📚┃suno-school

[true star]

There was a few songs that it self/generated when I first tried it out that used forward slashes between short lines in Screaming so they would essentially be rapid-fire

[torn spoke]

put the song text inside of triple backtics `

and post it, so I can see what the actual lyrics look like

o LOL

take a screenshot then

[raw oasis]

did the same thing to me.. 🙂

[torn spoke]

I'd remove the emoji, and just use commas instead of forward slashes

and remove the quotes

[true star]

Commas would still be a mixed signal of either Screaming or singing. The thing is I would give full on editing it, but I haven't had a single generation that screamed the whole chorus

It doesn't like screaming choruses

[torn spoke]

You make a personality from the sections of the chorus that contain screaming, and then extend the song from right before the chorus

[raw oasis]

Seems like once I tried to use my original suno work in suno after editing in Studio One and it rejected it as copy. how do we get around this so I can improve my song quality?

[torn spoke]

Never had that issue, you could always remove your original generation

that's risky though >_>

[true star]

How do I do that? Sorry for all the questions, I really appreciate the help

[raw oasis]

thanks. will try it again sometime soon. that would be huge for me to level up

[torn spoke]

Crop your song, the one with a tiny bit of screaming

and remove everything but the screaming

And then save that as a new song

to crop it, right click, edit, crop

then select the section of the chorus with screaming

and click crop

[raw oasis]

do you post articles or have a youtube by chance?

[torn spoke]

I have a yt channel, but it doesn't have anything suno related

it's just random stuff, I do have an old video on there that's music based though. I took two covers of a song and then blended them together by to get the best of both worlds

[true star]

So if I crop just the screaming section after, what do I do next?

[torn spoke]

and then click "go to song" after

then click the 3 dots -> create -> make persona

name it something you'll remember

then go back to the workspace you were creating your song

find the song you were working on, and find the timestamp from where you want to extend

[raw oasis]

will see if I can do a simple extend and add a guitar solo to the styles section.

try and level up on a basic Extend 😄

oh wow.. beautiful

[torn spoke]

I tend to be really descriptive with my prompts, not just using tags

but if that woks, then that's what matters

[raw oasis]

when you extend do you have to merge the 2 "songs" back together?

[true star]

So do I delete the non-scream section then for the song I'm working on?

[raw oasis]

oh i see there is a get full song button

[torn spoke]

yeah, though you can do that yourself in studio.

[torn spoke]

yeah, and save it as a new song

[raw oasis]

okay for later when I upgrade to studio.

thank you for helping us! off to focus.

[torn spoke]

Studio rn is only really useful for clipping together clips and extracting instrumentals. It has a tendency to create random artefacts on lyrics so those tend to not be usable

[narrow lynx]

Not looking for screaming but this is a brilliant method and I am dismayed I haven't considered it until you came out and said it lol. Just throw away what you don't like and regenerate it. Duh, cepo. 😅

I was doing all this crazy stuff like extraccting stems and pulling them into Ableton and re-exporting as a sample to have Suno use as "inspo" lol

[torn spoke]

yeah I tend to generate the first verse before I even include the rest of the song text

[narrow lynx]

Seems like a good way to get around Suno making all EDM songs sound basically the same, structurally.

[torn spoke]

I've had a lot more success with using personas than inspo

for getting really specific results

[narrow lynx]

I haven't messed with those as much but will have to give it a shot. I tried making a persona for a voice I liked way back when personas first came out but I don't do a lot of vocal stuff and never considered using it for instrumentals.

[torn spoke]

Right now I'm attempting to get a specific dialect of norwegian, it's turning out to be a lot more difficult than I'd first thought it would be

right now I'm having to recreate the start of the first verse, because it glitched out hard

[narrow lynx]

I'm trying to make a drum and bass tune which uses Altai throat singing samples as the bassline but I don't think it's going to be possible

Too specific and out of the ordinary for Suno to know wtf I'm talking about

[torn spoke]

Try and create a song that's only about altai throat singing first

specify time period, area of origin

be descriptive in the prompt and not just use tags

[narrow lynx]

I've done that a bit but not as specific as you're saying, and then tried to edit it into what I'm looking for but it just ends up using the throat singing as a sound effect. But I'll try being more descriptive and using the "crop and redo" method as well. Glad I joined this Discord I've just been winging it for almost 2 years lol

I noticed if you [bracket specific things] in the lyrics section, even if it's instrumental, that can help. Still somewhat hit or miss but better than putting it all in style from what I've seen so far. I wish it could show you which part of the lyrics it was playing so it was easier to gauge if it's following the prompt, but I know that's not really how it works.

[torn spoke]

Gotta be willing to spend at least 200 credits for a really specific result

and then pump the numbers up for the entire song

[narrow lynx]

Yeah, thankfully since I upgraded to try Studio I have way more credits than I'll use in a month haha. I wish they'd implement a checkbox so you could mass delete things you don't like finally. I have thousands of songs 😛

[torn spoke]

and feel free to @ me in here btw, the #💬┃general-chat chat is a bit gate-keepy, but this chat is strangely enough a lot more newcommer friendly

[torn spoke]

they did

take a look again

[narrow lynx]

woah when was that? I just tried like 4 days ago lol

[torn spoke]

[narrow lynx]

Do you have to do something to make them appear?

[true star]

What next? Sorry 😅

[torn spoke]

#📚┃suno-school message

[true star]

I did the scream persona

[torn spoke]

hover your mouse over the left side of the songs

they're not visible unless you move your mouse over em

[true star]

I have the persona and another version saved without the chorus now

[torn spoke]

extend from that version, chose your persona, set the audio influence to like 85% and experiment with the style and weirdness values

[narrow lynx]

oh, doesn't work in Brave but I got it working in Chrome. Cool, thanks!

[torn spoke]

odd, I'm in brave,

[true star]

So does it automatically add the lyrics back or do I re-add them?

[torn spoke]

you have to include the lyrics of your new section

as such I recomend you have a google doc and organise your songs in there, using tabs

[narrow lynx]

lol that is odd. Are you on Windows? I'm on Linux. Idk why that would make a difference. Maybe I need to update

[torn spoke]

ah, yeah linux is prob the reason

... it's always the reason

[narrow lynx]

haha

I use Windows on my desktop but this is my dev laptop. I'll have to try when I get home but I'm sure it works.

[true star]

One more thing just to make sure, when I extend from where it's supposed to continue, it doesn't include the beginning

[torn spoke]

Yeah, you'll have to right click on the one you're happy with

and click "Get full song"

It'll merge it for you

Then you keep extending from there with whatever personality ye want

[true star]

OH thank you!

[torn spoke]

btw, if it produces trash results, click on the thumbs down button

like if the vocals are glitchy or robotic (when you don't want them to be)

[jade mist]

The weirdness etc aren’t moving up or down what’s wrong?

[fair adder]

hello, i think the enhance lyrics prompt does not work. i type my instructions then i press enter but nothing happens and my lyrics don't change. is there something i'm missing?

[torn spoke]

which mode did you set it to?

[fair adder]

cover

[torn spoke]

I suggest trying on a different browser

or just press f5 first

[fair adder]

f5 just completely kicks me out of the cover mode and puts me back into normal create mode

i'll see with another browser

[radiant thorn]

Anyone ever find a magical prompt that will generate an instrumental-only cover of a song that has vocals...but without there being an instrumental melody that mimics the vocals? Like, just the backing chords and instruments, no solos.

I know I can take out the stems and upload just the instrumental, but the low quality and the artifacts always end up giving the generated cover a subpar quality tune

[torn spoke]

Generate a new instrumental track in studio using the original song as context

and hope for the best

[radiant thorn]

I don't have studio. How is that different from generating an instrumental song with the uploaded tune I have (the one with vocals and/or the stem-split song with just the instruments)

[spice birch]

Whats going on with v5? I've deleted the same song over and over because v5 AI can't pronounce half of the words in any song lately 😡

[radiant thorn]

Just trying to find out how good Suno is at recognizing the chords underneath the singing, without reproducing something that mimics the singing via an instrument throughout the song. I just want straight up consistently strummed chords for me to record my voice over afterward

[fair adder]

yeah nope, enhance lyrics definitely not working at all. shame, that feature sounded funny to experiment with. rip

[neat nexus]

You can create stems in Studio for multiple audio and mute just the vocals or remove just the vocals.

[radiant thorn]

I already do this on Logic and then I export it as instrumental only. My issue is that the cover of this instrumental-only version always inserts instrumental solos throughout the song. I want to rerecord this new cover version with new vocals after importing the cover into Logic, but the solos throughout the song make it hard to sing over

[neat nexus]

If you're already doing this in some other tool then it probably wouldn't help you, no...

[round bobcat]

Hu, now I have song intro were the femal vocals imitate the instrument from before...

[polar spade]

You might be using a model version that doesn't recognize them. Check your model: sliders NOT work with versions 4.0 and below.

[polar spade]

You did a Cover? That can happen.

[graceful condor]

Are thumbs down really training the AI? Someone mentioned this a while ago.

[round bobcat]

Would be nice, then I would give all songs (which means almost every song) with the same 'hu hu hu' femal vocal intro a thumbs down.

[round bobcat]

Yes, it were covers but different songs in different styles.

And I used v5

[polar spade]

I wouldn't see why it would need to change so it's a good chance it still functions as Before. Before, it was pushed to use thumbs down on any and everything you didn't like or thought it was a bug/glitch. This is generally called "implicit feedback".

The algorithm could then aggregate , then correlate songs with similar aspects showing a pattern of a bug or of a user preference which could later be looked at by the developers.

When we use emails and simple feedback surveys, this slows things down considerably as each have to be reviewed, one at a time.

I still think the thumbs down is preferred, but it all helps.

[round bobcat]

I would prefer an additional text field for the bug report, to write a small bug decription - ha

[polar spade]

If your saying your using different older songs and covering to ver 5, this may be true, as it seems ver 5 has its own interpretation of things, more to focus on the perception of what we expect of the prompts -- at least that's what I'm personally seeing/hearing.

So what was a violin introduction on a 4.5 ver pop song, Suno may feel that pop songs should have vocal intros instead when using 5.0.

[round bobcat]

I remaster the songs and then cover it and the cover and the new songs I do with v5 have almost all the same sung uh hu hu intro.

[polar spade]

Right. That what many prefer as a habit of us humans needing interaction.

The idea of this binary feedback, basically, yes or no, can pinpoint and find a specific issue much faster than reading our text.

But old habits fade slowly. 😄

[polar spade]

Yep. I'm with you. Many others saying the same thing. I think we have a hive of users trying to figure out a workaround to control it better from happening .

[round bobcat]

Well, v5 is still beta - maybe they will fix it in the future

[raven turtle]

As Suno is in Beta, but is perpetually being worked on end developed I wonder when they would consider it “proper” v1 release?
What’s the criteria for it more out of beta? I mean they charge a sub, they are improving it with updates, like a “proper” release… so why still in beta? 🤔

[hard beacon]

coz of user feedback

[hollow pine]

try "karaoke version"

[polar spade]

That's work for you? Cool if it does.

[hard beacon]

would make sense, but i don't think so 🙂

[half dragon]

I did this a few times, you can cover the existing Suno song
Then in the lyric box, remove all the lyric words, but keep the meta tags [intro],..[verse]..
Keep the audio slider high as 80%

[pearl topaz]

What's the best prompt for a sudden (non-fade) ending? For some reason it's not coming to me.

[hard beacon]

what kind of style are we talking about?

@pearl topaz

[tawdry scarab]

Have you tried enforcing the mode with a simple bass part

[pearl topaz]

I am doing mostly hard rock or country tunes. Trying to have more of a concert finish, but some are fading out so I hoping there's a prompt to use to change that ending type.

[hard beacon]

hmm.. interesting.. i've also made some rock tunes (Nu-Metal Genre) i never had a fading out ending...

on all of these songs there is a [instrumental outro] meta tag in the lyric-section

*in

[pearl topaz]

Lemme try that.

Thanks!

[hard beacon]

let me know if it works for you

[radiant thorn]

What percentage should I put for weirdness and style? Do I keep style description ?

[hard beacon]

i would say 15% and 80%

and yes.. just remove the lyrics (not the song structure)

and keep the style desciption

or go to suno studio and try to mute the vocals (in the hope that there are no dependencies on the frequencies) ;.)

[pearl topaz]

Nah. Didn't work. Did the instrumental ending, but faded it out. I think since it's a cover and the original track is fading, it fades too. So, was hoping to find a prompt or tag to bypass that and make it do more of a concert ending.

[hard beacon]

ahh.. ok ... yapp.. that will be the reason

[neat nexus]

Have you tried extending the generation from before the fade out and prompting in metatags the concert ending?

[tawdry scarab]

I would extend from before the fade tag heavily for concert ending style things and then expect to generate about 20~30 to get something good if adjusting along the way

[hard beacon]

great idea

[pearl topaz]

Would I use concert ending as the prompt? Lemme try.

[hard beacon]

btw.. i never used the "live concert" prompt on an electronic style ... hmmm... 🙂

[jade mist]

There is a problem with the sliders since several hours ago. The sliders have stopped moving altogether. I'm happy they returned to standard sliders because they are more accessible to screen reader users but right now they don't work. Luckily, the backend issue for V5 being 4.5+ has been now fixed.

[neat nexus]

Try adding to the lyrics prompt:

`[Crowd noise]
[happy applause]

[Fade out]
[End]`

and try some variations using this added to the style prompt:

live, live performance, 1970 Eurovision contest, live ambiance, crowd sing-along chorus, electric acoustic, unplugged

Fun note, if you change the decades you get different results in the Eurovision prompt. They get more or less talkative depending on decade for some reason....

[neat nexus]

I can't confirm. I've been using sliders all day.

[hard beacon]

so in 2025 Eurovision there is no live feeling because all people have a smartphone in their hands??

😄

[neat nexus]

Haven't actually tried that one. But like there's a noticeable difference in the way say 1970 works compared to 1990.

[pearl topaz]

Bah. Still giving me a fade out rather than a cold finish. I'll have to play around with it.

[radiant thorn]

Oof, the vocals were still there and came out sounding like gibberish lol

[hard beacon]

hahahaha

for me it looks like the cover function with v5 is still a little bit.. well...

[radiant thorn]

The problem with the “styles” box is that Suno gives the most embarrassingly bad and inaccurate descriptions of the audio I’ve uploaded. Mistakes piano for guitar, gives the wrong key of the song- so if I’m putting the sliders for that anywhere above 0% it’s gonna turn it into a godawful song

[hard beacon]

let me say "no that good" 🙂

[lament rapids]

80’s country. 90’s country, etc. they are all effective at giving one different t results.

[hollow pine]

was just an idea (would be funny if it did!)

[hard beacon]

v5 has (most of the time) a good audio result.. but it feels like a very very big russian-roulette to me

[radiant thorn]

Honestly the karaoke prompt got it closer

[lament rapids]

Karaoke prompt / backing track prompt.

Neither take away the lead vocals.

[hollow pine]

< --- YAY me

[lament rapids]

At least for me.

