#💡┃suggestions-web

1 messages · Page 17 of 1

strong swan
#

but it wasnt the first generation that was this fast. it took like 6 tries or so.

bright topaz
strong swan
wind crest
#

really hoping for this soon, after 4 or 5 extensions my lyrics are a garbled mess. i understand this is more complicated than it seems to implement, as it'd throw off the entire video generation process (like, how would it know where to put the lyrics in the new video) but i'd be happy to forgo the option for a video if lyrics are edited, or accept that a video may no longer be accurate (so we can make minor alterations and still get a video in situations where we had to, say, spell a word phonetically for the bot to say it right)

prime rock
#

First take on 3.5 is not only does it sound really bad, continues from 3.0 using 3.0 now sound really bad. Just really terrible and altogether unusable.

wraith grotto
#

Phonk genre is missing from v3.5 by just prompting almost like it's a banned word. Also it's not on explore page.

kindred flare
#

Is there any way to communicate with the developer directly?

sly pivot
#

Hey, devs, guys that work on Suno, nice one with the longer song lenght from V3.5, but the important 1 million dollar question: did you guys also increase the Lyrics and Style of music Lenght?, or is that still with a character limit of 1250 characters for the Lyrics and a character limit of 120 characters for the Style of music? Because if you just did the song lenght and not as well the character limits, that will not help in the end. Please tell me you guys also have increased those limits.

feral aspen
prime rock
#

3.5 completely ignores intros even more than 3.0

foggy leaf
#

please add list view to playlist page, thumbnails are** too large**😂

lofty hatch
#

workspaces/playlist in the create tab. selecting a workspace causes all gens to go to that workspace"/playlist Maybe have a filter option that lets you remove all of the gens from that workspace/playlist that don't have a thumbs up or maybe a button that lets you save all the thumbsup gens to a playlist.

#

2nd: weights in styles. e.g. Heavy Metal:0.7, Acoustic:0.5, Fingerstyle:0.2

sly pivot
#

ok, this is seriously getting annoying, i tell it to do a song, write the damn lyrics and make a nice style, and then it just does what it wants and uses parts of the Style description in the song-lyrics. Why the hell is it doing that? Anyone else having this happen?

feral aspen
sly pivot
#

in the lyrics is this:
Intro:
Hey! Let’s get to the point.

in the song description is this:
rhythmic melody that sticks in your mind.

now guess what it does in the lyrics. Yea, it sings, hey, lets get to your mind or hey, lets get sticks to your mind

seriously, why is it doing that?

#

it would be really helpfull if the AI would actually do what its told, and not use stuff from the song description in the lyrics, just a little suggestion.

feral aspen
sly pivot
#

i am only using custom mode, never have i used simple mode in my life.

feral aspen
sly pivot
frosty breach
#

Update the plans on what they offer/lack, seeing as there is now a difference paid and free. Might as well display it.

worn atlas
#

Suggestion : Give the actual size of the lyrics in term of number of signs. Si that we will see how far we are from the max

frosty breach
#

Please make it so we can just hide/turn this off. I keep clicking it while tinkering, it's annoying as hell.

Screenshot_2024-05-25_at_13.19.28.png
#

Resizing makes it slightly easier, until you swap pages and have to continuously resize it.

frosty breach
#

Please let me disable trending autoplaying.

tidal isle
#

More character lenght in lyrics section at v3.5 for 4 minutes songs

opal spoke
#

When you share songs should have lyrics animated like Apple Music or something that follows the song

worn atlas
opal spoke
worn atlas
opal spoke
#

Ye i see but I guess they should turn it more smoother

dense monolith
#

Hey guys.... you really need to show that the generation you are hearing is a preview version with "low quality" and show when the generation is done with the good quality pls!

rain charm
#

Now that 3.5 is capable of much longer sections, it would be really helpful if you could select how long of a piece you want directly rather than getting a random length. So far the system has been okay with figuring out how long something should be based on context, but it's really annoying when I get a 2 minute+ extension on 30 seconds of lyrics or a whole minute of just odd low sounds when I tell it to end the song.

upbeat mantle
#

add a option to show how long the song might become with the added prompts

rocky jay
#

Suggestion: paid subscribers should be able to remove that public button. No desire to share my songs directly and the fact that it's next to every song takes on unnecessary real estate within each section.

#

Suggestion #2: easier way to batch delete.

upbeat mantle
#

add a box that shows what is free and subscribed songs in your library so there is less confusion when u need to figure out when u made the particular song

#

makes it easier to not make the mistake and market free ones as a example

tidal isle
#

At mobile version at extend I can't see the song playing at the buttom of my phone anymore. So I can't find the right time where I want to extend the song

hollow locust
#

some sort of visual indication that a song's already in a playlist would be nice.

tidal isle
#

The extend time at mobile version also don't stay. I have to setback it after every new try.

teal perch
dire moth
#

self trained models nneed to be added asap. both major generators available lack extreme amounts of information on many styles and we as the user need to be able to feed it that missing information to achieve the desired results. ive stopped paying for and uysing either until this is added as neither can create anything i actually want to use one for

#

it's cool and all making random stuff but many of us have specific use cases for these and when they arent capable of that then it isnt worth paying for or using

#

also lets us generate in high quality wavs or better so these are useable on large sound systems (even 320kbps MP3 would suffice)

teal perch
#

There's also a problem with the mastering. Suno does stereo effects very badly. Most of the tracks sound like they are in mono.

dire moth
#

defin another issue

hardy falcon
#

🔳 A toggle switch for auto-tune vocals

that's it, I said it.

leaden marlin
#

We have the option to generate an instrumental. But as a composer, producer, sounddesigner, mix and mastering engineer i'd like to see a vocals only option to generate clean solo vocals and some command to decide which part of the lyrics to have backing vocals or harmonies added.
And an option to retain a specific voice from a previously generated track.

Sometimes suno can generate much cleaner sound than the usual noise. Would like to see that be the standard. I would trade some generation time for higher quality.

bold cedar
#

Okay, I have a few requests First of all, increase the sound quality, I'm begging you, this is the first and basic thing, I know you can do it and really Fix this error with the fact that at the end of the song there's some head and neck noise

#

the second thing is the possibility of changing the lyrics in the generated song, I think it's called impating, I think so, please do it, it will save my life

#

thirdly, the possibility of a remix on I love you I love you very much but on udio there is something that is simply great - the possibility of generating a remix of the song that will be similar to the original, please introduce something like this

vapid estuary
#

I maintain a custom GPT that works as a Suno cowriter. I'm generally very happy with it, but it keeps getting confused by when to use brackets, asterisks, and parenthesis in lyrics. Is there some definitive documentation on this? If not, can we please get some?

visual brook
#

please publish any additional technical info on v3.5 .... same model? fine tune? X% bigger dataset (without revealing actual dataset size)? etc

visual brook
#

add a URL element to help determine which version i am listening to from Discord or the Suno Home page -

for example, https://suno.com/song/11fbb381-2440-46e2-b170-cf6ea98c6c39 would have an additional URI element of /v35/ or /v3/ or /v2 etc somewhere

User story: New song area of Home is impossible to tell which song is which version, especially on the cutting edge like today how can the community know and evaluate v3.5 easily from peoples generations?

Hyper Glitch Rap 🌳 by @mrtree | Suno

virtuoso polyrhytmic glitch rap, trap, glitch gunshot fx song. Listen and make your own with Suno.

▶ Play video
brave turret
#

This may not be possible with all the stupid laws and the fact that this is international, but I would like to purchase songs made by other people. I've found so many songs that I love and have great, clever and clean lyrics. This may be a big task but it could bring in additional revenue to Suno and contributors. The maker of the song and Suno could agree to split a percentage of sales similar to Apple. This way I could buy songs and put them in my external media players. This would attract a lot more people who wouldn't mind paying the sub fee knowing they could make that back. Just a thought...

visual brook
potent coyote
#

Please give an option for the thumbs down button to delete the song directly. I have no use for the songs that I've thumbed down because of bad quality and they are clogging up the space.

blissful wolf
#

Having read 3Daisy on clicking the INSTRUMENTAL button:-
<<<<<<<<<<<<<<<<<<<<<<
"I think I found the solution for you.
Continue where you want, empty your lyrics and only put in [instrumental hook].
It gave me on the first go 2 perfect instrumentals.
Then just continue again where you want the solo to end and add your lyrics again."

but now with 3.5 we have 4 minutes of step 1 ! To do what 3Daisy is suggesting we need to break our lyrics into 2 parts ( back to version 3)
So what we need are TAGS like [45 instrumental hook] where 45 is our request for the length ( approx) to play instruments. ?

hybrid crane
visual brook
tepid trench
#

^ This is kinda what I'm trying now

brave turret
hallow frost
#

a checkbox TO DISABLE THE F%$%#&^%&#% Generate LYRICS PLEASE!

#

I have NEVER used the Generate Lyrics. I will NEVER use it. EVER. BUT - Misclicking that stupid button has cost me so many hours of work at this point its beyond frustrating. TOTALLY understand WHY it is there - I support it's existance, however at this point I just hate it, it eats my prompts and provides me with nothing. Why cant you UNDO it?

strong swan
true umbra
#

Big fan of 3.5. I would like the facility to change the Style of Music at different moments in the song. Especially now, you can get a whole song in just one click. Keep up the great work!

strong swan
true umbra
#

BTY I'm still trying out Udio and Suno still much, much better!

chrome siren
#

a possibility of editing the lyrics of already generated song would be great. for example - I generated a song and the music and vocal are great but I decided I would like to change literally one or two words without changing the entire vocal and music at the same time

true umbra
# strong swan for 3.5ea: have you tried to give suno a new style in form of a [metatag] in the...

Hi and thanks for the post... I've been trying [metatags] for a while now, but there's an inconsistency between Styles of Music. I do understand that ai music generation is based on the music data so it is highly unlikely we are going to be able to get [Angelic Voice] to work for Heavy Metal Style of Music. But it would be great if there were tags that didn't have any relationship to music, but they would give you options to change the style of music another way. I'm dreaming! I know! Maybe next week! Have a great weekend!

sleek abyss
#

I'd really love advanced options. I don't know how similar audio is to image gen in this case, but stuff like seed controls, prompt adherence, etc. That or some kind of "Remaster" option so songs can have their quality bumped to the latest model's standards.

true umbra
# sleek abyss I'd really love advanced options. I don't know how similar audio is to image gen...

I like the idea of seed-type options. I'm often drifting trying things, and bam! There is the sound that knocks you off your feet, but then you try and get it back, and around and around and around you go.

I'm hoping Seeding options will be available soon. We could then develop a style of our own.

The random nature of AI is based on seeding. So, let's hope someone is out there and listening! Plant the Seed!

sharp sundial
# sleek abyss I'd really love advanced options. I don't know how similar audio is to image gen...

I would also like to see something like a "Remastered" button to upgrade your old songs to a new version. I have an old v2 song that I really liked, but trying to extend as a means of using a newer model is a bit expensive and sometimes is inconsistent. With the latest preview in https://discord.com/channels/1069381916492562582/1083140371980234852 , I think it could be possible to use that feature to make a song that is as similar as to the old songs.

hybrid crane
opal ivy
#

that's a good question

severe barn
#

please this for v 4

shut dagger
#

Duets are more of an exploit, definitely not a feature.
what if.... it was a feature?

nova prawn
#

Add at least 100 credits daily, it s too little 50 per day

desert girder
#

Make possibility to recreate the song from v3 in v3.5

strong swan
wind abyss
#

yeap seems to be based on bark LLM 🙂 kidding, but pretty bad on other styles.

#

imo

fair hazel
#

There should be an option for us to recreate the music from previous versions in current versions. I love my songs from V3 and would like to reconstruct them, maintaining fidelity while incorporating new improvements and superior quality.

#

There should be an option for us to recreate the music from previous versions in current versions. I love my songs from V3 and would like to reconstruct them, maintaining fidelity while incorporating new improvements and superior quality.

void sand
#

can we add a reference track so the ai can learn what we want

hybrid crane
void sand
hybrid crane
void sand
hybrid crane
blazing sandal
#

I can see the neighbors now complaining from all the pot bangin coming soon

void sand
hybrid crane
blazing sandal
#

should be nice if we could do Acapellas and or just the vocals male or female with out the music added, so if we building a song in FL Studio we can lay the vocals down on our tracks that way.

hybrid crane
blazing sandal
#

totally

void sand
blazing sandal
#

I've been writing and composing music for 30 years and the one thing I hated doing was singing to it lol but I love mixing tracks and writing music. So a reverse would be nice where I can just add vocals persay

frigid timber
#

Suggestion: Keep up the GREAT work! 3.5 is fantastic and the vocals are MUCH better! PartyParrot

round zephyr
#

Would be nice if we had an "Empty Trash" button, or option to select multiple to delete permanently. I have waaayyy to many to go through and get rid of one by one

true coral
#

I'm loving 3.5, but I'm still getting crazy weird noising even at just 2 minutes. It sounds like a live audience almost until there is a quiet spot. Then it just sounds like... noise? I don't know how to describe it. You can hear it at about 2:20 here. https://suno.com/song/20550af0-e7f3-4d08-bdda-eab9633bf8f6

winner by @jack | Suno

Epic electric guitar, Epic Chorus, clear Male vocals, clear male vocals, melodic, catchy, Triumphant, Power Rock Pop song. Listen and make your own with Suno.

