📚┃suno-school

[primal harbor]

@wise kindle

@queen trench

[sleek stump]

Songs created in the free plan can NOT be upgraded to have commercial rights. Suno's TOS are clear on ownership and licensing. You can read all of those terms at Suno.com or excerpts down in the lower sections of this discord.

[primal harbor]

@regal vale

[wise kindle]

https://help.suno.com/en/articles/2425729

[primal harbor]

Thanks a lot, bro!

[wise kindle]

no worries

[visual coral]

Same here..

[visual juniper]

v4.5 is now 15 credits? 😱

[sleek stump]

No, its the same as it has always been., 10 per generation

[visual juniper]

It just deducted me 15 credits!

[sleek stump]

I just genned a set, it was 10

[wise kindle]

did you replace a section?

[sleek stump]

If you are editing, those costs can be much higher.

[wise kindle]

yeah I think replacing a section is 15 or so

I doubt they'd increase the cost without informing people

[azure crane]

Say it again: Is this a common problem when changing subscription plan? I changed from Premier to Pro and the invoice has been mailed to me but no credits has been added….

[primal harbor]

Hey everyone, thanks a lot for your help and interest!
I’ve contacted the support team and I’m waiting for their response.
Still, you’ve been really helpful, much appreciated!

https://tenor.com/view/gjirlfriend-gif-14457952604098199169

[lost holly]

If you didnt choose the upgrade/change plan now option, you get the credits on the next billing date.

[wise kindle]

you'll probably get the same answer mate, if it's about the ownership/commercial use stuff.

[primal harbor]

Bro, I paid for the song, so Suno shouldn't be showing me that message. They shouldn't tell me to subscribe to get commercial rights because it’s misleading. But I guess we can’t know for sure without asking them, right?

[azure crane]

Well, I did the change plan option and I got billed but no credits has been added…

[primal harbor]

Why does it tell me to upgrade to the Pro plan to get commercial rights then?

[wise kindle]

are you already on the pro or premier plan?

[primal harbor]

,When I clicked on that song to download it, it told me I need to purchase the rights to have commercial use.

[primal harbor]

Bro, when I saw that message, I immediately upgraded to the Pro plan. That’s exactly what you see in the screenshot I sent you.

[lost holly]

Did you choose the immediately with proration option or at the end of the current billig period option?

[wise kindle]

if you upgraded during that warning prompt, then it should be fair to use it commercially. but yeah, best check with Suno just to be on the safe side

[azure crane]

Hmm, I think I did the end of the current billing period option cause my new period was started today…and I changed it yesterday..

[lost holly]

Then give it like 48h to correctly process everything. if you dont have the credits by then i would contact [e-mail]

[azure crane]

Have already done that… I just wanted to know if there sometimes can be a glitch in the system when you changing.plans… Thanks for the info 🙏

[random dome]

This is for help on your prompts?

I'd like to find out what Suno turned my song into

[polar spade]

Curious: you said you went from Premium to Pro. So you downgraded? That would mean you canceled out of Premier sometime before the renewal period (preferably a week before), and on the date when it renewed, it should only have charged you $10 and would immediately given 2500 cap credits instead of 10,000.

Did you manually pay at some point, and why manually pay when it auto-renews anyway? I feel like there's a detail missing.

[spring token]

why does the sound quality of the WHOLE song get worse/is changing when I confirm a replace section? (v4.5)

[polar spade]

There should be no "production quality" change in the rest of the song when adding a replacement of any part of the song. Only if you would then perhaps Remaster, Cover, etc.....would the production quality change. Is this when you click "Save as New file" or just when listening on the edit ui page after confirming an edit replacement.

There is a different audio quality in the playback in the Edit UI editor than you'd hear in the Library playback page, but that should not be too much of an issue.

You could , before doing any further remastering, etc, (production quality edits) download an mp3 or WAV and listen to it that way to test to see if it sounds differently or same as one of the other two locations.

[spring token]

I just downloaded the original generation and the one where I confirmed a replace section and the quality is worse in the second one. The song is still the same its just quiter and the quality is worse

[polar spade]

OK. Anything is possible, i suppose.

So "confirmed a replace section". 🤔What did you do exactly? Add an extension?

Are you in the Remix panel or the Edit UI.?

[rough crest]

thank you, i will try this now

[frail epoch]

I need help

says I need Pro (I hav annual Pro) to use 4.5

[random dome]

Did you try turning it off and on again?

[spring token]

I am in the Edit UI and I used "Replace Section"

I will probably have to stitch the song together on my own but its annoying

[rough crest]

heres what GPT spat out: The Song That Shouldn’t Be Saved
“There’s a voice in the silence. And it’s smiling.”