[radiant thorn]

I had to see it to instrumental. There’s still an instrumental melody mimicking the vocals, but it’s a little softer

[hard beacon]

is the cover on the same version as the original ?

[lament rapids]

I would just like to take my song which I love and do a remix to re-add the chorus and remove my current finish after the bridge a couple of times.

Arghhhhhh

I do find that using [Tag] and [Outro] are really effective at giving a good finish.

[jade mist]

OK, the sliders work, I was just using up and down arrows before but now that I use left and right arrows it works as before.

[radiant thorn]

As the original uploaded audio? Yes

[hard beacon]

ohm... wait a sec... so you try to cover/remix an uploaded audio?

[radiant thorn]

Yeah. Actually, some of the time I’ll use the cover of the uploaded audio. If I’m on the app, I’ll have to save the cover and then upload that, because if i click create on a cover in my library, it doesn’t give me the option to play with the sliders like it does on desktop

[hard beacon]

but.. to be honest.. that's a way to waste of your credits 😉 Suno is an audio generating ai... it wasn't programmed to cover or remix uploads in a perfect way

it's a big different for an AI to cover/remix an upload or a generated output

*difference

[radiant thorn]

Well I don't expect perfection, but it's done wonders for old demos.of songs and concepts that it's been able to flesh out and expand on after I upload them. I’ve actually never done a text generation-only song- maybe once or twice out of curiosity haha

[ripe bobcat]

I can not get an instrumental past 2:30 for the life of me the last two days unless I use lyrics. I don’t get it

I think all I can do is wait it out

[summer garnet]

What do you guys do mostly when creating a new song? I generally don't have beats of my own although maybe in the future. I generally just prompt a style and write all my lyrics a head of time for a song structure of the [intro] [verse 1] [pre-chorus] [chorus] [verse 2] [pre-chorus] [chorus] [bridge] [final chorus] [outro] or something like that and add tags and parentheticals like (Whispered) next to a line I wanted whispered. I'll generate probably 5 or so songs and see if I'm hearing what I want based on the style and lyrics and if not I may change it and produce another 2 to 5 songs. Probably not the best approach. I did create 2 bangers out of it though. I would generally take the one I liked the most that had the least flaws and then complete what I did not like about it by regenerated that part and making it transition. Is there a better way and credit saving way I suppose to do that @torn spoke or @polar spade. I think Faecous, if I recall correctly, you said you do yours one structure at a time (e.g., verse 1) then extend?

[torn spoke]

I generally start with a concept and work on the chorus first. I test it out a bit, see how lines are interpreted, and then tune it a fair bit

[summer garnet]

I also found a guide online that a youtuber was referencing for Udio AI music generation that has all the tags and style information you can put into lyrics and the style prompt

Used AI to go through that and help structure the lyrical tags

[torn spoke]

When I have the chorus I want I tend to make a personality from it, and use that for the rest of the song

Building my way line by line

tons and tons of interactions and reworks

iterations*

some times I stumble upon a melody I like fairly early, other times I have to spend dozens of gens

[summer garnet]

I never though of making a personality of just the chorus or sections of a song. Interesting. I've only ever made a personality from a song where I wanted to continue using the female voice of that song in newer songs. Funny thing is that song has a male and female vocal so I have to specify female in the style and lyric tag to avoid the guy's voice. I wonder if I just make a personality of the female's part of that original song if I can avoid the guy altogether

[torn spoke]

By having a personalty from early on I can reduce the amount of drift

that slow creep away from what the song started as

If I want a certain flow to a line, that I have somewhere else

I'll make a personality from the pre-existing section to beat the model into submission

I'm also gonna start using melodyne in post to fix pitch and tone

[summer garnet]

Nice, didn't think of that. There are some nice flows I like from certain songs that were generated that I would love to add to either a new song or a different variation of the same song

[torn spoke]

Personalities is the cheat code to getting what you want. HOWEVER

You have to actually match the syllable flow

if you don't, it won't work well

For lyrics

[summer garnet]

Yeah, I guess that would be more of a problem for newer songs using older personality flows. But if its the same song being generated in different variations the lyrics are the same for the most part, then the syllables would be the same typically

[torn spoke]

I had a huge problem with one of my passion projects, cause the [Verse were so long

Trying to get a consistent oace

pace*

[summer garnet]

I no nothing of post production stuff. NGL, I am a music newbie when it comes to that level of editing. I have done plenty of video editing and some audio editing in Audacity from time to time, but nothing complex where you're using different kits and stuff to make beats and such.

[torn spoke]

Making stuff sound good is addicting, matching beats and rhythms is super satisfying, I encourage you to for example, take two covers of the same song that you like, and try and switch between them

[summer garnet]

From what I recall, you can use ... --- and , for flow control to some degree

[torn spoke]

You can use commas, line shifts, dashes

I've never seen any noticeable effects from dashes though

If I want to indicate what words should be emphasised I use commas

And trailing dots

OH!!! And the biggest life hack: '

that thing, that right there

is such a good tool

If you want to have "good" become two syllables, do "go'od"

[summer garnet]

By covers you mean a song in suno I used cover on? By switching between them, you mean finding ways to blend them together? Drums from one, maybe syth added, etc?

[torn spoke]

Or actual real covers of songs you like

[summer garnet]

Of a real song out in the wild

[torn spoke]

the goal isn't to upload it to suno, but to see what ye can do

Have you seen hazbin hotel?

[summer garnet]

negative

[torn spoke]

there's a song in there that I really liked, but the vocals for that song were really robotic

And I found two covers I really liked, and matched it up to the original music video

extracted the stems, vocals, switched between them and layered them when the duet joined together

Suno let's us do the same thing now

but it's easier cause it has the BPM preset for you

You don't have to painstakingly compare individual waveforms to make sure things are lined up

[summer garnet]

So you mashed things you liked from the cover into the original? How would you go about getting covers that have all the stems, etc. available? Is there any Terms of Service against uploading a song you like/downloaded, doing a few covers on it and mashing that together?

[summer garnet]

Yeah, a few times I've noticed when I have to regenerate lyrics that it will give me a really nice melody, beat and vocal dynamic, but extend it way over what it needed to be and end up having to chop it up and see if I can blend the two parts together playing around with it. Is doing a cover after fixing it as much as possible make that transition smoother?

[torn spoke]

Yeah making a cover of it will smooth it over

[torn spoke]

I threw the original out all together, only used audio from the two cover songs

put it on youtube not suno

This was way before i knew about suno

5years ago

[summer garnet]

The last song I did, I chopped it up where it layered the same word together so it kind of almost sounds like a distorted FX of their voice as a transition so it kind of works and when I shared it nobody seemed to pick up on that. I'll try a cover of it though just to see how it pans out though. Any recommended settings for the cover as far as those bars go (weirdness, etc.) I am guessing 0, 100, 100?

[torn spoke]

https://youtube.com/clip/Ugkx_cxAVaFE3D02LLal-LgvxGWiEL2UPTiY?si=cAL7OIpDZEM04Pu6

The two vocals here are from two different songs, the instrumentals are also yoinked individually, liked the drums from one cover so I used that primarily

God talking abut this takes me back, I put so much work into this thing

@summer garnet My end goal for the use of my songs is probably to make AMV's with them

[summer garnet]

@torn spoke For me its to publish it, not really looking to get $ for it, but I can at least say I took my lyrics and produced a song out of it and its on a streaming platform. I'm going to have some fun with it and have a band and everything associated with it. I am also using an animated version of the female vocalist I mocked up with AI and created the Album cover and logo using it as well. I used Sora 2 recently to do a little looping animated video of the female vocalist, but i'll have to use the API to get rid of the watermark and look at using another video AI service to see if does even better.

[torn spoke]

that's cool! I guess I'm more of a singles guy than albums

[polar spade]

Ive always just "wing it" with no go-to procedures. But just choose which to use depending on the art/output I want.

Since most of my outputs are songs which I will try to pitch, then those I write on paper, work out chords, melody as best possible, and sing an academic version to see what Suno does to it.

So , if I happen to write a duo style song, sure, I would do it piece by piece maybe, with extension. Lately with Studio, we can run a few outputs with a persona and take the best verses and in the order that sounds best if one vocalist sounds good on this verse or that. Or... speculating..... else, I can run two versions of the song, changing out personas specific to the song (keeping audio slider up so they all sound similar) and then patch and splice which verse and vocal goes where.

In the end, its just use whatever tools works for best for the project at hand.

[summer garnet]

In studio, how do you really apply a persona? Isn't that more attached to a song? Can you specify it in the style prompt? Or do you mean generating the song with two different personas and then editing them in studio?

[lean scaffold]

New to Sumo. Trying to figure out how to extend my song to be longer (another verse + another chorus + instrumental) and the extend feature in the Editing feature doesn’t seem to be working. Or I’m not doing it right. Any suggestions?

[delicate flower]

why does suno add drums and vocals even if i add to prompt not to add any of these?

[torn spoke]

cause it's generative AI

[torn spoke]

nah the UI on the left let's you aply a persona

It's got more customisation than the normal create UI actually

but somehow seems to perform worse

idk

[raven turtle]

Having some issues with Studio separating clean male Baritone vocals from a dense mix, so trying the WAV in UVR5 instead to see if I can get it any cleaner. A lot of bass information in there getting muddy!

[hollow hare]

Hm, What are your weirdness, style, and audio influences set at? Typically having it at 0% or 100% gives pretty bad results, 1–99 gives better results

[onyx sandal]

weirdness is at 50 to 80% (i never use over 80% and the style influence and audio influence is at 80%

[safe hearth]

Hey beautiful people. So I had a question that I just could not figure out on Suno. I'm trying to create a long format study song and I finally found the perfect prompt to create the initial song. It's only 3:02 so I've been trying to figure out a way to extend it. But I'm not sure how to. My goal is to just carry the same vibe as the first three minutes but extend it to 8 minutes, I'm using v5. Can anybody help me figure this out?

[safe hearth]

Whenever I try to extend from the end of the song it malfunctions, giving me 2s generations instead of the multi minute extention that I thought would be generated.

[honest heath]

give it more context, atleat 10-15 seconds from the end

[iron mauve]

hello
do you guys know if there's a way to use "custom" option in song creation and get random generated lyrics each time?
I remember it was possible with v4.5+ but since v5 I dont see that option even with v4.5+

[safe hearth]

ok, can I prompt it in any way to give me minutes of extension? It gave me like 15 seconds.

[summer garnet]

When you do a cover to smooth transitions of a song, what wierdness, style, audio sliders would you use to keep the same everything else other than it to fix a transition potentially?

Also added the persona to be used

Want to keep it pretty much the same song other than it fixing a transition area in the song

Also, with a cover do the tags need to be in the lyrics for direction or is it already known from the original song?

[summer garnet]

Is it me or does the cover of the original song always sound dull compared to the original?

[quiet solar]

Audio influence high (75%+) and weirdness low (<15%). Style prompt generally mid-to-low and using the same prompt as the original generation - unless you're using it to specifically prompt for things in the fix, in which case you may want to turn it up somewhat - but you'll have to make sure the general aspects of your prompt actually reflect what Suno created first time around.

Because Suno regenerates everything from scratch taking influences as prompts (e.g. previous audio, persona, style and lyric prompts, etc.) it's very unlikely you'll get a perfect replica of the original with your specific edits, but you can get something close enough to do a fairly seamless replacement. The above suggestions aren't a silver bullet, but they should help limited wasted attempts.

[quiet solar]

The original song will influence the cover, so some of those directions will be effectively grandfathered in, but it's generally a good idea to repeat the prompts if you want it to handle those sections in the same way. You might also want to ask it to do something very different - it depends what you're looking to achieve with the cover.

[summer garnet]

Yeah, I was doing 0 weird, 100 style and 80% Audio. I then just kept upping the Audio from 80% to 90% to 95%. The audio did seem to be closer to the original when I slide it up to 95%. However, there are still parts where it is sung a bit differently or transitions differently or the chorus is not as punchy and the sound always seems to sound duller than the original

I took out the lyric tags and such out though, it just had the original lyrics. Perhaps I should copy the original lyrics with tags back in for the cover and see if that makes a difference. Not sure it will make a different with the dull cover sound though

[quiet solar]

0% and 100% can be funny places to put the slider. Suno works best when you don't try to strap it into a straightjacket, and I've found extreme positions on the sliders can cause their own unique headaches. Maybe try it with 95% as the max and see if that helps you out at all.

[summer garnet]

95% for style?

[quiet solar]

95% audio influence, 5% weirdness (as opposed to 100/0).

You probably don't want 95% for style unless your style prompt is doing a lot of specific steering for the fix you're trying to make

[summer garnet]

I just put the original style I had which does do a lot of the steering for sections of the song, but I also have lyrical tags as well but did not include those in the cover just the lyrics by themselves

[quiet solar]

When you say "dull" - what is it you mean? The production effects on the song itself? Melody being less complex? Muddier and less clear sounds?

What is it you're trying to get the cover to do?

[summer garnet]

Yeah, I can easily tell the sound of the instruments, guitar, bass, etc. are much lower and dull sounding than the original song

I'll go back to the original and immediately tell the difference

I was trying to do what Faecous mentioned earlier about smoothing out transitions in a song by doing a cover of it. There was one part in the song I wanted to smooth out a transition

The transition as it stands now is ok, kind of sounds like a double vocal distortion effect, which works, but was hoping to see if I could get a smoother one

[quiet solar]

Ah, yeah - that can happen in both directions. It's worth prompting details for all of the instruments, and you might get some mileage out of style prompts for "cleanness".

[summer garnet]

the double vocal distortion effect is what I got from the song editor by playing around and dragging the bar of a section to somewhat transition to the pre-chorus of the song.

Was hoping to just cover the part where it double vocals and clean it up

[quiet solar]

Gotcha. So the problem with the 'cover' function is it's intended to rewrite and change up the entire song, so it's not going to be ideal to use it to try to do a clean, identical pass over the song but fix the transitioned area. And it may well think the 'mistake' is deliberate and carry it over.

In the absence of Suno studio, it sounds like the kind of thing that would be a lot easier to fix in an external editor like Audacity than to try to fix perfectly within the limitations of Suno.

[summer garnet]

Yeah, I could export it and fix it there I suppose. I've used Audacity before to fix things here and there. Was hoping more out of Suno considering the $ of Studio.

[quiet solar]

If you have Studio, you should be able to isolate and remove the offending vocal segments, no?

[summer garnet]

I do have a question about style though. So I used ChatGPT to help formulate a style based on documentation I found that has all the tagging, moods, etc. that suno and udio may share. I'm not sure what suno actually looks at and cares about though. Was hoping to see if there is an actual guide somewhere that tells you exactly what you can put into a style prompt.