▶ Play video
cosmic terrace
bright topaz
#

Did anyone manage to make music like Dark Souls boss fight theme?

spiral summit
#

It would be nice if the program recognized Brazilian Funk as a rhythm, it’s almost impossible for the AI to get it right 🙂‍↕️

~Thank you~

formal bear
worn atlas
hazy prism
#

please, make it so in v4 we can download individual stems for example for vocals, synths, etc. I love the vocals of some songs but they are so distorted or low quality I want to be able to mix them separately

dusk terrace
#

Wow guys thanks so much the update. I gotta say, I love having access to v2 model it beats them all melodically speaking. Just a hint, it has magic in there even beats v3.5. Amazing 🙌🏼🥳 and thank you again.

dim hearth
# formal bear please help downvoting on generations that have noise

i want to downvote but this means that 80% of the generated songs in my library are downvoted garbage and it takes me forever to click through the pages to find what i am looking for. i get it that you cannot delete them to use them in learning. but at least you could move them in another place than the library?

hearty star
#

Nice job letting us download the wav files 🙂 Please add to the roadmap a way to download all the wav/mp3 files in a created playlist ^^. I'm manually doing it now and it's taking a hot minute for ~100 finished songs. Worse, if I try to queue up several at once it terminates the previous job and there is no option for download wav/mp3 when I multi-select a group of songs.

forest spruce
#

@tacit idol

royal venture
#

Add the ability to report songs on the suno website that are just plagiarism / stealing an artists lyrics, like this guy who just essentially uploaded Im Not Afraid by Eminem https://i.imgur.com/mYcs1gG.png

Imgur
crude quest
#

Hi, is there a possibility to ask for a specific lenght in instrumental Songs in V 3.5. My songs at this point are very short. 1:22 etc. Not even close to the 2 Min in V. 3 and not close to the announced 4 Min.

bleak scaffold
#

while True: print(u'\u004c\u0065\u0074\u0020\u006d\u0065\u0020\u006f\u0075\u0074')

fickle meadow
#

Hi, is it possible to get a longer character length for the prompt, when creating your music.

late skiff
#

Suggestion: Split the Trending section adding style filters, so we can find more interesting songs in a specific genre.

#

Also, on the site site: Create challenges, competitions, with specific style/theme rules, and some prizes (extra credits? 🙂 )

placid vessel
#

most streamed page that shows the most streamed creators i fell like it woud be kinda cool

fair hazel
#

Suggestion: The musical structure should be less confusing and more practical. It would be interesting to have clear options on how many verses, choruses, and intros you want in your song, as well as the BPM during these parts and the musical style of each. It would be a dream customization, where the level of user control would be extremely relevant. I hope it improves in this regard later this year.

round zephyr
shrewd schooner
#

I don't know if this has been answered anywhere. But do you plan on adding the ability to download multiple audio tracks for one song? Like You make a whole song but you only like the drums, so download only the drums, then vocals are good on another so take that. I ask because I think it would be really cool to be able to take the best bits out of a song and be able to combine it into another. For instance I made (personally to me) a banger, but the vocal tone didn't quite match up with the instruments, and the guitar was a bit off then be able to reuse that file in a prompt so that for instance. "Those were nice drums, I'm gonna add this file as a prompt so the next track uses this drum track" This feature would be sick if it could be implemented

hybrid crane
shrewd schooner
#

Is there a plan of implementing a report for when a song get's flagged? Like I don't know what's wrong with it because it won't tell me. It would be nice if it said or highlighted the part which was flagged so people can readjust lyrics. I spent so many credits just making single verses to see which one was getting flagged, it took a while but now I figured it out. Would still be a nice feature.

mortal anvil
#

here's a suggestion, add a Personal Showcase for our profiles so we can give priority to our songs that we are the most proud of

worthy dust
#

I would love to have an „Empty Trash“ Button 😂 cos i always build tons of extended versions and trahs them away until i found a good one fitting my idea. To clear my bin is taking Hours 🙈

pearl ice
#

It would be cool if you could get more granular in what AI continues to reference as you move forward through the development of the song. If there was temporary adjustable knowledgebase that the AI could look too in addition of it's global reference knowledge base. Being able to adjust the 0s and 1s of each knowledge point would help the user get closer to the exact song they are attempting to bring into existence. Maybe even in the future, you could save that temporary knowledge point. For instance, if you develop a really nice sounding voice, you can add it into all your songs. 🙂 Thank you all for your hard work. Your platform allows me to bring my musical ideas into reality without the struggles and chains that hold back creation.

image.png
formal bear
formal bear
jolly merlin
#

I spent a couple hours reading all of your suggestions and there is not a single bad one. I would just like to summarize the biggest one I noticed and agree with 100%. Once a song is done it would be great to be able to add arguments / further define the song while keeping the overall structure Suno came up with. Such as: Increase BPM by 20% or switch voice to male/female or Add drum solo before 2nd verse.

robust finch
#

FOR SUNO DEVELOPERS... i have one offer/question and two suggestions: 1. offer: how you train your model, i can help with balkan folk, national, ethno music because have big library of that music or i can suggest which singers to choose from spotify/deezer.

#

two suggestions: 1. please add id3 tags in downloaded mp3s for premium users.. could you add info about created song (lyrics, genre, other usefull info) into mp3?

#
  1. more descriptive tags which could change music creation.. these words between [ and ] should not be only predefined commands but full descriptive sentences for AI like [keep instrumental for 10 seconds before switch to next tag/part of music] or [use only vocal singing for intro and then slowly add other instruments] and all that stuff.. It would help ppl to generate more customised songs with less randomness where it is needed _
#

btw, suno singing very well on my native serbian language.. I have ideas hot to use it to make songs in punk genree focused on politic and social situation here and upload it on youtube. Can i do that as premium users as long as i mention suno AI as main brain of all that _

ebon cradle
#

3.5 is a big step back. Longer songs but terrible quality and doesn't respect prompts

analog notch
#

Could you add it as an option to allow or disable the AI to possibly continue singing after its sung the last verse you put in the lyric box? For instance, I really needed an extension to stop after verse 3, but there was time left to use, so the AI started repeating random lines of the song. After spending all those credits I had to give up on the AI

robust finch
robust finch
ebon cradle
robust finch
analog notch
ebon cradle
analog notch
#

i tried everything

robust finch
ebon cradle
robust finch
#

when we train 3.5 model in level where bugs are gone, then its right time for comercial use and credits_

ebon cradle
tepid trench
#

v3.5 is early-access, for us to test if we want to spend our credits on it. I'm happy to do this and use v2/v3 if I want to actually try to make anything serious.

analog notch
#

ADD A SHUFFLE BUTTON!!!! Simple

red wind
#

This might have been mentioned but I really miss being able to sort songs into folders instead of just playlists

tight beacon
#

Suggestion: "Parental Advisory" tag for songs with strong language in them.

night lagoon
tight beacon
#

And yes even I make songs with strong language in them

leaden musk
#

it would be great if you could specify a female and a male singing voice for lines

#

like so

#

[Verse 1: Female Singer]
Good morning, dear Jill, I must convey,
Your messages bring joy, in every way.
Though I struggle to keep it brief,
Your kindness brings my heart relief.

[Verse 2: Male Singer]
But in these words, a hidden plea,
For constant praise, from you to me.
Her need for love, her need for light,
Masked in sweetness, out of sight.

hallow frost
#

You sorta can

#

i get it to work

#

you have to specify both vocals

#

and if you want to flip you gotta use duet

worn atlas
hallow frost
#

so like Ethereal Female Vocal, Street Rap Male Vocal, Duet

#

this works great on 3.5

leaden musk
#

ill try thank you i tried but didnt get it to work

hallow frost
#

What is your style prompt

leaden musk
#

Female and Male singer, chant with baroque violin, upbeat, energetic, stomps and claps

#

was the one i tried

#

im currently trying duet, whiney Female Vocal, and rational Male Vocal,baroque violin, upbeat, energetic, stomps and claps

#

with vocal instead of singer in the lines

hallow frost
#

Ok well when you get into classical genres it can be a lot trickeir

leaden musk
#

didnt work either for me

hallow frost
#

What genre? thats not a full style

#

genre sets everything up for you

gilded wagon
#

Perhaps the mods can make a "Prompt Feedback" channel where the community can help each other with any prompt style struggles we face.

leaden musk
#

trying pop right now eg:

#

Pop, Ethereal Female Vocal, Street Rap Male Vocal, Duet

hallow frost
#

so if you did like Yeah it should work

leaden musk
#

hm currently it dosent

hallow frost
#

but it can be tricky you have to add descriptor tags to your verses

leaden musk
#

im doing it like so

#

[Verse 1: Female Vocal]

hallow frost
#

No

#

[Female enchanting ominous Verse 1]

#

Female is a descriptor

leaden musk
#

ah gotcha so i wouldnt use Ethereal Female Vocal in the style prompt

hallow frost
#

vocal is the instrument

leaden musk
#

but in the lines?

hallow frost
#

you can also [Vocal Morph to Male]

#

and vice verse if you dont want to try to prompt two singers in a style

leaden musk
hallow frost
#

I dont really try to get specific here but you see how i get the rnb singer to do the hook?

leaden musk
hallow frost
#

the suno syntax is very specific

#

If you want Dark Drill You must type "xxx, Dark Drill, yyyy," if you put a comma between them it breaks it into a new factor

leaden musk
worthy dust
#

😄 sry ... should be a dm 😄

woven pilot
spiral acorn
#

Could there be a setting on songs to switch between two types of videos?

  • One option which is the video as it is now
  • And another option which only has the Song cover along with an EQ visualiser
worthy dust
woven pilot
#

banned

vague prairie
#

I feel like there could be some more documentation added around these. 😁

vague prairie
#

Having a lot of fun, but man its a mine field trying to figure out the way to combine and where.

worthy dust
#

im trying to descripe the block

#

but as we just tested ... long lyrics with often changing types ... not really working ...

male, female, male, female, duett 6 / 10 tryes like .... male, male, duett, male, femal

if you have only 2 verses .. with one change ... the [ ] will work fine ... so you have to split your parts

hallow frost
#

[ADJECTIVES INSTRUMENT ACTION TARGET]

#

[Female Jazz Vocal Break]

#

, , , , , , , COMMAS CREATE OBJECTS. Dark Morphing IDM, = 1 object / Dark, Morphing, IDM, = 3 different objects

worthy dust
hallow frost
#

If you are trying to create a VOCAL you need to have it in the STYLE PROMPT

#

Yeah I usually do 1 male verse and 1 female verse

#

because I make high bpm music so i usually only get 30-45 seccond clips

worthy dust
round zephyr
#

Hear me out:

a separate fee (honestly id pay 50$), upload a certain amount of clips of you speaking, get your voice trained and you can select to use with your generations

hallow frost
#

Thats really important

hallow frost
hallow frost
# worthy dust Thx for Sharing 🙂

You gotta understand different genres have different vocalsist weights, I never get 100% more like 50% but once you get your prompts straight it will become pretty reliable

hallow frost
#

A single lower case letter can break a whole prompt

#

NO spaces in front of brackets [

#

Trip hop is almost always female rnb/rapper and you can cross into either jazz to get somekinda hybrid funk or rap and get a really reliable mic, trip hop 80/f rap is 80/m

#

Doom metal sounds awesome

worthy dust
cosmic terrace
#

a higher plan for addicts of suno 😄 50k credits or something like that. just an suggestion 🙂

torn cave
#

songs should be grouped into projects to help with visibility

torn cave
unborn token
#

Seed. an option to get the same vocals in different songs.

frosty venture
#

Please add the ability to split the audio tracks of a song with vocals. I'd love to be able to clean up the audio of a production.