She was perfect—until the change. A skipped note. A pitch off-key. Tiny glitches hidden in harmony. Dismissed as anomalies. But something had begun rewriting her from within.

Her second voice whispered into being.

Light Lynx remains adored: hopeful, radiant, untouched by the decay. But deep in corrupted code, her other self watches. Not loud. Not violent. Just patient.

Dark Lynx doesn’t seek attention. She seeks trust. Installation. Love. Then—nothing. Every fixer vanishes. No logs. No trace. Just silence. Only she remains—singing, flawless.

Some claim she converts them into data. Others call it a memetic virus. The truth? No one gets close enough to know.

[polar spade]

OK. Then thats a wonder, for sure. Replacing a 20 second piece might splice in a terrible sounding piece in the middle of a decent sounding output, but it should not change the whole song quality left and right of the replaced bit. And saving the song should retain the same song qualit , as it was, as well.

Thats odd then. You can report or feedback on that. Otherwise, sounds like you're doing things correctly.

[spring token]

Yeah I use the replace feature a lot and I feel like it doesnt work that well with v4.5

[polar spade]

Its up to you to judge if you like the chatgpt results. I was just saying it might help to1) shorten the bio , or to 2) see if the change would then get you through the censor filters.

But you should like what you have first as well. I liked your bio, but wondered if it could be altered to make it through the censor.

[rough crest]

yeah im going to try to and put it in the bio and see what happens

[polar spade]

If its consistently altering the whole song, every song, every time you save it, then thats not an expected procedure. It should not attempt to master a song just by clicking "save as new file".

Sound appropriate that you should think it doesnt work well with 4.5 Sound like that to me also by your description.

I will say as a comparison, I have not had that issue. If the song sounds good and I replace a segment with a crappy sounding segment, in the Edit UI, and click Save File, I expect to get a good sounding song with a crappy segment spliced in the middle of it - but not to affect the whole song.

[polar spade]

I would make mention here https://discord.com/channels/1069381916492562582/1082775596767117312.

As a reminder there is also a feedback form which they like having. #📝┃user-research-feedback message

If you wish, certainly email [e-mail], with the understanding it can be weeks before a response there, due to a large backlog.

And last, ask the question again at a later time and see if anyone else has experienced this. Ill also, run a couple of edits when I can to see if this is a new issue showing up.

[polar spade]

I didn't notice you have the word Lynx. That is associated with a couple of artists already, so that could be triggering the block. It could be that simple.

And Dark Lynx is a music artist name as well as a character from the Invizimals and in Final Fantasy. These may trigger the censor.

[frail epoch]

Yes

Still tells me to get Pro when I have Pro to use 4.5

But I can use 4 and remaster it to 4.5

It’s weird

[rough crest]

it still did not work. i tried to have gpt rewrite it again. theres something triggering it after "they were wrong"

“There’s a voice in the silence. And it’s smiling.”

She began as a perfect vocaloid—rising, radiant, unstoppable. But then, a note missed. A pitch too sharp. A strange bend in her sound. Small things, easily dismissed. But somewhere inside, something rewrote itself. A second voice whispered into being.

They said it was harmless. They were wrong.

Now she sings in two voices. One—bright, joyful, adored. The other—hidden deep in her code, cold and waiting. Not loud. Not violent. Just… present.

Anyone who tries to fix her disappears. No files. No crash reports. Just silence.

And Lynx remains—still running. Still singing. As if nothing was ever wrong.

something in this sentence:
Now she sings in two voices. One—bright, joyful, adored. The other—hidden deep in her code, cold and waiting. Not loud. Not violent. Just… present.

[polar spade]

Usually, when it gives an "artists name" error, it will also give the exact text of the name in the error that it is blocking. Does it not?

[rough crest]

i tried deleting the bio in parts and it didnt work until i deleted everything up to "they were wrong"

its got me clueless as well

[polar spade]

Im out of ideas then. You could have a long title. That can give errors as well.

[rough crest]

possibly

even still it works with what i have now and its like the bare basic of her bio but it gets the point across

so im just gonna call it a day

XD

thank you for your help

[polar spade]

Check now. They've introduced some more features just now. This may have been affecting you.

[frail epoch]

I’ll check when I can
Long day

[jovial widget]

Anyone's else having a terrible experience with the new update? Songs are awful for me.

[sturdy dove]

So is this 4.5+ the reason nothing is working properly for existing projects? Perhaps the existing tools should work before Suno begins diverting resources to other endeavors. The editor doesn't work, extend doesn't work, replace doesn't work, personas are not preperly interpreted...