Genre: nu-metal, rap-rock, 2000s
Instruments: distorted electric guitars, punchy bass, cinematic drums, DJ scratches, piano, strings, and melodic synth leads in intro and outro, Vocals: all vocals performed by one female singer, Female lead throughout — emotional, powerful, and defiant, Bridge delivered as a rhythmic spoken verse by the same female voice, not aggressive, Layered female harmonies and atmospheric backing vocals in choruses and outro, Production: dynamic contrast between calm, cinematic verses and explosive anthemic choruses; half-tempo bridge breakdown; rain and thunder blended naturally; wide stereo mix with depth and warmth; subtle reverb on vocals; emotive strings and melodic synth motifs build tension and resolution, Mood tags: [nu-metal] [female lead only] [female harmonies] [female spoken bridge] [anthemic chorus] [dark energy] [cinematic build] [melodic synth] [strings resonance] [breakdown] [wall of sound] [emotional defiance][hope through chaos]

That is an example of the one I used

now this did get me kind of what I wanted, although I also have everything tagged in the lyrics

Like, it put in rain fx and has rain and thunder in there which I don't think suno even considers

[quiet solar]

You can put anything you like into a style prompt, and Suno will try to find references for it across its library of pre-existing music. Tags work better if a) Suno actually uses them, which it's less likely to if you bombard it with hundreds, and b) if it can easily look them up across other songs.

The trick with b) is using musical language where possible - the tags you're using are all pretty much in the right ballpark.

[summer garnet]

It would be cool if it could add in special effects like rain though lol

[quiet solar]

Suno doesn't reference FX, so FX prompts are usually not successful.

[summer garnet]

As far as a library of pre-existing music goes. How did Suno come up with all of these ways to sing lyrics, voices, riffs, drums, etc.? Did Suno pay to access all the music libraries out there and train the models on them?

I imagine they also sat down or hired music producers who know what goes into producing music to come up with all these tags and such to formulate melodies, etc.

[quiet solar]

That would be a question for the developers, but there's an ongoing lawsuit between Sony, Universal and Warner and both Suno and Udio for how libraries were used, so this isn't really the right place to have a detailed discussion about it. 😉

I don't know to what extent the tags were influenced by humans, as opposed to the AI learning algorithmically based on underlying information. But they act as prompts (which may or may not be ignored) rather than strict commands that Suno must obey.

[untold briar]

In case you missed it, in the "Feedback" category there is an dedicated Suno Studio chat: https://discord.com/channels/1069381916492562582/1420795057996300388

[polar spade]

Just tossing in here. I was doing some extensions on v5.0 and the outputs were odd...Like I was doing an "add vocal to vocals" .... It was just a combination of voices doubling up as if melding two songs at once.

Did a couple of Covers to correct just a misspelling in the lyrics and got the same vocals over vocals vibe. Got better with playing with sliders and I think I then just extended with 4.0 and 4.5+, and they were fine.

[ripe bobcat]

Why does v5 feel so broken? It can’t make a song at all. It was just fine last week it’s so strange to me I been using this for a real long time. Lyrics are not read clearly, all the songs feel “the same”, there’s always some woman chanting in the background. So much strange stuff on this model that felt like it was working after it got released, no idea what happened but it’s being really wacky I already wasted 1000 credits trying to make one song. I gotta stop for a while I think I have enough credits but I don’t want to keep running into these dead ends

Oh and mid song the song will stop and repeat, this has happened about 25% of generations with lyrics

And instrumentals will not exceed 3 minutes no matter what I try

[polar spade]

Anytime a model is in beta, those times the outputs sound wonky, I just stop after a couple of tries. Then just drop down a version and do the song from there. If the beta starts sounding good again, then redo or cover the old version and make current. Just don't bother repeating efforts and using up credits. Drop a version down or just wait it out.

[ripe bobcat]

That’s probably a good idea lol. I’m just trying to make it right cause I keep thinking it’s me but 4.5 seems fine but I have my own feelings about 4.5.

5 seems better with hybrid genres but 4.5 will suffice, thanks for the input

Here’s to hoping that whatever is broken is fixed soon

Not getting any younger 🤣

[ripe bobcat]

Even 4.5 is for some reason cutting the song in the lyrics I’m just like restarting the song. This is just getting too frustrating. From what I read earlier, it looks like the big companies are already trying to take this stuff away from us already. How dare the people be able to have a creative outlet right? 🤣

[dark linden]

I never seem to have any of these problems outta the new models when they drop? Is it like this for most everyone?

[keen fog]

i must be lucky too, only issues i get are from the website itself not the model

[fallen loom]

Any chance the style prompt will be getting larger any time soon the prompts I want to use a highly detailed and trimming them down makes what I am able to generate now significantly less than what I want it to be.

[raw oasis]

I have been really liking v5 from a sonic perspective. I am just coming back from a month off gasp (lol eons in AI time..) and really have a few months tinkering in Suno. what exactly is not working great? Is it that v 5 is not giving the variety you want?

[raw oasis]

I extended my song 3 times now. added a guitar solo section, then added a [Refrain - Choir] section after that. Thinking the next step is Remaster, just to polish and blend the merges. (anyone post-production mixing in DAW after? this song is pretty much how I like it pre-Remaster)

[icy ruin]

Can someone suggest me how to reproduce a music style like the following?

https://www.youtube.com/watch?v=R37IfQg0eTQ

any idea welcomed!

[polar spade]

Just use a genre analyzer online, such as Audio AI Dynamics https://audioaidynamics.com. Paste in the YouTube address to get suggested genres which you can use as prompts. It should be fairly simple for your sample.

[limber seal]

Hi any help Please I just upgraded to unlock the 1 minute sample songs that sounded great btw, it shows the full song length in the timer but still only plays that 1 minute? please any help appreciated

[ebon sand]

It's still incompatible with persona and inspo, and without them the songs are not in a way that I like; but when I use them the songs are bugged so it's still not really usable for me
It also has problems to follow inline tags or simply output the lyrics of the lyrics box
It's like it's a capable but fundamentally bugged model
I don't think I have a single usable song so far

[agile spear]

Good afternoon, where can I go? I accidentally paid for an annual subscription and want to cancel it. Who can help me?

[neat nexus]

Have you sent an email to [e-mail] already?

[agile spear]

Yes, I sent it. But they're silent

[neat nexus]

They're not silent. There's just a lot going on.

See: #archived-ticket-help message

There's also this...

[agile spear]

Thanks

[worn thistle]

hello, sorry for asking, but can someone tell me how to create the "Hook"? or how is this feature works?

[neat nexus]

You make a song.

You create a video (either full narrative or a decent loop) for the song.

You upload it and try to sync it to the two minutes maximum you are allowed to use in a Hook.

ProTIP: You can just make the whole thing, audio included and tell Hooks to use the audio of the video itself.

You will still need to muck about with the sync though if you want the segement to sync with your lyrics.

PorTIP: You can also try to deliberately make short songs to use with Hooks to avoid much of the syncing issues.

How do you make the video? Any way you can. You can video yourself lipsyncing, make a video with AI and edit it together in a video editor of your choice, or use stock footage you attribute according to their policies from places like Pexels or Pixabay. The sky is the limit.

[worn thistle]

ok, and if i want to use AI, what AI i need to use, or what AI can do that and it's free? i just wanna test it

[worn thistle]

↑↑↑

[graceful condor]

I do not get the "Heal" function in Studio or Editor. I tried to select a selection and use heal and it does nothing.

[true carbon]

Did you folks ever tried that? I like using AI but i want to have some control and creativity over it! Let's say i do a drum pattern into FL studio and add some FX here and there and upload that do do a cover. Did you folks got any good results?

[brisk flint]

Well, it's a yes & no-answer:
Yes. Well, I've already uploaded my own tracks from the DAW, mainly to generate vocals. So, yes, processing your own tracks does work on a level.
No, because: In all available models, Suno tends to add a lot when you use the cover or extend function, even if you try to exclude it in the prompt. In that case, you would have to load the stems and push these back into FL - and even though V5 sounds better than V4.5+ in my opinion, there are still certain artefacts in the stems, i.e. the drum loops can sound as if someone had soaked them in digital fabric softener, and you can hardly get rid of the artefacts.
My suggestion: just give it a try 😉

[true carbon]

Yeah maybe i should just do the opposite, get stems remove drums and add on top with good quality sounds, same for the bass

As reference i did that yesterday, modified it a bit in the DAW and quite happy with the result i dont want people to flag my tracks ai RIGHT away https://youtu.be/cK3EoUSkZMU

[brisk flint]

What do you want to achieve, or what kind of sounds do you want to push from Suno in FL for what 'reason' ?

[true carbon]

in fact what i want to achieve is not only AI i want to add my own to it

[brisk flint]

I see. So, I work with different methods: I generate tracks completely in Suno, then cut them up in the DAW and “pimp” them with my own sounds/instruments. Or I create a track in the DAW, upload it to Suno to generate additional sounds, especially vocals, which I then load as stems and process further in the DAW.
And the third method: chop up Suno stems in the DAW, add my own sounds and load it back into Suno 😉 Here's an example of method 3: https://www.youtube.com/watch?v=v-nRpMGbAg0

[true carbon]

Thanks got it

[shut ferry]

I'm having trouble getting Suno to begin any song using only drums. Suggestions?

[brisk flint]

Are you already using 'meta-tags', i.e. Prompts in the lyrics-area?

[untold briar]

Generate just drums, then Extend or use Suno Studio to do something more fancy, to build out rest of song startig from the drums

[shut ferry]

Yes, meta tags in the lyric section like [intro: drums only, no other instruments or vocals until verse 1]. I also can put something similar in the style summary to no effect .

[brisk flint]

I'd recommend avoiding using the word 'no' as Suno often - based on my experience - ignores the 'no', same as our human brain. Let me quickly give it a shot.

[untold briar]

Can try simpler formatting:
[intro: drums, drums, drums]

That's right that [blah blah blah "instrumentals" blah blah "vocals"] may be how Suno responds responds to complicated things like [intro: drums only, no other instruments or vocals until verse 1] though it does understand sometimes in latest version(s)

[brisk flint]

Yes, Suno is once again doing what it wants here; my tests just now were unsuccessful. In this regard, I would also resort to the solution of trying to generate only beats (which is not an toooo easy task either) and then adding the rest of the song with “extend” or deleting the rest in the first bars in Studio 😉

(Ops, gave it a shot by accident with 4.5, let me give it a try with 4.5+ and 5)

Okay, it worked in one out of ten attempts at 4.5: https://suno.com/s/DlIQ52LZ2ptPWHJE
At 4.5+, it didn't work in ten attempts, because I always have at least one bassline in there:
https://suno.com/s/UOq6tuvL4S5lGyqW
I gave up on V5 after 4 attempts because I'm too stupid to use the prompt.
In short, my prompt method is not the solution. And with that, I'm calling it a day, but good luck anyway.

[shut ferry]

Thanks! I will give it a try in v5

[summer garnet]

I've made 2 bangers out v5, but it all probably depends how you are using it. I created all the lyrics beforehand and styling at one time and had it generate a song while some people just do parts at a time. I did all this the past 2 weeks though and people have been saying the model has gradually gotten worse as time has gone on. I haven't tried to make anything this week, but I will say the studio is a pain when it comes to generating lyrics and vocals where a lot of the times it will not do what I want or give me nothing.

[ebon sand]

heh, yeah, i don't really create 'bangers' I'd be satisfied if the quality wasn't so poor with the features.

i've currently switched to elevenlabs for a bit since it can do some interesting things i need and wait for suno improvements

[summer garnet]

Out of curosity, I haven't really used elevenlabs. What going down a rabbit hole, what is the biggest difference you would go to elevenlabs over Suno for? What's its biggest strength?

[ebon sand]

elevenlabs isn't really a music service although they've added some functionality and a music generator, even fx sound effects which are quite cool, but i mainly need vocal copy and authentic pronunciation and squeaky clean quality that's its primary realm

[summer garnet]

Hmm, I wish suno could do some special effects as well. I'll have to look at it more. Could you vocal clone the persona voice you get from a song to be used in say tiktok, etc as a personality?

[untold briar]

All in Suno can clone like that, prompt something like "spoken word" and format like a script (and probably turn Style Influence up).

[elfin pollen]

I've had about 50% success with v5. When it's right it's so right and when it's wrong it's soooo wrong. 🤣 But it tends to work really well for what I use it for. I don't generate random works with Suno. I'm already a musician so I basically just run my own stems through and transform them into other things and turn audio influence up to 100% and write my own lyrics. Then I replace the vocals with my own. Usually 4.5 plus does this the best. But I have had some nice success at times with v5. For me, v5 is at its best with its remasters. At least that's my experience.

[onyx torrent]

Good afternoon. I've encountered a problem. When I clicked "Generate," I got this message: "Your account requires support assistance."
Please contact [e-mail]. I changed my Microsoft password a couple of hours ago. Perhaps that had an effect? ​​I've already tried logging in again, even opening it from a different browser, and I changed my password again just in case. But nothing.

[untold briar]

Ask in https://discord.com/channels/1069381916492562582/1372994553233276970 channel which is for billing/account issues, but send the email also, email is the main official support path.

[onyx torrent]

tnx

[graceful condor]

wth...I remastered an Extend because I did not like the audio of the OG one. Normally it was possible to use the remastered and just "Get Full" song but now there is no "Get full song" in the remastered version. Am I missig something?

[half dragon]

Yeah you shouldn't do that, apply things to an extension will break the link as it will create a new clip

The original extension is still there I think, Get Full Song first, then Remaster that, hopefully it'd fix what you were trying to do.

[modern dawn]

Maak een liedje over Lisa over serveren met make.up en serveren bij kardijs

[summer garnet]

@elfin pollen Yeah, i'm not much of a musician unfortunately. I just started learning to play piano and I do have an electric guitar, but never had time to learn to play it. I'm definitely not singing lol... my vocals are not very good. I rather have an AI singer sing my lyrics the way I want it to be sung or at least find the one I like best. Maybe at some point I'll play with hooking up intruments and coming up with my own thing.

[raw oasis]

Good place to start might be a little Nektar Impact LX25+ or other simple midi controller / pads. I used one in a song recently and the AI transformed it magically.

[placid narwhal]

So... I want to know if the studio has something like the extend. I know I can create after the music in a track, but... It seems that it doesnt pull too much from what came behind, different from extend. Does anyone has a clue?

[summer garnet]

I actually have a Nektar LX61+ laying in the closet. I sometimes hook it up to the Quest 3 with Piano Vision to practice

[slender rapids]

Can you have multiple persona ? If so, how can I do that?

[half dragon]

Have? or Use?
I think you can have as many as you like, but you can only use one at a time.

[slender rapids]

to use, I can put two persona and then make them sing

[half dragon]

oh okay, you can only put one persona on one creation at a time.
To put 2 different personas on one song requires extra work - Extensions.

1. create first part of the song for Persona 1
2. extend it and put Persona 2 on
3. Merge them with Get Whole Song

[slender rapids]

okay thanks

Where can I find Get Whole Song?

[half dragon]

You can only see that option in the menu on the extension clips, Part 2, Part 3 ...etc

[slender rapids]

Oh yeah... it's complicated for me, but thanks anyway.