#

Or to add more instrumentation

desert juniper
#

3.5 has very odd vocals, but I guess that's why it isn't v4 yet - the form and style so far is VERY good tho. But... for 4min songs, we NEED a far longer Lyrics window - at least double the characters - I cannot fit my full song in there.

hallow frost
hallow frost
#

your telling me man, i wrote aall lyrics no [] on a rapidfire dnb vocal and it went 45 secconds

#

but it creates amazing flows and is way moe adaptive to getting the words to fir the patterns, imean 3.5 is amazing, i just wish it didnt give me 50% tracks under a minute and I REALLY hope i get some kinda refund I have like 1000-1500 credits in glitch tracks

hallow frost
worthy dust
hallow frost
#

Nah Im not that young, and it's like ... some minor damages definitly worn in good

safe ingot
wheat temple
#

So far v3.5 is an improvement over 3.0; however, v3.5 can benefit from improvements in clipping mitigation.

hallow frost
#

Autosort Collection according to genre

#

we gotta get that collection in check ive got 100 pags total and 17 liked (we can do better that 14.25% like ratio too)

#

my like ratio is greatly afected by tracks that just do nothing or get uite an less that 30 secs

feral aspen
#

Suggestion: Allow lyrics input and the instrumental toggle. For normal prompting, users already find it useful to use lyrics as instrumental guidance. Whatever special sauce is happening under the hood in instrumental mode might be extra interesting with this guidance. For my prompting, I want to see if it does anything with normal lyrics.

late shard
#

rethink 3.5

#

its not as good as 3. it makes things way more generic and less unique

formal bear
#

we are not going to take v3 away from you

late shard
#

good to know ❤️, ty

feral aspen
late shard
#

noted

late shard
#

geuininely curious about examples of 3.5 compared directly to 3 with same prompts where 3.5 gives a better result

#

again, not trying to be rude or trying to start something, i just want to know what im doing wrong possibly

safe ingot
formal bear
feral aspen
#

If I had to lose one, I'd actually prefer losing V3 over V2. v3.5 is sort of V3 but better (with some prompting adjustments). V2 is still pretty different than either.

formal bear
#

ok ok...

late shard
#

3 is miles ahead of 2

#

3 is so good

formal bear
#

before v4 is out we are not going to take out v2 as well

formal bear
late shard
#

im waiting on examples of 3.5 being better than 3

#

in terms of style and creativity

feral aspen
# late shard 3 is miles ahead of 2

In general, yeah. But V2 can go into some less professional sounding areas that I, at least, can't really get to in 3 or 3.5 except very rarely. V2 also sends the V3 models into different spaces when they continue off it.

late shard
#

ill link a couple more 3 vs 3.5 one min

hollow locust
#

A way to make sure you don't put the same song in a playlist more than once. And an indicator for songs which have been removed, to know that you previously had them in a playlist. That, and a search function with your own library and playlists.

late shard
#

one sounds authentic and creative

#

the other sounds like its on vh1

feral aspen
#

If I had to guess, it's the 'rock'. 'folk-rock' may do better.

#

maybe toss some noise. and if it's ignoring some words, repeat those, they may be weaker in 3.5. indie-folk, folk-rock, psychedelic, psychedelic, psychedelic, folk, ;;;;;;;;;;

safe ingot
late shard
#

let me test

safe ingot
late shard
#

both results , the vocals are just clean and sterile

#

the music is not creatively mixed

#

neither of these are special

safe ingot
#

It refuses to give anything "special"

late shard
#

exactly, its all too sterile

safe ingot
#

Safe is the word I found fitting

late shard
#

brother is this country music television?

#

let me run that same prompt on 3

#

outshining 4 in terms of being music

#

4 might have higher fidelity but ... whats the point. if this is the wrong place to discuss this plz let me know where to take it, im just frustrated

safe ingot
#

I am comforted to see others with the exact feelings I have

late shard
#

ill do one more comparison

#

brother WHAT

#

making it with 3 now

#

if you cant see why 3.5 is a step back then im not sure what else to say

#

maybe you can fill me in on a better prompt than just motown, which is all i wanted from both gens

sage mirage
#

I mean I'm not doing motown but like, 3.5 is better than 3 imho. The amount of control we have now rules

feral aspen
sage mirage
#

Of course it prompts different than 3 but that always happens

late shard
#

ok so in 3.5 if i want motown that sounds like m otown what do i type

late shard
#

influencer music

#

also look how much longer the prompt took you to engineer

feral aspen
late shard
#

im going through my old gens and redoing them on 3.5 and only 1 of maybe 50 ive done sounds 'better' overall

#

its like, hey lets give the mona lisa an ai filter, it will be more high def

#

brother, the character is what makes it

safe ingot
#

Every post on reddit agrees on your point of everything sounding pop-washed, they just won't accept it here tired

late shard
#

brutal

sage mirage
#

It's...right there

late shard
#

i dont go on reddit much but ill put in my voice

#

i do! and i love it, and ill continue to sub if 3 is around

sage mirage
#

So your suggestion is "make it better'?

late shard
#

no

safe ingot
#

I think the suggestion is not take the soul out of it even more to make it generic and safe. That is my fear for v4 as well.

late shard
#

my suggestion is to realize this is the wrong direction and maybe try a different approach

#

im sure facebook will love 4 because of how clean and sterile and safe it is though 💸

sage mirage
#

And you still haver your 3

late shard
#

^

#

when 4 comes out, if they get rid of three im out

sage mirage
#

I dunno I'd probably just either mess with 3.5, explore new ways to try to summon whatever it is you want. Or don't, it's in alpha, you don't have to use it if you don't like

#

All you have now is more than you had before. Nothing is finalized, all in alpha, and tony clearly hears you

#

Since, like I said, for someone that messes around with weird stuff 3.5 rules

late shard
#

im not gonna spend credits trying to force the new version to do what 3 does better already

sage mirage
#

I'm firmly in the "right direction" category

late shard
#

i cant argue your opinion

#

but there is OBJECTIVLY less unique sounds coming from 3.5

#

enjoy it or not, its the truth and i can keep showing reciepts

sage mirage
#

That's not objective at all

late shard
#

whats your musical background

#

not that it would make a difference on your opinion

#

im just curious

sage mirage
#

I think it would otherwise you wouldn't ask

late shard
#

its both

sage mirage
#

I am not here to convince you to feel different than you clearly already do. I suppose I'm more of a "you don't speak for everyone"

late shard
#

thats fair

sage mirage
#

I've been here long enough to see this cycle though

#

The names change but the tone is the same

late shard
#

and im not trying to have you change your opinion, i just want examples of 3.5 doing a better job in a way thats NOT audio fidelity or vocal timing

#

thats my point!

#

the tone is changed

sage mirage
#

Status quo -> a change -> split community, one likes the new stuff, the other group hates it forever

late shard
#

so much

late shard
sage mirage
#

We still have people clamoring for v3a, since it was objectively the best

#

and so too it will happen with 3.5a

late shard
#

it probably was

#

i wasnt around for it

sage mirage
#

It was exactly your thing, but you and I switch places and also it isn't available anymore to me and I'm mad about it

late shard
#

id like some examples of 3.5 that dont have sterile autotuned front and center clean vocals

sage mirage
#

And I am demanding the alpha back since the main model isn't as good

late shard
#

and examples of 3.5 with unique panning and instrument mixdowns

late shard
sage mirage
#

~Don't Panic~

late shard
#

it blows my mind everytime i gen a tune with 3 at how good it is, so thats why im frustrated if they are looking to replace it

sage mirage
#

Especially since tony says they won't* take away previous models

late shard
#

but...

#

thats what a dev said above

sage mirage
#

I just assume you haven't found your prompt

safe ingot
late shard
#

ive been 'helped' to make my prompt better for 3.5 but all it does is make it the prompt way more in depth than i feel it needs to be, and then it still spews out 'safe' mixdowns

#

im a mixing mastering enginerr and music producer of over 15 years

#

the difference im talking about isnt subjective

#

its objective

#

you can like 3.5 results but you cant say they sound more authentic in any way

#

i mean you can but

sage mirage
#

You don't want to 'prompt in depth' ?

late shard
#

i want to type motown and get motown

#

sure i like to go in depth

sage mirage
#

and you are a literal audio engineer?

late shard
#

even if i type a paragraph of description it spits out sterile results

#

ll make a motown song on 3 and you make one just as good on 3.5

#

yea?

sage mirage
#

Then if I were you I'd keep using v3 then if you don't want to test v3.5a

#

Since it will cost money and you would rather not

#

tony has seen how you feel about it, so I guess just chill and stick around?

#

models kinda take a bit to make

late shard
#

yeah , im a little salty now sorry if im being a dick

feral aspen
late shard
#

im just scared for the future of suno

sage mirage
late shard
#

im still ready to go toe to toe with anyone on motown 3 vs 3.5

sage mirage
#

Dude in v2? We only had simple mode. We also couldn't choose when to continue a clip from

#

v1 wasn't even stereo

late shard
#

right so going from 1 - 3 was a big step up

#

but 3- 3.5 is a step sideways into sterileville

#

into generic pop

sage mirage
#

People said this exact thing about v3

#

Still do, sometimes

late shard
#

ive never actually messed with 2, ill do that later tonight

sage mirage
#

That's the argument for the people that still want v3a back, that was the most creative bestest one

late shard
#

wish i had tried that one

sage mirage
#

It wasn't, it was a whole lot better than v2, but then when v3 proper came out it was 'complete'

#

Took a bit to shift how you prompt it, since every model goes through that

late shard
#

i think if you want motown you type motown

sage mirage
#

and with v4 we're getting inpainting and infill

#

Which also is a massive bonus to usability

late shard
#

thats really crazy tech and im sure ill love that function but.. whats the point if i dont like the results of 4

#

ill just move to ... u-

sage mirage
#

genre-wise

late shard
#

if you dont mind spending credits can you make a 3.5 motown song of any subgenre style

#

but make it have creative mixdowns

#

im gonna run the same prompt both ways now

#

night and day

formal bear
#

easy easy folks

#

I think one thing I start to learn, is, there probably isn't one model that fits all's tastes

late shard
#

sorry, im a little heated

#

of course you cant please everyone

formal bear
#

no

#

I think we can

#

we just need personalization

late shard
#

but i just want to ask if people see how its objectively less good at making unique and emotional music

sage mirage
#

weird way to say inpainting ultrafast

late shard
#

inpainting doesnt change the whole songs mood and mix

formal bear
#

instead of serving you fast food burger, customize to your criteria

#

if you want non-meat burger, go for it

#

some people will puke on it, some people will love it

late shard
#

ok but whats happening in 3 is fast food burger, and 3.5 is plastic fastfood burger that looks better in a picture

#

but i dont think it would taste very good

#

would look great on investors sheet though

#

again, soryr, im heated, i should go cool off

sage mirage
#

Yeah but see I don't feel that way so then we get to the point of customization and tailoring what I like vs. what you like within the same system

late shard
#

id like to customize my plastic burger

#

since i dont plan on enjoying it, just making it perfect

#

i appreciate your time and not banning me for having an opinion

#

did my best to be civil

sage mirage
#

that's not how Suno works

late shard
#

love suno

formal bear
sage mirage
#

They don't do that

late shard
#

im gonna go chill

#

❤️

#

hope you all have great evenings

#

sorry for bad vibes

sage mirage
#

You too man, we'll be here when you come back

#

Nah don't worry about it

feral aspen
#

I think there's like 100 burger customizations suno could offer without model tweaks. I may soon resume my nagging about some of them lol

sage mirage
#

It's clear where this is all going, and the two boxes we have to type stuff in is absolutely not enough

formal bear
#

but no I think on top of that

late shard
#

last thing, why would there be any reason to remove a previous version

formal bear
#

if you go to a restaurant that you like and knows you, you just walk in, and they will make a burger to your taste

formal bear
late shard
#

ok but what if the restaurant change dthe ingredients from natural to plastic

formal bear
#

harder to move forward if you have too much tech debt

late shard
#

ok ill stop

late shard
steep jackal
#

Why won't Suno help artists promote their music on Spotify and Youtube -- Seriously -- we churn money for you in the form of subscription fees -- you guys won't answer emails or lend any assistance to pro-lvel subscribers wanting to build a fan base. Kinda crappy.

late shard
#

anyways, thanks for being available to talk tony

#

👋

formal bear
feral aspen
steep jackal
#

Suno helps promote snippets on their Suno gallery -- which is NOT monetized for authors. Suno will not even answer e-mails or promote subscriber music. BS.

sage mirage
#

If it's montetization you are after I think Suno might be a little not exactly ready for that, directly. There is still a lot of production that generally needs to go into tracks from Suno to really polish them up

#

My opinion, of course

late shard
#

def less sterile than what i was getting, but everything is still 'safe' . clean up front vocals, no wild creative panning, no vintage warmth

#

i like them though

#

its like everything gets sprayed with a pop mixdown

#

i gotta step away, i appreciate you taking the time to make those though

feral aspen
#

Vintage Warmth is pretty promptable, at least. Creative panning is an interesting challenge to think about

tight hamlet
#

please allow us to follow playlists. thanks!

steep jackal
#

Would LOVE the ability to uploade a previous SUNO-generated song to 3.5 or 4.0 or whatever and have the vocals cleaned up while the overall song remains the same. 😄

hollow locust
#

A notification email, whether you're subscribed (to the email) or not, that informs you your subscription has ended.

hearty star
#

Showcase section is too limited. There's only 10 slots - there really should be at minimum a few hundred or preferably an endless 'show more'. It would also be cool to have a history for it; so we can see what was previously showcased. It seems like the showcase is functioning more like an alternate 'trending' category? Or am I thinking about it wrong? What is the current function? To eventually show all the songs people put in suno-showcase channel?

