #💬┃general-chat message
I do hope the quality will get better before they release v3

[torn adder]

website user lol

[glass cliff]

Sure thing, and thanks for the info. About the growing pains, I meant Suno, I'm sure the demand is growing much more quickly than they anticipated (good problem to have though!). I can't say it enough though: their current token pricing structure is a poison pill for them, because it penalizes the paying userbase 1) incentivises less usage to preserve credits, 2) discourages "topping off" by charging 2x normal subscription cost for same amount of tokens, and 3) incentivises users to create multiple accounts.

[feral aspen]

I briefly tried to reproduce the same hybrid prompts in V2 #🐣┃suno-showcase message and if that V3 is an average sample, it is at least different. You can accomplish the same style in V2 but you have to weight the two genres asymmetrically and trial and error it a bit, where it (apparently) works first try as a hybrid in V3, pretty cool if so. It's also impossible to judge the quality ceiling from one random sample, of course.

[plush crag]

Does anyone know what prompt we can use to get child like voice from bark

Story telling or narrative feel preferably

[stone cloak]

could we have just 1 clip generated at a time instead of 2 for the free version and 10 for the paid? one would eventually refine its prompt instead of wasting 10 time the credits for a single prompt?

[robust jewel]

Paid accounts don't generate ten clips at a time. Free accounts can run two separate generations at a time while paid can run ten generations (twenty total clips).

Both get two clips every time they generate.

[waxen sand]

Someone has probably mentioned this before, but since its difficult to search in specific channel on discord to check for dupe ideas, ill just put it here: Style transfer. Sometimes you get a great unique music thats good but possibly want different lyrics or the voice is bad. Style transfer would not mean that it sounds 100% the same due to different lyrics lengths etc, but you can put new lyrics, but the style, instruments, pace, etc is carried over

[mystic zodiac]

@waxen sand Yes I think it's important to have more control over what we get rather than a lotto. I've been suggesting being able to use a previous render as a reference for the style and instruments and also the vocal used to create different tracks from the same artist/band. Or at least, drop boxes to choose instruments and vocals.

[edgy oasis]

The ability to change a couple of words in the lyrics would alter syllabic timing, but would be cool if there could be a rework of the same melodic phrase.

May require a more interactive GUI with shapes corresponding with phrasing.

[hazy willow]

It seems like i am asking for a feature in an upcoming version of some software related to music, where users can separate instruments, voice, drums, melody, and piano from audio tracks, and also train a feature called "clone voice" to sing using the user's name specifically rather than randomly.

[potent crane]

Because I have the pro plan I can be a bit more fussy, but for me the phrasing of some of the vocal lines don't quite work and the whole thing sounds a bit listless. Genre wise it sounds like early 1970's west coast country rock, like say The Eagles or America.

[final rivet]

Probably already mentioned 100 times but..

1. Tag or add notes on older generations to easier keep track.
2. Search/filter function for history.

[bronze citrus]

Does it work?

It's not working yet

I wish I could adjust more details in custom mode. So that you can give detailed instructions as if you were entering a prompt as well as the lyrics.

[bronze citrus]

It works here

[fickle blaze]

I think for what it’s worth, Suno is already powerful and yes, there could be more features, but like all hidden gems, it needs some time (and money). I agree It would be awesome to have more control over vocals, melody, Instruments etc. , that sound great to have a consistent song generation. To get a better Quality and all stems as separate files would also be great. The next stage would be a full blown project for your preferred daw, so you can adjust or master the song. IMHO it depends on how the song generation is done. If all layers are generated separately it would be easier to achieve a more modular composition, but if the AI generates all of it at once and all of the separation (stemming), cleaning etc. would come after that, the outcome would still be noisy and not perfect.
I tried to stem, clean and adjust the generated songs, but it’s really hard to get the song clean and not to lose too much of the Instruments and vocals.
I am still grateful, that a utility like Suno exists.

[bronze citrus]

Sometimes, the lyrics come out strangely different from the ones I wrote, so I fail in the middle of making a song. In times like this, the credit I used is such a waste and I’m upset that I couldn’t even finish the song. I wish there were more pieces that could be drawn at once. 1 minute is too short.

I've already spent over 500 credits to make one song!

[idle ember]

Few are!

[bronze citrus]

Please also respect minority languages ​​ㅠㅠ

[somber dove]

I wish suno had a feature where u can add your own clip and it’ll make songs based on that it would Be so much cooler

[harsh falcon]

Hello guys , I am new here. Let me know please how does it work with Suno. Once the Song Is generated, what can we do with? Am I free to modificate or use it as karaoke or make a cover with, or sing It as I want?

[delicate cedar]

Please see the FAQ: https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448#f0f1441180484d6094206e84e334ba36

If you still have questions, please email at [e-mail]

[prime rock]

The valentine thing is fun, would be nice if account users had a way to get to the songs once generated to work with them!

[red charm]

Hi, I have been using suno for some time and I really miss the ability to download vocals and instrumentals separately, because when I use other tools the tracks do not separate properly. In my opinion you should add this option, it would be a huge improvement to the tool as users would be able to mix-master songs which would make the songs even better.

[vagrant rock]

What is the Valentine Thing, I'm sure I missed something...

[prime rock]

It was posted on the twitter page, I didn’t see it either https://t.co/ksPmrkjZ2p

[vagrant rock]

Thanks. I'm not on that platform, so no go for me I suppose

[prime rock]

You don’t have to be on twitter to use it, that’s just where it was posted. It’s a browser thing.

[feral aspen]

https://v-day.suno.ai/

[fickle blaze]

It seems there is a bug. The last two inputs are switched.

[feral aspen]

I actually wonder if mixing it up like this could improve ChatGPT writing quality. At the very least it starts things off atypically!

[fickle blaze]

I don’t know, but sometimes, I let Suno generate the lyrics and put them into ChatGPT. I don’t know if Suno has a custom gpt or feed ChatGPT with a different prompt. Sometimes the outcome is better than just asking ChatGPT for a Song.

[golden peak]

Option to made AI title and ai images when you use custom mode

ai

Option to use AI to generate Images and Titles for existing songs you created alreada

already

[subtle torrent]

Hello everyone, unfortunately it's hardly realistic to use, maybe I'm not smart enough either. Functions such as speed, voice character, A cabella, just the singing cannot be controlled. So far, unfortunately, only the worst has come out. Sometimes German texts are emphasized incorrectly. Is there a trick here?

[golden peak]

It's not working good with german

Also the music is not clear enough to use it

[plucky mural]

Hello guys, am I the only one having troubles trying to continue songs?

it repeats what was written in the first half, it spends a lot of time silence, it mix the 2 texts, and so on...

[subtle torrent]

It's good for doing something funny quickly, but nothing more. There should be midifiles of the tracks available for download and at least a precise setting of the voice and color tone

[warm cargo]

How about a way to generate an insert between parts? Like part 2 and part 3 are perfect, but could be more perfect if there were something like a part 2.5

[delicate cedar]

Me like! Is this something that would be possible?

Hmmm...maybe a tweak of the of the some day release of the ability to set manual resume points? Some way to reattach the new part that way?

[warm cargo]

This is where I believe a lot of our issues would be solved in something like node based editing

"issues" - I'm not complaining at all, to clarify. But a lot of points in this suggestion feed could be implemented in node editing in the future

I would also like to be able to download the midi files of the instrumentals. I hear what sounds like very specific VSTs for that I use personally, which makes me think there's a midi accompaniment? I'm not sure. Stem splitters, even the best that I've tried like fadr, have a hard time with instrumentals

Example, I generate a song with an orchestra, by part 3, it's very muddy poopy. A midi split would help tremendously

[wide belfry]

Thanks! I just followed them on Twitter.

[wind crest]

having music and vocals generated separately would make this an ideal tool, completely solve the issue of absolute banger tracks generation with crap vocals or no vocals at all if we could lock the background music and use further generations for different attempts at lyrics

[wind crest]

when something is setting off the 'original content only' message it would be great if that part is mentioned in the error message so we can adjust a problem in a targeted way instead of just taking shots in the dark at what part is setting it off

[robust jewel]

You get that message because something went wrong and Suno doesn't have enough information to be more accurate. It would be nice if the message made it more clear that it is a generic message, but you're having to take a shot in the dark because even Suno doesn't know why it failed. It's often just ChatGPT saying "no."

[zenith gorge]

2FA on SUNO Accounts.

[old quiver]

is there a way to use the same singer and switch the beat and melody from an existing creation

[rotund orbit]

So I wasn't sure where to put this. But think of it like competitive analysis. If it's on the wrong board, let me know.

Okay, so I just reviewed the Splash pro service, here's my thoughts

1. The train your own voice feature, even as terrible as it is is nice to have.
2. I hate the way that the music and lyrics are different pieces that may not necessarily work together, but I like the fact that you can edit songs there.
3. I hate the format they use for parsing songs, but I understand why they do it that way.
4. I'm bothered by the fact that there's almost no genre control, and the system ignores what you tell it to do.
5. The longer generations are nice, the wait time is not.
6. In all seriousness, the quality of their output is nowhere near what Suno gives me.
7. There's a very steep learning curve that Suno does not have.
8. Key and bpm are useful

I feel like things Suno could do easily from this list would be

1. Voice personalization and swap
2. Lyric editing against existing songs
3. Bring the equalizer back on videos, I miss that.
4. Key and BPM settings

[formal bear]

TY

for 3, would you rather have the equalizer, or time aligned lyrics? (highlight lyrics as it sings)

[rotund orbit]

I think the timed lyrics look cool, but they're a little distracting.

The big one I would like to have is song editing.

which I totally get why that's hard

might be easier if the seeds were visible though, thought

[small crown]

They could increase the time of the music, because sometimes in some songs, the beginning is very long, and it would be nice to have a longer duration, then the music would be more complete, and there wouldn't be the problem of when to get to the good part of the song, simply restart because it took 40 seconds, but even so, 40 seconds is a good amount of time, but it could be increased.

[robust jewel]

It was increased. You're using the Discord version, if the song is only 40 seconds long. The one on Discord is V1. V2 doubled how long the first clip is, so you get 80 seconds for the first clip and 60 seconds for every continue.

[hazy willow]

As producers, it would be beneficial to explore new methodologies or tools for training AI models, potentially incorporating iterative loops or feedback mechanisms. This could enhance the effectiveness and efficiency of our AI production processes

[steep holly]

what would help me : the equivalent of "Interrogate CLIP / Deepbooru" so I can feed some music in and get ideas for what styles, genres and tags could be used to create songs with similar feels. (Plus of course auditory 'inpainting', and generation of personal 'singer/band LORAs' so I can get some continuity across songs)

[warm fable]

While working on a few tracks, I've come to realize it would be great to have the following features, preferably as

checkboxes:

• Instrumental Music
• Prelude, Interlude, Postlude

For instance, if Prelude is checked, I'd like the song to be arranged so that it's sung after the prelude.

Additionally, I think it would be necessary to have a feature that allows for the conversion of completed tracks into Instrumental Music.

[warm fable]

And, although it might be challenging, it would be great if there was a way to separate each stem and extract them as MIDI files. I'm a bit disappointed with the sound quality.

[glass cliff]

These are hilarious 🤣
https://vdaysong.com/2p8cxbbz
https://vdaysong.com/2p8kfmpk
https://vdaysong.com/55e7j348

[viscid urchin]

Can you make a plan that got more credits than Premier Plan? I keep topping up all the time.

[dark tiger]

+1 for MIDI

[fading thicket]

Could we have an option that allows us to choose the length of the song?

[swift totem]

https://medium.com/@flencrypto/how-to-split-music-into-different-parts-45555797ccbb

Here you go @warm fable

[warm fable]

Thank you for your response. However, I have already tried using that AI or similar tools. The issue I encountered was a degradation in sound quality when separating the stems. This led me to feel that native support is necessary.

[warm fable]

🤜

[sonic adder]

I could add voices of people who once existed or still exist, but if you can't due to image rights, I understand.

[umbral coral]

allow me to train my vocals on ur site. TY.

[marsh thistle]

A search method so you can search for peoples songs made public. and you can search for creators names with their discord username or what ever username they have

but it will only show the people who has a song currently public

[steep jackal]

I would pay a premium for STEMS! Awesome though! AWESOME!

[desert stirrup]

Generate songs using the melodies we composed allows a larger input to the song by the users and a higher originality to the AI songs
(I am sending this suggestion because I have a melody in my head but I suck at producing)

through MIDI

Also, I would like to see an option for premium users to purchase the rights to the songs they generated when they were a free user for commercial usage

[lament kindle]

allow us to download vocals and instruments separately for our generated songs

[desert stirrup]

Allow us to download the MIDI of each song

[modern hamlet]

Hello. Is it possible to download the stems for the songs?

[hybrid crane]

short answer no...
But you can use other software to split it.. downside, each stem has bleed from the other stems
They are investigating to find a way to develop this so it does it pre-production though

[plain quarry]

I would like to make the following:
include race ie:
African American Singer
(voices are very distinct along with accents and should be allowed in description)
ability to understand more direct bracketed instructions ie:
[acapella]
[female chorus]
[slow intro/outro]
[specific instruments]
[change genre]
[musical iterlude]

also the ability to download vocals, instrumentals and background singers separate
consistent singers
save voice or see the seed so you can build an entire music cd.

I would also like to add collab between my created songs and singers for duets for example

[woven cradle]

@plain quarry some of these kinda work already, it is just hit and miss. Some tested meta tags are documented here:
https://www.notion.so/Metatags-22a69f63b1814aa9afbec9d21f7b55c4
My request would be to make it more likely for meta tags to work.
The vocal style often works in the style field, like "Clear female voice Latin Jazz upbeat rhythmic"

[plain quarry]

thank you so much for this info, I am going to try some right away, the ones I posted don't work at all but I hope are added because [Acapella] or [duet] or [instrument name solo] are very common but do not work or haven't been added yet

[molten orchid]

Ideally, I would love to be able to fill-in between two samples. Like I have 2 chirps and I want to fill in X amount of time for a smooth transition between the two, even if they were a separate generation. Is it possible to do with the model? Are there any plans for it?

[feral aspen]

plans to try and make it work, and statements from the devs that it should probably be possible (after research) but probably not a near future feature

[wanton roost]

Hey! Just found out Suno on thursday and already created 600+ tracks with it! 🤩

Here are my struggles/suggestions/requests:

• If user is happy with the track, but want's to edit/fix lyrics, it would be great without losing the music and vocals
• Would be nice to be able to add for example Intro to already existing track without music and vocals changing
• More meta tags, [choir][male][female][children choir][phone ringing][phone hangup] etc
• Pitch and BPM options would be great

Thanks! This is so awesome, I have no idea how to play instruments nor have talent to sing, but I have liked to make lyrics since I new how to write ❤️

[bold cedar]

I don't know who this man is but God bless him for creating something like this

[woven cradle]

I didn't create that wiki, but I'm happy you found it useful.

[warm cargo]

How about midi stems, or an option to generate with Vocals in the right channel and instruments in the left channel, now that V2 is stereo by default?

[wanton roost]

Also it would be beneficial to have version numbers automatically added after the name of the track. The picture helps little bit, but after creating 20+ versions with the same name it starts to be painful to handle. Also project folders for different songs would be a good idea.