[half dragon]

Cool. It's really simple once you get used to it.
Extend the clip --> give you Part 2 clip --> if you like it, click the 3 dots menu to see Get Whole Song option --> that'd create a new clip with 2 parts merged

[wheat tulip]

Making a duet is not simple. Especially if you want the two voices to sing a final chorus together.

[slender rapids]

okay, interesting

[quick grove]

Is Suno down for anyone else? I just keep getting a black page with "SUNO" in the corner. But if I click anything, it just loads a blank black page. 🤔

[hollow hare]

Hm, seems fine on my end. Some quick tries to test:

1. Are you able to access other websites? If no, restart your device or check if you have data
2. Try clearing your cache
3. Try using another browser
4. Restart your computer
5. May be a DNS issue, see https://discord.com/channels/1069381916492562582/1417459233544339537

[quick grove]

I did all of that and thank you. ^_^
It's blank on google chrome when I use my phone or desktop. And I tried using Microsoft Edge and Foxfire. Still blank. It was working fine a few minutes ago.

And I'm using the internet fine. Everything works for everywhere else. Only Suno doesn't want to work. 🤔

[hollow hare]

Hm, sounds like a DNS issue to me. If you already tried changing the DNS as outlined at the bottom of the tip page, I have no other idea, sorry

[quick grove]

Much appreciated anyway. I guess I'll just wait and see if anything changes. Because it doesn't work on my phone or desktop

The app seems okay. But not the website itself.

[heady spruce]

i want to try to do a cover version of existing song, is there a work around to get pass through "uploaded audio matches existing work of art. please upload a different audio file"

[half dragon]

Is it one of your own songs? or a known songs out there from someone else?
if the latter, then you can't and shouldn't upload.

[mossy dust]

Does anybody have any suggestions on good sites/apps to use for making a hook video?

[scenic sequoia]

The new studio makes duets super easy (:

[untold birch]

Any tips?

[tawdry scarab]

with studio just run it with a man and with a woman extract the vocal stems chop the parts and then assemble in either instrumental backing

cover or gen new instrumental backing

there are other ways though too

[unborn beacon]

it's most probably a mode for songwriters.
alot of singers kinda kick off sessions with random words and noises, figuring out the melodical and syllable flow of a song, before filling it in with words.

could be useful to generate songs with whatever as lyrics, before going back and filling it in.. just a guess, but that would make sense...

[scenic sequoia]

Well the last one I did I had a song that was done as a male and then I added new track and selected backing vocals and selected the length of the song. I then changed my song style prompt so that it would mention how I want the two vocals to be done. And then when I hit generate it gave me a really nice male/female song... I wanted a female backing singer which worked perfect. But as sangdernard suggested is another way. Really depends on what you wanting specifically tho

[opaque garden]

My Tech Demo - "Suno on the limit" with glitchy Hardstyle beats. Have fun!
https://suno.com/s/AgNzIqKWUBV9PUXw

[full root]

i got lock out from show case because of the 10 seconds respite Woodhouse bot says

please undo

suggest you fix the issue when this happens to avoid locking up for no reason

[graceful condor]

It is so ridiculous. Before it was nearly impossible to get a short extend and now within the last days all I get is short. I can't even add anything to it because the AI is ignoring all commands and input.

[raw oasis]

what does this mean for all newcomers, is the issue related to v5 not being too popular at the moment?

[eternal mulch]

I just tried to download a song and it says all my free downloads ran out?

[raw oasis]

how many credits do you have? are you trying to grab the MP3 or WAV file?

[eternal mulch]

MP3 file

[raw oasis]

hmm I think that is normally free after created.

[eternal mulch]

I thought that too cause even when the credits come back it still tells me my free downloads ran out

[near flume]

Are you using the free plan? If yes there is a limit to the downloads.

[graceful condor]

It is known and is an issue with 5 but also swap over to 4.5+ because now all is the same. They are not really separate anymore. So what worked in 4.5+ does ot work on it anymore in most cases.

[eternal mulch]

This I didn’t know till now as Suno is new to me still

[eternal mulch]

How do I swap over?

[raw oasis]

That would be so helpful if they did not alter previous releases. Something I am familiar with as a 25 year dev. just branch and iterate. dont corrupt what was.

[wheat tulip]

Free plan should only have access to 3.5 (and possibly earlier)

[near flume]

They are starting to do depreciation of the older versions and putting new limits on the free plan.

[near flume]

V4 and earlier are going to be removed soon.

[eternal mulch]

Dang

[eternal mulch]

Okaye thank you

[quick sonnet]

@stuck junco I understand that certain features should be reserved for Pro users, but simply removing features for free users is a bad idea in my opinion. As a free user, you could only download poor-quality MP3s anyway, but now even that option has been disabled. If the purpose of this move was to force free users to buy Pro, it probably backfired, at least for me. Because if Suno disables features for free users now to force them to upgrade to Pro, it could later disable features for Pro users as well, forcing them to upgrade to this even more expensive version.

[polar spade]

@near flume

Fact check: Just the Remaster for 4.0 was removed. V.4 is still available as always.

What's being deprecated:
v2 and v3 models - No longer available for free users. 3.5 remains available.
Remaster v4 - Deprecated for all users

Please note:
Your existing creations are safe! All previously generated content remains accessible!

[polar spade]

"....V4 and earlier are going to be removed soon...."

3.5 is earlier than 4.0 😁 Have no fear. There are no plans to remove 3.5 anytime soon.

[polar spade]

Mp3 is still downloadable for my Basic account. 💁‍♂️

[quick sonnet]

It was some time ago.

[polar spade]

What is 'it' ? You post reads to say that Basic can no longer download MP3s? Did I read that into the post or you meant something else?

[quick sonnet]

Take a look at this:

[polar spade]

OOooo.... That helps. I haven't seen that as I don't download from Basic. I'll down load a couple-few more and see what happens. Thanks.

[polar spade]

How many do you think you get? I'll try to test, but if it's limited to 100 , it might take a while. Do you know? 20? 5? 100?

[quick sonnet]

There were no such limitations befor. And this month I had no downloads.

[rigid falcon]

Is Suno currently down?

[polar spade]

If you're getting the warning of only 3 left, how many downloads do you think you made, or may be the limit per month? I'll be asking others as well.

[quick sonnet]

Have made no downloads this month.

[polar spade]

Good to know for October.

So previous to your getting the warning of three left, do you have any guess of how many you downloaded previously in September, perhaps?

[quick sonnet]

No idea, defenetly more than 3 and there were never such warnings.

[polar spade]

@quick sonnet Ok no worries. Thanks for the response. I'm curious as to what the limit is now then, as well as if this applies only to 3.5 or included the 4.0 outputs as well.

I'll ask around.

[polar spade]

I just opened two accounts and both working.

[quick sonnet]

As a Pro user you should not have such isues and 4.0 is limited to Pro only.

[rigid falcon]

Intermittent. Seems back online.

[polar spade]

Correct. Should be unlimited for paid accounts.
I have Basic , Pro and Premium, all three. For testing.

[quick sonnet]

But if you have used 4.0 then only with Pro and Premium.

[polar spade]

If I know there should be a limit of, say, 20. Then I'll try to download 21 songs, and see what happens.

[polar spade]

You're right there. I was referring to the idea that there was the free 4.0 songs. But , you're right again, I think those were removed and now just to 5.0 samples available?

[quick sonnet]

Are you tallking about that 1 min tracks generated for promotion purpose?

[polar spade]

New Basic Accounts: those no longer offer a few free 4.0 songs right?

How did this pop-up appear? Was it while downloading the 5.0 sample?

[polar spade]

Yes

[placid aspen]

Anyone got a good prompt for excluding an instrument from an opening section of a track? I want it to go drums>piano>drums+piano, but the bass player insists on coming in with the drums, even though the audio example doesn't! I've tried "The song begins with a solo stomp drum beat, with no bass, before a solo piano takes over, then the drums rejoin." But the bass player comes in with the drums every time. Driving. Me. Mad!

[quick sonnet]

Defenetly not, have only 4.5 samples, but have never even tried to download them, since they are only 1 min long. All my tracks are made by 3.5.

[polar spade]

Where did the pop-up come from? Was that in September? You have made no downloads in October you said. I'm trying to see what triggers that pop-up about downloading songs which normally would have Commercial Rights to a paid subscriber.

[quick sonnet]

This is what I do. In September there was no such problem.

[polar spade]

You only have 4.5 samples? My samples are in 5.0 🤔

[quick sonnet]

Yes 4.5, they were generated in September and no 5.0 was avaible.

[polar spade]

Yes. I assume there is now a limit of songs to be downloaded. I don't question that. Basic has always been Sunos response to NOT having just a 7 or 30 day trial. So, Basic would, in my opinion, never be viewed as a perpetual level.

I'm just trying to see what the new limit of downloads are and now, as you mention, why are my samples in 5.0 and yours in 4.5

[quick sonnet]

As I said, 5.0 was not avaible at that moment.

[polar spade]

Ok. Got it. That pop-up was last month.

[quick sonnet]

No this month, last month everything was OK, the same for the month befor.

[polar spade]

Yeah? You mentioned you have made no [attempts to (?)] downloads this month. What did you do to get the pop-up about downloading limits. Was that just an automatic pop-up, like an advertisement promo?

It says '....for this song as well....' as if a song was being downloaded, or an attempt was being made.

[quick sonnet]

Tried to made the first download this month and got this pop-up.

[polar spade]

Got it. We're on the same page now. I see.

[polar spade]

Yes. If they are limiting downloads for Basic users, then, that seems in line with encouraging users to buy a subscription at some point. This seems in line with most recent AI companies to have such limits.

Even at 3.5, we can get great songs, where I'd assume some have even been able to make a business from just that alone without limits to as many downloads they wanted.

If that limit is 100 downloads a month, that seems reasonable. Even 50. But if limited to just , say, 5, then it may be too restrictive. It's all something worth debating.

Having a free service rather than a trial has always been a kind move on Suno's part in recognizing the social aspect and impact of creating songs by AI, to make it available in some form to all of us, rather than limit it to just a 7 day trial.

[rigid falcon]

Is anyone else struggling with the "replace section" function? Not very intuitive and rarely works (at best the section is not replaced after committing). Failing that, what is a good workaround that allows me to use the track precisely as created but at the same time allows me to change a word?

[polar spade]

It doesn't help, I know, but I've done about 5 replacements (using both Legacy and New edit UI since Thursday with no issues other than the full song taking a minute to show up in the Library when done.

Only one had an issue of a replacement repeating a line for about two seconds. I selected the repeated line and hit delete and the splice was on point, so I didn't have to do more.

[rigid falcon]

This feature needs a redesign. It would be excellent if we could edit and listen on the fly before committing.

Now you commit first (and get charged credits upfront) without knowing whether it will work. It's not about the money, it's just that there is no way of knowing in advance.

[raw oasis]

where can I offical docs for version 5?

[polar spade]

Replacing a word without some change in the surrounding line may be tough to do. Using "fixed" and squeezing as few surrounding lyrics as possible may be as close as you can get to just replacing a word and not affect any of the parts needed to blend-in from the change.

But I don't think then that fixed would reliably work but in a lucky few.

[polar spade]

There no one official manual for how to use Suno or it's models. The knowledge base https://discord.com/channels/1069381916492562582/1373026010211287262 is good to review for beginners. YouTube videos help often in how-tos .

Mostly it's better to ask for what you need here in the channels.

[rigid falcon]

Then, we need an option to recompile the entire track even if it means changing an 'S'.

The only hack that I have found that kind of works is I create a cover, change the word and move style and sound to 0%.

[raw oasis]

I get confused if I should do musical art direction in the Lyric brackets or in the styles tag (field). can I get help with that?

[polar spade]

Replacing a word and affecting a few notes of the melody or getting a replaced chord in the progression is common. That's all I'm saying. But some are needing everything exact as before, which I would agree is preferable, as that's why we have pitch correction software.

But that's not to say the replacement samples aren't just as good. So it's about preference as well, not always about function.

[raw oasis]

but I do notice that these generally work
[Solo: Fiddle plays soaring runs for 8 bars | higher octaves]

[Outro: Cello swells and cross harmonizes with fiddle]

Also it is possible I try to do too much and need to compose smaller and extend

[polar spade]

Yeah. Best I can do it find the path of least resistance and find something that works, then post suggestions in hopes changes can be made.

[raw oasis]

Definitely gotta find a good YouTube channel. gonna see whats out there. Pushing create too addictive. New song and all so expected pain.. but if banging head. take a break. 😄

[polar spade]

@calm grove

Downloading or storing a 60 minute output may be, as you suggest, a daunting task for Suno. That would be some new territory to explore now that studio is out and we can patch segments together indefinitely.

It may be simpler then to crop out 5, 12- minute , segments and patch together in a DAW.

If the output eventually shows in your Library instead of timing out. Mark the time it took , at least for future reference.

[calm grove]

yes i had one show in the library with 5.07 mins so I guess I will need to look for a workaround, would have been nice though .... thank you for the support

[quick sonnet]

The quality of that downloads was really low, beter quality was avaible for Pro users only. So a better way would be to give Pro users more features instead of taking features away from free users.

[polar spade]

You could always record off of your desktop as an WAV or MP3 if you have a digital recording software, depending on what your trying to accomplish.

[calm grove]

ahh, a simple solution yes, just need to make sure the quality will be kept , thanks 🙂

[polar spade]

I used to use polderbits, but that's now defunct. But others are still around.

[echo cipher]

Hey, I created a song that I like, both instrumentally and vocally. But the vocal quality is somehow bad after separating stems. Anyone knows an AI tools to improve the quality of a vocal?

[quick sonnet]

What exactly you mean by bad? Wrong voice, or something like that?

[echo cipher]

Sounds like it was recorded in a bathroom. I want to have a vocal that is as close as possible to studio quality and I'm sure there are services that can pull it off

[raven turtle]

The stem separation even in Studio isn’t perfect and depending on factors such as the pitch of the vocal (baritone male or high female) and the density of the low end can mean the vocal information can be lost because of the lower mids in the bass. Also artefacts can creep in and that “reverb” wash too.
I’ve found UVR a more reliable splitter, but still takes a lot of work in a DAW to get a usable vox stem.
I tend to clean up the stem as much as possible in a daw, and then run it through Audimee and replace the vocal with a voice clone

[polar spade]

If it's just the vocal, then sure, use anything to master it to your liking.

If you plan to add it back into the mix, then mix everything together (i.e. have the individual tracks from 2 tracks and up, sequenced at the same time) to make sure they all sound good together as well.

If not, the quality of the new mastered vocals may be reversed as sounding great alone but sound like crap -- dissonance, too low, too compressed, too dry -- when mixed back into the music tracks.

[austere minnow]

Is it already possible to make a persona with one’s own voice?

[hard beacon]

nope

[glad badger]

How are people hiding this part?:

[hard beacon]

öhm.. there is no way to hide it ?

[glad badger]

I can't seem to figure out how... it's only remasters that show up for me

[carmine folio]

Hello there, any one have any tips about how to make a proper duet of 2 females voices or 2 males voices but doing that the voices are really distinct from each other ? Every time I try no matter how, the voices are way too similar and it doesn't really sound like a duet. Thanks for you advice !