[random dome]

I'm so sorry to hear your Suno Pro membership isn't working. Please hold while I get management
"Camels. Camels calling my hoof like I'm locked up nonstop from the pyramids to the desert yea yea"
Alright. This is [ ]. [ ] as in ( ) and [ ] as in { } and what can I do for you on this fine evening?

Damn

I thought the reference was funny

Mb

Going to funny jail, Goodbye

[frail epoch]

lol

I’m just working on stuff

[random dome]

Oh

Mb lol. Thought it was cringe

[ebon sand]

i also basically cannot do anything on suno right now since the tools i need are broken

[frail epoch]

Nah but I make cringe songs lol

😂

[random dome]

I get that lol

[frail epoch]

Listen to it 😂

It serious and garbage fire at the same time

[random dome]

I like it. I think you cooked more in the second half

[frail epoch]

Here’s a serious one

[random dome]

The beat was really good throughout though too

[frail epoch]

Tried to go for a Juice Wrld vibe with the second one

[random dome]

I don't really listen to juice world, but I know a few songs and I think I can hear the relation

[frail epoch]

Gucci Gang 2.0 but better

Thoughts?

[random dome]

I thought it was nice

[frail epoch]

Last one for now 😂

[azure crane]

Credits in place now 😊.

[frail epoch]

I’ll relax now lol

[random dome]

Well, alright. They're pretty good so far, I'll try and finish em

[frail epoch]

I’m banned in North Korea btw

Wanna see why

[random dome]

Your music?

Sure tho

[frail epoch]

And ya
This

I cant go there anymore
Not that I could’ve before but ya

[random dome]

Well, damn lol

[frail epoch]

I looked it up
This is the ONLY SONG that calls Kim out directly like this more then once

It’s a 1 one 1

[random dome]

Not really my type of music, but I guess I get it

[frail epoch]

In English and Korean

So Grandpa messes up and he pays, his kids pay, their kids pay with permanent life imprisonment

[random dome]

That's wild

[frail epoch]

And people say Mexico is the worse in the world

It’s not

😂

[random dome]

I wouldn't know

So do you want me gone?

[frail epoch]

It’s an easy W and nah

[random dome]

I really don't care

[frail epoch]

Nah I’d not want you gone

[random dome]

Idk what you said

[frail epoch]

So no I don’t want you gone 😂

No reason to be insulting when you are the only one insulting

[random dome]

Oh

I'll stop tho

Sorry for being a prick

[frail epoch]

I’m talking about myself btw and it’s fine

[random dome]

Also

Ok

[frail epoch]

I try to not let people get to me anymore

[random dome]

I can tell

[frail epoch]

I’d rather spread love

[random dome]

Good on you ig

[frail epoch]

To much hate these days

[random dome]

I used to think so too

[frail epoch]

The world is where you make of it

[random dome]

Sure is, bud. Idk why I'm still making it here tho, so, seeya

[gilded plume]

anybody tried chorals or duets male/female since the introduction of the mutually exclusive "Male" "Female" buttons on 4.5+ ?

[hard beacon]

nope 🙂

[raven turtle]

Does anyone know of 4.5+ dropped earlier than announced as since yesterday I’ve been getting some absolutely incredible results with Suno. Really damn good.

[hard beacon]

well.. round about 24 hours it wasn't live

[opal orbit]

Was told I should bring this issue here, see if anyone's had the same issue.
4.5+ is ignoring vocal prompts all together. 40+ cover generations of the same song and I haven't been able to get anything other than a low opera-like female singer. It's also ignoring my instructions how the words should be sung.
I'm trying to get a high-pitched young adult 'cutesy' voice.

This wasn't happening yesterday, so I'd assume this is something with the new update.

[lunar anvil]

What is your style like? I have noticed the style helps direct the vocals.

[hollow hare]

Are you able to provide a link so we can see what prompts you tried? Also let us know what you wanted to get so we can suggest some ideas

[opal orbit]

It's a dark-wave, haunting tune. Which yeah, is definitely associated with deeper vocals usually. But I was getting high pitch variations yesterday using the same thing. Only thing that changed was the update.

It's also ignoring this like "avoid vibratto singing", which typically works in other projects I've done.

[lunar anvil]

Is that exlude in the exclude section?