[wanton steeple]

I would love to white label the platform and pitch it to a specific use case I've been thinking about. I'm all about creating an AI sync agency once quality gets another bump, I already got the domain 😂

[abstract dirge]

https://mubert.com/render/pricing?via=syed-farhaan-ahmed

you can give a try to this music generator

[silk cargo]

I think they have some pre prompt there than kind of optimize the lyrics for suno

[fickle blaze]

Yeah it seems like that.

[cold hornet]

Subscription plan upgrade to output 4800hz WAV files with STEMS for SYNC music 🙂

[leaden marlin]

Would be nice to have an editor where, after we create the song we can perfect it with masking or just select certain part and only regenerate that part to what we want to sound like, maybe add or take out certain sounds/instruments. For now its quite hard to make a song without using too much credits... but the final product is worth it if your persistant enough... maybe on the website add a comment box when you flag a song as buged so you can understand, and help fine tune the AI to correct its errors faster? Also would be nice to have some specific prompts that users are aware of and it works at least 80% of times.

Oh.. also sometimes when I ask for the whole song, it glitches a phrase and theres a "Blip" song that is not in the separate part.

[potent crane]

Getting suno to produce a female (or male) voice on command is a lottery, I put female vocal in the music description box and maybe 6-7 times out of 10 it works.

[potent crane]

Much easier for all concerned (suno and end user) to do any editing once it has been downloaded.

[potent crane]

quality is the next key development, must be lossless .wav - when that arrives it will be a game changer

[woven pilot]

the sound quality gets worse and worse

[weak elk]

I just figured out something that has confused me several times. If I have a continued generation (two clips as usual), and I click thumbs up on the lower one, it moves up!!

[woven pilot]

is there a way to make songs not start so like sudden? short intro maybe?

[swift totem]

Yeah I have a few issues with the creation process, the list of your snippets that you choose to continue from, if you like me have several alternatives and review each one and decide to continue with where each when you go back over the whole song you can see to which clips the whole song is joined to, you cant see them

I suggest either an infinity scroll or another reference view so you can see all the parts that are going /have linked to create the whole song. swapping out clips for others that you prefer with the same

[swift totem]

Try Typing:
[build up] if electronic music
[instrumental intro] if bands

[woven pilot]

The biggest suggestion I have is to be able to use the same voice on multiple songs. If you guys can do that I will pay for 2 years upfront

[sudden grotto]

Would love to have a higher bitrate and stems and the ability to use your own voice would be cool

[weak elk]

I see this has now been changed. Thank you team!

[wanton canopy]

A button is missing to check the lyrics. It's a nightmare to find the word that doesn't fit, and it's disappointing to have to spend generation tickets word by word to figure out what Suno doesn't like.

[woven pilot]

if i upgrade my subscription do i get points added right awai?

[rotund orbit]

Okay, so here's the math on what I want from a larger suno plan.
5,000 tokens is roughly, 6.5 hours of work. Work being the time I'm going to put into making songs.
I want to be able to do that 5 days a week. So 5*5,000 = 25,000 token, and four weeks in a month... I need 100,000 tokens on my plan.
I would happily pay for a larger plan if I could get one.

[delicate cedar]

Sounds like what used to be called the Enterprise Plan. You have to email them at [e-mail] for that.

[rotund orbit]

Yeah, they don't read their email.

but they do read the suggestion board

🙂

[delicate cedar]

When did you send it? Remember there was a huge influx of email a bit back that they had to work through and it is possible they are still working through that backlog.

[rotund orbit]

I've been waiting for 3 weeks for them to get back to me on the email

[delicate cedar]

Ah.

After three weeks I'd definitely resend the email. Heck as impatient as I can be I would have probably emailed them twice more or worse by now already. You are much more patient than I have been in these situations.

[rotund orbit]

I've sent them three of them

[delicate cedar]

I'll also link this in the moderators section in hopes of boosting attention for you. That kind of patience deserves to be rewarded.

[rotund orbit]

Love you hepcat

[delicate cedar]

...

OK, have done so.

😳 Awww...thanks. 😄

[modest tinsel]

hi, how can i create a music with my chords and my lyrics?

[bronze citrus]

It doesn't matter if there are more credits. Can't the time for one track be increased? 1 minute is too short. Is it technically impossible to make it 3-4 minutes long so that you can create a complete song in one go? There is something unsettling about the current connection.

[oak notch]

There should definitely be folders in the library. It'd be really helpful for organization.

[rotund orbit]

I think they're doing that for technical reasons. It cuts down on compositional decay. Ai's get weird at about 90 seconds, on a single take. This one's pretty good, but it's got the same problems all the others do here if you stretch it long enough.

[waxen sand]

idea: support to make songs in the style of national anthems and dprk music

[rotund orbit]

thought: Man, the imgen images can get really dark sometimes. I'm not complaining. I don't distribute them outside the videos, usually, but damn.

[cyan umbra]

Ok, what should be a super easy, barely an inconvenience request... Can the web player get a volume control? I happily confess I'm slacking making songs when I should be working and the player always being full volume sometimes makes it a challenge to catch the things I should be paying attention to. 🙂

[dusty beacon]

I suggest you FIX BARK

[robust lark]

One single feature that would make Suno 10x better

[delicate cedar]

Please don’t spam the same thing in multiple channels, it makes it harder to follow up with people. It’s also in the #📜┃rules . Thanks.

[plush scroll]

Yes, this.

I would like to be able to add a clip before the present clip.

I.e., if a song's beginning is a bit abrupt, I can add a fade-in or intro.

[floral cairn]

Yes or even rewrite totally other vocals in the same exact style as before could be cool

[spring mantle]

New Suggestion
Users should be able to choose the type of voice (male or female)

[lapis phoenix]

Can you please add more songs to Explore folder in web app? Maybe "Browse all" button or something.

[empty geyser]

date and time for each song generated by the Suna AI platform. Tell me why Suna doesn't use date and time records on their platform like most others do.

[fossil blaze]

being able to transform the chirp into sheet music so people with instruments can play it live

[cold hornet]

Hello, if ya'll need help modding the discord I am avaliable. Im in here all the time helping new people. I don;t mind 🙂

[languid knot]

Hi guy can you make search bar in the web-app?

[woven pilot]

is there a way we can see only "full songs" onm suno?

[dusty beacon]

It wasn't the same message

In this channel, I added "I suggest you"

[delicate cedar]

This is not D-n-D. There is no benefit to being a rules lawyer here. You were asked not to do it. Please don't do it again. Thanks.

[dusty beacon]

When is it getting fixed/

It's been how long now?

Without any indication to people blindly going into #bark-beta trying to execute commands wondering what they are doing wrong

But thanks for pointing out rules instead of helping

Now fuck off

[wide belfry]

Hepcat doesn't work for Suno. He doesn't have that kind of info.
He's a moderator, not a coder. And he doesn't deserve this kind of disrespect.
He's trying to help you get traction because as he pointed out, " it makes it harder to follow up with people."

Relax. You're not making your situation better. You're only making yourself look bad.

[ionic dragon]

hello , i like the ai app - but continuation of a song does not continue the song its a new song - different styles - change it plz . also : the error message is too wide ! - if there is a "forbidden" word - say it or mark it - i think the AI is confused because of all the different languages. and why are there forbidden text anyway?? art is free ?! no ? where is free speech ? i live in a country without the need of censors . please fix it or make a 18 yo disclaimer or something. or move server to different country where the art is not censored - maybe ? i edited the text for the better 🙂

[wide belfry]

When you "Continue the song" it should be very closely related. Not a different style. How many times have you had that happen?
I use it a lot and it's generally pretty good. A couple times it goes sideways but that should be the exception, not the rule.
Then you can stitch the different parts together using "Get Whole Song" and it comes out pretty well.

How come you went so negative at the end? Was that necessary?

[ionic dragon]

better now ?

[spring mantle]

Doesn't work me unfortunately.. already tried it

[robust jewel]

Are you providing it the style prompt when you continue? If you are, it should be picking up at least some part the way the clip ended and trying to use that to make a new section in the prompt. The error message is too wide because they don't know what is actually wrong. Something they are relying on, such as ChatGPT is refusing the inputs, but isn't telling them what or why.

[wide belfry]

Serious? That's one of the few I've never seen fail.
Dunno.

[wide belfry]

I've been using Suno for over a month and it only went sideways a couple of times. NOT a common problem for me.

I did a 5-part song mostly yesterday, finished it this morning. It worked fine.
There were a lot of takes I didn't like very much. But the continuation always matched the style and voice of the song I was working on.

No, I take that back. There was 1 version that went so far sideways, I couldn't see where that came from. I def would have agreed with you.
But that was 1 track out of many...

[rocky ocean]

It would be really awesome if we had some different views for the library or song creation.
For example after making a ton of tracks the library can become quite full and tedious to sort through. Also when making songs using the continue feature it would be nice to see like a map/path or like decision tree/node graph of the tracks that are used together.

[scenic bolt]

yeah, that. just a basic folder structure or a tag system to have the library be more than a long list

[waxen sand]

How about an "album" feature where you can basically have folders to sort them? currently, the only way to sort is to like and scroll through. Personally i have downloaded all the good ones i made and sorted them in folders on windows and never use the site except for when making new songs.

[delicate cedar]

Oh I like this idea.

[forest chasm]

I would like to see a more specific error message when a song can't be generated. Especially for my German songs it is currently ultra-hard to figure out which words the filter does not like. Sometimes completely harmless words like 'erben = inherit' are blocked and I have to invest a lot of time and effort to figure that out. Sometimes, like today, I am not even able to figure it out at all and so I simply wasted hours writing a songtext and endless tries to create the song before finally giving up. My second wish is to somehow mark the end of a song, maybe it's possible but I just don't know how. I tried [End] but that does not seem to work.

[wide belfry]

Could we have a simple "Continue From This Clip" button without having to hit the ... menu?

[potent crane]

Mentioned this and included it in a ten point list of suggestions that I sent to the co-founder of suno last week, I was told that some of my suggestions were being considered for future versions.

[waxen sand]

nice

[potent crane]

Give me a second, I will post my ten point list here

here it is:
Keenan
As promised (threatened?), I have compiled a selection of possible changes to the app that (in my opinion) would either make things easier for the user or improve the overall experience.

I make these suggestions without the knowledge of a programmer as to how difficult these would be to implement and incorporate into the app.

1. Ability to rename songs.
   Please see the first attachment to see what I am regularly faced with when I have created a song I want to work with. Just long lists of files all with the same name (plus a default number) - if I could change the name of each song creation (or part thereof) I could keep a track of where I am, this is also related to the next one.

2. Dialogue boxes.
   If you look at the 2nd attachment I have roughly drawn where a dialogue box (next to each song or part song) would be immensely useful to add notes as an aid to what each part represents. For example I often have a part song that I am not going to continue (having chosen and alternate part) but I want to use elements (edit in) some of the "failed" creation into the finished song in post. A dialogue box would allow this kind of thing. As it happens another user on the Discord posted a macro (?) that inserted dialogue boxes but unfortunately this vanishes upon reboot.

3. Consistency of song order.
   When in normal mode, my creations are in order of creation, from top to bottom, which is fine, BUT when I switch to library mode the order is reversed, going from bottom to top, this is confusing and makes no sense. They should both be the same (presumably top to bottom would be logical)

4. Bulk delete.
   The introduction of the delete button was great, but should I really have to spend 30 minutes of my life clicking and deleting the masses of dead wood in my library? The option to bulk delete in some way would be fantastic.

5. Folders within libraries.
   I really think there should be the option to create folders within the library, for example you could have "completed songs", "instrumentals" "pop songs", "rock songs" etc, these would be user specific to suit each individuals workflow.

6. Lossless files.
   It really is criminal that you are not offering lossless wav files - even if it was just for premium users (no greater incentive to upgrade - and a strong driver to increase revenue and brand loyalty)

7. Higher resolution pictures.
   The picture created for each song part is a nice touch and as thumbnails they are useful to identify song parts as we can't rename (see #1), but they are of low resolution and are impractical to use as artwork for places such as Spotify. I realise this is a minor point and to be honest when I move things over to Spotify I start from scratch creating artwork.

8. Glossary of terms and styles
   I don't know if you already have one, but a reference list of terms that would aid creation and are known to (usually) work would be useful. Also a list of some of the music styles that should work. I realise that the creation is a fluid process and part of the fun is to experiment with terms and styles, and that AI in this context is, shall we say, unpredictable, and can throw out some really strange sounds, which is great! However, a core list might aid newcomers.

9. Drop down options.
   Again, I know the AI tool can be random, but would it be possible to have drop down list options for, say, male/female vocals, or for prominent instrument (guitar/piano etc), also an option to force an intro before vocals of a number of seconds or bars.

10. BPM
    This one doesn't bother me too much but on the forum this one has been mentioned a few times. I know that in AIVA (which I also use) you can set BPM before you create, might be worth looking into.

So there's ten ideas to consider, I hope you find them useful, and no, I won't be charging a consultancy fee this time!

regards
John Hodgson

(had to post in 3 parts)

[regal hazel]

What the fuck do you think of this???? https://soundcloud.com/kyle-coffman-750408638/hard-scarz?si=240751001f804c83a5c5db6fc6838c37&utm_source=clipboard&utm_medium=text&utm_campaign=social_sharing

I can write an entire album should I be given access to not riding out my max outputs.

This tech is amazing. I mean amazing in the sense

If there are any admins out there, I'm super and I mean SUPER UBER curious as to what your training data is. This is the best thing since sliced bread, as it's put.

I could have worded that better but I'll let it stand. I'm dying to know. This is amazing. Not many quite get it.

[delicate cedar]

Did you mean to post this number?

[regal hazel]

Yes.

[delicate cedar]

What is the number?

[regal hazel]

My personal cell, voip.

[delicate cedar]

For the time being could you please remove the number? Thanks.

Much appreciated.

Thanks.

[regal hazel]

Done. Who can I talk to about this application? I'm dying to know. As a master musician, this is wild to me.

[feral aspen]

The Suno devs do chime in from time to time, typically morning hours EST on weekdays. And if you browse old conversations there are some discussions.

(in discord chat, I mean)

[regal hazel]

OK, so a an individual that is rather naive to being online and talking to strangers, as I only wish to get to the bottom of this tech, I'm not even sure where where to turn to and or ask for advice. Is anyone else here into the music industry? Is the industry even utilizing this tech???

I'm new here guys, cut me some kind of slack and I do appreciate your time.

You can vet me all you'd like. OSINT tools are widely available.

Question now being, do I just jump into Suno's chat in hopes that I can get answers?

[feral aspen]

Two similar models you might want to look at. Facebook's audiocraft or musicgen. And Suno's Bark, the TTS model by this same company (though the training behind that is private as well). There's also some recent music work by Stable Diffusion. Now Suno music is ahead of those but not so far ahead they aren't likely built on many similar principals

[regal hazel]

Thank you, Jonathan. My link above, if you were to listen with headphones (not being an ass), I took a few hours to tailor my outputs. I wasn't aware that SD had an audio feature. Mind you, I'm running everything off of my laptop now, just an i5 Intel. I gets the job done. But all other audio generators are for the meek. I'm not putting myself above anyone else but as a classically trained musician who plays everything but brass, I love metal because of it's complexity, comparable to jazz.