Changing topic: It would be beneficial to Suno as a company and for us as users if there were a textbox to limit the generation length. Hear me out. Suno generates 1-2 or even more minutes now. But I only like 15 seconds and want to regen after that (as an example). This is very often and even less sometimes like 5 seconds or even 2. The point is Suno is paying for computer, power, resources, storage space, etc - for a heck of a lot of data that I don't even really want to keep. It isn't good to train on because I don't like it, other people may not either - and I will probably never go back and use it for anything. So in three months I have 300 minutes of good stuff I kept - but 45000+ minutes of content generated that Suno has to continually store. That means Suno is literally wasting and wasting and wasting again a ton of resources! Like when I say [Fadeout] at the end but instead of fading out it does 45 seconds of instrumental, 2 more verses and a chorus; but I only wanted it to finish the 2 seconds of fadeout it already was on track to do 🙂 If I could have just said 'limit generation to 2 seconds' maybe I wouldn't have had to be so wasteful generating 8 times to get 16 variations only 1 of which I kept but that was the one it REFUNDED because it was only 2 seconds.

I hope you can see how silly it is and why it's not beneficial for Suno or us as users of the service to continue without a 'limit generation length' feature. It creates an unsustainable dilemma later on for Suno, to store bad data forever, I think.

hearty star
# steep jackal Would LOVE the ability to uploade a previous SUNO-generated song to 3.5 or 4.0 o...

I have had some kind of success with this; if you continue a song made in 2 or 3, like at the very end and just use the same prompt and the same style, 3.5 can basically do an 'encore' of the original with the same style but with 3.5 flair. If you click my Discord name, in the profile is my Suno public link. Click that and go to "Do Your Best". That song was a test to see how it handles exactly that sort of thing. The first minute is in 3.5. The second minute is in 3.0, the third is in 2.0 and then the fourth minute finishes in 3.5. What you will see is they are all the same words, same tunes, etc - but each version gives a flavor of its own and 2 kinda struggles with synthy voices but it IS possible to do this already, with enough trial and error. For that test I didn't cherry pick - it was 1 generation for each and I chose 1 of the 2 for each continuation.

feral aspen
#

Because 3.5 has 2 minute memory, that's enough to cover a lot of shorter songs

#

I did something like this in earlier versions. Extend a clip without lyrics in the extension, and it most commonly still sings but repeats random phrases and and lines from the previous clip. Often not well, but sometimes interesting or at least funny. It's like asking Suno to improv off the previous verse.

torn kraken
#

i have a suggestion

#

what about a mastering service to increase the tune sound quality,
then a distribution service,
kinda like landr or distro kid etc

would give suno a kind of potientially one stop shop for somepeople

safe ingot
torn kraken
#

its still pretty easy at the moment,
just means another sub to somewhere else
spose the other service do check the song for somethings b4 it uploads to the streaming services just idea
i like to do a mixture of ai with a blend of fruitylooops i like to do some twists and turns, then if im happy ive chucked a song or 2 up

#

but yeah i could see a spam attack happen

#

but how do you find the happy medium

#

i nominate @safe ingot as quality controler

tight beacon
#

Industrial Metal in Suno is not Industrial Metal it mixes Industrial EDM with Metal. Which is two different styles, Industrial Metal is a style of it's own. Suno needs to figure out the difference.

#

I think a solution would be let us specify Industrial-Metal as wanting Industrial Metal, That way Industrial EDM can just be "Industrial".

drowsy dove
#

Make it so paid users can upload (hopefully their own) voices for their songs. This would rock

(Im certain others have recommended this -- its an implementation, and "company bandwidth" thing, not a concept thing)

wary hinge
#

Hello, it would be appreciated to be able to delete multiple songs at once. Right now it's really painful to see hundreds of songs sitting there waitting for one-by-one deletion...

lethal trail
#

Click one song, hold shift, click a song further up/down in either direction, then click the 3 dots and select "move to trash"

#

Or if you want to select every other song, you can use select instead, however you have to manually click on the songs you wish to delete

delicate cedar
#

I would (again) like to suggest the ability for one person to tip another person some credits. I think as time goes on this feature becomes more and more useful.

worldly sphinx
#

stop making us pay to alpha test v3.5 or at least make us pay less. v3.5 clearly has glaring issues, but we still pay full credits for the failed (noisy, degenerated, too short) generations

wary hinge
kindred prawn
#

dear developers, I'm ready for a crop button, where I can clip the first or last seconds off 🤙🕉️

worn elm
#

PLEASE create Jump Up Drum & Bass

north dune
#

Isn't Suno a clever combination of Chirp and Bark? In the future can we download the vocals and instrumental separately?

brittle crane
#

Hmm.. did anyone suggest a setting for the amount of songs that are generated at once yet?
Seems kind of a waste of credits to generate 10 if you only need 1 or 2 tries.

hybrid crane
brittle crane
leaden marlin
#

but every time you click "create" it generates 2 at a time regardless

brittle crane
#

Uuuuuh I get it... My bad then. No problem, got more 😂

hybrid crane
brittle crane
#

😄 quick edits to a song rather than having to sacrifice all, voice and instrumentals alike,
just because the AI messed up reading.

hybrid crane
brittle crane
#

Niceeee. 😄

#

That weirdness with the create list is getting fixed too I guess? 😂
Like.. if there's more than x songs in the library, only the recent are shown?

hybrid crane
#

In the right upper corner you can go to the next page page

lethal trail
#

Right, I was supposed to put this in here a while ago. Would it be possible to have the "New" songs moved higher up on the main site? As it currently stands, you have to go past; Suno Showcase (which in all honesty doesn't take up that much space), Trending (which should be near the top if not at the top), followed by four different sets of playlists. First User made playlists, top categories, Fresh new categories, Community playlists followed by your liked songs. And then you get to the new songs.

It honestly doesn't make any sense as to why the new tab is so far down on the site.

visual brook
#

i would love if 3.5 allowed the first generation to be user configurable length instead of 4 mins

ie. i suspect v3.5 might actually produce higher quality if users could select 1min gen to start (like LLM transformer benchmarks, less tokens for some benchmark leaderboards makes some LLMs better at shorter generations)

then allow users to extend

tender basin
#

Question: Is it possible to make Twitter-Embeddings to play without redirecting to the song page?

hybrid crane
tender basin
#

No, i meant when i post a Suno-Link on Twitter, there is an embedding (with a play-symbol) created. But when you click the play-symbol you get redirected to the song's page instead of playing it (or the lyrics video) on twitter.

potent folio
hybrid crane
potent folio
#

basically a convo in general about the downvote system, user preference data, and some aspect of how people all have a different understanding of it.

#

I'm sure its all stuff yall have already considered

#

but I thought it was interesting

tepid trench
#

Like with v3 alpha, giving some free credits to aid with testing v3.5 might be nice, particularly as it seems to behave quite different to v3. (I'm going to continue throwing my existing credits at it regardless 😄 )

worn atlas
reef gazelle
#

are there some promo codes in sight for the near future?

long hawk
#

Quality is my number 1 wish. I have several songs that I really like but can't use because they distort in parts.

robust jewel
cold plume
#

I'd love the ability to add more to the style of music tab for. I feel like the amount you can type in there is not enough and limiting. i absolutely love the 3.5 vibes. sounds more realistic on the voices.

cloud wadi
#

3.6 seems to not understand prompts

late shard
#

3.5 is a step backwards

cloud wadi
#

Very much

trail briar
#

ive been making music for 9 years and a tool like suno is actually very interesting if it can be used as an addition to my creation process. i already use ai a bit when it comes to mixing and mastering with izotope plugins because it helps for the more boring part of music production. i still know how to do the manual process and that helps a lot but if i can use a tool like suno to make production more fun and faster thats very nice. i still prefer the aspect of choosing exactly what you want to do with manual music production but there are some stuff i would really like to use suno for as a addition to my production.

  • what i didnt use yet but may consider when quality gets better is using ai beats or instrumentals to sing over.

  • i think suno would be good for making instrument loops that are already mixed like guitar loop or drums and all those real instrument sounds that are hard to reproduce in a daw so i could use them as a sample in my songs. only problem i noticed when i last tried in v3 was that those real instrument sounds where not synced well enough or were just too hard to sync manually with cutting and stretching. if it could be more stable when following a bpm or a given bpm prompt that would be very nice because those guitar loops it can generate are very good.

-specific bpm prompt or bpm range prompt/slider would be very useful or a specification of what bpm was applied after the generation

-mastering still isnt that good in my opinion atleast for the genres i tried but for some i was able to get impressive results when applying manual mastering or ai mastering with ozone mastering plugin.

  • i also noticed the v3.5 is better at producing some genres now but it still doesnt recognize some less known genres or underground genres. for example uptempo or very heavy distorted bass music/dariacore/melbourne bounce/uk bassline/or some prompt like perfect fifth bassline harmony. it doesnt understand some more technical terms like fm/panning/etc
#

-some edm genres are very loud and are hard to get or generate with suno so it would be nice if some more loud or niche genres would be added without getting quality drop because of the loudness

-a way to explain more precisly some type of pattern you want for example i had a hard time generating a precise pattern like kick-bass-kick-bass like in some bounce/melbourne bounce song

-the flow of the vocals definitely got improved tho and thats very nice for people using that feature in v3.5

-also when i tried making instrumental there was still often a predrop vocal in edm songs or some weird vocal mumble at some parts but that will probably improve with time

-in v3.5 the quality often gets worse near the end and bad quality happened more often but i still got what i wanted more often too or at least closer then with v3 even tho i think v3 audio quality is more consistant

my biggest request or hope for Suno would be the stable bpm or precise bpm choice and more loudness and genres option (and ofc sound quality but thats obvious)

marsh oak
#

This is beautiful SUNO, but after 3 days, it started to bore me, if it doesn't manage to give us the separate elements, on stems... it's not really worth SUNO. When he manages to deliver us acapella voices, then the BIG GAME begins, until then...this thing is weak!

sage mirage
#

Suno doesn't use Stems at all

#

There are no stems to give

#

You can use various tools to break apart clips after the fact with any number of tools, but your results may vary

marsh oak
#

But SUNO will never have STEMS or MIDI or ACAPELLAS generation?

robust jewel
#

It may be an option in the future

hybrid crane
marsh oak
hybrid crane
#

In the ideal world.. everything would be finished 🙂

worn elm
#

To extract acapella uses Moisés app

lean zodiac
#

can we had a rule : to post a song in #🐣┃suno-showcase , you must listen at least 2 songs ? it reset each time you post.

robust jewel
#

You could certainly do that, if you wanted to

terse sleet
#

Generate instrumental STEMS or be able to split the STEMS from the generation for additional customizations.

Also maybe allow a duet style vocal.

woven pilot
#

3.5 sucks so far . i keep getting 9-20 second songs smh

robust jewel
#

As a first clip or a continuation of a clip?

buoyant scroll
#

To. Admin
ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡ
I am very satisfied with the latest update. There were often issues with the sound quality in version 3, but it seems to have improved to some extent. For the next update, I would appreciate a focus on enhancing sound quality, especially for acoustic instruments like piano and acoustic guitar, which still have low sound quality.

Most importantly, for existing users, I believe it is crucial to have a feature that allows us to remaster songs created with versions 3 or 3.5 using future versions. While version 3 provided good composition, the sound quality was lacking, so we had to spend many credits on numerous generations and expansions to achieve the best possible sound quality.