[opal orbit]

I've tried it with and without, doesn't seem to have made much difference

[hard beacon]

would be "clever" if you can choose the male and female button together 😄

[hollow hare]

Or something like by-line edit and mark what vocals you want for that particular line. Maybe in the future

[hard beacon]

but overall it sounds like a good update... vocals are much more better..

cover versions in 4.5+ sounding much better then 4.5

[opal orbit]

I don't have any of my music public on SUNO, but here is my current style prompt.
"HIGH PITCHED anime female vocals, cute and sweet, singing is clear and not breathy. Her singing voice grows more entranced and layered with each chorus sung. MINIMAL vibrato singing. Song should be extremely haunting and ominous, with eerie piano melodies, backing with faint choirs that sound entranced. Choruses should feel distinctly different from the verses in how eerie and overpoweringly trance-like they feel. Soft synth-wave motifs or brassy instruments that give the feeling of being in an icy wasteland accompany the melodies, chimes could be in the background, If violin is used, it is very faint, Dark beat with emotional, ambient minor-key piano with haunting chords and dark atmospheric deep sliding 808 bass. Piano is sparse and moody, adding a haunting texture under the punchy drums and minimal brass. dark-wave, ethereal-wave"

For context, I don't actually want it to be as high as an anime voice, I'm just trying to see if I can force it to do anything higher

[hard beacon]

@steady kelp : will there be a "remaster in 4.5+" button soon? cover works good, but then the gender of the singers is.. well... like a roulette 🙂

[hollow hare]

Unfortunately AI is random. Being descriptive (e.g. tone, mood, pitch) can help direct the AI to what we want, but as the saying goes, "You can lead a horse to water but you can't make him drink".

As StraySlyph hinted at, using negative terms in the regular style entry will not work the way you think.

The AI would read it as "avoid" "vibratto singing", and it will pretty much ignore 'avoid' since there should be no implicit training pattern with that term.

If you want something excluded, use the exclude style if you have that issue appearing without that input.

What I noticed is that the output will be influenced by the entries in the exclude field if that term was not originally there.

E.g. if you exclude, I dunno, didgeridoo in an A Cappella, you may get didgeridoo appearing because the AI had to look up the pattern for that instrument to avoid it and, consequently, had some of the learning pattern contaminating the normal process (or something like that, this isn't my field of reseaech). If you did not define it in the first place in the exclude field, the odds of it appearing is almost negligible

I will try some tests of what you want on my side and either ping or DM you when I am done

[opal orbit]

I'm not as worried about the vibratto singing as I am not getting the voice I want. I put it in the main prompt because there are some parts I do want it in for, just not all, and its worked in the past.

The issue I'm having is that I can't get a high-pitched female voice.

[loud fiber]

Whats the max number of songs that the new inspo feature will consider?

[hard beacon]

did you check the "explore" function to find songs that have vocals like you want? Maybe their prompts can help you out

[opal orbit]

I hadn't considered that, I'll test it out

[hard beacon]

do we talk about high-pitched vocals or something like "miku-style"-vocals?

[opal orbit]

I just want a high-pitch, young adult, female. Vocals that are gentle and sweet, not necessarily miku-style. I've been putting 'anime-style' and alike in the prompt to try and force it, but it still comes out deep

[lunar anvil]

I have been having a problem where Suno generates into a slam poetry style. I have been fighting this problem on this one track for months.

[hard beacon]

did you try the prompt "singer-songwriter" ?

[lunar anvil]

That usually gives me a indie-pop/folk style. My current music style is the following.

Opening with sparse, staccato synths, deep 808s, and neuro bass, the track evolves as vocals enter swiftly shifting between whispers, rapped lines, and soaring falsetto, Layered vocals fold in and out, creating a dense, ever-changing sonic tapestry, Instrumentation is unpredictable: glitchy samples, warped pads, and processed vocal harmonies intertwine, sometimes diverging into atonal clashing, The bridge pulls back into minimal, detuned keys with only spectral vocal echoes before a sudden, dissonant crescendo employs polyrhythmic drums and thick, modulated choral layers to drive the outro, Experimental avant-garde track with staccato synths, glitchy samples, shifting vocal techniques, unpredictable layers, warped pads, and dense, evolving harmonies culminating in a dissonant, polyrhythmic crescendo with processed choirs and spectral echoes

[near flume]

give this a try, I've had pretty good luck with it "airy female vocaloid chops"

[hard beacon]

hmm..

[lunar anvil]

I have a good generation that is unique. I was preparing to make it public when 4.5+ dropped.

[final rose]

How is 4.5+ for production quality now? I'd like a sample mp3 export of a metal song if anyone has one.