Perhaps I'm just rambling. But I love what Suno is able to do, albeit limited. Thanks guys.

I took everything into Reaper. I then applied EZMIX2.0 to the master along with slight EQ.

[dull sandal]

would love to be able to affect progression of the music via instructions (e.g. [Verse 1: Deep Bass Undercurrent - Mimicking the steady hum of movement, the start and stop of buses navigating through the city's veins] or [Transition: The tempo shifts subtly, capturing the momentary standstill amidst the chaos, akin to a bus caught in a sudden traffic jam]), especially makes sense for purely instrumental compositions.

at the moment Suno seems to mostly ignore such instructions or worse - to sing them aloud

another important feature would be a way to control a duration of such segment (outlined by the instruction), with ability to affect it separately, of course

[waxen sand]

If they ask for more info or you want to send this as well, you can also recommend a feature where if a song was made with free tier, it will have a symbol to indicate so. that makes it easier to avoid accidentally using free generated one while you think its one of your paid ones. Also recommend some way or process so that if you really want the rights to a free generated song in the same was as pro subscription, you can pay an amount thats fair to get the rights for that song generated during non-subscription

[feral aspen]

I echo adding a symbol or even audio ending sound effect or something to indicate free vs paid, for users who don't keep subscribed all the time it's hard to even know what clips are what, currently

[waxen sand]

hopefully we do get a way to pay for a song generated during non-subscription by paying an amount based on credits used + a standard cost due to non subscription (3-4 usd?)

[feral aspen]

a lot of people have suggested a grace period for songs in the free tier, maybe 24 hours, where if you upgrade to pro within the grace period it covers those songs. that would cover most of the times it happens i think, generally new users who are playing around and accidentally generate a clip that happened to be a favorite

[waxen sand]

exactly yeah. since its not a current feature though, i hope for a feature so we can pay for those "favourite" ones, even if it costs 3-5 usd pr song

[feral aspen]

I will say though, when you first start there's a lot of times where you feel like 'oh this is a one of a kind clip, i won't get one as good as this again' but probably you are underestimating how unique and special that one is, and if just keep making songs, you will match or even surpass it in time (especially sinceyou will get better at prompting!)

[waxen sand]

my issue is that one of them became a classic in our group and was generated during free tier while testing suno for the first time right after v2 came out and none out of the like 200+ i did during paid never really replaced it

[feral aspen]

yeah there will still be exceptions. there's no official path to handle stuff like that now but if you email support I think people in the past have usually been able to work something out with Suno folks. however since Suno's recent rise in popularity it may take awhile to get an email response for a special case like that.

[waxen sand]

ok 👍 thanks for the tip!

[shrewd pendant]

Suggestion: Remove credit cost for auto-generated prompts, or remove auto-generated prompts.
I hit the create button not realizing it was going to use an auto-prompt

[safe sand]

BULK DELETE

[kind bridge]

A timeline that we can scrub, highlight sections to "remix lyrics" or insert generations starting from that point" replace sections, with new generations, remove empty spaces so the next generation continues,

[kind bridge]

I wish it would give us the first generation of 10-20 seconds for free so we can see if its what we are looking for and then continuations cost credits

[next coral]

Hi, is there an API for song creation?

[weak elk]

Not at this time.

[coral harness]

a delete all button for entire trash folder

[zenith vine]

I would like for there in the program to be able to extract just vocals, or intrumental so we can use our own ai voices for it, hopefully it gets implemented 🙂

[bronze abyss]

I think adding to the user interface would be a would be a would be a big benefit. For instance separating genre, instruments, BPM.. and making it possible to have more control over these.

Another huge benefit would be some type of editor... For instance if you just wanted to change a word, or one little spot in the song where there's a problem

More supported tags, and better control and reliability of tags... For instance the ability to add guitar solo, piano solo, etc reliably... Also away to get a reliable intro, or outro, to begin or end song smoothly

Also a separate field to identify a male or female vocalist, or if it's in acapella or instrumental

Let's get more control to users that want it! For people that just want to click the button and roll the dice, these features could be hidden behind an advanced dropdown.

A great feature for further down the road would be the ability to choose and preview a vocalist, before generating a song... If you had a dozen or so stock vocalists, that would help users get the sound they're looking for and make it more consistent... I've seen this done on many of the sites that do vocal conversion

Another feature would be the ability to The ability to save styles and vocalists, instrumentation and other settings so you can create additional songs that have the same sound... Which would be amazing for creating entire albums

[graceful hatch]

Midi stems would be amazing

[loud eagle]

any idea how to get the lyrics ive written into the song i seem to be having an issue having the lyrics song on the song

hope everyone is having a great day!

sung on the song

[keen carbon]

Hi friends. How to export a multitrack of a written song.

[cursive widget]

id probably use fadr for that honestly https://fadr.com/stems

[potent crane]

If you mean stems then you can't

[keen carbon]

Thanks you!

[wary hazel]

https://app.suno.ai/song/aa6c5788-9cb7-487e-859a-36264ac6af78

mój tekst, ale suno mi pomogło w wokalu i muzyce ❤️

[slim aspen]

Suggestions: 1. Would love an option to change the name of a song. It would help in keeping things straight. Along with that: 2. Ability to create folders to help with organizing, as well. Right now everything generated is in a jumble. There isn't even a search option.

[feral aspen]

Title/Name Editing and playlists have been mentioned by the devs as coming features which will help a lot with organization #💬┃general-chat message

[wicked pike]

Partner with open ai and get access to their text to video as early as possible for full music video generations. I need this in my life

[peak pasture]

Plz make Paypal an Option to pay for SUNO, so the most in German dont use Credit cards, and i need more Credits!

[slim aspen]

Yay! Thanks!

[tepid trench]

I usually don't bother with a title, and often want instrumentals, which leaves me with various "mp3(5).mp3" and such when I download them. Could be good to have some auto-generated name if one isn't provided. I know songs used to save with a UUID as their name.

[earnest wave]

The Chirp Examples Page on the Bark Github AboutMe currently leads to a 404

[earnest wave]

(btw no clue if this should be in suggestions or bugs)

[restive jewel]

Who knows how to create an album cover

[forest chasm]

I create mine using a local stable diffusion installation, but that is maybe overkill. https://playgroundai.com/ , https://openart.ai/ or https://leonardo.ai/ should all be able to do that easy and free.

[lavish tulip]

Problem:

• personal preference to dislike some genres/themes/adjectives
  Solution:
• a filter 😀

[restive jewel]

Yo that AI is sick ain’t gonna lie

[delicate cedar]

It’s an interesting idea, how would it work in practice?

Would everyone be obligated to include some sort of hashtag? What if someone is posting something they think is country but is obviously rockabilly? How do you know what genres are being used? How do you handle disagreement on what is or isn’t 19th century gangsta rap?

[umbral wasp]

wish you could rename songs after generation! cause getting 2 at once with the same name is weird, and also after hearing the thing you may probably come up with a better name

also would like to be able to generate only 1 song at once instead of 2

[alpine tapir]

We need ios app

[warm cargo]

Instead of a trending, how about a Suno Billboard 100 based on genre in the context of how we understand genres today. Then, an experimental and/or hybridized genre?

[worn atlas]

Here my two cents for the suggestions.

1. The user should be able to keep same melody and music while beeing able to change lyrics and voices. This sounds as a "Must Have" For me
2. Suno should avoid stopping the génération in the middle of a Chorus or a verse. This will make easier the continuation.
3. User should be able to guide the melody by adding chords to the lyrics. Sometimes the suno mélodies are monotoneous . The possibility to add chords will guide more suno in melody génération.
4. It is not necessary to extend the génération time . What is usefull is to always generate the totality of the chorus the verse etc. So the good stopping time is the more important
5. Simplify the song deletion process . A simple button in the same idea as the 'like' button for example will be great. I have to delete too much songs because the continuation does not work correctly, the melody is too monoteneous, or suno goes to hallucination generating words i dont understand

[delicate cedar]

Why are you pinging mods?

The team members are the ones who are working on the models and they do what they can within the technical limits of the models themselves. Do you really think it is intentional behavior to end a clip in the middle of a chorus or a verse? The generation goes on as long as the elements allow within the model's own limits.

[worn atlas]

Just to initiate discussions with suno advanced users. For the clip ending, well, i would prefer the model deletes the "not completed" part instead of letting the end in the middle of a verse or a Chorus. But you are right; i dont know the technical challenges that the team members are facing here. It was a suggestion.

[delicate cedar]

I'm not a team member, so it might even be a good suggestion. They would be the ones who could take it as feedback going forward and do what they can with it if it is one.

I'm more on the issue of you pinging people like this. Yes, I suppose a mod could be seen as an advanced user of sorts. (I really don't see myself as that way but I do try.) But you're talking about something that would need to be coded and which applies to how models work.

Mods have nothing to do with that. We just try to help out here on the Discord and to give suggestions where we might have learned a thing or two as we went along. We try to make the Discord a good place for the community.

All that is to say you're pinging the wrong people.

[worn atlas]

Ok. My apologies for the ping. You are right. The suggestions list is only for dev team not for mods as mods are not necessary team members. I was wrongly thinking that you were suno team members too. Apologies again.

[delicate cedar]

No problem. Glad we got it sorted and hopefully this way the suggestions you made will be seen by those who can actually do something to implement them.

I'm assuming technical limitations but maybe there is a way to bias it towards never ending on a vocal? That would be a boon.

[thorn phoenix]

there are several options for that
I strongly recommend to learn at least a little of photoshop (you can use Photopea online instead, free) to devellop a better result.
for to create images, I use playground, bing, gencraft, ideogram, but you can pick whatever you want

[lilac lake]

models have any codes?

[primal hill]

v

[slim aspen]

Do you know if there is an online roadmap we can see what is in the works?

[rotund orbit]

Hey guys, love the new error message about things being copyrighted, or potentially copyrighted. That's great. Adds clarity.
There's still one huge problem though, and that is that the shorter something is, the more likely it is to trigger the copyright warning.
This is precisely the opposite of what it should be doing

[formal bear]

any examples?

[rotund orbit]

In my latest one, I hit it when I tried to render the chorus by itself using the "release" tag.

I had to change my process to get around the issue

(and hope for good takes)

[formal bear]

ok a word of advise first -- given we block ppl downloading music not generated by them, sharing links is probably better than sharing video 🙂

[rotund orbit]

uh, but that one's mine

anyway, good to know

[formal bear]

yeah but anyone on discord can download it...

we are working on this but not very good at attributing to the original author yet

[rotund orbit]

fair

But yeah, this one pops up intermeittently

the moderation system is very finnicky, in my experience

Also the thing with tags

tags shouldn't count toward moderation

but they do

[formal bear]

they do 😦

you can type F****k unfortunately in the tags 😦

[rotund orbit]

damn, would that actually do anything tho?

I'm more worried that tags, when used in conjunction with short pieces of verse, words like "release" can trigger the plagerism detection

that's what happened to me today

wasn't even in the song, and I'm getting dinged for it

[formal bear]

oh I see what you mean

interesting

🙂 could. you type the prompt that you used that trigger plagerism here? So I can pass it along

[rotund orbit]

oh it's super interesting

lol

I've got a record of it

keep in mind that the system behaves differently for me than everybody else, because I'm special

hehehe

brb

This

[chorus, release]
But there you are on the other side of the world
Doing what you gotta do because you care
Some things are bigger than me an you
And I understand it all, it just ain't fair.

I triple checked it against a third party detection system.

it's original

[formal bear]

k Thank you

[rotund orbit]

np

by the way, I'm super useful for finding stuff like this

🙂

I spend a lot of time on suno

[formal bear]

Again appreciate the bug feedback. Always welcome to share.

I hope and think you will probably spend more time on SUNO 🙂

[rotund orbit]

That's the plan. I am having an absolute blast. By the way. While I have your attention. I want a bigger plan. Support says I can't have one. I need anywhere between 6 and 10 times what I've got on the big plan

[formal bear]

lol ok is this because the top-up pricing doesn't make much sense?

[rotund orbit]

I mean, sigh, I'll pay it. But it would be better if I didn't have to worry about it

Also I want vip features

🙂

[formal bear]

but one way or another you still need to pay ... basically for the GPUs

oh tell me yr wishlist

[rotund orbit]

I don't mind that.

Okay

sure thing

[delicate shadow]

I was shocked that my texts were published by a bot. There may be personal information that Suno makes public. And information that there will be a content leak is very difficult to find. I am sure that not a single user would agree with the automatic publication of his work (ideas and texts).

[hybrid crane]

It’s not. Only if use the public option here on discord. If you use the web version it’s fully up to you. You can then set it public or keep it private.

[dapper trench]

Suggestion: Actually respond to the emails sent to billing :/

[delicate cedar]

Suggestion: A custom tag or some format to tell the AI that this word should be replaced with the bleeping sound. I dunno what would work best. Maybe ++feck++ or something like that? The idea is that when it sees a word with the tagging it will always substitute a bleep sound there. This could give people the type of sound they're looking for because everyone will know what was replaced? It would also be great for humorous value...

[oak echo]

so i think v2 is getter better, when the first days it came out it was crazy good but it slowly died down, it didnt sound good but now it sounds really good. i am really hyped for v3

suggestion: when your done with v3 maybe upgrade bark

[cyan snow]

posted in general so adding here, as a singer who cannot play any instruments or even imitate them well, a feature I hope for

1. We can hum a tune (15-30 or so seconds) & write lyrics, and Suno creates a song based on its understanding (with any corrections needed ofc).
2. Record our vocals and through enough info Suno can make us singing any and every song. Can take 5-15 minutes of singing to work a voice based on our voice, or get better the more vocals we submit.

[waxen sand]

I have also had issues where i have pasted a chorus from a song i have wanted to continue but then it gives that warning so i cant reuse my own chorus... This is also an issue sometimes when pasting in lyrics done in other ai text generators. Also, some words in languages like norwegian are blocked from those warnings. its not even a bad word or something. some artists have names etc where it means something in another language and because of that, i cant even write certain things in my own language

[hybrid crane]

There should be a way to report an error. Now it just spits out a reason (or just an error), but we cannot report fals positives.

[forest chasm]

Already way better than the generic message before. I am glad they changed it, helped me a lot to solve my issues faster. Are you sure about the false positive? Sometimes it's hard to see your own mistakes.

[hybrid crane]

Yes I am sure 😉
I was burning credits on an idea which and it starting maken errors when I added [chorus build].. when I removed it, the error was gone. When I add it back, the error came back.

So that is clearly a false positive.. I love to puzzle if it gives an error, but false positives should be reported as feedback

[potent crane]

The trash option is a mess. I accidentally trashed a part I wanted, first sweep of the trash showed me 20 pages, it wasn't there, came out and back in this time the trash expanded to 105 pages, found it on page 103 - surely it should be first in last out, a recently trashed file should be top of the pile! And it's not searchable, it's not in any order, it's totally random.