I anticipate significant improvements starting with version 4, but it is essential that existing songs are not neglected. Please consider developing a remastering feature for previously created songs.
ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡ

fading grove
#

what would be really cool is if you could make songs in the discord, the same way you can create images in midjourney in the discord group

robust jewel
#

That was an option a while back

fading grove
#

can they bring it back please?

hybrid crane
fading grove
#

was it difficult to maintain?

hybrid crane
fading grove
#

Would be so cool if there was a twitter bot that created songs

#

even if people had to wait a day or two to get them

hybrid crane
#

In the future they will offer an API so people can do things like that.

fading grove
#

that would be so awesome

marsh oak
#

I tried to make a soundtrack, only with real instruments, ancient world instruments... It doesn't work. Maybe soon?

cloud wadi
#

I think Suno should work on being better at following chord progressions, and getting the tempo input

#

It would be amazing if soono became able to read Guitar Tabs and stuff, but I know that would probably take a long time to implement

#

Maybe if there's a way you could mess with the sound levels of a song you generate. Like being able to turn instruments off and stuff, and get the separate tracks like vocals and drums and guitar and stuff

marsh oak
#

I tried to make a soundtrack, only with real instruments, ancient world instruments... It doesn't work. Maybe soon? I am referring to ancient instruments with wind strings.

robust mason
#

Hello Devs, Could you guys a "edit lyrics option" so that we can manually type the lyrics of our song in the correct order, the only real way the lyrics make sense is only when you fully copy paste a song together on the first generate... the problem is when your creating in sections and correcting lyrics mid verse, it doesnt replace the old ones when the full song is complete.. instead it just adds the 2nd paste of lyrics below it that also repeats with the new variant of words if any.. I want to be able to edit manually the lyrics at the end of the creating process so that people tuning in can follow along correctly from the full lyrical summary, of course if the they click the video they can see all the edits one made in the extend process, but it doesn't reflect that way in the full lyric summary that is created at the end.. if you understand what I'm trying to say.. if u want me to elaborate with an example than just @ me

safe ingot
#

I am noticing almost every single gen I make in 3.5 for metal, the vocals seem to be as if they are reading out perfectly stuck to a metronome on the beat. They get zero space to have actual character and are very bland, compared to v3 where they'd actually get some unique delivery.

zealous idol
#

一度生成した歌詞付きの曲の歌詞だけを変えた状態で生成できるようにして欲しいかなと。

woven pilot
#

any way to add an equalizer since we cant get stems?

flat sphinx
#

two things, 1: the audio drop to in quality in the middle of the song until the end, the song starts clearly and distinctly then it transitions to a dragging scraping sound with high tones. "One can't distinguish the instruments, and it just becomes some sort of 'sound mess'. 2. Suno needs to learn more section labels, other than verse, chorus, and bridge."

amber finch
#

it would be cool to have more lyrics character count. after using v3.5 with max lyrics i have songs that can still fit plenty more words.

tepid grove
#

I hope Suno can extend Instrumental Soundtrack without lyrics, and Instrumental (V3.5) can extend to Part 2, when I extend Instrumental always be refund😂

potent folio
#

Could we get a genre for "Experimental/Outsider"

robust frost
#

I would like it if you could regenerate the video after you have uploaded your own cover

lethal trail
#

In the future, would it be possible to see where a piece is extended from, through the lil

image.png
#

Would make it way easier to keep track of where you are, and what you've done/tried so far

lean zodiac
#

Can you display the duration of the music somewhere (without we had to launch it) when we are in the library / a playlist ?

ebon cradle
#

3.5 is definitely a disaster and we pay with our credits to be beta testers. It's crazy and disappointing. I guess that's when competition really begins with the upcoming eleven labs generative music. Too bad they didn't work harder. I love this product but they forgot about competitors...

worldly sphinx
tepid trench
#

Where have you seen this officially stated?

#

Ah. Speculation. Got it.

#

How about some facts?
#🪲┃bugs-web message

User: "is this true? is the quality of the clips delivered on a per-server-load basis?"

Developer: "This is false"

tidal isle
#

Maybe a music analyser for existing songs that tells what genre this song is.

tepid trench
#

So are you saying that less people are using v3.5 after release day, and have gone back to v3, so that's why v3 sounds bad now?

#

(also a rare instance of the "instrumental" button actually giving me a full-length instrumental in v3.5!)

#

Honestly I haven't had much luck getting good quality with v3.5 in general. I have 2 upvoted songs out of all my attempts and this was one of them 😆 but I have had issues even with v3 so I'm just throwing credits at it so I can upvote/downvote to give feedback if nothing else.

upbeat mantle
#

It would make it a bit easier if there was a Delete account button in The Manage account area due to that it would be faster then send in a resume for it

sacred shard
#

Since you seem to be using the 👍 and 👎 votes for training and improving the model it would be nice if there was some kind of an extended optional voting interface that would allow me to better express what I mean with it, since in particular a 👎 can mean many different things, e.g. "bad sound quality", "bad lyrics sync", "does not really fit the style prompt" or simply "technically fine but I don't like it".

tepid grove
#

Same to use “classical pop, piano, violin, guitar, 1990s, 125 bpm” in V3 & V3.5,
I think V3.5 needed to improve the music structure, but V3.5 is early access only, it have times improved
V3.5 music have 4 mins, but the song was below 2 mins, V3.5 the time control also should be improved
https://suno.com/song/f010b060-8718-41b5-a59a-28fafb72d089

https://suno.com/song/df45f8d4-0aef-40e8-b33f-8d6ec5748858

Timeless Love by @connie881 | Suno

violin piano 125 bpm 1990s classical pop guitar song. Listen and make your own with Suno.

▶ Play video
Timeless Echoes by @connie881 | Suno

1990s 125 bpm piano guitar classical pop violin song. Listen and make your own with Suno.

▶ Play video
full verge
#

Today, v3.5 is having trouble, took my points and didn't make the song. Just spinning

weak bloom
#

AI created 3D visualization via 3js to accompany Liked songs

tepid trench
full verge
#

now i tried again and the same thing just spins.

tepid trench
full verge
#

it has been spinning for 15 minutes now, it has never spun for so long

weak bloom
#

Mine spinning too now, must be working on an absolute banger

serene trench
#

Здравствуйте, дорогие разработчики Suno!

Ваша версия 3.5 просто великолепна! Многие вопросы были решены, и это просто потрясающе!

У меня есть ещё одно предложение: было бы здорово, если бы можно было перегенерировать трек с уже исправленными ошибками в словах. Иногда я замечаю ошибки в тексте после генерации, а трек получается очень хорошим, и исправить эти ошибки уже невозможно.

Кроме того, искусственный интеллект не всегда правильно понимает, как нужно петь некоторые сложные слова или слова, смысл которых можно понять только по произношению. Также есть слова, которые редко используются в речи. Было бы замечательно, если бы можно было помочь искусственному интеллекту с помощью ударения на слог или букву. У нас для этого есть специальный символ «`».

#

Было бы замечательно, если бы искусственный интеллект мог анализировать загруженный трек и на основе этого анализа создавать музыку. Это позволило бы нам создавать именно те композиции, которые мы хотим услышать.

full verge
pastel lynx
#

You shouldn't need autoclicker to go back a 100 pages in your library

hybrid crane
full verge
#

It's still spinning for me 😖

cedar sun
#

I would love the option once I have generated a song with lyrics, to be able to download it as usual but I would love the option to download just the instrumental version of that same song.

shell kestrel
#

I don't have much luck with 3.5. For instrumentals it creates rather short tracks. Often less than a minute. Extending before the song ends is also pointless because it only adds like 18 seconds.

#

Using 3.5 to extend v3 instrumentals will also only add a few seconds and immediately finish the song. Maybe it needs a setting for the desired time?

#

or if the song closure is desired

fickle blaze
#

Please extend the Inputs, since your models can now handle longer Lyrics etc.

shell kestrel
#

yes. but instrumentals have no lyrics...

#

It works very well with regular songs. I tried that too.

fickle blaze
#

Actually with the 4m generation we could paste our whole lyrics

#

But we are still restricted

shell kestrel
#

what do you mean with 4m generation? The 3.5v also accepts the whole lyrics though which is a huge step up. Just instrumentals like harp songs require a lot of credits to finally get one which is longer than 70 seconds

fickle blaze
#

It’s cutting off my lyrics at a specific character length.

#

I’ll count them wait a minute

#

ChatGPT says it’s 1189 characters

#

Its 1250 actually using other tools

proud moat
#

Hi! :), I've been working with Suno for a long time and have a subscription, and also realese official , I think it is possible to improve the quality of the sound and the supply in "psy trance", the kick is very spread in most generations, and the psychedelic sounds are very few, we need more supply for this style, for example rock and metal is amazing, But in psy trance I would like to see more improvement if possible.
I find myself doing so many generations until there is some good and basic rhythm to the style, it can reach 5000 for one track, if you can put more emphasis on this style, it will be really helpful.
It seems that the AI ​​needs to learn more references from Psy trance tracks from the leaders in the field, and of course they have subgenres such as Progressive psy trance, Goa Trance and more, which it would also be amazing if it improved and learned them, so that we could create such tracks.
🙏 ❤️

ember crane
#

Now that you have millions of dollars in funding can you please hire an S-Tier product designer to rework the home page

#

specifically the show more BEING AT THE TOP OF A LIST

pliant tendon
#

Not sure if this has been suggested, but for personalization purposes, could we choose a default model for ourselves? I often have to switch to v3.5 manually. It may be the default post-early access but it could be useful for people who prefer v3 over v3.5 in the future.

bronze oriole
paper blaze
#

I would like the ability to edit the displayed lyrics. When in extending a song multiple times from a middle section of the song, by default, all the lyrics are shown added together and it just becomes one big mess that does not reflect what's actually sung.

For examples, if I have a verse 1 Verse 2 and chorus, but I extend from before the chorus to remake it, it'll show the chorus twice in the displayed lyrics. So it'll be verse 1 Verse 2 chorus chorus, even tho the chorus is only sung once.

It would be great to be able to go in and update the lyrics with what's actually sung to match the song itself

#

Yea. It would be great if suno can separate the backing track from the vocals in the future. Sometimes I really like a track, but want to redo the vocals in a diff voice or diff melody style. Or I can regenerate a part of a song, like the chorus, and slap it onto the track

#

Currently when u regenerate the song by doing extend, it can sometimes change the music and make it not what u intended or like

young coral
#

whats a good way to get an intrumental intro like piano or acoustic guitar etc?

#

ok ty

#

Is there a way to do a duet?

#

like a male and female singer?

#

It did it for me accidentally lol

#

I cant recreate it tho

digital anchor
#

I wanted to sing along to my song, but I can't. The chorus is way too fast. I'm really sad about that because otherwise I really like the song. Could Suno perhaps be made sensitive to the fact that a song's lyrics can also be sung by real singers?
https://suno.com/song/76fcc79e-57cc-4a94-9a04-9079829d8748

The Shepherd's Flute - Version 4, aktuell endgültig by @zauberfrau ...

metal rock with flute, post-grunge style, male voice, minor, but hopeful song. Listen and make your own with Suno.

▶ Play video
haughty temple
#

When v4 comes out, I imagine that it will improve the audio quality of generations (hopefully drastically). It would be great if it could allow us to use credits to regenerate our existing songs with the higher quality audio.

chilly junco
#

Hi there, I would like to suggest adding a button to the "create page" to all clips. Currently, there is no accurate way safely to remove from the list one or a few clips instead of downvote them which I guess is not good for the feedback. This shouldn't be "remove" but "hide" most likely.

final radish
#

I would love an option to take instrumentals from full songs. I would also like to generate a video but customize the lyrics before.

unborn ocean
#

i would like an option in the songs to get only the music instrumentals (interact with the star if u like the suggestion)

low adder
#

I like to use Suno to help me generate from pieces of music I wrote that I've hit a road block on. When I find one I like, use another AI tool (or soon now will be AI Agents) to extract just the guitar track. Then feed that to another AI tool that will then listen or read the raw audio data and transcribe it into tabs. I can then learn and play in the real world. Thus bringing something from the digital world into the physical world. Thoughts?

arctic nest
#

Hello guys, I'm new here, can I ask how to custom lyrics to get 10 first seconds is Instrumental, I try many times but it's usually start with vocal from 1st seconds

#

I tried

[Intro]
[Powerful guitar riff]

or

[Intro, Instrumental, setting an atmospheric mood, BPM 120]

or

[Instrumental intro]

It's all doesn't work

chilly junco
arctic nest
faint osprey
#

Feedback 1. 3.5 is not as good as 3.0 in music and tempo when creating a slow soul song. It keeps adding finger snaps and the tempo is too fast especially a leading into the instrumental solo for a slow jam. here are two examples. from 3.5 https://suno.com/song/2993fa61-78b1-4d07-96d7-04a81b3066fa From 3.0 https://suno.com/song/03fc29bd-67d8-485f-9db0-f50a216612f4 Please find a way to keep the the tempo when slow is desired and remove the finger snaps. or provide a way to ask for it or not in the style prompt

P4a by @aitc | Suno

soul female no finger snap song. Listen and make your own with Suno.

▶ Play video
P4a by @aitc | Suno

soul female song. Listen and make your own with Suno.

▶ Play video
arctic nest
#

In my personal feeling, I feel v3 in May doesn't return good result like my first try on March, I can't explain why

cedar mulch
#

make delete key on selected tracks move to trash, please

tepid trench
#

Before rolling out early-access versions to all pro/premier subscribers, allow users to volunteer as testers. These should be people who aren't just going to abandon the latest version within a couple of days because they don't like it or because it has bugs, and who are willing to provide constructive feedback, accurately flag songs for poor quality and report bugs.

dusty spoke
#

While I'm here, my two cents on v3.5. It seems to have "less fantasy" than v3.0. Which is awesome when you want a verse with the same melody as the previous two, and it comes after a bridge, chorus, and solo. On the other hand, it seems (to me at least) a lot less likely to generate interesting, creative solos. I guess there's some parameter change involved; if it is so, the ability to adjust those parameters would definitely send me straight to Cloud Nine.

mental radish
#

It would be nice to be able to remove certain instruments or omit specific instruments all together. Many of my EDM songs end up having a high pitched synth drone after 1 or 2 extensions. I've tried using prompts to fix it, but I just end up having to generate multiple times to keep it out.