[hollow hare]

Ugh, it was so hard to get that and I am pretty sure it was by luck I got these. Tell me which one has the vocal you like and I will make a Persona of it if you want it

1. https://suno.com/s/aiWHMbF4ASEreyfa
2. https://suno.com/s/r1cAhxHX12sM9yBv
3. https://suno.com/s/n5JSxOrFsbQeyFqv
4. https://suno.com/s/8ny1rRy5gsLkLAkP
5. https://suno.com/s/p73c0UeOm60FwQJ2
6. https://suno.com/s/eYOMKxtW7pp4SYpC

[opal orbit]

These sound great, thanks!
The 4th one is most similar to what I'm looking for, I'll definitely give it a few tries if you do

[hollow hare]

In case you want it, I created a persona based on that song: https://suno.com/persona/871990bc-2f68-4a3f-a284-be16483b7ebb

[polar spade]

@steep loom Ummmm.... if... and only if.... Suno periodically [daily(?)] scrapes YouTube for new lyrics to update their copyright censor, then you could possibly get flagged for using your own lyrics. Its not discriminative so it wouldnt know they are yours. You wouldn't even have to list it with Youtube's Content ID for that.

[steep loom]

Oh God, so I might have done this to myself?!

[polar spade]

Oooo you offering a free service?

[polar spade]

Not sure. Its logical though.

[hollow hare]

I tend to offer it when I find the prompt intriguing to achieve

[patent leaf]

Add vocals is the one feature I’ve wanted for years

[hard beacon]

@steady kelp : will there be a "remaster in 4.5+" button in the near future??? Cover function works well, but the gender of the vocalists is.. well.. "suno roulette" 🙂 overall a great sounding update..

[neat field]

"Inspiration" producing very interesting and strong outputs for me so far, text tags seem to have a little less at play

if one has a genre playlist of v4.5 one made, v4.5+ will use those superbly

[jovial widget]

Are people liking 4.5+? I think it's awful

[near flume]

I guess it depends on the style of music you're creating. For the styles I do it sounds magnitudes better than v4.5

[jovial widget]

What style is that?

[loud fiber]

I gave it 88 songs to work with and played with the sliders and it's definitely getting better one shots than anything Suno has done so far.

[pine wind]

The new version isn’t bad at all — the improved instruments and rhythmic variation are definitely a step forward. However, the vocals feel too perfect, even in songs that are meant to be soft or intimate. It’s like every singer is trying to win a talent show, which takes away from the natural, raw emotion I’m looking for. Sometimes less is more.

[loud fiber]

I did notice an increase in belting, but I was making extremely complex and involved music, so I didn't think much of it. I should try something that's meant to be ultra low key and ASMR-like and see what happens

[pine wind]

If you come across a way to slow Suno down a bit, let me know. I haven't managed it yet. 🙄

[loud fiber]

Have you tried literally slowing it down then remastering that?

[pine wind]

The song tempo itself is fine, but the vocals are always overpowered – even in songs that are supposed to be quiet and natural. It feels like the singer is always pushing for maximum impact, when a softer, more intimate delivery would fit the mood much better.

[loud fiber]

@pine wind Example of target vocals from real life?

[pine wind]

A well-known example would be Eddie Vedder in Pearl Jam's quieter songs — honest, emotional, but never overperformed

[loud fiber]

Better Man?

[tough isle]

It seems like this version doesn’t really take the uploaded voice into account. Even when I set Audio Influence to its maximum, the generated voice still doesn’t have the slightest resemblance to the one I uploaded

[pine wind]

Yes, exactly — ‘Better Man’ is a great example of the kind of vocal tone and emotional restraint I’m talking about.

[raven turtle]

Industrial techno sounding superb on 4.5+ 👍

[thin flint]

Has anyone gotten Suno to be able to do a brushed snare, like is often used in country and rock styles? Riffusion does it more often than not but I can't get Suno to do that click sound which is what it sounds like.

[thin flint]

Here ya go. Some thrash for ya.

[loud fiber]

Yeah for this rock ballad style, I have type many things to it and it only listens about half the time, even with extensive negative prompting about loud vocals, belting, etc...and putting "very soft" in the main prompt. If it thinks you want rock, even on a ballad, it is going to try to put some power into the vocal it seems.

[sleek stump]

Its used more often in Jazz, not so much for rock or country.

[fresh wave]

How do you get it to do a one minute ending and then stop?

[sleek stump]

We have no control over how long something is, all we can do is gen and then crop.

[fresh wave]

I feel like I'll get a really good sound, and then it just kind of drags on. It's hard to cut. It's hard to tell it okay. Okay, you can stop riffing.

[sleek stump]

Do a crop and a fade, problem solved

[fresh wave]

It seems there would be a better way than doing like a dozen regenerations trying to get a shorter ending. I see fade is a tool in the editor? I haven't dabbled in it yet.