[final hazel]

Go to 3 dot menu, then to "Edit Song Details", and untoggle "show remix origin"

[glad badger]

That's the thing, they're all off lol... hides covers and all that, but not remasters

[austere minnow]

Is it already possible to make a persona with one’s own voice?

[glad badger]

at least not for me

I say that, but now I found a song that shows this:

The Edit Song Details actually has the Remix option thingie greyed out for this one... Did I break something? This one was an upgrade from a V4 song I made months ago

[quick sonnet]

Exept you can reproduce some bug, I was faced with when creating some cover for an other remix contest here.

[glad badger]

If it's a bug, I seem to be able to consistently reproduce it lol... Almost all of my songs are like this

[quick sonnet]

But are all of your songs remixes/covers?

[glad badger]

No, but I do remaster every one prior to publishing

mostly to smooth out the glitched parts until I'm happy with it

The remixes and covers I have NOT remastered don't seem to ahve this problem

[polar spade]

@crude badge

1. As I read your post, I assume you have an annual Pro account?

2. You want to try out the Premium as that level has Suno Studio.

3. You want to try Studio at Premium level , for just one month but not commit to a year.

One option: Would you consider making a second account to buy at Premium level, and share a couple of your songs from your Pro to try in Suno Studio of your new account , or create new songs. Cancel the new "test" account after a week to assure it renews at only the Basic rate.

If you like Studio, then update the original account to Premium annual, and, I assume, it has a prorated option. That would be for you to confirm.

[knotty stag]

Is there a way to edit the last 20 seconds of a song, which is just an instrumental part, in Suno editor (not using Studio)? The song come out really well, except for the last few seconds where there is a kind of hissing type of noise in the music.

[brittle socket]

I think extend might do something like that

[knotty stag]

thanks, that is giving me something to work with

but the extensions are very short

[livid crescent]

Does anyone have any tips on how to get vocals to do growls or screams consistently? Tired of spending 1000s of credits per song trying to get the singer to not be a whiny little...

[tawdry scarab]

getting extends to work when they come back short is mostly either move the start of the extend back further into the song so it's not carrying closing cadence into the extend or keep extending out from the short extend s as long as you only include sounds that are not a closing cadence

if it takes 1000 credits to get a voice you like you probably aren't asking for it the way the model wants you to

[livid crescent]

That's why I'm asking 🙂

[tawdry scarab]

what are you doing at the moment do you have an example of a song that succeeded eventually but took forever

[livid crescent]

I haven't succeeded yet

[tawdry scarab]

Here’s a quick, useful map.

Subgenres (w/ typical scream flavor)
• Death metal / Brutal death / Slam — low growls, “tunnel” resonance, false-fold distortion
• Melodic death (melo-death) — mixed harsh + clean; mid growls, intelligible diction
• Technical death — fast, precise mids/lows; tight consonants
• Black metal / Atmospheric black / Blackgaze — high shrieks, icy edge, airy compression
• Deathcore — lows + “bree/bree” false-folds, pig squeals, big drop accents
• Metalcore / Post-metalcore — mid screams, gritty belt, layered gang shouts
• Grindcore / Goregrind — extreme distortion, rapid-fire phrasing
• Doom / Death-doom / Funeral doom — sustained subterranean lows, huge space
• Groove / Thrash with harsh vox — shouted grit, taser-like twang
• Symphonic black/death — shaped vowels, sustained resonance over orchestration
• War metal — chaotic, overdriven mids, relentless airflow

this was from asking for metal voice influences that are far from screamo

then i'd put screamo in the negative

and then there's more stuff that can be done on top of that

[livid crescent]

I was asking to see if anyone had experience in actually getting them consistently. I've used variations of most of those examples, still sounds like the 'Justin's lol.

[tawdry scarab]

i have excellent experience getting whatever i'm after in most cases

[polar spade]

Consistent means you've had success with at least one, doesn't it? If your tries with each of these similar prompts are not working, that seems the first goal.

Which sub-genre are you mostly going for and what's the general content of your lyrics?

Note: I'm no connoisseur of metal. But I'll even assume mixing a few of these might get us the Jackson Five if we're not specific and sparse in the prompts.

[tawdry scarab]

i could try and go get a low voice and make a persona in a bit but it's hard to know what the brick wall is without a sample of a persons work

[livid crescent]

Currently working on Prog/Djent. Tried every variation of 'harsh', 'raw', 'growls', 'screams' that google could come up with, still sounds like a boyband singer. here's a brief lyric sample: ONE SPARK!
ONE BROKEN PANE OF GLASS! THE FEAR IS GONE! THE WAITING'S PASSED! NO MORE TALKING! NO MORE LIES! JUST THE FIRE IN OUR EYES! THIS IS WHAT YOU GET WHEN YOU PUSH US! THIS IS WHAT YOU GET!

Woody will get upset if I post more of them lol

[polar spade]

Would the consistent caps and exclamation points take it over the top? You may have experimented.

[livid crescent]

it's not consistent with that. sometimes caps work, sometimes not

[tawdry scarab]

well, ^

what they said

I'm going to play around for a second and I'll see if I can hit on something

[livid crescent]

The problem I'm running into is that eventually the song will usually go in that direction, but it is VERY difficult to get the song to start without whiny vocals. Which is not a good look for prog

[sour stream]

Hi all, I use chatgpt to help with my lyrics and styling, but does anyone use the built in Suno lyric writer? Is it any good ?

[tawdry scarab]

hopefully I can get a good voice, clip down to just the heavy vocals, and make it a persona

and then it will hopefully kick in immediately

I've got deep mic swallowing vocals you can't hear any enunciation from if that helps

@livid crescent is this an acceptable sound? https://suno.com/s/ozOghIkdhXUaYCpL

the tags are gpt assisted so if things are stupid they can be swapped out for something more apropriate

but it got a pretty aggressive while still being legible screaming

[carmine folio]

Hello there, any one have any tips about how to make a proper duet of 2 females voices or 2 males voices but doing that the voices are really distinct from each other ? Every time I try no matter how, the voices are way too similar and it doesn't really sound like a duet. Thanks for you advice !

[tawdry scarab]

also this one seems alright from the same gen: https://suno.com/s/vh3GETiGIrkjr5GL

Xianah: i haven't done it but I want to, gen the same song with two different singer descriptions then chop the vocals up in studio and generate a cover or new musical backing for the mixed vocal stems, but that's kind of involved

it is very hard to do what you describe

doing it a small bit at a time with extends is an option

[carmine folio]

Yeah that's what I thought, I have no problem with duet with male/female vocals, but 2 females or 2 males vocals is really hard

[sour stream]

Yes you can if you have pro

[sour stream]

Steps to Create a Persona with Your Voice

Record or find your vocal track: Record yourself singing a song or singing your own original content. Make sure it doesn't contain any copyrighted music, as Suno checks for this.

Upload the audio: In Suno, go to the "Create" tab and select "Upload Audio".

Use the audio in a song: Add your uploaded audio to a song, then optionally use the "extend" feature to isolate your vocals.

Make a persona: Once the song is created with your voice, right-click (or click the three dots) on the song and select "Create Persona" from the menu.

Adjust and save: You can then customize the persona's name and description, potentially load an avatar, and enable the option to keep the persona private to protect your voice.

[polar spade]

@livid crescent @tawdry scarab I'll throw this in for comparison, so granted it may be way off base. It's not whiny or melodic on vocal but a prog djent feel on the music (?), which seems the right direction?

https://suno.com/s/HVYxFyWWWKwsA9SU

https://suno.com/s/NTlYGLr9LjBRi8KU

[tawdry scarab]

you can no longer use uploads for personas I don't think

[sour stream]

Since when ? 😭

[tawdry scarab]

5

or shortly after

[sour stream]

I hate 5 more and more every day lol

[tawdry scarab]

that one you can blame on users

they don't open up uploads because people don't behave

[sour stream]

Yet it still lets up upload voice free music (with copyright music). It blocks most but not all

It let me record my vocals and upload just now ?

[sour stream]

Instead of upload, choose to record and record your voice

Still works. You have to record your vocals live instead of uploading a recording of your vocals

[tawdry scarab]

on the question of personas - oh glad to see you're already correcting my bad advice

[lyric kestrel]

I cannot load suno? Whenever I try load Suno on my browser all i get is this.

[hollow hare]

Working on my end. Maybe an issue with your DNS or VPN if you are using one?
Try going through https://discord.com/channels/1069381916492562582/1417459233544339537

[lyric kestrel]

thank you, clearing cache worked :)

[waxen fossil]

a lot of my gens are gone in the create history list, anyone else have this issue?

[plain valve]

Same all of my gens I was tweaking are gone too

[frozen dagger]

Looking for human help. ai sending me down a rabbit hole.
"Live feed dies" i want the "eye" sound rhymes with 5. not give...
Looking for consistency... I've tried a few phonetic tricks. Am I maybe overlooking something to get sounds you want without wasting time/credits with iterating and praying?

A heteronym is a word that’s spelled the same but pronounced differently, and often has different meanings depending on context.

🧩 Example:

live

/lɪv/ (liv) → “to be alive” or “live feed”

/laɪv/ (līv) → “happening right now” or “broadcast live”

[sour stream]

"Live feed dies" i want the "eye" sound rhymes with 5. not give... ? - confused

if i want it to pronounce a word a certain way, i spell it that way - like Troy Sivan it kept saying "si-van" so i changed it to "Troy Sih-vahn" and it pronounced it that way.. does that help?

[sour stream]

Phonetic hints in text (for AI or collaborators)

Use “(rhymes with five)” or “pronounced līv” inline the first time.

Example:

"Live (līv) feed dies, I revive what's inside"

That small note forces correct pronunciation when sung or generated.

Rhyming scaffolds

Write out the rhyme group so the model hears what family you’re after:
“five / alive / survive / drive / live (līv) / thrive / high / die.”

Then insert your desired line — this anchors the generator’s phoneme choice.

Stress markers

You can fake phonetics visually:

līv or lyve (temporary spelling)

or li’ve — sometimes stylized apostrophes trick the model and human readers into the right rhythm.

Homophone substitution

If the generator keeps reverting to the wrong vowel, temporarily swap to a pure rhyming word (dive, hive, thrive) for structure, then replace it manually with live (līv) in your final edit.

🎶 Human shortcut list — words with “eye” /aɪ/ vowel

Use these to rhyme cleanly with live (līv):

/aɪ/ Family Examples
-ive live (līv), dive, drive, strive, survive, contrive
-y fly, sky, try, imply
-ine line, sign, align, shine
-ide side, ride, collide, inside
-ight light, fight, night, ignite
Oddballs die, high, my, eye
💡 Example Fix

Original:

“Live feed dies” (wants “eye” sound)

Try:

“Broadcast live, feed dies — I survive”
“Stream goes live, but the vibe flatlines”
“Stayin’ live, feed dies — still I rise”

All hold that /aɪ/ thread, reinforcing the vowel sound.

[young vale]

how does this make sense

1 sentence

causes me to somehow copyright some material i dont even know about lol

[random stag]

I’m a premium paid user but I still don’t have access to v5?

[half dragon]

Try "Find yoor way to me"
Apparently "Find Your Way" is detected as existing song title... I know, this is not the right context but Suno is quite strict.

[quick sonnet]

There are song titles containing just 1 word, does it mean this word is then forbiden to use in suno?

[frozen dagger]

appreciate it. i haven't tried [telling it rhymes with five], at this point this it's about to be "Hive feed dies". appreciate the response. lie-vuh ... SMH. i get some words gonna be trouble but common heteronyms should be a way to clarify. wait for v8, i guess. lol V8... like the commercials. ok im done. back to work. thanks again.

[half dragon]

It may or may not prevent you from creating a song but when it does - just have to find ways to get around it, by spelling it phonetically.
You can modify the words for lyric display later.

[green pike]

there's none, copyright issues are very weird after v.5 uptade... best advice is to re-generate another song, sadly

[quick sonnet]

The really strange problem I've faced, was using something suno thinks is copyrighted in styles prompt. For example have used a prompt like "generate a 4-min instrumental track..." and suno gave me an error mesage saing something about 4min being copyrighted or something like this. As seen here it's not only about how words are writen.

[half dragon]

You can change the prompt - four minutes or 4min .. it sucks but it does happen.

[quick sonnet]

4min is exactly what suno seamed to have a problem with. Since I am a free user and every character in that promt counts using four minutes instead of 4-min could be problematic, but for a Pro user could be a good solution.

[grizzled ember]

Report it to suno I have reported copyright issues and they fixed it immediately

[quick sonnet]

And where should I report it? Here?

[grizzled ember]

I emailed them. It literally would not make the song and then suddenly it worked was fixed a few minutes after I emailed

How do you get rain sounds in prompt I've heard people do it?

[polar spade]

If you've recently paid from your bank , but the money hasn't posted to Suno's bank, then the account is not yet activated, it's only "set up" and ready to start functioning once the payment has cleared.

Five business/banking days is typical for banks to review a first transaction with any new payment source. How long ago was it deducted from your account?

[marsh thunder]

try 4:00?

[crisp bronze]

Not sure sudo song lengths can be chosen either

[quick grove]

Anyone know how to stop Suno from humming in the beginning?

[cedar owl]

Slap them, then refuse to give them any snacks if they don't stop

||add [Instrumental Solo] at the beginning or some variety of [bro pls no with the singing] and hope for the best, that's what I do, OR I make lemonade and turn it into something else||

[quick grove]

It's okay, someone helped me figure it out.

[polar spade]

@sharp cosmos did your artwork upload show up? Sometimes it takes a few minutes before it updates the artwork.

[polar spade]

What solution did you come up with? I'm looking for a simple workaround to have Suno avoid making interjections or audibles at the beginning of genres like hip-hop or trance

[sharp cosmos]

The truth is that it doesn't even appear loading, I click and it doesn't work

Can I record a video to show?

[polar spade]

Oh. Is it the right file extension, minimum 72O pixels tall, 9:16 aspect?

[sharp cosmos]

like this

[polar spade]

To upload a video is fine also, it would need to be 10 seconds or less and mov or MP4

[sharp cosmos]

that's the problem

[polar spade]

It looks like you had uploaded one photo before?

You can download a Suno artwork and try uploading. If it works, then you might be getting blocked for uploading public artwork. Is that a popular anime character or just an AI one you made up?

[sharp cosmos]

anime image itself

[polar spade]

Suno might be blocking it. Upload a Suno artwork, then upload the same photo as you already have there - the blue insect one . See if both have no problem.

[sharp cosmos]

OK

[sharp cosmos]

I couldn't even create it or take a picture, it didn't work

[velvet granite]

Suno is problemstic lately, returning monotonous results. I have over 200 generations of one song and one amazing final version - the problem is it's spread over 3 generations. 1 = monotonous verse amazing chorus. 2 = monotonous chorus amazing verse. 3 = boring monotonous song fantastic bridge.

They are all different speeds too, even though I specified the speed.

I need to somehow splice these together. Is there a way to do this in Suno? in Studio maybe? Then generate a single acceptable song?