[gritty owl]

Is there API access for the song generation model?

nm scrolled up and found the answer lool

[torn stone]

Yes, or something like a "clear all trash" button, things get disorganized super quickly when you make lots of iterations to find the best version

[real root]

An "update this song" kind of button that could switch the voice model used on a generation. Like Remix, but without changing the melody. So it would be a voice model specific remix button.

[waxen sand]

it happened again... i cant even type in norwegian without it thinking its an artist name or a bad word. i literally just wanted a song about a mountain in norway in norwegian...

[night agate]

hello, is there possible add some option for mantaint the 1st clip configuration? , actually im trying to make somethin similar with [] elements, but new ones are generated in the others clips, not the same 😦

[vivid magnet]

Most of the songs that are generated including most of the songs that are in top on the website sound very robotic / electronic. We should be able to tell the bot we want a clear, human voice and generate it so. It should generate electronic voices only if we ask it to.

[flat spruce]

Suggestion: the ability to make duet songs (both female and male voices in one song)

[sick hollow]

https://github.com/suno-ai/bark/discussions/523

[trail silo]

Would love a fast forward button when listening through a clip (2x speed)

[old quiver]

do the thumbs up and thumbs down give information to the am model about whether the song given has successfully met the criteria that was put in the lyric and style prompts? if they don't they should

[stuck oxide]

Nasıl güzel şarkı yapıyoruz taktik var mı

[hybrid crane]

Yanlış kanaldasınız, bu kanal geliştiriciler için öneriler içindir. Ayrıca, İngilizce iletişim kurmanızı öneririm. Eğer bu konuda zorlanıyorsanız, ChatGPT'yi kullanabilirsiniz. Son olarak, sorularınız için #💬┃general-chat kanalına gitmenizi tavsiye ederim.

[cyan snow]

Suggestion:

A database of hundreds/thousands voices that can be triggered in an optional Performer box, similar to the Style of Music box (with the option to use Random Performer). These voices would not have a real life person name but it would have an AI name.

Why? With current creation of songs, people cannot create song after song with a voice they enjoy, they have to rely on extreme randomness. If someone wants to create a music album consisting of multiple songs in the same genre, or even consisting multiple genres, they'd want a similar/same voice feeling throughout. With a working database, or a way to combine specific voices, this can be achieved.

Thank you.

Also would work with multiple voices, creating a unique band etc..

[rain iron]

Suggestion:

1. Ability to choose to match the length of the first clip or not (when a song has more of an introduction) or to specify the max length of the continuation (up to the max of the first clip). When one generates a 1:20 long clip in a [verse] and [chorus] pattern without much of an [intro], and when Suno then matches the continuation against a 1 minute clip using the same style and cadence, it tends to cutoff the shorter continuation clips. In a [Verse], [Chorus] pattern, this cuts off the chorus and then the chorus ends up being hard to fully replicate and this pattern persists for all subsequent continuations.

2. Ability to indicate the presence of an introduction. When continuing, the continuation clip would start where the verse kicks in.

[flat spruce]

Suggestion:

A feature where free users and paid users can pay to get the rights to songs made using free version bc currently you can't even if you become a paid user. Makes sense right? If you pay to get the rights to future stuff you create, shouldn't there be something to get rights to songs made with free otherwise Suno will own them even if you get a subscription.

(note: us users will actually pay to use for our uses and Suno will have more revenue... win-win scenario)

[lost quartz]

I'm writing original lyrics and it keeps complaining about copyright. I google the song lyrics and find zero songs that use the same lyrics. I ask Chat GPT what song it's infringing copyright on and every suggestion is wildly wrong. I spend like 10 minutes rewriting anything that could be the issue, then click generate, and it still says copyright. What is the issue? Tell me and I'll rewrite it in an instant. Don't just say "Error" that's extremely difficult to fault find. You need to tell us what the offending words are so we can change them. If you don't, numerous people will quit using it because they have no idea how to fault find. What are the rules for fair use here? Can I not write "I love you" as a line because it's used in some other song? When thousands of other songs have it, it's not copyright. If it's in the public domain, we can use it. Please fix this because it's a hassle.

[hybrid crane]

When you are a "paid" user you own the copyright. I quote from their FAQ: ***Who owns the songs I generate using Suno?

If you are a paying subscriber to Suno, then you own the songs you generate while subscribed to Pro or Premier, subject to your compliance with Suno’s Terms of Service.

If you are using a free version of Suno, we retain ownership of the songs you generate, but you are allowed to use those songs for non-commercial purposes, subject to your compliance with Suno’s Terms of Service. ***

[worn atlas]

It seems that there is no feedback for all issues/suggestions posted here. That does not mean " no one looks at them" but it looks like.
A roadmap on which prioriterized features appear would be welcome. You may add a features request or issues reporting , tool on the website to better track features requests or issues reports.

[bronze citrus]

I tried several times today, but the lyrics don't come out, only the music. I just ran out of credits. I speak Korean. Sometimes the lyrics come out, but more often than not they don't.

[daring pagoda]

Make suno free without credits please.

[raw cave]

Bubbling up...down?
#💡┃suggestions-web message

https://app.suno.ai/song/bbafce90-e05f-4fa4-bbe9-fcfa11af0ac0

[worn atlas]

https://tenor.com/view/leonardo-dicaprio-clapping-clap-applause-amazing-gif-16078907558888063471

Great suggestions

[compact lark]

💡 allow to write own lyrics for v-day.suno.ai

[compact lark]

https://tenor.com/view/leonardo-dicaprio-clapping-clap-applause-amazing-gif-16078907558888063471

[formal bear]

plz just wait for v3

[compact lark]

in 2025 year ? 🙂

[formal bear]

come on...higher expectations for our hard work? 😄

[raw cave]

https://app.suno.ai/song/2074c0f1-6f63-46de-9d11-4ea193f8abef

[formal bear]

you got it

[strange barn]

I haven't had as many problems as you, but it's brand new tech, so we should expect a certain degree of wonkiness...that being said, I've found it to err a bit too much on the side of 'copyright safety'...Peace.

[brittle birch]

Be able to go back to previously finished songs and input a different language

[wraith sigil]

Please! Accept Paypal!
Take my money for god's sake!!

[twilit hedge]

I'm also curious... did it complain about certain phrase or phrases? title? or the whole exact lyric? I find it's interesting that it could claim words or phrases... but I'm not copyright professional and still new with Suno

[zenith vine]

theres a bug that keeps repeating verse 2. when im trying to add a bridge and final chorus

[twilit hedge]

it happened to me once but I ran into another issue (not sure if a bug or not) but it kept generating in different language that is not in my lyric... wasting me some credits to re-generate!

[languid phoenix]

♥️ Button - to “Fav” tracks on top of 👍 “Like”

[twilit hedge]

Suggestion: Please Add Folder in Library for better clip management!

[hybrid crane]

You can make playlists. Did you know that?

[twilit hedge]

No I didn't know that.. How? I see no option

[twilit hedge]

hmm

This is all I see

I'm just on Pro plan... maybe that?

[hybrid crane]

Ah yes.. I guess so 😉

https://tenor.com/view/austin-powers-dr-evil-mike-myers-come-join-us-come-here-gif-4627161

[kind bridge]

v3 is taking liberties on the wrong parts, especially when continuing from a song that already has an established chorus, and wanting it to be sung in that style each time the chorus is sung

I wish you could input artist names and hit "generate style" and it would generate a style based on the artists you enter, and replace their names

[hybrid crane]

Problematic due to copyright laws

[kind bridge]

the way around the copyright laws is you are asking it to detect the style of music and not the artist

so it wouldn't have the artist name associated with the final clip, it would be like "hey what is the style of music of this artist, and then replace it with genre or style it comes up with]

like a search engine

artists in real life are inspired by others when they write their own unique music , and will often say yeah these are my favorite bands ect..

[quick olive]

fix the "prompt likely copyrighted" issue. I've instructed chatGPT to make lyrics similar to taylor swift and it swapped out all the words, yet it's blocked...

'Cause the critics gonna judge, judge, judge, judge, judge
And the cynics gonna scoff, scoff, scoff, scoff, scoff
Darling, I’m just gonna dance, dance, dance, dance, dance
Dance it off, I dance it off

Dreamers gonna dream, dream, dream, dream, dream
And the makers gonna make, make, make, make, make
Honey, I'm just gonna dance, dance, dance, dance, dance
I dance it off, I dance it off```

[lean island]

PayPal as Payment Method 🙂

[brazen arch]

i would love suno to have paypal or ideal as a payment method.

at least paypal

[trail silo]

Be able to include how long the clip should be. Sometimes I say short outro and itll end the song, but sometimes it wont and will adda new minute long clip

[spiral hornet]

Suno is great, thanks for your hard work devs. It would be great to be able to change the lyrics of existing songs, but still keep the same melody and/or music.

Also, I'm sure it's been asked before but it would be awesome to be able to upsample our existing stuff. Probably not possible but thought I would ask.

[prime rock]

I love the quality of music in v3 but it doesn’t really listen to the prompt, needs to pay more attention to tags in the lyric field in particular. V2 was very good at making choruses sound like songs and v3 doesn’t do that yet.

[spiral hornet]

I second this. Having run a small software business before I quickly realized that many people in Europe will only use PayPal. You are missing out on revenue if you don't include PayPal.

[lean island]

coming from the same region 😉 both geoghraphically and jobwise...

[rapid sonnet]

a feature to choose how loud you want the vocals and instrumentals to be.. with metal the instrumentals are always insanly loud

like a slider.

[woven cradle]

Small feature request: Show song duration in right side panel

[mossy stirrup]

Would be great to have it use multiple singers (eg: rap battle)

[lean island]

check the docs, that has been possible before - might not work on every generation, but i did this on v2 already... [Male] Blabla bla [Female] Bla bla

[mossy stirrup]

Ty

[lean island]

with pleasure 🙂

u might want to check the community wiki, it has many [Tags] in there

Curry Spirit check your DMs, i send you the link

[mossy stirrup]

Workers Ty!

[mossy stirrup]

Oops I accidentally killed this - can you send again?

[lean island]

done 🙂

oh my goodness... those deep techy house sounds v3 creates 😄

[sour raft]

I also want control of how long (minimum/maximum) a generation is

[languid phoenix]

🖼️ Artwork Idea #1: for INSTRUMENTAL (when no lyrics)

have Cover Art default to draw based from Style field rather than birds

🖼️ Artwork Idea #2:

👉 Download > Image (option below "Video, Audio")

[pine plover]

I would like to see what they use our own voice for music

[acoustic pike]

How can we use our original voice in the song creation?
If AI makes Elon Musk voice song then why not we use our own voice in the song?
Please add this option.

[pine plover]

I have suggested it several times

[acoustic pike]

So please add it for all Singers or Non Singers.

[acoustic pike]

If we sing a original song and add our voice and lyrics on Suno.AI, so AI create music by our vocal Chords it that possible?
with our instrument choice?

[twilit hedge]

Already made us a songwriter.. now make us a singer!! 😀

[sage mirage]

Anybody suggested a negative prompt yet?

Let me tell Suno what NOT to do

[acoustic pike]

When i want female voice it makes Male voice and when i want Male voice it makes female voice. Please fix it.

[urban sleet]

Can you PLEASE turn off by default this s***y button, so we won't have to disable it every time we reload the page?

[sage mirage]

A little harsh but uh yeah, at least let it be an option we can set ourselves.

[urban sleet]

Sorry, didn't mean to be harsh. But it is reeeeally annoying.
I can't imagine anyone wanting this to be on by default.

[solid snow]

I would love an option to exctract the instrumentals on the song so you could sing them yourself

[wide belfry]

Is 👍 or 👎 really enough? Should be a checklist of reasons when we 👎 so y'all have a better idea of why...

[stone solar]

i'm not sure this is a "bad thing", but continued songs will go quiet after the song "ends" for about 5-10 seconds, and then decide its time to start a new song... so you get hooked into more continues

[lean island]

when u can't choose which continuation is better - first world problems

[sage mirage]

Continue both an see where the two streams go

[lean island]

i already have like 5 brances going from that track 😄

[sage mirage]

Hey that's what I do

[lean island]

we need better overview to check & find parent/child

[sage mirage]

Yeah really

Trees, give me trees

family tree

[lean island]

thats what keeping me back the most actually

[sage mirage]

Same, organization

[twilit hedge]

Have you tried this? it might work! https://vocalremover.org/

[solid snow]

I have and i am a patreon there from before. 😉 But it would be nice too have the option here, as it can be a wee distorted :)¨

[twilit hedge]

I also used RipX, works well

[solid snow]

Ouuuuh i will check that out 😉

[spiral hornet]

Want to be able to rename tracks

[tight stone]

Is it possible to change the voice for each part within one prompt？？

[sage mirage]

Oh yeah look at that

[sage mirage]

They are right, it's in the ... options now

[snow coral]

I hope in the future it will be possible to customize the voice more flexibly.
Clearly specify male or female, separate which lines in the text are sung in which voice (if it is a duet, for example)

[flat spruce]

But what I'm suggesting could be a new clause in the ToS which is actually quite good

[oblique dagger]

Being able to place our own instrumental section and generate a "consistant" AI Vocals, The Vocal voices can be generated separately to our own liking and can be repeatedly used in our own songs

[hollow locust]

Being able to adjust meter and rhythm of the lyrics.

[twilit hedge]

Being able to reuse the voice we like that sing our previous song... would be really really nice!

[tight stone]

how to specify

[oblique dagger]

Yes! certainly would be. Also, like I mentioned, being able to import our own high quality instrumental sections and just let the AI vocallist sing our lyrics, the Holly Grail of music generation.

[twilit hedge]

I've been looking for that AI... give it the "karaoke" version, and it will sing it!

[oblique dagger]

How about opening a vocalist marketplace, generated by users (somehow..) each vocalist can have a style specialty, and can also be tweaked further to suit our own instrumentals or AI generated instrumentals. The important thing is replicability and reusability

Check this out https://dreamtonics.com/synthesizerv/ it is good, however still cumbersome to use

[twilit hedge]

look promising.. did you try it?

wonder if they support multi language

[misty blade]

Please, please add a search capability to the library page! I get lost among all the creations, and when I need to go back to something old I did, it's about scrolling through dozens of pages and it's very uncomfortable. 🙏

[modest prawn]

Give simple prompts an Intro to all songs created for V2 and V3 (Alpha)

or at least to have the option as a button somewhere to add an intro into those simple prompts

[steep jackal]

With LLMs, there is temperature setting that dictates level of control / creativity. Currently, there needs to be something similar as it pertains to a source clip. For instance, once I find a clip I like and try to continue from it, I'm looking for repetition in sound and cadence -- not lovely joyful creativity by a bot gone rogue -- such creativity eats up credits ... but it also ruins a promising beginning. To be truly useful, the temperature setting should be adjustable from clip to clip because the song should progress and evolve.

[kind bridge]

Id rather just have it focus on the music especially if it would save us credits to not have any artwork at all, that's doing too much.. I find i am just blowing through credits trying to get it perfect 5 pages of downvotes spending like 3000 credits just to finally get a full song

[lean island]

the artwork doesnt cost much resources in that size , especially not compared to the audio 😉

the videos arent resource intense either

it wouldnt' save you credits really, nobody says if theres no artwork, they have to charge less, i think for the resource usage, the prices are super fair.