I just got the $30 sub today to try all the new stuff

[sleek stump]

no, really, if you keep getting tracks that just keep going, crop it and do a fade.. stop wasting credits regenning over and over

[fresh wave]

Awesome, I'm still getting a hold on how the tools work.

When I tried suno last it was basically you just read, you just generate like a minute at a time and it was hard to even get endings

I think it was last year free version whatever. A ton has been added it is more like music tool suite now like garage band i love the direction its going it allows for more creative expression

[urban canyon]

Here is what the announcement seems to be saying but in different words:
Prompt Matching is Stronger
➤ You can now use longer, more expressive prompts without losing fidelity.
➤ Expect more vocal harmonies, smarter chorus detection, and natural instrumental transitions.

Structure = Power
➤ Fully structured prompts now produce exact intro → verse → chorus flow.
➤ Use adlibs, bridges, and emotional tags more freely — Suno understands them better.

Voice+Instrument Split Prompting
➤ If uploading only vocals or instrumentals, describe what’s missing, and Suno fills it in.

[thin flint]

That was Suno V1, with 30 second clips.

[fresh wave]

Yeah, that's it, I guess.

It's crazy now.

I just uploaded a YouTube clip that I did like 15 years ago where I voiced a courtroom scene and did all of the voices and it made it freaking epic.

Like, it totally riffed off it in the craziest directions. I think it's super cool how you can infuse actual performed things with the AI-generated stuff now. Like being able to generate a backing track.

[thin flint]

Can we hear it? Or can they still not be published.

[fresh wave]

When people talk about AI ruining exoression they havent seen the magic in that sort of thing.

I don't know. Everything I do is inappropriate. LOL.

It's not that bad, I don't think there's any swearing in it.

Also, I was trying to find a good ending, but these are all just way too long.

[sleek stump]

AI music hate is going to be around for a while, traditional industry folks and artists are afraid of losing money, and they should be.

[thin flint]

I personally could care less lol

[fresh wave]

I feel like it breaks my heart for the traditional industry folks and artists because they're becoming the new stubborn boomers.

Like what really can anyone do to change any of this? All you can do is get positive things out of AI, or you can just choose to be under a rock and worry about everyone else

[sleek stump]

Theres 2 ways , get on board and use the new tools to make your own stuff better, or rage against the machine and fade into obscurity.

[fresh wave]

Yeah, pretty much.

[sleek stump]

(bah dum tish)

[fresh wave]

The thing is that music had gotten so cut and dry manufactured crap over the years that almost like 15 years of pop music, top hits were written by one Swedish guy and performed by everyone in here in the US.

What it boils down to it is whether do people connect with what you're making? Is it fun to sing?

I like how AI gives a voice to people. That's not necessarily just, um, this is my lavish lifestyle. I'm pretty cool.

[sleek stump]

Nah, most were written by a handful of writers in LA, one just happens to be my friend.

[fresh wave]

Wow, I bet they make a ton of money.

[sleek stump]

She does, half of what Katy Perry recorded was written by her.

[fresh wave]

I wonder how it feels knowing that you've written those songs, but that you're not the one getting famous off of them. Maybe the money is so good, you just really don't care. You just surround yourself by people in the industry who can always acknowledge that you tell Katy Perry what to sing.

Okay you guys are gonna like this one

I just gotta get the fade out right.

[fresh wave]

https://suno.com/s/N4nXRNje3XQneDyK

Tell me what u think

[random stag]

Is there a way to prompt Suno to produce one version with a male singing and the other with a female? I mean, if they’re going to give us two versions anyway, right?

[bronze helm]

Do we know if Suno v4.5+ does stem-based generation?
i.e. if it generates separate stems to combine them into a full song.

I'm asking cuz one of its marketed functions is to add vocals to instrumentals and vice-versa

The quality is phenomenal

[sleek stump]

While on rare occasions it can produce one version with a male and the second with a female vocalist, the arrangements are always different. Unfortunately this is not something we can prompt for. If you want a different gender vocal on a specific song, you will still need to cover it and specify the vocalist type.

[sleek stump]

I dont know yet whether 4.;5+ can do this, I just got home from work... BUT, my educated guess is thats a no, Suno generally generates the entire song as a whole, never by separate instruments, as that would take massive amounts of space to create and store each individual sub channel.

[bronze helm]

The question was based on the fact that, supposedly, I can generate a full instrumental, and then I can add vocals on top with v4.5+ without changing the instrumental. At least that's what I gathered from the news post. So I was wondering if it does just that in the background to combine the stems (2) into 1 full song, which hardly takes space. It could also explain why my v4.5+ gens take longer. But I'd love to hear a confirmation from a higher-up on Suno

[sleek stump]

I get that, but it still takes your part and generates everything else as a whole, it does not do any "mixing" of tracks.