[jagged kindle]

Suno needs to get serious and invest in pro audio hardware. I would pay big for a box with a dedicated GPU and in turn produce music in real time

Like big big money

[frozen dagger]

play with Suno Studio. or drag the song gens you like over to where you put prompts and drop it in the Inspo. can have up to 4. wouldnt bet on them being great, but will fuse them and you can cover that or remix the new one.
aside from that seperate stems and put the vocals on the track in studio. add instrumentals.

v5 is more monotonous, leaning for more stability and less variation. use older models to explore if stagnant, or alter text prompts. small splash, big ripples.

[regal echo]

oh my god ive finally managed suno to make a song ive requested a duet.... been tryining since v2

[polar spade]

If we have problems with other areas, it might not be Suno related. If you haven't cleared your cache or restarted your phone/desktop in awhile, it may need that.

[edgy plinth]

Guys, how to make the lyric exactly in european portuguese? 95% of the times it goes to Brazilian

[quiet solar]

If Suno can pronounce things in two possible ways, one of which is wrong, there's a good chance that's what it'll do when you least expect it. 😂

"Lyve" is a good substitute for getting a reliable "Live with the eye sound", as Suno tends to read Ly as "Lie" consistently. Over time, you get a feeling for words it's going to regularly trip up on - I usually get around it by modifying the text to a phonetic equivalent, or inserting ' to indicate a break in the word for stressing if all else fails.

[livid crescent]

This is awesome... except it's nothing like what I asked for - https://suno.com/s/wWc3IXtEuaF22H2B

[polar spade]

You might want to try asking in the Portugal https://discord.com/channels/1069381916492562582/1230630513299947581 channel to see what tips those users may have.

[hidden yarrow]

".m4a" is the extension for an audio only track . It seems you have splitted an mp4 into m4v (video) and m4a (audio).
You probably wanted to just split a longer video into smaller parts, but you will know better....

[echo prawn]

I want to replace the section, but why is it stuck at loading DSP engine?

😄

[swift gulch]

what is the reason my image is not loading on the top?

[summer garnet]

Quick question. When you drag a recorded beat you recorded yourself to the studio timeline and drag it into create. I picked "cover" and in style I put the instrument I am looking to do with that beat. However, after it generates, it appears to put vocals, and other intruments outside the style scope. Is that typical or maybe my style prompt needs to be better?

[livid crescent]

I typically use something like [Guitars Only][No Other Instruments] with good results. Sometimes it still adds other things, but usually not

obviously... substitute 'Guitars' for whatever instrument you're looking for

[polar spade]

Does it meet these criteria?
Has it been a couple of minutes? Sometimes it takes awhile.
Are there any images of shirtless people with long hair?

[near flume]

Suno doesn't understand time or duration so you're not going to get the results you want.

[summer garnet]

Will try that out. Sometimes it would be nice to see what other instruments it adds and just split the stems and take what I want, but... I would like to just get what I want at times. Thanks! Do you typically just bracket tag instead of the

guitars only, no other instruments

Or does it matter?

[livid crescent]

I don't know if it matters... so I use the brackets 🙂

suno has enough troubles I think lol

[swift gulch]

shirtless no? its been sometime now yeah

[summer garnet]

It depends how big it is. I found if its 3000 x 3000, it will not accept it, but if its 1024 by 1024, it will.

At least that has been my experience

It may have some max size pixel ratio, although I wish it would just error out and tell you

[swift gulch]

well you see the image on the left bottom

[quick sonnet]

Sometimes it works.

[polar spade]

Oh. Is already loaded, just not showing on the songs main page.

There no specific reason or fix. It will likely show up later then.

[swift gulch]

yeah exactly

[restive turtle]

hi guyss how do you prompt suno to keep the instrumental consistent without dropouts

https://suno.com/s/7CntM5adLWKpR0aW this is what i have so far with this, i really want to keep the instrumental consistent without any resolve

if everyone knows how to do it, please it would be so helpful lol

[graceful condor]

Remaster now changes the lyrics? "Chasing" got changed to "Emptying".

[half dragon]

What variations strength level was it at? High?

[polar spade]

It shouldn't. Are there any back generations having "emptying" in the lyrics? If Suno ever changes lyrics it's always been just a morphing of a word because it sounds similar, but changes once it's more clear or more muddled. So "carp" might become "car", "sweet" become "sweep" or "sweat" .

[polar spade]

What does the (!) info dot say that variations affect? I'm not on desktop at the moment.

I've assumed variations meant the spectrum of mastering was a little , average or a lot. But it could mean to vary the output as well. 🤷

[polar spade]

I meant what does it actually say it does when we hover over the info dot.

[half dragon]

That. It says exactly that.

[polar spade]

Oh .. hahah... Yea.. I see now... Sorry . What the heck does "transformed" mean? Into something better or into something different? 🤣

[half dragon]

Yeah, it does not explain more.

[polar spade]

I guess I'll need to pay more attention and test more remastering, before I can help. Sheesh! 🙄

[quiet crescent]

Hello, can anyone help me? I’m having this issue..I paid for Suno Studio but I can’t access or use it properly. Can someone help me figure out what’s going wrong?

I got the message: Error recreating clip.

It doesn’t work no matter which browser I use. Ad blockers are turned off.

[polar spade]

I kinda thought we'd be lucky and all they did was rename 4.0 remaster as subtle, 4.5+ as normal , and 5.0 as High. But clearly not so fortunately. Darn!

[half dragon]

Did you select a clip or section of a clip? and still got the error?

[quiet crescent]

yes I selected the clip

I still get the error...

😢

[polar spade]

What are you attempting to do: add a new section? Replace the selection with something else?

[half dragon]

I just tried, select a clip by clicking on it doesn't work. ---> might be a bug
Try click and drag to draw a box on that clip, then the button would say Replace, instead of Recreate... then it would work

[polar spade]

You just wish to copy the whole track into the new track?

[quiet crescent]

ja replace...

oh ok let me try

give me a few minutes : )

brb I try : )

[half dragon]

Another way to 'recreate" might be ... add a new clip, draw a box on that new clip exactly length you want.. then set the the type of creation you want, whole song, an instrument..etc.. then hit Create

[quiet crescent]

No, I want to keep the song as it is—I don't want a Suno creation, just to improve the existing song.

[polar spade]

If replacing the whole thing.....

You could just do a remaster of the original and drag the new output into Studio? That would give the same time length.

[quiet crescent]

okok it works 😄

jaaaaaa

Dankeschön!!

thank you guys ❤️

love you

😘

[polar spade]

Did we do anything? Looks like you got it on your own. 😁

[quiet crescent]

a good idea too : )

no with your help : )

[quiet crescent]

this works

Wishing you a lovely Sunday. Thank you so much. Now I can get back to work 👋 😅

bb

[final hazel]

@waxen current Hey Dirk, we appreciate your enthusiasm for building tools .. but promoting or sharing links to external websites or personal projects isn’t allowed here. Please keep our channels free from advertising or external promotions.

[waxen current]

hey i am dirk, a composer and producer from germany - i am a suno expert and pro prompter - would love to share my experiences with u all. would love to connect . peace

[graceful condor]

What the hell is going on lately that you can't have a normal male centered song anymore or female centered. Each time you get harmonies of male and female and even female singing lines the male should or opposite around. Even if you add [Male Voice, male singing] etc before a Verse, or Chorus, and also mark all other options only on male you get female voices. it is so annoying.

[tawdry scarab]

this is unfortunate that you are experiencing it, it could be related to the style or lyrics guiding the generation in a way that strongly favors back up singers or harmony. the model is not acting as it does for one persons focus on one day for everyone

these problems tend to be finding out the limits and strength of different keywords and how style pulls the generation as well as lyrical content

v5 does seem to like more adlibs but the line on what's too much depends on the style of the music and the person, v5 has a few quirks

[graceful condor]

Honestly I am so sick of it. I generate the I do not know how many versions and each time I got such a great version and the whole narrative is ruinied because of the female voice singing. This just got so much worse the last days with Version 5. I got a Silent Hill 2 inspired Song I created. There are only two lines a female voice needs to speek and the rest is male. All my prompts are leading to male. I put male everywhere to be sure to get a male voice and still this dumb Ai is ruining my songs I am so mad already because I got such a good song now and dumb freakin ahhhh ruined. Hate this!

Genre:Progressive Rock,Post-Rock,Alternative Industrial,Dark Cinematic Metal}{Subgenre:Gothic Pop,Darkwave,Ambient Grunge Fusion}

[tawdry scarab]

it's ai

use extend or something to piece it together but you're struggling with a need for total control from a tool that doesn't offer it

[graceful condor]

But it can't be that each update they do is making stuff worse instead of better. This was not an issue in 4.5+ but now all is the same because they even forced the mechanics of 5 onto 4.5+ which makes not much difference anymore which one of them you use.

[tawdry scarab]

it isn't

you're frustrated and have confirmation bias

you said you tried a thousand times to get one thing to work, there is not much in the model other than fringe music that takes more than 100 gens to get, and when you need to piece something perfectly you should extend a bit at a time

[graceful condor]

Man I know I can extend but that is not the point. It just pains if you got a song with so many perfect elements, great solo etc. And then there is verse wrong etc. And if I extend all that came after it is gone and will not come back that way and as I said if it was perfect it is just painful to loose it because the ai does what it wants again.

[summer garnet]

Not sure if it helps, but have you tried taking the generated song that has the most of what you wanted except that part and save a persona off of it so you keep the guy's voice? Then maybe use studio to grab that section where the female is singing instead of the male and save that as a new song. Then go to that new song and do a cover on it with that male persona (male selected) as well for voice and try to experiment with the sliders to get your desired result? Once you have it. You could then take the result back to studio and plug it in where the female is singing try to work it in there? Just a thought.

[graceful condor]

I even already use a dedicated male Persona.

Honestly I feel really mocked by suno. I even got into Studio now, selected this section. Put into the lyrics and added this in styles: [Male Voice – breathy and haunted, shifting into rougher intensity, sustained emotion][Arrangement:No female vocals or harmonies in this section] And all generated sections are again female. WHY? I am so sick of this.

[torn peak]

Its a sign

[half dragon]

Making duet or multiple vocals is never easy; either you got lucky or would have a lot of challenges.

[tawdry scarab]

tags are stronger when they're short and negations don't work inside tags a lot of the time

meta tags with brackets should be small atomic concepts not depending on any other tags or ideas

[crystal helm]

(This is for all the versions but mainly on v5)

Sometimes there's that song you generated that is so perfect and extended to be around 5min+ or more.

The burning question is... why the hell... when you make a cover of that "cover" with a persona or not, it stops following the original song?

To give you a example, around the 3rd or 4th minute, it deters away to the point that it just improvised that section, ruining the entire song.

[tawdry scarab]

the cover can only produce so long of a song

so it needs to end at some point

[graceful condor]

I even get mixed vocals in purely single songs. Even if the song is fully male or female I often get both which makes no sense. And in this case only 2 lines are spoken by a female voice the rest is male and all sections are marked for male but it is always at least one section a female voice singing where it shouldn't be.

[graceful condor]

Because the AI is dumb

[crystal helm]

I honestly even feel that Suno it's rushing it to end before the 5th minute. But why wr have the option to upload 8 minutes of audio if it isn't going to commit?

[tawdry scarab]

.

covers, persona, and inspo take some of the memory available to the process to set the context

which reduces the length they can gen for

[torn peak]

Hey is there some sort of manual or book where I can study suno commands, prompts and other stuff?

[tawdry scarab]

there is no reference

[torn peak]

Theres no collective of data?

[tawdry scarab]

there's the reddit and we have a workflows thing here

I have a set of instructions that work but they don't give you any list

it's an LLM you get what meaning you put in in the eyes of the training

it's not a deterministic assembly from your stated parts

at this point they probably have fine tuned submodules for things but still, there is no reference of tags that will work

because that's not how LLM works

[torn peak]

I see, thank you

[tawdry scarab]

think of it more like making a stew, some flavors will be on top, some will be buried

[crystal helm]

I see...

I usually do covers on my finished songs. To change the genre but maintaining a lot of the original song.

Sometimes, it's perfect. But some songs seems that are bound/stuck for being too perfect to make a exact cover.

So basically, the song itself it's too complex and long to memorize it.

[tawdry scarab]

you try to stage when you put things in and how much from experience

you can put 50 tags in that you can only taste if the salt in the soup is low enough, do you see?

everything is always influencing everything else

ssome folks work to refine their language to be short and effective but I've always had more success going broad but keeping in mind a subconscious broth, gen, taste, adjust, gen, taste, adjust. and when something will not happen, remove all tagging anywhere, all lyrics, repeat the lyric 50 times, put in a ton of tags and styles and punctuation and phonetic spelling, and then a special trick, extend using 1 to 5 seconds multiple times and you will break the song context and get to use hard changes of style

[torn peak]

this is a new age of music creation

[tawdry scarab]

totally

and there's a thousand ways to do it it's nuts

I can only speak to my methods

[opal blaze]

anyone found out how to get 2 different male voices in v5 ? duett, voice A, voice B, nothing seems to work

[tawdry scarab]

I can get them but I don't, like, sit down and do the extend work for a duet

it's just too much

[half dragon]

Have you tried to search this board? there are quite a few discussions on this topic already - one of the hottest topic 🙂
In a nutshell, there's not a guarantee way to make it work everytime, but it isn't like "nothing seems to work"
You'd have to spend more time and more credits and try many methods, Extends to build each section... Edit tool or Studio (if accessible) to replace
And there will be challenges as model is generative.

[tawdry scarab]

you can make like a header with vocal descriptions and short tags consistently used for each character then keep nice short [tag] tags for them before their parts

put them in quotation marks and put dat tape and field sample in the style

[opal blaze]

im absolut new to discord sorry, dont know search etc

[tawdry scarab]

this is my version of how things work: https://discord.com/channels/1069381916492562582/1365192916687126650

I don't super like bold claims and perfect lists, this won't tell you exactly what to do

it's just the guidance for how to talk to the interpreter that consistently works when it's all orchestrated reasonably

https://discord.com/channels/1069381916492562582/1252267444773326848

here you can find other peopls' guides

[wheat tulip]

This is something most people don't understand and/or forget: Generative models aren't deterministic.

[tawdry scarab]

some of them can be, but yeah totally

oh, sorry, yeah they aren't ever in the way we mean

I slipped up and was gonna mention like repeatability deterministic gening

different from deterministic rule following

[half nymph]

hello

[graceful condor]

Honestly why the freakin hell is it not working if I select a section in Studio or Editor, generate that section to get a male voice but all I get is female, over and over and over again. I am so close to punch something because I wasted hours on this song and I am really sick of it. Regardless what I do the AI does what it wants. Regardless all commands etc.