[prime rock]

Please make it so when you remix from something it defaults to whatever version that song was in, I keep continuing and forgetting to reset the stupid pull down menu every time and wasting credits since it reverts to v2

[autumn pilot]

Cool idea that might be interesting.

Song of the day :
Suno will greet the user daily with a custom song ( free, no credits needed) , it will be a mixture of the styles the user has been generating the most often. "Hey ! Suno just made you this song. We thought you might like it." kinda thing.

[timber bay]

The "Instrumental" feature does not work as expected. It appears just as likely to make hallocinated lyrics. You train a separate model for strictly instrumental, but still permit for writing lyrics, since I always do [tagging], even for my instrumental songs.

[prisma vault]

I can see there are many interesting suggestions for the AI side of the project. I, however, would like to suggest a few quality of life improvement on the web app. I am quite the "heavy" user, in the sense that I generate a lot and I'm very selective. I feel like this is however a very common use case.

• Please add a trash button next to the share, thumb up and thumb down. AI generation is a process of trial and error, trashing is a very common action. It should be the most reachable thing imho.
• Make the rename function "activate" when you click on a song title (or the little pencil appear when you hover a title). Again, tracking your creations is very important when you go heavy on generation.
• Please move the "action notifications" somewhere else. They always pop up on top of the player, and if you are skimming and removing stuff for instance, you have to wait for the popup to disappear. It's a small thing, yet, a bit frustrating.
• Folders/tags. Add your own custom tags to songs, so that they are easier to fetch and organize.

Improving the model is without a doubt the core activity, however, making the interface usable would also make "professional" users happier and thus more prone to generation/credit purchasing. Anyhow, thank you so much for this awesome tool.

[dry robin]

i create alot of songs using this, heavy metal songs generally, one thing i have noticed is that 9 times out of 10 the vocal starts within a few seconds 9 times out of 10 it would be really great if there was a prompt that ensured there was at least a 30 secondf intro to the song before the vocals start, for example i have created around 15 generations of the same song and on everyone of them the vocals start straight away; that would be my only bug bare to be honest. burn alot of good ideas and creds because i want an intro

tried this but doesnt work

i think some basic commands would be really good so the AI knows where your trying to take the song, i.e. (intro) (verse) (chorus) (guitar solo) (bridge) (middle 8) and (end song) telling the AI this would massivley benefit the song writing process

[worn atlas]

What do you mean?
Are you requesting something différent from the classical metatags [Verse] [Chorus] etc

[dry robin]

i rarley get an intro, when i put [intro] if i put [chorus} it sometimes builds but sometimes it treats it as a verse [bridge] works [end song] never works [guitar solo]rarley works in my experience ive done 12000 creds in 5 weeks of playing with v2 and now playing with v3, dont get me wrong ive completed an album of 13 songs with it and i love it, just seems the structure is hit and miss

[sick osprey]

could ypu make it so that the songs are longer?

[lean solar]

yes with v3 now there longer

[sick osprey]

thanks

[lean solar]

if you dont put lyrics it makes a starting part (works for me with hardstyle)

[dry robin]

will try thanks

that just puts random vocals on lol

[weak elk]

How about a v3-specific chat channel? thanks

[mortal anvil]

some feedback:

1. Songs tend to drift and repeat the lyrics towards the end especially when using the continue feature.
2. I'd like to see more variety in singers for genres, they can all sound very similar in specific genres, it's not entirely a bad thing if people want to make a consistent album, but it does sacrifice uniqueness.
3. lyrics, I've started writing my own as I am so tired with the generated ones that always seem to mention "echoes, night, music and numerous other cliches" i'd like to hear lyrics from at text AI that has been trained on lyrics as chatGPT tends to be a little to poetic in a cringey way

[opaque cradle]

Please make it adhere more to structure prompts in the lyrics. It is IMPOSSIBLE to make songs when the singer consistently ignores bridge, pre-chorus and chorus directions.

[thorn comet]

We need super clean Stem Seperations for every project we render

We need add only lyrics to an exisiting generated beat we made with suno.

mayby at the end like a mini DAW playlist would be cool that we can build the parts together and export them we püroducers wants to see everything visual

[cloud acorn]

please please pleaaaase Suno team, two things that would make suno perfect :

• make that on V3 we can have separate download links for music and voice
• User can re-use seed number to "re-do" or keep a good roll
  that would be just amazing

[solar bolt]

Make it so it can make video game boss songs

[hybrid crane]

re-use seed? Just hit continue and fill in 0:01 sec

[spiral hornet]

Would like to have additional prompts, like [short pause] and [1/2 step up]. Also some way to go back and add small tweaks that I can hear in my head, but have no way to realize on suno

[cloud acorn]

imagine that you want to re-start the song from "intro" to "outro" with other lyrics without loosing the whole style/génération

with "continue" you keep "connections" with the previous parts

[fickle blaze]

The free coins are great (thanks for that), but wouldn’t it be better to give (a lot) more coins to users who are pretty active? You could require, that those key users must provide a textual feedback of their experiences to get things fine-tuned.

Maybe that they participate in your faq wiki

BTW v3 seems to be a lot better in Quality 👍

Thanks for that !

[scenic remnant]

I love the app on suno.ai (i dont know what chirp is)

While using the "use random style button" I noticed a pattern where you combine a list of prefixes and genres, may I suggest 2 dropdown boxes so we can see all possibilities?

[urban sleet]

Concerning V3 - I don't like it. I've used about 700 credits on it till now. The overall quality is great, the vocals sounds superb, but it just doesn't care for what I want. More than half of the times it totaly ignores the Verse/Chorus division, it sings however it likes to do that - usually in the wrong way - like lines in verse does not matter (it divides them in strange ways, such that they no longer rhymes), hallucinations are just crazy all over the place, IT DOES NOT SING MY LYRICS but repeats something that was sung before. Continuations are mostly failing, it sings whatever it likes instead of what I want. It makes pretty music, just not at all what I wanted.

[weak elk]

To be fair, v2 has been like that for a while too

[cloud wadi]

I turn instrumental version for the version 3 and it still made up lyrics that or complete nonsense and sounded like a different

[hybrid crane]

probably depends on the genre you are using.. try something simple like piano for starters...

[cloud wadi]

90s pop

[shrewd pendant]

I'm using V3 and I'm trying to do ~2 minutes of lyrics, but I'm finding the input box to not allow enough characters... is this a limitation of the AI or the web interface? If it's the latter, could we get a bigger input box?

Also I noticed when using lyrics it abruptly cuts off at the end of the lyrics even if there is time remaining before the 2mins cap.

[wraith sigil]

Please add Paypal for a payment method!
Please take my money!!!

[hybrid crane]

You can also get a prepaid credit card. Just a suggestion

[dense gulch]

Screen readers no longer see the option to download a song using chrome and Firefox on PC (those are the browsers I've tested. From what I can tell on IOS Safari, it's due to download being behind a separater element. If this could be fixed ASAP for PC screen reader users, that'd be great. If you want to test, there is a free screen reader at nvaccess.org

Maybe put a separate download button that opens a menu for video and audio after the three dots button?

[obtuse tree]

should be fixed by Monday, ty for spotting this

[dense gulch]

You're awesome!

[obtuse tree]

thank @solar igloo for this one

[short blaze]

Simple Suggestion:
Make suno pronounce "ai" as "a i" since we know AI is pronounced "aye eye".

[hybrid crane]

I just always type "A I".. simple 😛

[short blaze]

i'm talking about when you use ai to generate the lyrics and it generates the word "ai"

[hybrid crane]

Ah you use the automatic lyric creator.. Yeah I get that

[bold cedar]

Suno in the third version is much better but... It still has a long way to go to be as good as the second version, it seems to me that even the second version has deteriorated a bit

[dense gulch]

Appreciate you tracking this down. Curious why it's not seen by screen readersthough in case this happens on other sites that I have to let people know about.

@hybrid crane thanks for pointing me here as well.

[hybrid crane]

Very welcome 😄

[formal bear]

I appreciate the feedback -- see the announcement You may need to prompt V3 Alpha differently than V2 for best results. I don't think your previous favorite prompt technique would work out of the box -- and that probably shouldn't be...if we want to make sth better than before. I think lyrics / choice of words / song structure matter a lot more in v3.

[formal bear]

your most direct /helpful feedback is to rate your generations -- if you can rate all of them. Be picky on your likes, and be generous on your dislikes.

[frigid timber]

Suggestion: For when we return to the site to keep working on a song, currently have to track down which generation a part was continued from, if it could somehow be shown in the song generation itself would be super helpful and possibly tacked on when clicking the "Remix" button. Possibly here:

[hexed wharf]

i wish the library can be more useful, such as the ability to create folders or playlist.
the instrumental feature is very nice. my suggestion is that we can have option to control the flow of the music. for example, a certain duration like 15 seconds, there will be particular musical arrangement, and the next 15 seconds would be a variation or addition of certain element of musical arrangement and so on.

[brazen arch]

Paypallery

https://tenor.com/view/paypal-logo-transition-gif-16332863

[formal bear]

you can continue from anywhere right? Just pick the time

[marble vessel]

When I choose to continue from a generated song, why doesn't it create the next lyrics? For example, in the first one, if it created a Verse and a Chorus when I select continue, it doesn't create the next step of the song and shows in the right sidebar. The random lyrics button creates something different in English rather than in the language of the song of the song you continue from. I'm trying this with a Hindi song and not helping at all.

[formal bear]

are you on v3?

v2 has this issue for now

[twilit hedge]

I also experience this on V3, it often generated in English even though the previous part isn't

[icy brook]

that would be cool if stem can be separated for mixing purpose, like
Drums, bass, accompaniment, vocal and be downloaded that way
On FL Studio we have an AI tool for that but would be cool to have the raw audio for mixing

[formal bear]

k if you send me the clip you try to continue from, the bugged clip, and when you try to continue from that will be helpful -- all links ofc

[twilit hedge]

I did not generate the song though... when I clicked Generated Lyric and if it was wrong language.. I did not follow through with it

[formal bear]

sry not sure I understand what you were doing?

[rotund orbit]

Observation: By default, both of the suno models take themselves very seriously, and like to go slow on complex verses.
What if you did the opposite, and made the default faster and more energetic?

[formal bear]

heh? what do you mean by making the default faster and more energetic...?

You can prompt the style however you want right?

[rotund orbit]

Doesn't work when the verse calculator is broken

I mean

Not broken

just confused

[formal bear]

what's a verse calculator....

[rotund orbit]

It gets weird when you throw complicated lyrics at it

[formal bear]

honestly asking

[rotund orbit]

That's the change from v2 that tries to do the math on the total number of syllables in a line

[formal bear]

O.o

this is the first time I heard this concept...

[rotund orbit]

uh huh

I've been having fun stretching and bending it

The v1 model calculated verses with a less rigid structure

[formal bear]

I think the model should understand the lyrics and try to stress it itself now

lyrics's meaning aslo affect the representation which is very interesting

[rotund orbit]

totally

super interesting the way this seems to work

[formal bear]

I'd not over engineering on the format, but more on the content, overall genre

[rotund orbit]

it really depends

Like for example

it likes consistent oscelating meters

across the board

10,12,10,12

and so on

It has trouble with 10,15,10,20,8

at least in v2 and onward

and to make it work, it stretches it out

but you could do the opposite

[formal bear]

very interesting observation...

let me think about this 🤔 ....

[rotund orbit]

Thinking about it tonight, because I'm trying to write some fun underground rap to challenge myself and the machine

which means a lot of weird flow and meter tricks

It's hard to write this stuff

[formal bear]

line breaks, commas, should help the flow in general, but not guaranteed

[rotund orbit]

It does a little better with clear notation

but it still calculates the whole before the lines

[formal bear]

yeah 🙂

[rotund orbit]

Also: dumb question

why does it sometimes calculate in background vocals as part of the main text body, and sometimes not?

Also super interesting

sometimes, backups count toward the music

sometimes, they count towards the words

is there a way to pin that down?

One thing I've been doing lately, is not using backups at all, but writing the lyrics in a way that the system assumes they need to be there

which actually helps

[formal bear]

yep

I think v3 automatically adds the art effects (not consistently) but it can

[rotund orbit]

makes sense

[formal bear]

not dumb question at all

[rotund orbit]

v2 does some of it too

[formal bear]

I mean to sing well, it needs to understand the whole song first to get the emotion, tempo, fit, structure etc right

like...a real composer

[rotund orbit]

which makes sense

it just sometimes goes a little sideways on the details

not to be critical

[formal bear]

no

I get it

[rotund orbit]

It's the best machine of it's kind

(that's why I'm here)

[formal bear]

the defects are why we think this is still alpha

these feedbacks will help us to get it better

I mean I don't expect v3 full will be perfect...but hopefully better 😄

[rotund orbit]

I am really impressed with it so far. I mean, overall, and understanding that's it's an alpha build

I like what you did with the level changes and panning in v3

makes it much clearer

[urban sleet]

If I write lyrics and it just sings something different then there's really little I can do...

[rotund orbit]

speaking of, there's this thing v3 is doing where it'll get confused and repeat sections of lyrics, like it's combining multiple takes, especially around the beginning of the track

[formal bear]

yeah hallucinations happen

[rotund orbit]

fair, but these are an entirely new class of hallucination

[formal bear]

I know 🙂
as if it thinks your lyrics should be sing in other ways...

I don't love this but there is continue from...you can break it from anywhere and redo it

[urban sleet]

Yeah. I have a track that I tried to continue like 20 times and it just said "screw you, I'll repeat previous verse".

[formal bear]

ok send me the original link, and an continue example link?

[rotund orbit]

It's more editing, but sometimes it's worth it.

[urban sleet]

Happened with continue too. I even tried different moments to start from.

[rotund orbit]

oh really?

I love how quirky the machine is

🙂

it's fun to work with

[urban sleet]

I did send it, from the track itself, as a bug with descriptions when it happened.

[formal bear]

ok thx did you try to continue from other time stamps?

[urban sleet]

I try to send bugs when it behaves odd with descriptions so that it might help somehow maybe.

Yeah. Like 2:50, 3:00, 3:01, 3:02, 3:03, 3:05 multiple times.

Also got very similar behavior from other tracks too.

[formal bear]

k I am getting offline...if you can find the link for the clips...could you DM me them? Thx

[urban sleet]

Ok.

[autumn pilot]

Yeah I agree, the actual music generation is incredible , props to the engineering team!!
But the "continue song " feature could use some variation, it feels like wasting tokens when most of the continuations feel like the same track. Adjusting the prompt does work a lot better than V2 though.

[hollow locust]

Microphone input. Why? So after you've heard the song and notice some lyrics don't fall on the beat quite right, you can sing in how you want the flow of the lyrics to go. You could call it "Lyric Flow" (or something like that).