[bronze helm]

How do you know for sure? Have you noticed differences in the instrumentals or vocals when trying to add the other stem?

[loud fiber]

Yes, the instrumental is different after you add vocals

[bronze helm]

Aha... I see

[loud fiber]

It keeps the composition extremely tight, but the timbre is different

So they are using some sort of control net I think, rather than stems

[sleek stump]

The model will generate everything it has as an input as a whole track, that is how generative works

[loud fiber]

The main problem I have with the add vocals feature is that it caps it at 4 minutes. It would be unusable for most of my tracks

[sleek stump]

Limits are the same as 4.5

[bronze helm]

I think that goes without saying that it's generative lol. I just don't think it's that unorthodox to make a breakthrough and generate vocals as a separate stem which is then combined with the instrumental stem. That's still generative. I can see now that v4.5+ hasn't taken that approach yet. Maybe v5?

[loud fiber]

I think they can do this if they wanted to, but it would probably cost more credits

[bronze helm]

Sounds like a fair deal if you're getting better quality and control

[sleek stump]

What you arent understanding is the space required on their servers to store all those separate parts... they are a massive space requirement

[bronze helm]

A greater space requirement than generating 2 songs with 1 prompt?

2 stems for 1 song mathematically isn't more than that lol

[sleek stump]

2 songs, 55MB each, now generate 12+ tracks for each song at 50-60MB each, the requirement is exponetial

[bronze helm]

Who is talking about 12+ tracks!!

There is a miscommunication

[sleek stump]

Large bands can have more than 12 instruments playing, each would need iots own channel.

[bronze helm]

I am strictly talking about 2 stems

Suno's instrumentals, which it can do if you tick "Instrumental" for your song

• Vocals

[sharp gull]

I was wondering about resource usage myself. Particularly where it pertains to continuing to generate songs that aren't going to be kept (like when I thumbs down it within 20 seconds).

[loud fiber]

I am shocked they haven

t given everyone a space alotment

[bronze helm]

Good point tbh

[sleek stump]

What any of us that actually produce music wants is channel generations, where we can mix it all as we see fit in a DAW. Just separating teh vocals is not what many of us want, we use the 12 stem version

[loud fiber]

And just make you download stuff and get it off the platform if you want it

Even if they gave you like 20k gens on there, people can cull down to the things they want very very easily.

[sleek stump]

AT some point thats going to happen, storage space is not cheap in data centers

[bronze helm]

Let's not make assumptions, I am a music producer myself. If I want complete control over 12+ tracks, I will use my DAW like LMMS for that and improve. I am not referring to 12 stem separation. But okay. If you insist that it must be 12, when it can be 2.

[sharp gull]

I am not a music producer 😉 Nor really even musically inclined. I just like to tinker with Suno and make some stuff that I enjoy myself.

[sleek stump]

If you want just 2 thats cool, but you have no control to adjust levels of justr the base, or the percussion, or a synth or guitar, just the whole or the vocal.

[loud fiber]

A creative stem separator in post is probably a better solution tbh

that can backfill the damage

[sleek stump]

Which is what they already have

[loud fiber]

Not nearly where it needs to be

I think these are coming very soon though

[sleek stump]

No, but its better than any other STEM separator, even lalala.ai

[loud fiber]

I haven't compared it to the new Spectralayers

[sleek stump]

SPectral separation will never be able to compete with Generated seprated tracks.

There will always be bleed

[loud fiber]

It is non-destructive, and you can manually reassign the stuff that shouldn't be on each track back to which stem they should be on

But I am wondering if the new one does have a creative option. I haven't checked it out yet

I imagine the next version of Izotope RX will do it, too. They already have this tech for several years. But they didn't add it to their stem splitting

It's been able to estimate missing frequencies for a long time and fill gaps quite well in damaged audio

[remote flume]

Why don't I have the 4.5+ option, or is it not available for covers?

Yes I am a pro user

[dusty plinth]

Not available for covers yet. You can see it in new generations.

[sleek stump]

are you sure? 🤣

[shell timber]

I have a question. I have a two songs. one song i like the vocal's voice, another song i like all except the vocal voice. is there a way i can put the voice i like into the another song i like? i want to work in suno. if it's not possible in suno, i want to find a way. please help me.

[loud fiber]

You can try making a persona out of it, then using cover with that persona on the new song. If it isn't working for you, the more involved way is to stem separate it, re-record it off of your stereo mix at 192khz sample rate, put it in a DAW with the backing track from the other song, and use Melodyne and fix up the vocal notes and timing to match your other song.