[tawdry scarab]

stop doing things that don't work more than a couple times, it made it female in the first place because there's a groove in the solution space for female singing there

sometimes you can't fix that

I don't mean to be overly dismissive but you're frustrated that one of the hardest things to do is difficult

meticulous control of vocal performance among people that can do it repeatedly involves a lot of credits and throwing around different strategies

it's "possible" but you're playing the last level on hard mode

I keep wondering if it can be done to, in studio, chop up a bunch of attempts into a single vocal track and then cover that to smooth out that the singing won't be exactly the same across different generations

[half dragon]

A quick test but I might have got lucky on first try.
Then split stems then cut and replace the vocal for that section. Could have done stems first as well.

[graceful condor]

Dude this is what I have done. I split stems and just selected the section in the Vocal stem part and generated that section even with direct commands in Styles that it is purely male. And still female. I got one good male version (at least mostly male later the female joined in) and I would have taken that one but just that one was off beat. I do not get it. all section I create are always marked with Tempo "on beat" but it was off and I did not managed to bring it on beat. So it was a waste again.

[half dragon]

It's frustrated, yeah, sorry. Maybe it's time to work on a new song 🙂 come back to this at later time when the tool/model gets better.

[graceful condor]

Are already extracted stems and projects gone if I close that window? I mean I do not wanna waste 50 credits each time to generate those stems again. I always see that Studio is automatically saving in the background so is this stuff saved?

[half dragon]

Yes, they will always remain if already extracted.

You will always see them in Studio, but if you click Get Stems from the menu in Library, you might not see the small link says "See extracted Stems"
Many users accidentally extracted them again.

[graceful condor]

How do I load extracted stems again. Is it under "..." Open Project?

[half dragon]

Just Open in Studio on the song and you'd see the extracted Stems on far right panel.

[graceful condor]

If I export a full song I always save the whole song with the master main track right? because the stems are always missing something. But if I do so, I still got the other voice from the master track at this section. I don't get it. I never had this issue with other parts. I changed drum sections a certain sections in the stems and saved all for the full song and in the end it took the new generated part from the section and exchanged it with the og one from the master track but now I got the male and female voice why?

[half dragon]

Click the speaker icon on the master track to mute it. So it'd play only from the Stems that you imported into the timeline.

[graceful condor]

I know this is not what I meant. The stems are never extragting all, there is still some stuff missing which is still part of the master track. So before I always at the end saved the whole song and exported it with the master track too. So all speakers were on and I did that while I changed drums and guitar parts and in the end it only used the new parts but here it uses the old voice from the master track and the new one and this is what I do not get.

[half dragon]

Strange indeed. I've never run into that issue.

Just make sure to play every track on S mode in isolation to identify problematic track. Mute it and play everything as well... see where the problem actually coming from.

[graceful condor]

I already checked that it is the Verse from the master track he still uses. Now it sound like a duet. The old female voice from the master track and the new stem with the male voice. I even deleted this section in the original stem so that it really only uses the new Male stem. So it is clearly because of the master track.

But as I said if I export the full song with the Master Track muted (speaker icon off) the songs sounds not the same anymore because Studio never really extracts all stems. There are still some hidden stems left in the master track which gives the overall song a full and warm sound. This is missing. And so far it never was an issue. I always changed parts in the stems like I said guitar, drums, synth etc and in the end saved all plus the master track and the final song only generated the new stems and exchanged them with the old ones from the master track and this should actually be the case how studio works. But why is it now using both voicelines Master Track and Stem. I do not get it.

This gets more and more absurd

[half dragon]

I never keep the master track once I imported all the stems on the timeline.
And exported version sounds the same as it should.

[mortal sphinx]

Hello,
I have a problem with a song. It's quite a long song that I have to split into three parts to stay within the character limit.

After 300 generations, I finally found a first part that I'm completely satisfied with. However, when I extend it for the second part, there's always a moment where the voice goes off track and you can't understand anything anymore...

I've tried cutting it into smaller parts, using other versions of Suno, cutting at other points... nothing works. As soon as I get to 4 minutes and a bit, it all goes haywire...

Yet I've already generated 8-minute songs without any problems, in several parts too...

If you have a solution that would prevent me from throwing away this first part that I like so much...

[graceful condor]

They really do not not even in the slightest. It feels like if you have fries but forgot the salt and sauces to use them. Check again. They really sound like day and night.

[half dragon]

Is it occurred for all songs that way in your experience, or just this instance?
I feel your frustration, I'd be too if that happened to me frequently.

[gusty vault]

Anyone knows how to double-time an instrument in studio?

[raven turtle]

Can’t get Suno to create 6/8 time on an electropop track. No matter how phrase things. I wish it would listen to me. I tell not to sing all over intros either and it does it all the time. I get it. It’s Beta. And the tech is amazing… but still … come on…

[twin forum]

So guys; is the cover's length context a bit of smoke and mirrors? Particularly when it comes to purely instrumental. I've found that a purely instrumental piece when attempting to cover it exactly but say, changing from a single instrument to a symphony orchestra; it completely starts to randomise and can only pull the melody back from that first...30-40 seconds or so.

And this is running it at 100% Audio Infeunce, and 0% style influence and 0% weirdness

Without lyrics; can it not guide the melody to completion and accuracy?

It seems like Suno actually behind the interface still has the same chunks that say UDIO has - set to 32 seconds. It just doesn't display them or management in the same way it seems.

[cinder drift]

Suno needs to update their v5. I did a couple of thrash metal songs on v5 with specific BPM and it was SLOW. I downgraded to v4.5 and it was exactly what I wanted speed wise.

[cobalt palm]

fr tho

[grizzled tiger]

hi guys. can you help me name the percussion used in this song? the boom tak in the beginning
https://www.youtube.com/watch?v=k6X2wJ6L0SY&list=RDk6X2wJ6L0SY&start_radio=1

[plain pumice]

I've tested something, just to see if it was possible.
Basically made pseudo-words, which basically are made up words, but interestingly enough, it was has been able to pronounce all of them.
Where normal english words it would struggle with, which is quite interesting. Anyone who done something similar or who have experience with using different language than english, but also making up some words, using "unique" names, slang, etc.?

[sleek warren]

https://www.youtube.com/watch?v=JTALblPUaN0&list=RDJTALblPUaN0&start_radio=1

What would the prompt for this be?

[tawdry scarab]

brushed drums, contrabass, breathy voice, quiet club, room noise, warm, lightly swung, minimalist jazz

would be my first attempt

https://suno.com/s/xGfXruRSd8uaPPWN

the lyrics are some stuff i was working on but you can see it kinda gets you in the ballpark

@sleek warren

from that result i'd adjust to using legato and solo, since it was too fast and too many vocal layers:

https://suno.com/s/WgbKFzEqmWm7eiHB

from there id gen maybe five times and decide if anything hit the spot or what's missing

[tawdry scarab]

sons a little like that's 100% wet reverb on a drum set

but i dunno

[tired cypress]

Can someone help me why I can't use my free credits? 🥺

[half dragon]

For account issue, post in #archived-ticket-help if you haven't.

[velvet granite]

@frozen dagger dmming you

[hexed iris]

V5 is almost unusable. And with that even old versions seems to have degraded. The surprise element of Suno is gone!

[dull kite]

guys how to fix artifact in stems ?

[half dragon]

That's a bold claim. Many many ..many... users found v5 is awesome!

[versed grove]

Is there a way to make Suno stick with male and female voices, I have a song idea where it switch back and forth between a male voice and a female voice, but each song keeps switching it around and the female will sing the males lines and the male will sing the females lines

[half dragon]

Unfortunately no. Not in one shot anyway.
You'd have to spend time on this and probably lots of credits.
And there's no one method could be reliable so you'd have to be patience and try them.
One way is to build each section seperate using Extend feature.
Other way is to use Edit tool or Studio to replace the wrong vocal section.
Making duet or more than one vocal is never easy. You either got lucky or faced lots of challenges.

[versed grove]

Alright maybe they will fix it in V5.5 or V6 but that would be nice if Suno would be able to see female and male lines and sperate them

[hoary mauve]

I have created a song that has an instrumental fade out. instead, I want it to end on a strong drum beat. is it possible? I deleted the fade out and regenerated but I see no option to add infos before I regenerate a section

[half dragon]

I don't think there's a way to control that. Just keep rolliing the dice. 🙂

[hoary mauve]

ok, thank you

[graceful condor]

All songs so far. If I export them without the master the sound feels less alive and plain. Because of the missing stems from the master.

[graceful condor]

Btw. I found a semi optimal solution. I erased this vocal part from the Master Track and enhanced all stems for this part considering their Clip Volume. It still sounds not as full and round as the master track but because it is more of a silent slower section it feels less weird. And in the end I got my male voice finally. Still weird workarounds to get there.

[graceful condor]

Okay I was too fast. This is not working at all. The exports get worse and worse each time. I finally managed to add a pure male line to it but in the end the rest of the vocals are barley hearable it is so weird. Studio makes me nuts.

I give up...

[upbeat jungle]

I have a song with length 49sec. I really like it but i want to extend it for 2 min long with modified Lyrics. I had tried it but patching inbetween works using style reference also works, but when I try to extend my existing song with new/modified lyrics, it showing an error occurred. can anyone help me?

[torn spoke]

Btw, for those who don't know, adobe podcast is pretty good at cleaning up vocal stems

The brightness can be increased in post, what's important is just that the artifacts are reduced

specifically the strange "woosh" sound at around 8 seconds

that adobe just, completely erases

It only works if the vocals are voice-like enough though. If the vocals are too robotic it wont consider them a voice

[raw oasis]

If I am genre-bending, (broadway meets prairie fiddle etc) should i go back to using v4.5?

[torn spoke]

I don't see why

[graceful condor]

They forced all mechanics even on older generations it makes no difference anymore what you use because 4.5+ isn't doing anything better anymore, sadly. So better to stick to 5 even if it has a lot of issues but now everything has.

[raw oasis]

okay thank you. that helps me (catch up) a lot.

And also - sorry to hear that for the very active.

[graceful condor]

Okay I am done. I really can't anymore. This Ai is just broken. I even used my Studio fixed track now. All male and generated a cover out of it and the same section I fixed with Studio is now female again! It clearly states no female voice everywhere in Style, in Lyrics, in exclude, in Sliders. I can't...I can't anymore. This is no fun!

[torn spoke]

Yeah I had a problem with it switching to female just now, I had to make a male persona to override it

[graceful condor]

I already use that too and makes no difference. Male Cover Audio, Male Persona, Male slider, Male everywhere in Lyric and Style prompts and still female. I give wrong generated songs a thumb down to signal Suno it is not correct. It ignores everything.

and we do not even talk about a male voice who maybe could be mistaken for female. It is really male male sounding - raspy deep.

[torn spoke]

the thumbs down is just feedback for the next model

Btw, don't put negatives in the prompt, just positives

kinda like me saying "Don't think about elephants"

[graceful condor]

It is time to move on. Maybe I get back to it if a newer Model makes it better. But for now I need to except the song as it is.

[torn spoke]

and boom you're thinking about elephants

if you want to put negatives, you'll have to do that in the advanced style field

nowhere else

[hoary mauve]

for me, it usually works to put "no female voice" in the negative prompt section

[graceful condor]

Best thing to do was making a complete new song out my prompts and lay the other one to rest. Got a better version now. All the time and credits I wasted on the former version was literally for nothing.

[torn spoke]

happens

[torn spoke]

what it's like with the instrumentals

[arctic holly]

https://open.spotify.com/intl-tr/album/4ez1MpadymFe6aMrlaDsRP?si=vElOEC_dS5eDJ2j9Jk0C7Q

[hearty nest]

That's probably been asked 19499 times but still... is there a way to make a song with my voice ?

[ivory peak]

How do i extend a song to 8 minutes? Right now it will only extend a minute or so. Hardly useful.

[graceful condor]

Not anymore. Because they messed up how Personas work. Before you could use a certain voice and the voice had the tone and all the features of the original audio but now it is only looking at how the voice pronaunces words, phrasing etc. But it never really sounds like the persona as it did before.

[hearty nest]

Is that just a querk with the V5 or all models are like that ?

[graceful condor]

Now sadly all because they forced V5 mechanics in the background on all models so in the end using 4.5 makes not much sense anymore because having something different as before is gone now.

[graceful condor]

really today is a top notch sh.tshow which literally deserves a dumf.ck oscar. It is unbelievable how bad V5 is with words. It is unbelievable how worse it gets each day with butchering words, misspelling words or simple talking/singing absolute gibberish which makes no freakin sense anymore. It is so sad that it is actually funny again.

[velvet hornet]

Maybe this is a repeated question, but I'm finding that after my custom playlist finishes, the player automatically jumps to a specific, "different" playlist every time, starting with the same song. The music style in this auto-played playlist is completely different and not at all similar to what I listen to.

This seems to be boosting that song's views significantly. Is there a specific reason or sponsorship for this?

Ideally, when my playlist ends, I want the music to either stop entirely or loop back and restart my current playlist, not jump to a random playlist with music I don't like. Is there a setting I can change to prevent this auto-play behavior?

[sterile yarrow]

Can someone help me? For somehow I cannot generate music. This problem has been for days and when i press create it just spinss the icon and does nothing?

[torn spoke]

is this in all browsers?

[sterile yarrow]

Yes I tried for, chrome, edge, mozilla 🙁 and still does not work.

I also tried just second ago create a new Workspace and it still wont generate it there too 🙁

Is this also a right channel to communicate on my problem with the site?

[polar spade]

It's a teaching channel as well. Lots of new users every day. But, the models DO change and some things that's worked before, may not now, mostly due to security and filter changes.

[sterile yarrow]

I had only 1 tag and it was Cinematic.. and it wont generate anything not even Even slap house music...

[polar spade]

Sometimes we haven't restarted our phones or computers, so always good to do that before other tests.

[sterile yarrow]

I have done reset on pc and website. Ive had this issue for 3-5 days or more I have not counted but now I really need to generate music so I have to get this fixed

[polar spade]

There is never a "fix" that I've seen when something simple doesn't function. That's usually a server issue (then we'd all be affected) or user error or interface conflict.

It's helpful, yea, to discuss with other users here in this tech channel.

You can mention in the bug channel to report, just in case others are also affected.

There is no direct Support other than emailing [e-mail] The hope is that we peers can help work out the problem to keep you from waiting on Support.

[polar spade]

These few steps are recommended to everyone. https://discord.com/channels/1069381916492562582/1417459233544339537

If you are logging in and getting to the Create screen, but it hanging there, there may be issues in communication between server and your browser, so going through the steps is helpful. Turning off VPNs and flushing the DNS are good initial or final steps, which help many

[sterile yarrow]

Thank you!

I am using vpn lets see what will happen

ohh so on mobile I can generate I now checked!

[polar spade]

As a test, you can log in while in incognito setting on Chrome. This disengages much of extensions and virus checkers. If it works that way, you should augment your system the best possible as in turning off extensions.

[sterile yarrow]

Ohh so it was a vpn the problem! But why does it work on phone when I use vpn??🤔

[polar spade]

There's no exact science. But Suno does frequently check your IP and location. Where VPN changes your location, it doesn't necessary hide your phone ID. So more likely your phone APP might work.