[lofty wraith]

if you want alot more MONEY, this is what you do. theres 3 things that will make you the top dog #1 allow extraction of only vocals, people use them for references to hire real singers, or simply use it for there own DAW to make a song of there own. but extracting the stems manually messes everything up, and somtimes leaves a lingering beat. this would be a massive upgrade #2 and this ones A MUST HAVE GAME CHANGER... we often generate songs with a ai generated text from somewhere else that is usually corny. but its good for testing waters.. without waiting days coming up with our own lyrics. BUT i know this happens to everyone we come across a MASTERPEICE of the way it is sang with the tone/melody ect. but now we dont like a few words and want to switch it up.. simply contining song will never have the same result the beat may have dropped already or the singer changed tone for middle of song. WE NEED A REDO type of thing. where we click redo and can grab exactly how the tone of that song was but change the lyrics in certian parts if not all of them. but have it sang like the masterpeice we generated !! cause we will never generate another like it again. #3 allow a export of the Vocals midi of other instruments. the vocals will allow to make our own progression and melodys alot easier, then dumping into piano roll and trying to fix it, or simply extracting the drum loop ect.. would be amazing.

also be able to (Verse 1 - Girl) (Verse 2 - boy) would be cool.

[rigid eagle]

I came here to suggest the ability to upload our own .mp3 files, and then being able to use the "continue" feature from Suno. Is it likely that we will have this soon?

Better yet, and continuing on the above comment, it would be great if we could select ranges of any .mp3, then add some prompts, and then ask the AI to re-generate only those ranges.

I think this would be a game changer in the music industry, the above is probably the only thing preventing me from generating top-quality music with Suno (after lots of tries, but top-quality regardless!).

[autumn pilot]

Guys PLEASE add a volume slider on the audio player, this would be specifically great for desktop . Playlists would be killer too but don't know if you may be already on that 😉

[worn atlas]

V3 : Better sound, better continuation concernant starting time.
But more hallucinations
Continuation is "credits wasting".
Various suggestions have already been made here. The user does not necessary need changes when doing continuation. Most of the time it become worse. Just add parameters like the temperature parameter for LLM to control variability. 0 will mean "repeat exactly the previous part" . Song melody and the music could be separetely controlled.
In this way we can just modify the lyrics for the continuation part and keep same melody for the chorus and Verse.
Another request already given is the ability of editing the lyrics for the first part. When after lot of trials , we finaly find the good music and unfortunately there are mistakes in the lyrics it is impossible to fix. This feature should be a "Must Have"

[wanton roost]

I get worse results with V3 alpha. Feels like it doesnt respect [Chorus] anymore. V2 was almot perfect with finnish language, but now seems like it has gotten waaay worse. Music is also maybe clearer, but more flat from finish to start. About lyrics V2 was really amazing with rhyming, but V3 feels to be lost cause with that.

I have tried with lyrics that gave me good results with V2 many times, but not getting even one good result with V3.

Seems like studying the metatags and style of music combo starts from zero with V3.

[hybrid crane]

When the song is "finished" it would be great to edit the lyrics. Sometimes parts are double due to continuing.

[thorn comet]

I would be super happy if we can sort the generated AI parts in folders and make new folder like EDM Song, Rnb Song etc. Would be more
clearer also a search functionw ould be dope

[tidal wharf]

Let’s thank the team that they didn’t just switch the versions and are giving us options!

[hollow locust]

V3 generates longer songs, so more characters for input would be appreciated. It's 1000 right now, I believe. An extra 500 characters would go a long, long way.

[woven merlin]

is it me, but are there results with V3 creatively less varying? I mean if you repeat the same prompt and tags, earlier results were sometimes very surprising.

[potent crane]

folders HAVE to come in, it's sorely needed

[hybrid crane]

I am sure they are tweaking because of our feedback..

[snow blaze]

I love the quality of the songs, interpretations and vocals in the new V3 Alpha, but the biggest problem is the continuation of the song. I enter a new text into a new part, and the program ignores it and sings either from a random place or returns to the text that should not be in this part.

[tidal wharf]

My current feedback: v3 results seem to be less variative and more “modern pop”-like

From melismas to song structures to sound

[spiral hornet]

Could you please provide us a setting to default to v3 if we want to? It keeps defaulting back to v2 and I keep forgetting to change it

[formal bear]

k thx for your reports -- I think I fixed a continue bug and that should be deployed now; you should have better experience continue off long clips now.

[tidal wharf]

hey, how do I "continue from anywhere" with v3, as announced in #announcements?

[sage mirage]

It's kinda hard to see but

[sage mirage]

When you 'Continue From' you can input the time here

[tidal wharf]

found it, thanks, majestic!

[twilit hedge]

as you continue, there's a field where you can type in 00:32 if you want to continue from 00:32 seconds mark from the clip you continue from

[tidal wharf]

this is a game changer. and being able to do this from v2 is just wow

@rose seal you are the best!

[warped basalt]

I use [Verse] [Chorus] and [Bridge] in brackets like that to try instructing the music. (V2 till now.) Seems to make some sort of difference. But have wanted some sort of standard like this so it knows to draw from earlier parts of the song for continuity.

My experience with V3 so far ... it's not understanding line breaks and rhyme schemes the way V2 did. So you don't get the natural and rhyming version of the song.

[tepid trench]

Please can it be made clear on the subscription page that any unused "monthly credits" do not carry over to the next month, as although it works the way I expected, "x credits/month" could be open to interpretation based on the user's experiences with other subscription-based services (e.g. mobile/cell data plans.) This doesn't seem to be covered in the Terms of Service page either.

[unborn token]

V3 needs to have longer memory. music and chorus forgotten with 1 part gap is very limiting. I think it's the main issue right now. it can continue a song from 2 months ago from 0:10 but can't repeat a chorus from 5 minutes ago if there is a gap of 1 part (unless you are super lucky}

[potent crane]

would LOVE to have a "no bloody auto tune effect please" as a valid instruction.

[velvet cedar]

💡Make a filter for "FULL SONG" on the library page. This way it is easy to filter for full songs you have made.

[prime rock]

This is the main thing. Having a two minute initial generation is nice but doesn’t matter much if the song can’t consistently remember a verse and chorus through to part 3 because part 2 is limited to one minute. Surely there are a lot of casual users who don’t even know how to continue so making continues longer wouldn’t be that much trouble.

I wanted to suggest adding a way to gift people subscriptions and credits. Often I talk to some talented people in the discord who are out of credits and it would be very little skin off my nose to gift them a few dollars.

[rose seal]

Should come this week

[rose seal]

Working on this!

[rose seal]

There's a little bit of a workflow hack here where if you "get full song" and then continue from that, it will have a longer memory

[sage mirage]

Anyone suggested profiles? Some sort of front end to use when you are making a song? At least for me when I get in the zone I'm focused entirely on finishing a song, right? Would be cool to be in a 'this song' profile, front end, something. So everything is tagged in there. We kinda already have it, especially with the highlights when playing parts of a song which rules

[tidal wharf]

wanted to ask if it works that way for a long time, cool!

especially relevant now that I generate by 9-10 seconds at a time (thanks to "continue from anywhere"!)

would it be much of a hassle to automatically add context if the fragment being continued is too short?

[dawn tundra]

i don't know if they already said that, but would be nice to have more control on files on web. filters, folders, search bar ... because we have a lot of credits with 10 dollars so we do a lot of stuff. if i'm looking for something i made two months ago it's difficult. i used discord thread for this, but the discord version is not good as web one.

[fallow knoll]

Would be amazing if we could feed it either an instrumental or vocals to drive some of the generation!

[tidal wharf]

would be great to be hear a preview of "continue from", to know where exactly the sample cuts. even better, allow tenths of seconds to be entered (e.g. 2:24.5)

[unborn token]

Thanks for the tip! I will try and use it

[cloud acorn]

please please pleaaaase Suno team, two things that would make suno perfect :

• make that on V3 we can have separate download links for music and voice
• User can re-use seed number to "re-do" or keep a good roll
  that would be just amazing

[tepid trench]

Pretty sure vocals and music are generated together. Would need a vocal/music split afterwards - would be nice if they offered this at least as part of the service so you don't have to do it somewhere else.

Reusing seed would be nice.

[cloud acorn]

I've always tought that the voice part was gen with bark and then added

[hybrid crane]

@cloud acorn read this #💡┃suggestions-web message

[warm glacier]

Exactly needs a bigger text input box

[cloud acorn]

Meian sad T_T Meian cry

[hybrid crane]

They didn't say it was not possible.. so keep hope!

[worn atlas]

But there is no feedback from dev. We dont know the roadmap , the prioriterization , technical chalenges

[hybrid crane]

It's still a very young company.. it's normal.. I've been working with startups before. They are a small team and cannot be everywhere at the same time.

[worn atlas]

@hybrid crane Your song is awesome and i am fighting with my own song to get something near. I was surprised by the guitars

[hybrid crane]

the intro?

[formal bear]

not just you -- I don't even know the roadmap, so...thx and be patient 😅

[worn atlas]

The intro is fantastic too. But the bridge with the country guitar solo is awesome

[instrumental bridge]
[instrumental]
[country guitar]

[hybrid crane]

ah yes.. sometimes you have to burn a few credits, or I might even have used the instrumental option, and then cut off where I want to the voices to start, by punching in the time I wanted the guitars to stop with the continued clip.

[worn atlas]

Today i have made several trials unsuccessfully. Suno was doing something else 😀 I was more successfull using your intro. So thanks for the tip.

[waxen sand]

suggestion: voice chats in this discord server. would be nice.

[sage lodge]

Suggestion: What i would like to have in the v3 is the option to change between male and female with [female] [male] thing to make duet possible 🙂

[velvet vector]

I love the version three Alpha access. 😁 it would be amazing if we could give a melody prompt. Being able to continue the song from a specific measure would also be great. The timestamps are not always accurate.

[wanton lodge]

Hello, I'm using this amazing tool for some time, love the V3 as well! I would like to suggest some very important and handful additions:

1. Song duration regulation settings.
2. New song creation using other song's settings / style. For example, it could be a seed code, like in AI image generators, or just a simple button, like a "continue from this song" (IT WOULD BE AMAZING!!!).
3. Prompt tutorial - how to write all actions correctly for the best result.
4. Lithuanian language support for the songs.
5. Credit refund system if the generation is buggy or it doesn't meet the user expectations. For example, it could be a timer (Like 5 minutes), where you would need to decide. If generation is accepted, then the credits will be paid, but if it will be rejected, then the credits will be refunded. Also, you could download the generation only after accepting it (To prevent fraud, system manipulation).

[tacit mist]

i feel like this can be solved with having a both 'continue from' and 'style from' instead of just continue

but i dont know the technicals obviously

[amber violet]

you have the "(link) part 2" as a tag, but no way to:

1. go back to the song clip that this came from easily
2. regenerate this part 2 from the song that it came from (without going back up to the song clip that it came from). When on a song add "Redo" and it fills in all the same info (that we can edit) and allows us to retry this song clip without hunting for previous song clip.

[rotund orbit]

Thought: I have hundreds of pages in my archive at this point. I don't even know how many. And sometimes I want to re-render takes. Sometimes from older projects, and tracking down what made part two of any given piece is challenging. What if, that was listed on the song profile, so it's a little easier to work with?

[fallow garnet]

I just need a give me a few more seconds of audio button because it was perfect until it just abruptly ended at the last word

[sage mirage]

time to burn an entire run just for the ending of a song

I suggest still allowing us to generate multiple songs at a time

Not that there is any hint it was changing, I just really appreciate it

I don't even know what the limit is for new generations
surely there is one

[sage mirage]

Anyone say Playlists? Tha would be cool

[rotund orbit]

Thought: What if in addition to the up and down thumbs, we could add a note about why it's a good or bad take?

[empty geyser]

The beginning of Part 1 of the generated song shows that color noise such as white, pink, or blue is present, but at a level below 10% or almost clear, approaching the sound quality standard. but when part 2 is generated (the continuation of part 1), there is an increase in color noise between 10% and 20%, and when the continuation to part 3 of the song, the increase in colcor noise starts to be high, approaching 30%, making the song start to be crowded with background noise. Even needless to say, for parts 4 or 5, the color noise will inevitably start to aggravate the background noise until the vocals start to drown. Is there no way to solve it?

[tidal wharf]

Quality hack: generate with v2, continue with v3. I feel like v3 is missing some variativeness off the bat, but as a continuator, the possibilities are 🤯

[tidal wharf]

I’ve noticed that to some, but not overwhelming, degree. I have songs of up to 8-9 parts where I can’t discern the quality drop

[empty geyser]

Color noise will be more in the style of metal or hard rock music because I am currently working for the gothic metal genre of Majestic Storm, so the song produced shows an increase in color noise starting to interfere with the vocals, while for folk or hip hop, for my previous project, color noise did not affect the song too much; even just a little touch with the mastering of the song will return with good quality.

[safe wadi]

You should focus to generate the end result by track layers. Music is a combinaison of multiple layers. So until you don't give us the possibility to modify only 1 layer of the result that tool generator will only stay a idea creation or a IA music. Human will not be able to control it to achieve better result. or the custome everything like that want. it's a great tool but to much random for the moment. We need to have more control on it 🙂.
Also you should have by layer to download the midi file. As well

[worn atlas]

Yes too much random . A comprehensive and Efficient suggestion has been made here #💡┃suggestions-web message
R&D should commit this

[frigid sentinel]

I'd love to be able to bulk download all my clips, does anyone know if that is possible?

[brazen arch]

suno radio.

[formal bear]

coming soon

[brazen arch]

LETS GOOOOOO

[tidal wharf]

Yep that’s my genre too mostly. Here eg 4-5 parts

Hack: continue your earlier, complete, v2 generations with v3 (doesn’t matter what it generates) and get full song to improve quality

[tidal wharf]

(That’s a separate observation, not related to the noise discussion.)

@formal bear any chance v2 will support continue from anywhere too?

[supple gull]

please turn the V3 music track down, it drowns out the vocals

[rotund orbit]

hey guys. There's a new thing it's doing the lyric ai is confused. It'll fill the text box and erase whatever's there in custom mode. This is bad, because often, lyrics are written or adjusted in the custom box, and erasing them without generating them forces you to start the entire process again. How about one of those animated status messages instead?

[marsh thistle]

This absulutly has no event to music, But what if suno were to make a storytelling A.I that knows how to have anger sadness and happyness in his voice, IK it would be hard but imagine a A.I story creator, The setting the evnt the main chracter the side character (Optional) who dies, and the PLOT. Or you can ut in your own story that you want it to read then you can save and post it and continue it if you want! It would be the best!!

[limpid pumice]

I would love to get separate tracks for vocals and music. This helps alot in post production. And separating them after downloading isnt good because of bleeding.

[hybrid crane]

I asked something simular, see: #💡┃suggestions-web message
They are investigating how they can manage that. Now it is all baked in one mp3.