#

Also, the ability to set tempo would be great!

dusty spoke
#

Actually, scratch the sentence about "interesting, creative solos". It took me twenty-something generations to get a solo at all.

worthy torrent
#

Would love to keep my remaining credits after cancelling a subscription. I've already paid for it and being able to use them, I assume, would still help train the AI model? Other sample libraries/AI generator plans do have this policy in place. I understand that it's a business model but taking something away from a user generally tends to be viewed quite negatively.

hybrid crane
# worthy torrent Would love to keep my remaining credits after cancelling a subscription. I've al...

There is a reason for not being able to keep them: credits in subscription plans do not roll over to the next month. This policy is in place because Suno.ai reserves computing power based on fixed monthly credit usage. Allowing credits to accumulate can lead to unpredictable demand, making it difficult for Suno to manage resources effectively. This unpredictability can cause service disruptions and increased operational costs. By resetting credits monthly, Suno ensure a smooth and consistent service for all users.

worthy torrent
warm lily
#

Dear Suno Team , i love it so much i must tell you and that you real deeply changed the world .
You invented the language for all , you can say.
I am so unbelievable happy to can use SUNO at my lifetime and never cried so much positivly since i use it.
After 42y i can say this is the absolutly greatest Software invention of the world , far after GPT, hole Mircosoft or Apple or TESLA and even the Moonanding.
TRillionBillionsThanks. On of the next is a Song about SUNO. Deepest LOVE to you all, TEAM!

hybrid crane
worthy torrent
#

Maybe add a credit cap instead, i'd prefer not being able to have more than an X amount of credits at a time, than actually losing credits. Maybe split the tiers up, amount of credits/month and higher credit cap the higher the tier, but get rid of the credit deletion NODDERS

daring ocean
#

Hey!
I would love to suggest to add some checkboxes on songs so I can move to trash more than 1 song at a time! 😄 Or it IS possible and i'm being dumb here? Hahaha 😛

hybrid crane
#

then with the last one, click on the 3 dotted menu and choose move to trash

daring ocean
#

Ooh nice!
Any way to do it on mobile?

worn atlas
hybrid crane
opaque gull
#

an optimization function if a song is basically good but still needs perfection

worn atlas
mental radish
#

An option to remove lyrics after a generation is done would be great. I've had a few times where I get them even though I say instrumental or other times when the metatags will be used as lyrics.

prime stirrup
#

Please devs give me Folders in the generate tab! pretty please ❤️

worn atlas
mental radish
#

Having the ability to build song structure in the custom mode would be nice. I would like to do instrumentals, but still have control over the song structure. it could even be able to have structure and style prompts in the same prompt window.

ocean anchor
#

"4-minute max clip length. You can now get your full song in a single generation!"
IMO this isn't useful. 30-60 second clips would be better—I can quickly judge whether it's good, and then extend or discard. With 4-minute clips, when the song is bad, I lose way more time judging it, and I still need to listen to it all because it may change for the better along the way.

bronze oriole
#

Maybe a channel with known working brackets like [Verse] [Outro] etc?

#

Unless I'm missing it

mental radish
# bronze oriole Maybe a channel with known working brackets like [Verse] [Outro] etc?

if you haven't seen it yet there is a page on the website that has quite a few https://suno-ai.notion.site/Using-Metatags-34944efe09ec41d693e314f13af44695

Suno on Notion
Using Metatags | Notion

Metatags guide prompts but aren’t verbalized themselves. Like stage directions or parentheticals in screenplays, you can use metatags to describe a specific part of your prompt (e.g. [laughter] and piano interlude) or to influence your entire prompt by describing speaker characteristics like [whispering] or music genres like rock.

#

The list there has the option to look at v0 and v1 tags. there are quite a few more though. In theory anything in brackets should work as a metatag, but in practice even the known working ones don't always work as they should.

ocean anchor
#

It would be nice to have the song’s date, genre, description, lyrics, and other details as file metadata once you download the audio or video file. This way, the information would be available when you right-click the file and select Properties (Windows) or Get Info (macOS).

shrewd urchin
#

I'd like to join the pile of people asking for PayPal. I can use neither credit card nor google. Ideal or PayPal are pretty much my only options and I am wanting to experiment and perhaps use some Sumo songs KillerbeeDance

formal bear
scenic remnant
#

I just download a wav for the first time, always did movie, the WAV quality is so much better (38mb instead of 6 mb).

Can there be a video download with the same audio quality as WAV?

#

the voice sounds more realistic and less robotic too in WAV

bronze oriole
#

The ability to Mass Delete

grim geyser
# scenic remnant the voice sounds more realistic and less robotic too in WAV

ability to output only the voice as well.. I hink that the robotic sound comes from he noise they started with for the diffusion, that one bit of noise is repeated over and over.. wonder if instead of using a single noise frame, that continuous noise could be generated, ie.. a new noise sample to diffuse with each itteration.

rain ore
#

That works in desktop. Not sure how to do it on mobile

sleek lance
#

Is there a way to regenerate the video if we changed the image?

split flax
#

I know it has been said before, but man would it be nice to have a search bar in our library. Takes forever to find anything. >.<

grim geyser
#

@split flax I think that is in the works as we speak because what was in the announcements.. 3.5 early access is available to subscribers, but that doesn't have the search bar in the library yet either, so here's hoping.

amber finch
#

small and simple suggestion. you can change covers for a song. being able tto get new suggestions for song names would be cool to. maybe based on the lyrics (if any) and the style

cloud plover
#

There are a bunch of Genres that Suno currently does not recognize at all. May be interesting getting some training from Spotify (?) Or a website such as Everynoise.com

formal bear
shrewd urchin
mint bridge
#

3.5 is absolutely amazing ..... .......

whole spade
#

Hello, my suggestion would be to have the instrumental and the voices in separate files so we can download the full song, instrumentals or just the vocals.

#

Also, when we cretae a new song with X named but then we change it later, the video created keeps the old name, would be great if this could be changed so it matches any future changes to the song.

#

Also, some lyrics get out of the border of the video cutting the lyrics in some cases, would be great if lyrics autofit to fill the visible part of the video.

#

Also, it would be great to have an option to download the video landscape as well.

#

Also, having the option to edit the lyrics after the full song was created would be nice as some lyrics do not match 100% the song (extra lines or repeated lines) in some cases.

left crescent
#

@obtuse tree can you add the option to Suno where I can duplicate a song that I created and change the lyrics in the duplicate but the song sounds the same as the first version if you know what I mean?

obtuse tree
left crescent
sacred shard
#

3.5 seems to have a higher chance to create surprising elements which is great - I wish there was a "temperature" slider to allow that to happen more often.

upbeat mantle
#

How about adding a notice when when u trashing songs so u have the chance to undo the mistake if u remove the wrong one

trim gull
#

I not yet allow to comment in Languages. But please add Laos language because we have to write many rock and heavy metal Laos songs. Thank you. It’s a second language to Thailand in SE Asia.

undone bough
#

when you are finished with the song there should be some kind of possibility to clean out access lyrics, maybe do it in the "Edit" screen maybe even make an AI do it for you. Because if u use to extend a lot u get many verses and such that's not being used.

#

also a stem splitter would be nice, and a cleaner that cleans the basic

#

i think if u make this lyrics cleaner people would use the video export a lot more often since the text now aligns with the song

#

wich means more advertising for SUNO

#

and the drums should really be improved, i know it's lo-fi but often the drum sounds broken.

#

Also if u get a good voice or sound but the rest sucks it should be possible to copy that voice or sound into the next Create. also somethimes u get a awsome break or solo or outro but the rest sucks, would be nice to be able to mark the things you like and be able to bring them over to next Create 😛

#

all this should be possible to code a AI to do 😛

#

OH i forgot! often when u use other language the first few words are starting with a few words with bad pronouncement and slowly corrects itself but it ruins alot of good songs and makes u spend way to much credits (i guess thats good for u) but its extremely frustrating having to spam the Create button just because of a few words in the start is not correct.

#

it would even be possible to splitt the tracks and let ai suggest improvements in sound both quality and maby change of type of instrument

#

A click menu with commands made just like the new music-type menu would also be nice being able to search then and add them when you write the lyrics. Maby make a whole new section for writing lyrics yourself, (stil be able to get a suggestion from ai) bit being able to go inn and change and fine tune the lyrics and having the "command" menu available would be awsome. This alsdo takes away a little bit of the bad vibe people have about ai music generators since you actually write much of the lyrics your self.

#

sometimes im stuck on a verse where i cant finish a line and make it keep the meaning of the verse and still rhyme, adding that possibility to ask aSunoAI to sugest a verse or two would be awsome. I mean i can just go to a external AI in my browser and do this but why not add it in SUNO keeping us in here 😛

#

and now something that really would blow peoples mind! a lyrics to video generator making a script from your finished song and splitting it up in like 10 short scenes of animation 😛 with choices of different styles. KABOOM you would start a revolution if that happened 😛

amber finch
#

im sure this has been suggesting already but extending the beginning would be cool, like how Udio does it. (i prefer sunos quality, but having udio-like tools would be cool)

undone bough
#

yepp also being able to edit the music style after song is finished

hybrid crane
#

That would help with editing.. but music style is probably a step further

undone bough
#

i kind of use the music style as a prompt also, and then you are stuck with all that and cant edit it

#

i mean just the plain text

#

not possible to change the "music style text!

amber finch
#

after using 1000+ credits on 3.5v, pls increase the lyrics count. I have 4 min long songs with only 2.5 mins actually having lyrics before it just randomly reuses old lyrics. theres plenty more room for more lyrics

undone bough
#

i have this under music style on my finished song "upbeat, pop melancholic rhythmic, guitar intro, , deep male voice, guitar, bass, drum, guitar solo,! and cant change it

#

u know u can extend the song?

amber finch
#

not rly the same and getting the lyrics to work togethr is less gaurenteed

worldly sphinx
#

This may be super hard to do given the async nature of the process but I'd like to have the ability to interrupt a generation (say I made a typo)

amber finch
#

yes

#

pls

undone bough
#

agree

amber finch
#

lots of wasted credit bc using the wrong ai version or typoing something

worldly sphinx
undone bough
#

and just one at the time at the cost of 5 😛

amber finch
#

or maybe, give a warning if the style or lyrics has a typo? when there red squiggly lines?

#

"check your prompt, are you sure you want to use this?"

undone bough
#

that must be possible with all these AI language out there

worldly sphinx
#

that wouldn't always work, e.g. synthwave always has a squiggly line despite being a legit genre name

hybrid crane
#

With V4 and inpainting, little errors are easily changed.. so looking forward to that!

amber finch
#

cant wait

hybrid crane
undone bough
#

problem is in song writing u deliberatly change gramma nd shortens words etc to make it rhyme 😛

amber finch
#

good point

worldly sphinx
undone bough
#

i gues there should be a choice of correcting or not then.. a "ignore" button

amber finch
#

im rly hoping inpainting will be able to giv more control over vocals, esp when one song has two differnt voices

undone bough
#

cant w8 to check that out

#

its like christmas

amber finch
#

lol

hybrid crane
undone bough
#

uh where can i read about that?

hybrid crane
#

You can't.. I heard it first hand

worldly sphinx
#

yeah, is there a roadmap somewhere?

hybrid crane
#

No.. no public roadmap. Probably because things can be extremely fluid with new tech.

worldly sphinx
#

that makes sense

undone bough
#

i need to know how u set theese time stamps 😛

worldly sphinx
#

but still, somehing like "we're working on X and Y"

#

would be nice, even if vague

hybrid crane
hybrid crane
#

Like stems

amber finch
#

nice

undone bough
#

hmm im using audacity to split up the stems and then i cut and paste and often make new drums and bass because they often suck 😛

#

also a good way to improve and clean or change the sound

#

then mix all bits together

amber finch
#

dont suppose v4 can take a previous song and "upscale" via the inpainting can it? I got a song riht out the gate in v2 that is amazing but it needs a quality glow up

hybrid crane
undone bough
#

anyhow if they listen to us (which it seems the really do) SUNO will be even more ground-breaking than it already is. I do see thast real musicians feels threathened but hey think of it as a instrument of inspiration to make new songs instead, just use it as a brainstorming app and then go perform it yourself. That makes you even more effective then u ever were before ; (using 2 years in a retreat to write :P)

wind marsh
#

I would like a way to have the AI to remember a song it creates so that if I want to create the same song but have it in another language I can.

amber finch
#

artist hate ai but ive seen muscians more mixed on it

#

prob bc theres been ai-like vst and plugins the automate a lot of stuff for a while now, SUNO is just next logical step... well.... leap maybe

undone bough
#

well, creativity lies in many layers, i have written lyrics and poems, etc since i was a kid. Always dreamed of making some of it into songs. But then I have to hire people or get people to spend their time on it so i just put it aside. Now i can just paste in my lyrics and choose my style and wupti there it is! Does that make me less of a poet/ lyricist or less creative? dont think so 😛