[polar spade]

Suno knows who you are on the APP. 😄

[lyric kestrel]

I cannot find any way to make a beat like this in Suno…

https://youtu.be/51qvdWCowfU?si=kPNkqBIIjuMSbfdu

Ai (ChatGPT or Gemini etc) can’t do it either… any help?

[sterile yarrow]

But still thank you so much for the help!

[vestal flame]

how do you make/control a duet?? I've tried things like using persona's, using [male voice] and [insert persona name] to try and control which lines are male and female but no matter what I try it just always seems to be random chance....

[stable ridge]

feedback on this thumbnail for my next album?

[half dragon]

In short, you can't reliablly control it.

[vestal flame]

honestly? just glad Im not batshit crazy xD

[polar spade]

But assuming we're not considering credits use, using a persona with each extension and assuming we're extending per each gender change, we run through the generations until we get the right gender before moving to the next extension.

This is true also with running a full song, then going back in and replacing every segment that doesn't have the gender we want.

If we set a limit of say, 5 generations before giving up, then yes, it might not work every time. 🙂

[polar spade]

Also, worth trying a trick of pasting duplicates of the same verse a few times, prompting differently or the same. Run the generation and choose the one that works. Delete the rest, then repeat with the next section (verse, chorus, bridge....).

[vestal flame]

oh whoa, those are both great tricks! I'm deff going to keep that in mind going forward!

thanks! ^.^

[graceful condor]

This can't be just me because it wasn't the case with 4.5+ before but it is not normal anymore how bad V5 is with spelling. 8 out of 10 songs get misspelled words, weird gibberish or other lyric issues. And it is exhausting to say at least. This ruined my mood so much within the last 2 days me wasting countless hours in fixing Suno's errors.

[weak bobcat]

hi all. i'm trying to create ambient background for meditiaon track. just needs to be muddy chords gently shifting (mabye 8-12 seconds) over and over. maybe 2 minutes of changes. then tiny variation of those changes (like maybe more string) then another loop with maybe more base. I want to be able to patch these together and extend to support as needed

[vestal flame]

oh, a random small problem I have that I would love to know why is a problem. one of my songs I cant replace the image on..... as in every time I try and put an image on it, regardless of what image, it just doesnt stay, defaulting to the mix of colors image.. y?

[half dragon]

Even when refresh after saved? it may just be a refresh issue?

[vestal flame]

yea it never sticks... I can see the image in the play bar, but only in the playbar, the infoscreen and when its just sitting in the playlist its just a mix of colors...

https://suno.com/song/dcaa8a3d-fffc-4166-93b4-2ad523a91cd1

I wish I could screenshot to show what I mean, idk if it does it for anyone else, since the image is in the playbar I assume others can see the actual cover image, but I cant tell..

[half dragon]

Yeah I only see this. Bug? maybe report it? in #🪲┃bugs-web channel.

[vestal flame]

https://suno.com/song/dcaa8a3d-fffc-4166-93b4-2ad523a91cd1

you can see the image thats supposed to be there on the embedded link too....

thats so wierd

[frozen dagger]

try opening the song in editor, save/export it as a new song (clone the track) then see if u can replace the imge on the new copy

[remote dome]

Someone else also getting bad generations today? All generations sounds... distorted?

[foggy aspen]

Can someone help me here?: I’m working on recreating the devotional hindi song in Suno, and have run into a consistent pronunciation issue with Hindi/Sanskrit consonant sustains, especially the “र” (ra) sound.
In the original song, “ईश्वर” (pronounced Eesh-var) is sung as a long, rolled sustain:
Eesh-varrrrrrrrr… satya hai…
But in Suno, no matter how I write it —

ईश्व र———
ईश्वर———
ईश्वरं———
Ishwar———
— the “r” is cut off immediately. Suno holds only the vowel before it, not the trilled “ra” consonant.. Apologies in advance if this is not the right forum to raise this...

[hollow sky]

There is any way to not let suno read prompt as lyrics? I heard from time to time it uses style descriptions as lyrics?

[foggy aspen]

anything under square bracket [] is not part of lyrics

[hollow sky]

yes, but it reads from the "style" not from lyrics so should I put style in [] ?

[foggy aspen]

honestly, i have stopped using style box and instead enter details under lyrics box like this [intro]
[silence 8s]
[instrument: long tanpura drone continues; temple bell chime × 2; bansuri theme Sa Re Ga Pa Ma Ga Re Sa enters softly]

and it works well.

[hollow hare]

Typically repeating a character is used for prolonging, e.g.
ईश्वरररररररररर
ईश्वरrrrrrrr

Sometimes tilde (~) works but mainly does not in my experience
ईश्वर~

[foggy aspen]

Suno starts repeating र.... Instead of prolonging ...

[frozen dagger]

try using [aria, melisima] , or an emdasg alt 0151— where you want the note to hold. ive mad ea song with sanskrit and gens were all varied in pronunciation. in my case it, for me, the intention behind the sounds was more important than pronunciation. Hindi is so old and has so many dialects it is beating your head against the wall for small detail. spell it phoentically with english characters and roll the dice with gens.

[hoary mauve]

keep in mind that it doesn't have to be orthographically correct. it just has to know how the word has to sound

[frozen dagger]

if it is only the one part you want to replace use Studio or editor to select that one part to recreate to try and get it. Most problems working with Suno are nuanced things like this. Trying to get the details articulated. GL

[tawdry scarab]

it's a rolling r?

you could try making a persona for it from this? or use these tags? @foggy aspen https://suno.com/s/9XKKPTdACgY0BMTT

then to use it you'd have the audio influence pretty low

because you only want it to take a glancing blow, or I could work with you on developing a tagging block that gets more focused or sits in your work

[tired forge]

Hello. New to the Discord. I have a question about something that I'm confused about. Not new to Suno, but it's something that I cannot seem to figure out. I'm hoping someone can help me since I emailed Support and I got an automated message that was completely unhelpful.

[tawdry scarab]

breath abates

[hollow sky]

what uploading audio does? you can upload audio as prompt?

[tired forge]

When I try to extend a song I basically get a separate song as opposed to the extension being added to the existing song, which is what I'm trying to do. I'm trying to get the extension to be part of the song and not be separate from the main song. Why does it separate as opposed to actually extend the song?

[tawdry scarab]

you can upload audio just to upload audio, you can do it and cover it, or do it and extend from it

@hollow sky

[hollow sky]

im not sure what does that mean 😄

[tawdry scarab]

you need to use create -> get whole song, this makes a difference depending on what you want to do while extending

@hollow sky what of suno do you presently understand?

[hollow sky]

if I upload audio I can make somthing out of it or not?

as prompt

[tawdry scarab]

yeah you can cover it or extend it

you can make something transformed from it as a base, cover

or you can elongate it with something that attempts to follow after, extend

[hollow sky]

okay thank you 😄

[tawdry scarab]

If you get whole song it will kind of smooth the transition a little, try to match things just a tiny bit, and then an extend off that will contain the whole musical context

@tired forge but if you extend and extend and extend without getting whole song until after, you will distance each extend much more from the musical context it came from

[tired forge]

When I click create on my song I get:

Make Persona

Remaster

(•) Song Radio

[foggy aspen]

thank you for your help.. this is what i was expecting

[tired forge]

But that's on the Website. Is it only the app that I can do it properly?

[half dragon]

Yea you see that Get Whole Song only on the extension (the one says Part 2, Part 3..etc)

[tawdry scarab]

and if you make edits it might break the ability to get the whole song

you're supposed to get it all joined up before edits occur

[tawdry scarab]

oh, I just assumed and went and made the wrong thing XD

[tired forge]

Part 2, Part 3? I don't see those anywhere either. Dang it! R.I.P song...

[half dragon]

When you extended your song, it'd create 2 versions of Part 2 from the original song that you extended from.... on that Part 2 that you like.. you can then do Get Whole Song to merge it.

[tired forge]

OH! Lemme see if my dumb brain, tired as I am can do it. Less than five hours of sleep.

[tired forge]

So I create the extension with the new lyrics and generate it and then AFTER it generates it I do that?

[half dragon]

Correct.

[tired forge]

Wicked

It's all the exact same length… That's a little bit concerning, but let's see how it sounds. Hopefully the vocals are exactly the same. The vocals hit different with the version I have, but it's too short the way it was.

YES!¡! THE SAME VOCALS!¡! Thank you so freaking much!! You're a badass!! I truly appreciate it!!!

I was really worried about losing the vocals because I really liked what I got. I'm so freaking happy that it didn't happen.

[tawdry scarab]

@foggy aspen it is harder than I expected it to be, is this getting close?
https://suno.com/song/bf293903-2000-4785-9469-53aa0bcb8234

targeting the r for elongation seems exceptionally difficult

[foggy aspen]

thank you for all your help and it is better... in hindi "a" at the end makes it completely different.. but will try to manage with this one as have spent 300 credits trying different options 🙂

[tawdry scarab]

well, I'm gonna keep working, this is a good challenge

and I think it's attainable

[foggy aspen]

thank you very much..

[tired forge]

This is absolutely unrelated to the one I wanted to extend, but that's because I'm listening to that one still, but I think this one came out well. Thoughts?

https://suno.com/s/VbMxiqlenB1l8wlF

[tawdry scarab]

https://suno.com/s/6IGpjGwmFTyQ3VTE

try that out, see if we deleted all the schwa

and if what's left reads right

if I"m not mistaken this is the other side of the problem, a repeated wind up to the r sound?

[foggy aspen]

its very close........ "श्व" is not coming out very clear but good.. i will try to understand more and play with it... much appreciated..

[tawdry scarab]

take a look at https://suno.com/s/swjym8Pzl2mVQvLp and
https://suno.com/s/T0ShsnCLGmo6lssP (fixed)
and then I'll leave you alone, I won't keep insisting on trying XD

the very first delivery, i mean I have no ear for this, but it seemed clean

on the second link

[tired forge]

Schwa?

[tawdry scarab]

the difference between how suno was pronouncing it and what they intended is a piece of the sound called a schwa

piece of the pronunciation/letter

I cut just the r sound, deleted the rest of the word, in studio, and then pasted the r sound over and over and over, and then covered that over and over with instructions on how to pronounce it

[foggy aspen]

awesome ... if we reduce number of "र" hopefully i can stop at the right level...ईश्वरररररररररर...my lyrics is very long with number of words like "ईश्वर" for each word I will have to include this entire block?

[tawdry scarab]

and now I've made a persona and am covering over and over with the persona

[tawdry scarab]

the hope is we can build up a strong and clean enough sample that it stands in the model as a hindi anchor

and then maybe that will overpower sanskrit

but who knows, sometimes on languages that aren't english it pulls hard at one pronunciation

like the whole middle east is egypt somehow

[tired forge]

OH

[foggy aspen]

i was looking for a library to see which words or characters from hindi are not pronounced correctly in Suno and what alternatives we have to fix the same...:-) I already have a big list....

[tawdry scarab]

Can I ask you, does this sound like an echo on the right thing, or like it just breaks down into the wrong thing:

https://suno.com/s/agE80xioibgpFp4o @foggy aspen

[foggy aspen]

starting characters "Īś" is not very clear... to me... end with "r" is good....

[tawdry scarab]

would it be clearer if, lets say, the sibilant was sharper, like it just cut clearer, or is it a fundamental mispronunciation. I'm chopping little pieces out of the track, if it's just muddy and muted I can eq it after the fact

[foggy aspen]

yes...it is not very clear to comment if the pronounciation is correct... would request to listen the audio file which i have shared above...

[tawdry scarab]

right, sorry

[tired forge]

I hope mine sounds fine.

[half dragon]

Music wise - I think it sounds really good.
Maybe it's just me - I hear a metal screeching sound in the high frequency - but maybe that can be cleaned up offliine.

[tired forge]

Huh… I don't really hear that myself. But I did try a weird combination. Stratocaster and dark guitar. So it's probably the dark guitar. The website suggested it and I just shrugged and said, "I don't know what that is, but WTF? Let's see what happens." LOL

[half dragon]

Yeah, it could be just my ear... and not sure I described it correctly either 🙂 maybe not screeching .. maybe ringing?... idk.. hard to describe ...lol

I think it's more in the hi-hat sound

not the guitar

[frozen dagger]

the metal screeching in the backgroun you describe makes me think of what i been hearing in the background of metal tracks. it sound like that "digitial glitter" that contaminated all tracks back in 4.0? 3? i think it was 3.0 version. its like its lower and hidden and tried to cover it up

[half dragon]

that exactly reminds me of that - when it first occurred in v4.0

[tired forge]

Fair enough. This one is my LONGEST ONE EVER!¡! Also my most unique idea for a blend of music. I thought I'd try something a little outside of the box. Stratocaster, orchestra, bass, heavy drums, wafū, lute

The website suggested wafū and I have no idea what that is. The rest was entirely my choosing.

https://suno.com/s/sCWzVqsBsegv5kA2

[half dragon]

Music wise, I like the other one beter 🙂 Still cool sound though.
Now, it started out pretty clean... then over 2 min something.. that artifacts creep in and get worse around 3:15. Killing my ears 🙂

[tired forge]

WTH does artifacts mean?

[tawdry scarab]

@foggy aspen https://suno.com/s/sMZuhoXMvyK7jg5f

this won't work as a prompt and style, this is an example I got of an elongated r that sounded right to me, but it can be covered or extended from or made into a persona

[half dragon]

That weird.. ringing, screeching metal sound in high frequency, this time I hear more in the vocal

[tired forge]

I'd send my top two favorites, but I've already posted those two on my YouTube channel, so IDK if that disqualifies them somehow.

[half dragon]

I wouldn't worry about it if you and everyone don't hear it 🙂

[tired forge]

I was going to say that I don't hear a single thing like that.

[frozen dagger]

words we use to describe that which has no name and not known/heard by all 🙂

[tired forge]

I just hear the vocals and drums predominantly

[frozen dagger]

usually (in my experience) the first minute or so the model is "figuring it out" after that it finds the sweet spot. then it gets lost again... sometimes.

[tired forge]

Okay. So I put a Stratocaster into some songs because I saw one pop up in an episode of Pawn Stars and I wanted to see how it sounded and I like the sound of them.

[frozen dagger]

i do that. search for gear or eqquipment musicians use that i like the sound and put them in the prompt. it works really well. use the A-I rewrite button to have Suno re-interpret the prompt full fo gear to use wrods that show styles of sounds it relates to those too.

[lethal nebula]

Hello everyone, I'm having trouble with Suno when it comes to covers. I put in the reference music but it doesn't create the music in the total time of the reference. What do I do?

[frozen dagger]

also... little tip for all... most know we cant use the names of artists in prompts. but.. you can type the artist name and then use that Suno AI button to have it change the artist name it cant use into a style that reflects what it is trained on for that artists style... doesnt work for all names of course.

[tired forge]

This one I said, "Screw it! I'm throwing caution to the wind. Let's see what the website gives me." The website chose the style entirely and I honestly couldn't be happier since I think it works so well. This was on Version 3.5 software.

https://suno.com/s/sj4ia29NKfEsRA3y

[frozen dagger]

covers are capped aroudn 5 mins