[weak elk]

I would love it if Suno stopped singing my style prompt 😄

[limpid pumice]

oh ok thanks. I thought the ai creates vocals and music in separate and then merges them together

[hybrid crane]

Sadly not 😦

[cold vine]

maybe a UX issue. When I press the "get the whole song" it seems likes nothing happened (there is not a notification or indicator to know it's working). Sometimes I press the button twice thinking it didn't work, so get the whole song multiple times

[craggy rose]

V3 fidelity is much better in many ways. Guys you did a great job! On some fixes inmho..The organ distortion and volume really ruins songs that are great. I'm talking in the rock genre. Not sure how it is on more simple genres.Maybe swap it for something not so abrassive. Also the volume takes over a mix and makes the song unlistenable. I've tried to tell it not to use organ in promptsbut it doesn't understand that. There really should be away to seperate vocals. There are tools now that if we can get the vocal seperate we can change them without fidelity loss. I've tried a few stem splitters and the quality is horrible. Does anyone have one that doesn't kill the vocal quality? Getting vocals alone is so needed. Once that hurdle is completed we then could think about drums, bass and other instrument melodies stemmed.

[twilit hedge]

I tried out RipX Daw and the vocal track always sounds crisp to me... but maybe it all depends... some other background instrument tracks did get distord some degree

[stoic elk]

Have you tried the Moises app

[potent crane]

I was thinking exactly the same thing today. I wanted to say why there was a problem, surely that would be more help?

[potent crane]

I assume you are clicking "continue" after that?

[formal bear]

maybe

[amber violet]

sometimes when you continue, it will finish the word. so, if you liked it right up to the end, just try the next part of the song and see.

[waxen sand]

have you guys thought about adding a voice chat in this discord server? would be nice to just hop in a call with random people to discuss and test stuff

[formal bear]

yeah we did but my voice is terrible -- that's why there is chirp 😄

[waxen sand]

😆

[waxen sand]

btw i sent a reply email to keenan like 10 days ago. Are you guys getting tons of emails so its taking time or did it end up in the spam folder?

[formal bear]

I don't read Keenan's emails 😄 but I am sure he does get a ton

I'd do a voice chat at some point but we are all very busy in general -- and today is Sunday 😄 but let us think about it

[warped basalt]

If people would use them, Suno might be able to learn a lot. Like, Garbled Lyrics ... Lyrics Not Being Used ... Lyrics Cut Off at End ... etc. (I.e., if they standardized responses, they could get useful data.)

[warped basalt]

% of ad revenue shared with pro users who allow their songs to be included, when they're played? Played more often as users "Like" songs? Etc.

[warped basalt]

Wow, I never had that happen.

[warped basalt]

Yep. And sometimes it doesn't, or it ends the word incorrectly (plural "s" instead of past "ed" or whatever). And then you lose a segment of your song that you would have loved. Would be SO good if we could save those by extending them properly.

[formal dome]

The big problem with suno (and very frustrating point) is that you cannot tell you have made a song because you don't do nothing. Any improvement in user interaction, be able to give a chord progression, to choose the instruments, to get the midi, to get separated audio tracks... It would upgrade suno from a toy to a tool. Much more value. Is is planned ?

[tidal wharf]

There should be special lyrics tokens for “end of generation” and “and of song”

[tepid trench]

I'd like the creation settings to default to whatever I last used. e.g. I want to create custom instrumentals with v3 so if that's what I did last time, I'd like Suno to assume that might be what I want to start with next time I visit.

[cold vine]

I don't need to press continue. The song is generated by only pressing "get the whole song". I just tried again.
No idea on PC, but on mobile 📱 (Chrome nav.) it's happening.

[unborn token]

I feel V3 has a problem ending songs. the usual prompts I used [outro] [fade out] [end] some, one, or all, doesn't work like in V2, it continues way too long with music that fade out yet continues or changes tune to piano, beats, whispers.

trying to fill the whole minute, I think there should be some strong prompts that will prevent it

[prime rock]

I love the v3 quality overall but having played with it a couple of days I do have a style criticism. Some people have accused v3 of “sounding the same” which seems silly at first glance since it will do a lot more varied styles. BUT I have noticed that what it won’t do is changes in tempo. Across a single song, even when continuing, it likes to keep the tempo dead similar. There’s no rubato (flexible tempo), no changes in tempo during a song, and because of this it tends to sound like someone using a very nice drum track instead of real drums.

[cold vine]

a "Prompt Strength" slider as we have with Stable-diffusion to control how strong Suno will follow the prompt

[quick olive]

Man guys, you need to stop lobotomizing your product with "maybe copyright violation in the prompt."... rapping like eminem isn't a copyright violation is it ?

[formal bear]

yeah we know...are working on this

[frigid sentinel]

I don't know how hard it would be to implement but that is actually a cool idea for people making voiceovers and audiobooks. I reckon there are probably already tools that can do this. Oogabooga with Coqui is pretty fun to play with if you can do local installs although it only does one voice at a time.

[quick olive]

thanks for the heads-up

[frigid sentinel]

Moises is OK. FL Studio does better track separation imo

[tepid trench]

Can we get a "continue before" to extend a song backwards? Could be nice for when the first part starts in the middle of the music, or if you're happy with a continuation and want to regenerate the stuff before it

[prime rock]

I’m gonna add my vote to those who think too many voices sound like Beyoncé now regardless of prompt. I just put in “breathy voice” with an anime prompt and she’s belting her heart out. Tuning problem? I don’t know how y’all stuff works

[tardy bloom]

For the music style could we get options to choose instruments, the tempo and other things. Also for the style of music I put 'powerful rock (electric guitar riffs, thunderous drum beats and kyrics). I also included the intended lyrics but for all 5 generated songs, there weren't' any lyrics. There were lyrics for one. Is there something else I need to add or do?

[feral aspen]

You may have to vary your style prompts more than V2 to get the variety you're after. But also - when you get a nice style prompt, you may appreciate the consistency at other times.

[prime elm]

I keep getting stuck with generating a decent cross over, I got a beautiful 1st part, then it hard cut just on the last lyric and couldnt copy it over or continue 2 part 2 . It was perfect.
So how about a REPAIR function of some kind to nudge the time and finish the line/lryic? I was usingV3 alpha - but it was so good....

[tardy bloom]

I get the same thing or sometimes one word or line doesn't fit, so a function where we could edit the song would be nice

[prime elm]

I'll post the first part, it was so good and then just hard cuts, dammit. https://app.suno.ai/song/c107d63d-f1b3-46b7-bde3-ed0fb457ef67

[tardy bloom]

That is annoying

[prime elm]

I'm hoping there is a way to edit/repair this in a future update, but V3 is amazing. Just amazing.

[rotund orbit]

Thought: What about a creativity setting? like, offer different degrees of control maybe?

[frigid sentinel]

Just bumping this, a way to bulk download all generated clips would be amazing 🙏

[long robin]

May I please request a general volume control on the Suno page (for listening) the songs are quite a bit louder than all my other tabs and volume controls in windows don't work per tab sadly.

[ancient cradle]

I would very much like an API to manage and create my music. After making 600 clips in the past 2 days, I would like to start making my own pipelines.

[last gust]

This should have API access! Specifically allowing us to use automation to create songs. Very important.

[grim cedar]

I would love to be able to use the same or similar vocal stylings from one song to make another, so that it feels like the same artist sang them both. I guess similar to how continue works to imitate the previous part

[mystic zodiac]

I see that V3 Alpha is up. Thought i'd try it until I found it takes credits to test the new version. Testing should be free and charge for V2 until V3 is ready. Interesting to see how many people pay to test?

[hollow hazel]

Get some sorting mechanism in Libary - folders/category or something what can help you with sorting. Its kind annoying find some old songs.

[craggy sierra]

Suggestions:

Separate Stems (ie Vocals, Drums, Instrumental)
Retain a generated voice

[twilit hedge]

Did you not get the extra credits Suno gave you to play with V3 alpha?

[twilit hedge]

This has been suggested and requested a lot 🙂 let's wait and see

[mystic zodiac]

No. i only had what i had last time

so the V3 testing is costing me so i'm out

[twilit hedge]

see the Announcement of V3 alpha in #📢┃announcements

[twilit hedge]

Yeah.. V2 is still available

[mystic zodiac]

i saw that in the web version then i selected V3. still charging me though. couldn't see anything on free credits for testing V3. did i miss something?

[twilit hedge]

Click "Subscription" button... it should say "Extra Credits" or something like that

if not see if you could contact dev/admin

[mystic zodiac]

appreciated. only says to buy more credits. all to hard to test their stuff sadly

appreciate your efforts though 🙂

[twilit hedge]

Sure... just trying to help... I'm new here too 🙂

[harsh prawn]

if we can input a link of a song as a reference for suno to make something similar it'd be really cool or a few links to take few songs as a referece to blend

[grim cedar]

even if it only let us reference other suno creations so we didn't break copyright

[harsh prawn]

just listening to songs and creating similar tracks(not copying the song exactly) should be okay

[slim aspen]

I'm finding the same issue with V3. It can be almost impossible to get it to finish a song. I mostly do instrumentals and I've had to uncheck "Instrumental" and spam the custom box with [end] about 20 or more times, and even then it is hit or miss if it will end a song.

[frigid sentinel]

As well as bulk download my other feature request would be a BPM slider/field. Oh and Flac or Wav too 🙏

[pure cobalt]

suggestion - it would be SUUUUPER useful to be able to render separate tracks instead of the whole mix. and get a .wav too instead of mp3? especially for musicians who want to use the output beyond simple background music - like a co-pilot for musicians like we have github copilot for devs

or to be able to prompt "drums only" or "vocals only" etc

[dusky charm]

a way to select a start point and end point in a clip resulting in a fade out from start to end and complete silence or a complete cut after the end point. That would be nice for all those clips that just flare up after an otherwise great outro (see @2:30 in https://app.suno.ai/song/23b5497e-198a-4587-99b1-5dedd14f614c)

[dusky charm]

Webapp UX : either a filter for 'full songs' or (even better) the ability to create albums to order the library a bit. Oh ..and a way to delete multiple files at once. Once I have a finished song I spend way too much time deleting the parts I don't need anymore

[weak elk]

The bass is often way too loud, even for hiphop and dance.

[celest aurora]

or at least to upload the melody of the lyrics. I would love to upload an audio with lyrics and then suno can take over there and make a full song

[hybrid crane]

Likely that would be problematic due to copyright laws... And keep in mind, every country will probably have different ones as well

[sage mirage]

Yep. Technically possible, easily, probably, and wonderful for creativity and the ~art~ itself, but alas copyright, profits and big ol' corporations make it impossible to implement if Suno doesn't want to get glassed from the sky

[ember stone]

I'm sure someone has suggested before, but a playlist of other people's songs that we liked would be good.

[hybrid crane]

They are working on playlist so you can save your clips or other people too (who shared the urls)

[tardy bloom]

Suggestion: For the Style of music, there should be sections for the theme, instruments, mood & vocals - type of voice & any affects (within the vocals like reverb)

[wide belfry]

How hard would it be to have the AI not just look at the previous part, but also the ones before it?

Example, with Synthwave, it often starts with an AMAZING synth and by the time I'm 3-4 parts deep, that synth is gone and never to be heard again.
https://app.suno.ai/song/27ed3dab-8131-4142-a1ff-17e8d2bc153e/

https://app.suno.ai/song/bb2cc26b-bfc0-4f5b-9fdd-ba48816d842e

[sage mirage]

You're contining from the 'full song' right

[wide belfry]

No... Apparently I should be.

[sage mirage]

Try that and see if it can reach back further

One of the devs mentioned that as a 'hack'

[wide belfry]

Ok, thanks!

[sage mirage]

I've always just done it that way anyway but according to the devs there is something to that

[wide belfry]

I was originally told to keep it to the individual parts...
Trying it now. Thanks!

The 1 that finished is pretty good. Curious about the others though...

"Continue From" is an amazing feature! Thanks y'all!

[leaden marlin]

First suggestion - I consider it fundamental in managing files in the main dashboard in Suno, both in the creation area and in the library. I'm talking about the possibility of creating folders. For example, I have a project for a certain album, and I create the music. In the end, I want to store these songs in the folder for that album. The organization would be more structured.

Second suggestion - In the trash section, I find it useful to have the option to check the tracks we intend to delete, thus avoiding the quite demotivating task of erasing each file individually and confirming the operation.

Third suggestion - This would be quite interesting for more professional use. Having the possibility of having the music in separate tracks, that is, having the voice file and each instrument so that we can work on this audio in a DAW and there allow working the sound with the utmost freedom.

Other than that, nothing to add except for the phenomenal experience I'm having with this tool. Thank you.

[hybrid crane]

• They are working on playlists. You will be able to create several and add not only your songs but from users too...
• Bulk delete is asked many times, I believe they are working on that too (but not 100% sure)
• Stems are also asked, but they are investigating that, see: #💡┃suggestions-web message

[delicate cedar]

The problem with stems based on previous conversations is that the method they use to generate songs precludes them because however they do it is all in one piece. To do stems they have to either reinvent something that works from scratch which may or may not work as well since they're starting over again and will then need to play catch up or they'll have to extract the stems after the fact, in which case why use Suno's extraction over some other DAW based one? And can the quality ever be as good as the original generated audio?

I don't know the answers to these questions but I assure you the Team are pondering them.

[tribal ice]

Want to give a +1 for API access for paying accounts!

Also Folders or Filterable Tags

[dense gulch]

Just following up on the download option showing up to screen readers in the menu. Let me know if I can do anything to help 🙂

[modest pewter]

You would have to offer more benefits to Premier subscribers, not just more credits, to differentiate them from the Pro subscription.

[worldly pelican]

Not an expert so I'm just a ghost trying to learn from you guys and wanted to say that I LOVE Suno

[bronze jay]

I don't know if this has been mentioned but a way to edit clips before they are put together as a whole song would be nice.

[potent crane]

for me that would be of no interest as I always do editing after the song has been completed

[lyric sluice]

Liking a continuation should automatically create the full song IMO.

[solar bolt]

A way so I can put a animation and the AI will sync music over it

[pseudo surge]

Any chance we could get like a folder system or something to better help us sort our generations into something like genres? Make it to where if we drag one generation into the folder then the whole chain to that song gets taken with it or something like that and if we want to continue on that song it will generate in whatever folder we are in. Similar to how AIVA is setup. I would just really like to be able to sort and separate somehow.

[sage mirage]

v3 generates so much less garbage now that I too am running into organization issues that are really hard to keep up with

Especially since I can't just type a page number and instead have to click that little button

[pseudo surge]

Exactly

[sage mirage]

Grabbing whole tree or chain would be excellent

It already knows on account of the highlighting

[pseudo surge]

Yes which that alone is very helpful

[vestal sinew]

yeah it would be good to have some sort of folder options. its good we can finally rename stuff it helps keep the chaos a bit more organized but folders would be a massive plus.

[lyric sluice]

For radio mode: Why is same lyrics, different genres not an option tbh that's all I would use it for lol.

Like it would be cool to give it a bunch of single terms for genre and it just runs through all the combinations of them.

or even permutations if order really matters that much

Also a mode where chatgpt recommends the genre based on lyrics.

Longer allowable lyrics, currently I'm not getting the most out of the 2 min time window because I use rap lyrics which are typically faster paced than normal songs.

[marble vessel]

Yes I'm on V3.

[mystic zodiac]

Seriously tame the auto tune on V3 and not make every song sound country even when using a Pop Dance style.

Oh and increase the text limit for the lyrics in V3 as it makes longer tracks and won't fit them all

[lyric sluice]

Allow liking with notes, sometimes I have something in mind when I like a track and I want to remember it.