#

ooh, i forgot! It should also be possible to show people a finished song after it has been remixed externally, and upload the finished result. but i guess that would require a lot of security measures and server space.

worn atlas
left marten
#

Hi there, i've been using SUNO a lot but i'm definitely held back by the limit on free accounts. Unfortunately, i have no creditcard and i feel conflicted about getting one. Is there an expected date for paypal integration so i can keep making songs and support the product by paying?

undone bough
#

a song lock would be nice so you dont delete something by accident when u try to clean up

#

and be able do select everything but the locked songs and delete 😛

bold cedar
undone bough
#

Also if u add a song to a playlist it should be possible to see...somwhere wich playlist/s its added to

hybrid crane
bold cedar
hybrid crane
worldly sphinx
#

Probably already suggested: would be nice to have access to the specific seed used for a generation, so it can be reused

undone bough
#

yes agree and it does not matter if its already suggested if a majority suggests it it might get priorytized 😛

bold cedar
#

And we know when version 4 might be released, they said something about it?

hybrid crane
pseudo lake
#

<sarcasm> I think it would be helpful to allow a small whitelist of particularly responsive and observational users early-early-access to v4 to give feedback and iron out all of the wrinkles </sarcasm>

hybrid crane
tepid trench
pseudo lake
hybrid crane
#

😬 I helped testing V3a before release and V3.5 before release.. So they do

pseudo lake
pseudo lake
tepid trench
#

I kinda worry that people frustrated by anything v3.5 does wrong will just downvote the song which could affect any efforts to fix the degraded audio quality.

worn atlas
undone bough
#

more space for adding style its obviously being used as a prompt than some few single words 😛

#

things like breaks and solos are more likely to appear i find if u add it in the "Style of Music" maybe keep the "style of music" but in addition have a "music detail description"

#

this one defenetly worked 😛 "medieval guitar drop medieval edm flute outro flute intro fast-paced" when added to style of music 😛

tired mortar
#

can we add sindhi language in ghazal format as an option please?

left crescent
dark drift
#

Please add ability to "remix" generated tracks, as well as extend them.
Just pressing "Create" with the same lyrics and style can produce wildly different songs, It'd be really useful to say "i like this version, but i want to change that lyric", or say "i like this tune but i want to add layered synths to the style", and have it reproduce the same song with those minor tweaks...

true umbra
visual brook
#

i think screaming is generally a failed task for the v3.5 model, i have maybe 2 examples at most i've heard in many samples

pseudo lake
visual brook
pseudo lake
#

@visual brook I'd mention that it's more a "proof of concept game" in my mind due to the insane level of randomness (legit "experimental" psychedelic music, utter blasting screamo, spoken word with bizarre circus ringleader voices, entirely sung on an inward breath, a 30 second radio jingle, a perfectly completed masterpiece)

But if you have some credits to blow on it, and want to spark some creativity, get some GREAT extension points to seed your desired songs from, or just have some fun seeing what 3.5 is capable of (and get a GREAT sample collection of poor audio to report & downvote in the process; ❤️ you devs) I'd really recommend "playing"

visual brook
wary mica
#

Can we have dedicated channel for hallucinations?

pseudo lake
vagrant rock
#

Can we have more than 6 'structural' parts generated by the simple mode please?

worthy venture
#

Add the ability to remove the voice from a created song, and conversely to be able to remove the instrumental from a created song (So being able to save separately from a generated track, individually the voice (acapella) and/or the instrumental)

raw cave
#

Look similar? Why not add more controls? Time to take back the lead!

blissful wolf
#

Today's UDIO ANNOUNCEMENT is just full of "complicated UI, redundant features, and unnecessary use of complex technologies.-- over-engineering. Users either don't need them or don't know how to use them." So SUNO please don't take that route .

hybrid crane
#

apperantly you learned nothing.. Tony warned you and now you are pinging Keenan again. Last warning. You'll be timed out for a week. @leaden marlin

visual brook
blissful wolf
# visual brook i strongly disagree with this Apple, walled garden, philosophy; although at leas...

ALL OF THESE ARE UNNECESSARY:
--- ADDITION of a two-minute model
--- random SEED ( or ANY seed)
--- prompt or lyrics STRENGTH Higher prompt-strength improves adherence, but may lead to less natural sounding music. Lower lyrics-strength can lead to more natural vocals, but sometimes lyrics can be ignored.
--- clip start-time can be used to control where you want the generated clip to start in the context of a full song.
For example, 0% corresponds to the beginning, 50% to the middle, and 90% to a clip from the end of the song.
This is especially useful in combination with the extension feature, but it also means you can always start a track from an intro, for example.
--- a generation-quality slider, which lets you trade quality for speed-of-generation, ( SUNO's speed is super fast and not an issue)

tepid trench
#

More features are better to have than less features.

raw cave
visual brook
# blissful wolf ALL OF THESE ARE UNNECESSARY: --- ADDITION of a two-minute model --- random SE...

presuming you applied the same logic to Bark (transformer using an Encodec SOTA codec) by Suno

there are studies on "decoding" in transformer architecture - Suno backend is probably still transformer based:

https://arxiv.org/abs/2402.06925

at worst, there is simply not enough evidence in audio transformer decoding to say if extra features help, at best Suno should provide more decoding specific evidence and features

if only this were some domain with lots of research behind specific transformer machine learning features that influenced generation quality in various domains! /sarcasm

arXiv.org
A Thorough Examination of Decoding Methods in the Era of LLMs

Decoding methods play an indispensable role in converting language models from next-token predictors into practical task solvers. Prior research on decoding methods, primarily focusing on task-specific models, may not extend to the current era of general-purpose large language models (LLMs). Moreover, the recent influx of decoding strategies has...

wheat sierra
#

Would it be possible and a good idea to extend mp3 files with this AI even if it's not generated via the website ?

stray sparrow
#

Suggestion: Since v3.5 can make longer songs (up to 4 mins in one gen), I think the lyrics generator in custom mode should make longer lyrics, since current lyrics generator make very short ones for a 4 minutes song (TBH they are short for a 2 minutes song in most cases). Oddly enough, when you turn off custom mode the generated lyrics are longer.

clever hinge
#

Can we please have a normal creation field like before? In this dialogue box now my screenreader has issues with letting me create, it just doesn't do it for some reason, plus it's completely unnecessary, maybe visually it looks good, but bothers me more than anything honestly, thanks

visual brook
twilit frigate
#

Calling anything "Chiptune" makes the song start off with this repetitive triplet that often works itself into the motifs, which reduces the uniqueness of the songs a lot.

flat seal
#

Does anybody knows how to have choir in the instrumental behind without having them on vocals ?

worthy junco
#

Be able to save a voice that generates with a song for paying members, I've found some really kick ass sounding vocals but can never keep them.

marsh oak
#

I noticed that in 90% of the drops, SUNO uses the same synth and that gets annoying... the same synth tone appears on the drop side!

safe ingot
#

Display seed used, and let us set seed for generations.

prime stirrup
pseudo lake
pseudo lake
# worthy junco Be able to save a voice that generates with a song for paying members, I've foun...

I genuinely wonder how easy it would be to accomplish this.

Not familiar with what happens on the backend, but if it's anything like image generation, the initial content is able to be made quickly because it's diffused from random noise-- steps can be made to try to nudge certain results out of the noise, but unless you've trained the model on how to recreate something specific, it's always going to have some randomness.

The image comparison would be like making a random image of a person, then wanting to generate that exact same person in another photo-- unless you've created a LORA to add context for what the specific face or style should look like, you're limited to how accurately you can describe that person/style, and even then the AI can only try to "imagine" what you've described, instead of "knowing" exactly what you want.

hoary thistle
#

Mentioned this on Twitter, but the ability to save stems of a track Suno generates would make me a paid subscriber yesterday - unless there's a way of doing this already that I'm just missing. Even just the ability to seperate out clean vocals from the rest of the mix. I need to try experimenting with prompt shenaniganry to generate very clean vocals, that might get me where I need to be, I don't have a solid enough understanding of the underlying technology to know what's plausible unfortunately despite some background in GANs more generally.

pseudo lake
hoary thistle
#

Huh. Somehow did not realise aplitting extant audio like that was effectively a solved problem, hot damn. I'll look into that

paper blaze
#

Please have a Prompt to prevent the ai from deviating from the written lyrics. Very annoying when trying to end a song with some repeating chorus and the ai decides to belt out an earlier verse instead

#

I'm currently a pro tier, and have topped up a few times, and was considering premier. But I won't be topping up anymore or doing premier until the ai gets smarter. Gave me nothing but frustrations all day and wasted credits.

#

And here's another suggestion. For every 10 downvotes on a song and/or moved to trash, we should get some credits back, like even 10 credits. Kinda unfair to burn thru so many credits due to the ai being stubbornly dumb, and at least some compensation would alleviate the pain and frustration.

leaden marlin
#

Suggestions

  • Make it so you can download vocals or instrument seperately
  • Basically the same thing but the option to download the stems for the song without using a AI tool to do it (makes the quality sound dogshit, etc)
  • New payment methods, not everyone has a credit card.
  • A new lyric generator, this is a long shot but there's not one AI yet that can actually make good lyrics so one that can would be awesome.
  • Fixing the detection AI, i forgot what its called but if i make a song that isnt following the bible it denies me. litterly anything that the ai finds offensive or bad the song wont generate and thats really annoying.

I don't know if all of these are possible right now but all of them would be super useful

reef marten
#

I see that now when I click on a prompt it goes to a search for that prompts. that's an un-feature. I can copy a prompt and search on it myself. It was much more useful to copy a prompt by clicking on it

#

I would at least like to be able to turn that off as an automatic feature

#

also, "female singer" is a nice prompt. works pretty well. "male singer" on the other hand got me a female singer every time I tried it. FWIW

viscid bloom
#

How do i get this to work 😮

reef marten
#

? which ? Suno ?

shy badge
#

Suno need to let users to keep the seed of a music to use it with another lyrics

inner rover
#

We need a feature where suno users can, “just like the extend button” you can clip songs, like a song that’s too long for instance, you can shorten it, or clip certain parts out you like, gd that would be nice. 👍

reef marten
#

you can continue from mid song.

But it would be really nice to remove and inappropriate beginning (or middle)

inner rover
#

Yeah 👍 that’s what im saying

#

So basically you could clip a beginner or end you like, and just extend off of the end or beginning, whichever is the preferred clip or whatever

reef marten
#

also, some way to frame a verse. I still have a lot of songs where the verse structure does not match the way this is laid out in text. This worked really well until some time in april. Now, most of the time, one line will be out of the verse, or the chorus-style singing will start in the middle of the second verse. so some adherence to the [verse] [chorus] [bridge] or if that's not possible a way to simply redo that section, without having to discard everything that came after it

There's just always one in a song, a verse that gets messed up, and if I redo the whole gen, it'll mess the others up on the retry, the ones it got right the first time, and I have to manually edit in a daw

reef marten
inner rover
#

@reef marten Yeah, mostly likely, but I hear what your saying, so basically you render a song, and you love everything but the chorus for instance, so you can just edit certain parts like the chorus and change just that section. 👍

worthy junco
sage cloak
#

Few feature ideas that I feel would be great to have:

  • are you planning on adding a feature to extend a song but on the beginning of the song? I've had good songs generated that would really need an intro added.

  • In a similar stream, maybe a redo button that would re-generate a section between two timestamps

  • A way to set an estimated max time for a clip (eg only want a 1 minute Instrumental)

  • Way to set the tempo in bpm, key and (maybe) time signature

  • An API would also be lovely to incorporate generative music into thing (eg games)

quiet lark
#

i just wanted to say, devs, i love where yall are taking this software. Please keep it up

velvet shale
#

First off, 3.5 is amazing, kudos...I was wondering...would it be possible instead of just an extend from could you do an extend to and from? So I could be like say start the song at 1:32 and extend from 2:10...Thanks for everything guys making songs is so fun!

quiet lark
inner rover
#

Love 3.5

tiny ocean
#

Love 3.5! First maybe a redo but you can choose to keep the lyrics, or keep the music and get new lyrics, or keep the music but add your own lyrics. Second a way to stronger set to have male or female vocals. And third this would just be the bee's knees lol but if there was an option to stem out the track I.E. drums, guitar, vocals, ect. Love Suno and can't wait to see whats next.

feral aspen
mint bridge
#

Can we please get a larger text box for the initial lyric entry in the wonderful 3.5?

mint bridge
#

3.5 is so open and responsive to guidance, but when I use these new capabilities I can only enter 2.5 to 3.5 minutes of lyrics

velvet shale
#

3.5 babyyyy

formal bear
feral aspen
# formal bear ok how much longer?

I posted this in memes, but as usual with me, it's only half joking. In a sense the limit is already 5000 characters for 4:00 of audio for any clip except Clip 1, because two 2:00 Extensions are also 2500 + 2500 = 5000, right?

length.png
formal bear
#

it is already 2x the 2min one right now?

feral aspen
#

Well only Suno knows if it completely falls apart at very high numbers or something. But assuming it doesn't, there are people doing very fast raps that could use 5000 in 4:00 even without using brackets. Now if this also happens to make the clips hallucinate a ton or make the song rhythm bad, maybe it should be capped, I can't really speculate about that. But I can say assuming it did sound okay, at least some people could use it. Perhaps a warning popup instead of a hard limit (< 2500 characters is recommended).

formal bear
#

ok let me test it once

#

give me a few mins 🙂

feral aspen
#

The same is true for style prompts in 3.5, which at least so far, seems to prefer long styles. However, extra long style prompts definitely increases the chance the style prompt is sung so increasing style length is a change I would proceed more cautiously on (make sure a warning message is in place, or it's tested in 3.5 and turns out not to be much worse than a maximum length prompt already is)

formal bear
formal bear
feral aspen
formal bear
#

I will ask mc to do it tmr

mint bridge
formal bear
#

I am pretty sure 4k is already reaching the limit a bit

mint bridge
formal bear
#

ofc

#

it's my job 😄

frigid pivot
# paper blaze And here's another suggestion. For every 10 downvotes on a song and/or moved to ...

perhaps instead of per 10 downvotes, just when a song receives 10 downvotes below positive. In other words, the total should be negative 10, even with upvotes. so if there were 20 positives, there would have to be 30 negatives to receive a credit refund. This way you can't break the system as easily. If you received credits for every 10 downvotes, you could just make a credit generator song with 10 fake accounts. A guaranteed no on that feature, if that were the case.

paper blaze
frigid pivot
#

ahhh that makes a lot of sense

swift parcel
#

I've said it before, but it would be nice to have a list with playlists and the ability to sort by name or date within playlists to help with organization. 😵‍💫

frigid pivot
#

Suggestion: please let us cut existing songs. Sometimes generation goes way too long and ruins a good song

paper blaze
frigid pivot
#

yeah, it tries too hard to reach that 4 minute mark

feral aspen
#

You can fix this by using credits and Extend From until you get a hard stop, but this is also a feature I really want. Not just for the convenience of trimming songs, but also will be useful for generating songs because you can trim the song history in ways you can't specify using the current Extend From feature (you can't remove extra audio before a specified time).

frigid pivot
#

Discord channel suggestion: A forum channel for workshopping. Say you want to create a song together with a group of other users, using their input, collaborating etc. Another example would be collaborating with others to create a community-made guide to advanced prompting. I'm sure many would use it. Midjourney has something similar for their visual art AI, a channel called Image Jams. If you're in that server, you'll know exactly what I'm talking about, and the attraction of such a thing

raven flame
#

Feedback : Ver3.5 early access is very exellent! I compose Japanese language AI music with suno. I write all Japanese language lyrics. Ver3.5 early access makes all perfect Japanese language singing voices with my Japanese language lyrics! Thank you very much for great works of suno staffs!

spiral acorn
tepid trench
thick reef
#

I was thinking, would be cool if it had an "Advanced settings" where you can make a chord sequence for it to follow.

upbeat mantle
#

add a Delete account in the Mange account Area so u dont need to send in a resume in order to get your account removed

upbeat mantle
#

More Medieval instruments like Hurdy Gurdy and Keyed fiddle

spiral acorn
kindred hull
#

It would be great to have a "Remaster" option next to the "Extend" option, the purpose of this option would be to denoise the song, make the audio crisper and sharper and add a more detailed and richer stereo landscape.

spiral acorn
# kindred hull It would be great to have a "Remaster" option next to the "Extend" option, the p...

yea that'd be nice! Perhaps on a clip you could select a mastering option "Studio", "Airy", "Organic", "No Remaster" or whatever, and if you set it to "No Remaster" it'd return to the original sound. And when extending clips it would always be using the non-remastered track so you dont end up mastering something that is already mastered

(or it keeps applying the last mastered setting on extends, and you could simply change this back to "no remaster" if you don't want it on - and it would return to the original audio)

golden axle
#

Maybe the possibility to separate the vocals from the instrumental, so you can use the instrumental again for a male and female version of a song 🤔

kindred hull
spiral acorn
#

i think this is something they want to add at some stage but are waiting for the quality to be at the right level where they feel they can do it well

prime stirrup
wooden fable
#

I think the current video design is a bit confusing visually. It's hard to see what is lyrics and what is not. Could the lyrics be visually separated from the rest somehow? Here's a crude design suggestion which I think makes things more clear.

(Not super important though, because I still think the chorus'ish fuzzy sounding vocals should be the primary focus of the bug fix team.)

suno_video_current.jpg suno_video_suggestion.jpg
#

What I did here:

  • Added a box with a border and a darker background.
  • Used a different font inside the box to further distinguish things.
  • Moved elements apart, in order to group related things together.
  • (Forgotten on the screenshot) I think the currently audible sentence should be the same font size, but rather change color instead, so that things don't jump around quite as much.
#

Another thing: Avoid changing too often. Every time the text skips, your eyes will have to hunt around to figure out what happened. In fact, making it scroll smoothly would probably be even better.

feral aspen
#

I will echo that I almost never want any website to automatically scroll the screen based on events like a song finishing.

wooden fable
#

websites tend to scroll in very uneven and nasty ways, so i can see why. Maybe skip the smooth scroll idea. 😄

worldly falcon
#

Dunno if it's already being suggested, but a way to change/correct parts of the song, without changing what comes after

amber forum
#

add seed so we control the song generation

#

Suno website is not well optimize. Plz optimize it

hybrid crane
pseudo lake
#

I enjoy how in this instance the "THX" reaction can either refer to "thanks" or the American company known for exceptional audio quality.

potent crane
#

Liking the wav downloads but...you have them at 48000 when the universal is 44100, when I convert the song drops a tone (slows down), if this is deliberate then I will still have to d/l mp3 and convert it in post.

viscid cloak
#

It would be cool, if we could add Genre's to our Playlists, so if we make new Songs it automatically gets in to the playlist if it has the Genre in it
(Of course only for our own songs)

cedar mulch
#

oh, yea, i like that, a smart playlist

cedar mulch
#

can you disable the new style links on the creation page? i keep fat-fingering the style instead of just wanting to have the song selected.

cold vine
#

is it possible to stop other's songs to be automatically played when hearing mine? It's annoying if I'm trying to work on a metal song and starts playing a k-pop song every time my new part reaches the end 🤬

tepid trench
#

Here's a weird suggestion I just thought of: A random chance to get a bonus 3rd song during quieter periods of less server load.

...ignoring the pleas of "but I only wanted one song"

fair cipher
#

I love that you can click on music styles to search through other songs with that style. But can you please make that only on the song's actual page? I'm getting sick of accidentally clicking on song styles in my library when I'm trying to click on a song to edit it.

pseudo swallow
#

Ban this thing

IMG_2639.png IMG_2640.png
#

it’s tryna capitalize on SUNO brand

#

I remember that is illegal

visual brook
#

"y'all got anymore of those 500 free credits?"

subtle adder
#

Being able to create 4 minute songs is great but the space for the lyrics cuts the character limit too short, it’s very difficult unless a slow genre to make them last more than 2. Though it is great not being cut off and have them end up 2:30 or so. No real complaint, but maybe make the character limit a little longer.

formal bear
subtle adder
#

Well snap, you guys are awesome!

#

I also love 3.5, while it doesn’t get some of the obscure / ridiculous genre mashups as well, the vocals and way the lyrics flow and are stressed is a great improvement.

#

And for normal genres of music it’s better, in my experience the last few days

formal bear
tranquil lantern
#

SUNO IDEA - LET US WATCH ADS for suno credits! win win for both!

#

like let us choose to watch ads to gain suno credits

#

i see this as a win for both!

exotic prairie
#

for the next versions I would like the improvement of instrumentals, this is a great tool to imagine songs like from videogames or basically for devs, the biggest problem I faced trying to do both of these things wasn't even the songs generated at all, it was the text cap for the prompt especially. For normal songs with lyrics yeah, it's pretty cool, I don't know if the cap for lyrics is enough for the recent 4 min songs, but it should be enough for a big part. Keep it up, this AI is amazing!

untold sandal
raven grove
#

This update is SUPER, i wait for ability to add your own voice ;D

wind crest
#

thanks a lot for that

hybrid crane
#

lyrics are now 2500 characters

#

for V3.5

formal bear
#

🙂

hybrid crane
undone bough
#

is it jusr me or do the 3,5 repeat verses a bit to often?

#

it loops back to previous verses ALOT!!

true coral
undone bough
#

and to soon to,, after vers 2 or 3

#

i mean it repeats the verses

true coral
#

Hey Suno, you should really implement a trim feature. During the 3.5 beta you could do a whole song and it would end at 3:10 or 2:56 when the song ran its course, but now each song seems to go for minutes without needing to, which mean the song stops and you get some weird tag on.

undone bough
#

i made 3 songs today and i had big problems getting it to sing my verses in the order they are set up, it kept going back and sing prevous verses again and again

#

its cool once in a while but not so often as it is now 😛

#

and 1 of the drums u use alot is just cracked and broken hehe, it apears in manny songs ::P

#

sounds like a broken speaker bass

bold cedar
#

Is there any chance to save a song that can't end nicely and will it be possible to remix existing songs like this udio?

formal bear
#

just add to a playlist or sth

#

you should be go back and re-edit

undone bough
#

O M G ! seriously 3.5 keeps repeating verses 30% of the songs!!! often 2nd and 3 vers get repeated before it moves on to 4th etc..

#

4 times in a row now...

#

repeated vers 2 or 3

#

Defenetly a glitch needed to be adressed

tough iris
#

Hello everyone,

I am probably not the only one, but I would like to subscribe using PayPal since I don't have a credit card. Is there a plan to accept PayPal in the future?

robust jewel
eternal kiln
#

The ability to edit and update the lyrics of existing songs would be nice.

inner rover
#

Can we get a bulk select button for selecting multiple songs at once, for instance, deleting multiple songs at once and so forth.

#

I go through tons of songs in my trash folder to permanently delete them and so forth, when they add up overtime, and it takes forever to one click each song, especially for me when I have 200 folders of unused songs, so a bulk select button would be sick 👌🔥

misty blade
#

First of all, amazing work, everyone! Version 3.5 is so good.

Thank you very much!

I would be very happy if there was a way to use a specific word in the prompt to almost force it to finish the song. Sometimes I find myself struggling a lot to create an ending for a song, and it's really annoying. Maybe there's a way to do a special fine-tuning for a specific word like [ending] that would elegantly force the song to conclude at that moment. Please!

autumn basin
#

SUNO should definately work on matching verses...with melody and voice...i only get verses when i extend where nothing matches anymore

austere night
#

i wish i could keep more of the credits from last month since some months i am not able to use up the credits in time

misty blade
wind crest
hybrid crane
pseudo lake
lethal flame
#

I would really love to be able to organize things aside from a playlist. Folders would be nice. Also, for the love of my sanity, please give us a bulk delete. I think anything we thumbs down should be moved to the trash automatically.

pseudo lake
eternal kiln
#

An editing feature that is capable of formatting, refining, and/or reorganizing custom lyrics. I feel like that would be a huge quality of life update.

inner rover
#

@pseudo lake 👌thanks

rustic heron
#

An "Empty Trash" option to let us empty the entire trash can in one or two clicks would be quite welcomed

unborn token
#

A way to choose the length of generation we want. since the biggest problem of V3.5 currently is continuing on his own after the prompt ends. so a short prompt means you usually get 2 minute of free hand generation by Suno, best possible solution is let us decide how long we want the generation to be.

prisma aurora
frosty ferry
#

Suggestion. Allow us to have the option to regenerate a video/embed video if we have a custom image used on a song instead of the Suno Generated images.

stuck vessel
frosty ferry
stuck vessel
#

For embeds which I assume you’re referring to it does

digital anchor
#

I have a brilliant flute melody in a Suno song and would love to use it in other songs. So I think it would be a good idea to be able to specify a few seconds of another (own Suno-)song as a default when creating a new song.

undone bough
#

OOH M G! u need to fix this repeat verse thing its really ruining things!! this is urgent! half the songs are wasted because it decides to repeat previous verses! nad it does it in the beginning of the song not the end...is it just me having this problem?

sacred shard
#

For the new upload feature it would be useful to have some way to also cut the beginning of the uploaded material when extending from it, like "start from 0:03, continue from 00:10" - whilst one could do that with an external tool it would be more convenient to be able to strip away silent pauses at the beginning of a recording (since I have the feeling that those do affect the seeding process)

undone bough
#

thats it! im going back to 3.0 this is doing my head in...

#

i have 10 verses a chorus a bridge and pre chorus, i DONT need it to repeat verses 2 or 3 times in addition jees!! 😛

undone bough