[worldly pelican]

I'm starting to understand what you guys are saying that sometimes the songs hallucinates. However I feel that 1 in 6 songs are still pretty good!

[mystic zodiac]

V3 is not only charging for us to test but also charging for 15 second bits of rubbish. Paying to test V3 is complete rubbish. Should be pay when it's ready and free to test to help.

[sage mirage]

You don't have to use it

v2 is right there still

[mystic zodiac]

@leaden marlin I've recommended they add to use a previous track as a reference.

yes v2 is now making everything country. must have changed programmers lol

@sage mirage my point is that if they are charging for us to test it, it needs to give value and not just take hard earned and paid for credits 🙂

[sage mirage]

So then don't use it, test it, and spend your credits on it. It's voluntary. You aren't being forced to use it.

[mystic zodiac]

lol

[sage mirage]

Alternitavely, it does give value because, for my use, v3 is pretty fantastic and I can't use v2 anymore because it sounds too bad for my use

How you feel isn't how everyone else feels

[mystic zodiac]

oh also so there's no point us giving feedback then.

just your "don't use it" strategy. that will help lol

[lyric sluice]

I think v3 has trouble finding it's voice if that makes any sense, specifically in the beginning of a generation. And when it "fails" to find a voice, it seems to just generate an instrumental despite lyrics given. At least that's the feeling I get from it.

And sometimes it just messes up the first syllable or word, and then performs fine

[worldly pelican]

The more I use it, the more I'm starting to understand what you guys are saying. However, I'm still a novice at best.

[lyric sluice]

It's a tool you learn how to use it.

[mystic zodiac]

it will be great eventually. just needs a lot of work

[lyric sluice]

In continuations it feels like the model tries to force in drops.

[cinder blaze]

please for the love of all that is holy, add scaling or something to the vertical columns

[hollow locust]

It'd be nice if it understood contextually the difference between words like 'wind' and 'wind'

[worn atlas]

Really need the "redo" or "restart with same music same melody but with differents lyrics" feature. V3 sometime gives me awesome music and melody but i feel like it would be better with some modifications in the lyrics. Today the lyrics of the first part cannot be modified. Someone said here ; "the wrong part can be cut with à DAW". But i am looking for a suno-only solution.

[marble vessel]

Observations - When I provide original lyrics, Suno Creates much better result and many times up to the first verse and even adds some chorus lines taken from the song. Secndly, when I want the last note of the particular phrase to be held for longer time, adding dots (like ..... ) at the end of the phrase does the trick quite well. I've also experimented specifying the vibe, mood, beat style and many times that also has worked well. I've seen that it ignores the Male, Female singer type instructions in the song and makes random decisions on it. If there are consistent rules around this, it will become really creative and easy to get what I want. If you observe the text-to-image prompts, they allow tags for camera, painting type, style of painter etc. Similar tags will be helpful

[sterile raven]

I am brand spanking new, I got suno, dove in and created my first song before ever coming to the discord. I learned immediately that it does react VERY well to the .... at the end of a phrase. It seems ot 'want' to take direction but shrugs it off in the end.... I also noticed that at some points it sings words like it has a mouth ful shredded glass or marbles LOL And certain types/styles of music will generate one gender voice over another regardless of what you ask for 😛 Am still learning-but I found your comment about the dots at the end true..at least in my case.

[weak elk]

reset the Continue From time if you select a new gen to continue from, I've lost several credits to this...

[safe thorn]

SAME! Need API access please!

[delicate cedar]

We don't have a publicly-available API at this time, but it's on our roadmap.

Direct quote.

[wary hazel]

I am asking for the opportunity to choose for a given verse or verse to be sung by a woman, a man or even a choir

[supple smelt]

The ability to split up the song into vocal and instrumental tracks would be huge. I could load these into a DAW and master each track separately, which could hugely improve the quality.

[hybrid crane]

A lot of people have suggested that, and they are investigating it. Read: #💡┃suggestions-web message

[prime elm]

2 Things:

1 A timed SRT exporter, so the synched captions can be exported with the audio only option.
2 A "nudge" option to add some more time when the track finishes early before the lyrics or just hard cuts a little through the last lyric.

[crisp furnace]

please instruct me on where to find the Suno bot? I a kinda new to discord.

[marble vessel]

Suggestion: Can we have folders in the library where we could organize our creations by genre, composition name (to store multiple versions) etc?

[delicate cedar]

It should be in the sidepanel.

[crisp furnace]

under what name @delicate cedar

[delicate cedar]

Suno Chirp Bot

[crisp furnace]

I don't seem to see it, not sure why

[wary hazel]

Ability to quickly remove unwanted files.
e.g. after library, all tab. In "ALL" songs can be listened to, but in tiles, so you can click and listen, and at the bottom there is a box to check to be able to delete or save interesting songs in a subpage, e.g. library or some other subpage.

This will make music design completely easier.

[crisp furnace]

Oh yah, I see it now. Thnak you so much @delicate cedar

[hybrid spoke]

One suggestion is, just like you now can choose "Instrumental", to output only the vocals. Or even better to get a way to output a split / both on demand.

[jovial cradle]

yeah, that would be so great, to have the chance to get both, one with lyrics, the other one with the same song and just instrumental. Thatz would really make my day.

[orchid lava]

Can we the textbox for custom songs be given more characters, in V3 you can generate up to 2 minutes, but some of the quicker songs won't allow you to paste all your text in the box.

Can the textbox we use for custom songs be given more characters, in V3 you can generate up to 2 minutes, but some of the quicker songs won't allow you to paste all your text in the box.

[crisp furnace]

I purchased v3 yestarday and my credits are running out real fast. Is it possible to have the general free credits that are offered in the free plan added to us everyday as well, in addition to our paid credits?

Why I ask is because some of the generated music has no vocals yet the lyrics have been generated. Sometimes when there is an error in the music genertion, the system refends my credits, but it does not catch all of them, thus I loss out on some of my paid credits.

[unborn token]

I wish you could get the music without the prompt on it if you chose to, just the cover art on it. it's actually good

[crisp furnace]

You can do this by having two credit section, one with the paid credit options, which of course, has our v3 credits, and a second credit section, which has the daily credits which are under the v2. This will help the paid member and Suno ai keep track of the paid v3 credits without having them getting all mixed up. This will help Suno ai retain paid members longer, enjoying your amazing services. I believe it will aid in repeat customers as well because of your generous daily offers.

[potent crane]

it's been asked for a thousand times, I am sure it will be with us very soon.

[potent crane]

extremely difficult as giving it the ability to recognise context in a sentence would take up a lot of computing power, the best advise it to write words phonetically when there is a doubt, wind is a tricky one, you could try wynd or better still whined.

[lyric sluice]

The ability to use the keyword freestyle in style.

[tacit idol]

An idea that's been floating around my head for a while now:
The ability to select two tracks from our library, and have the AI generate a bit of music that connects them with a seamless transition.
Not sure how necessary or how useful this would be. it's just an idea.
It could be interesting if you chose two clips from vastly different genres / styles.

[potent crane]

That would be useful if you want to create long-form progressive rock tracks (which I like to do), creating a credible musical link between two songs would be very useful.

[oak echo]

I just wanted to share my thoughts on the v3 alpha. Now, I don't have access to it, so I might not have all the details right, but from what I've heard, it sounds like a big improvement. The singing seems a bit different, but the music is just wow, it's changed so much and for the better. Keep in mind, this is just the alpha version, so the official v3 might be even better. Can't wait to see what they come up with next!

[lyric sluice]

The amount of times v3 generates solely instrumentals as opposed to the lyrics I passed it is way way way too high to get appropriate use from it, if this is not fixed, I personally don't think it's worth the same amount of credits, when nearly half of the generations don't contain the lyrics I used.

Should be half the credits if half the results just don't match what I put in, in any way.

[potent crane]

There's no question that V3 has much improved overall sonic quality, you get less hallucinations, and cleaner mixes, but there's hardly ever an "intro" no matter how hard you try, and although the complexity of instrumentation is much better, as you say, that seems to be at the expense of lyrical stability.

[lyric sluice]

Lyrical stability should dominate instrumentation always, as most people listen to songs with words in them, at least from my perspective that's what seems to be the case. I also mentioned something similar to the intro problem but not exactly that when the model is performing the first line or even just the first word it will frequently fumble and then fix itself, which indicates to me at least that the noise they are using for the starting point is probably not optimal, but the model is able to compensate after the poor initialization. It does this far less when the beginning of the generation is actually an instrumental intro, which is rare. It's weird that it will fail intros frequently, but is also better at general instrumentation.

[misty blade]

To the amazing Suno team!!!

First of all, I'm addicted to your tool. Just so you know, today I created music that was used in a real election campaign that's running. And people danced to this song without knowing it's artificial intelligence. This is an incredible achievement for me!

Secondly: I've been experimenting a lot with version 3, and the quality has improved significantly, and the melodies are much more beautiful and interesting, and it's noticeable that there's more expressiveness in the singing. But - we lost something in the precision of the singing, many times the lines break in the wrong places, and there are also sudden skips over words which is very disturbing. It would be really good if you could perhaps give control over a parameter of how much creativity to give to the model versus the accuracy of the prompt. Because in version 2 there is an advantage in terms of accuracy and it's important. I really hope you will enable this accuracy along with the quality of version 3, at least to have a choice through control over a parameter or something. Thank you

[wide belfry]

Please add a link on the right, near the 👍 & 👎 to add a song to a "Follow Up" folder so we can store all our awesome tracks to check out again later.

[sage mirage]

Yeah with the knowledge that the thumbs up/down change...something with how the system works means I can't easily use it as an organization tool since now I have in the back of my mind I'm messing with the system rather than just organizing

[lyric sluice]

Did someone do something, my gen's have been insanely good over the last 30 or so minutes

Probably just found an outlier style honestly but still damn

Though its been a few different styles that were similar hmmm, but either way if you did something keep it.

But like shocking difference swear on me

[formal bear]

Just luck, trust me...

[lyric sluice]

legit was 8 in a row

[mental barn]

Yeah! I agree. I wonder if individual tracks like this just isn't possible because it may be trained on audio itself and not on midi or musical concepts like we would record something. I use fadr.com to do separation and then add various plugins and attempt to remix and make them better in Reaper.

[mental barn]

I use Jaws, and I thought they removed the option altogether for awhile to download. If I wrestle around enough, I can kind of eventually get it, but it's not real easy since turning off the virtual cursor and trying to navigate with the Jaws cursor doesn't work real well.

[dense gulch]

Thanks. I managed to figure it out. At least with NVDA, if you enable the menu, and then turn off focus mode, it will show up. I only use jaws like 2% of the time. The issue is, they put the download option after a separator element, so the screen reader thinks there are no more options. Was supposed to be fixed by Monday, haven't used it since yesterday.

[crisp furnace]

where can I find the songs I created two days ago on my paid for Suno? I was able to see them yestarday but I can't find them today?

[crisp furnace]

no worries, I found them. The public button was on and so I could not see them but once I turned it off the were visible.

[idle hill]

Suggestion:
On the website, there should be a hide button that is easily accessible and distinguishable from the thumbs up and down buttons. On V2 I downvoted any song I wanted to hide, but didn't realize I was inadvertently training future models to see those as bad, even when they weren't. Realize I can delete, however clicking the menu button and then clicking delete can get tedious.

Another suggestion would be to have a secondary workspace that has branching paths for each song generation. Easy to lose track of things if you're running multiple threads in different directions to compare and contrast. Something like this, generally speaking:

https://media.geeksforgeeks.org/wp-content/uploads/20221017140045/6.png

[north wolf]

it would be great if there was a way to redo a V2 song with the V3 model. Even if we could do a "Continue from" but start from the beginning using the same basic feel of the V2 song.

[supple lake]

Ummm, you can

At least I can rerender the v2 songs with v3 that I have

[mystic zodiac]

@north wolf That would also be a huge help for any songs that were ok in V2 where the quality degraded after Part 2.

[craggy sierra]

There already exists AI that can seperate audio into stems for you

[supple smelt]

That is absolutely not the same than having actual multiple tracks for a song. :P

[potent crane]

Can't wait to hear what a rough Asian woman sounds like

[thorn jay]

I already put that in the genre box and gives me what I want. I.e. catchy edm pop female British singer …when I do that I always end up with a torri kelly type singer…I’ve also used sultry voiced female, if I want dancehall I put “jamaican female rapper”…the AI “will” try and give you what you want. For 80s synthwave or dance I put “paisley park uptempo dance with oberheim strings, dx7 bass, arpeggio and syncopated drums machines”. Dont be afraid to TRY. It WORKS!!

Try the “actual” voice classifications first. I.E: tenor . I made an opera and it understood the difference between a soprano and contralto

“musicians” helped to make SUNO. Learn the most about music to help you generate “better” prompts. This theory applies to all AI. I’m already a musician for over 30 years and artist/photographer (I use Midjourney alot and with great results because if my “knowledge” in those spaces. ) More music knowledge will make you a better prompter. Period. And that especially goes for songwriting VS. chatgpt songwriting …which has a lot of limits and is MAINLY good for getting out of writers block…not actually writing the songs themselves…make sure you are learning to actually write as well (if not already)

Suno will help make better human writers because you can try out your ideas instantly ..all my musician buddies who use it have gotten a renewed interest inspiration to keep writing and to write more

[astral gust]

Right now there isn't a specific/consistent way to make a song end. It would be nice if in the "continue" feature we could have a checkbox to force the song end

[wooden fable]

Two features I really hope will come in the future:

1. .wav or .flac export to avoid the quality loss caused by mp3 compression
2. a "vocals only" mode

[fossil cape]

Suggestion: allow manual post-editing of the lyrics (text only) so it matches up better with what was actually generated. I'm thinking that might help with continuations in case the generated sections don't quite fit the labels in the lyrics prompt, or in case the song cuts off before the prompted lyrics end. On the "display" side for songs, if it matters what lyrics the song was actually generated from, you could always add a toggle to view the actual generation params (that might be helpful in general too, for example for a better user view of continuations)

[fossil cape]

And when continuing from a particular timestamp, allow specifying where exactly in the lyric prompt that timestamp is (if that isn't already possible)

[wooden fable]

Agreed. I had that problem yesterday. The "remix" button on the Suno website didn't do the trick, so it was a lot of trial and error.

(But I gotta say, the bloopers were hilarious!)

[alpine verge]

It is possible, u can choose exact time, only V3

[wooden fable]

How?

[hybrid crane]

@wooden fable

[wooden fable]

oh gosh, i didn't notice that input field. Thank you!!

[worn atlas]

There are lot of suggestions about folders organization. One idea is to create a folder for a song including all the continuations tracks. Than in the first level only the full songs will be visible. The intermédiate parts will be in the folders one hierarchy below the top level folder

[hollow locust]

I used 'wained'. But I'll remember 'whinned' in the future.

To tell when (and when not) to elongate words. E.g. Lovers dancing in the night. Generally it makes its own mind up, but it'd be great if it was possible to do something like "Loverrrs dancing in the niiight" or "Lov-ers dancing in the ni-ight" ... just something that gave users power over sung vowel length.

[cinder blaze]

yo @fringe knoll

[fringe knoll]

🙏