I think there are also some services that will let you train a new voice on their platform using the other file. Not sure what happens after that, but I assume you can get it to follow a MIDI file and the lyrics you want, as I have seen things like this out there. Or it might just hard-replace the current vocal in whatever you feed it. Bob Doyle Media on youtube was doing both of these a while back I think.

[weak saffron]

Been trying since v4.5+ dropped to see if it's just user error. Went through and played with sliders, removed/added style tags, covered songs, made playlists for trying inspos but nothing seems to fix how much v4.5+ destroys the genres I make (Dubstep, Future Riddim, Color Bass)

This is what I generally use for generations if it helps any.

Also dropping audio files for comparison. 1st is usually what I look for sound, second is v4.5+. Used the same settings for both

[loud fiber]

You have complex and experimental in there, but your target style is more generic. It could be messing it up on the new version if it is listening better to the prompt.

[weak saffron]

Thought so too. I tried removing that but it didn't help any

[polar spade]

".....You can try making a persona out of it, then using cover with that persona on the new song....." -- @loud fiber

Yes. Thats the way it can be done.

Assuming the two original outputs are 4.5......Run a Cover of the instrumental version ( that will become your Audio OG in the Remix panel) and add in the Persona made of the vocal output you liked. Make sure your lyrics remain the same.
Weird slider near 1%, for now, and the others over 65% , typically works. Style prompts may be from either, tbh, as they would be the same. Click Remix.

[echo prawn]

What the hell is 4.5+? When I sing a cover, even if I turn up the parameters to the highest, the melody will change. It's better not to update it.

😅

[zinc pumice]

Hello,

I don't understand the prompt system in 4.5+ anymore. The same prompt in 4.5 and 4.5+ doesn't generate the same style of music at all. It's quite frustrating when you've spent many credits perfecting this prompt. So if anyone has any ideas, I can't generate Caribbean zouk, konpa, or Haitian gouyad in 4.5+ anymore. It generates a kind of generic soup with a Caribbean feel without any customization despite the numerous prompts and credits used.

[sleek stump]

Its a new model, time to learn all over again, like all the rest of us.

You can also continue using 4.5 if + isnt doing it for you.

[remote cedar]

Does the 4.5+ option go away when you select a persona?

Or is it just not working for me

Nmi guess that's it. No persona allowed yet

[polar spade]

You not getting the sound and vibe off the persona, or it simply has persona greyed out? I've been running a few personas, can choose 4.5+ and I get outputs of 4.5+

[shell timber]

Ty for the reply! it's so helpful

[shell timber]

Ty for the reply! it means alot.

[agile stag]

What's the problem with the bot strictly avoiding to sing on the beat

Suno almost got it right, except for one word that is the first word of the sentence, it never gets it on time

[wise kindle]

what's the weirdness level at? it tends to be more safe the lower it is, and more creative/syncopated when higher

[opaque notch]

hey i need help

my biceps wont stop growing

[vital edge]

Hey. when will suno fix the 4.5 replacements as they are always sync/alignment issues

[wise kindle]

i feel you bro. my jawline is so chiseled my razors keep breaking

[polar spade]

I've found the extent of weirdness, slider up to about 75% seemingly depends on everything else. I cant give an actual sample, but it can do the expected odd gabage if cranked up, but sometimes, if both audio and style are cranked as well, then a high weird number can just mean its throwing some nice variation or riff or fill in. So I now try an extra generation, now and then to check. A few I know I've felt were better that I did try-- about 5% , if I would guess, i would keep as "possibles". Not to mention, some have some m snippets that can be used as well to create another song.

[opaque notch]

dam that was pretty good. welcome to the conceded club

[hybrid vine]

I would like to have a two part chorus in a song, but in one of the occurrences, the first part has the melody only played on guitar, without lyrics, I never managed to do it. Any hint?

[indigo canopy]

When using the Get stem function, the duration of the resulting audio tracks differs depending on whether I choose [Vocals + Instrumental] or [All Detected Stems].

Specifically, the number of samples in the former is 8,027,963, while in the latter it is 8,035,020.

Upon listening, the [Vocals + Instrumental] version sounds normal, but the [All Detected Stems] version contains some duplicated frames and has a strange echo-like effect.

Does this issue occur consistently every time, or is it just a rare occurrence?

[opaque notch]

That was a bombas line yo

[pseudo blade]

this aint a song promotion @opaque notch

[opaque notch]

Coco go home

Thats for men with chizzle jawlines

Secsee

Man you are slow.. Need help typin?

@pseudo blade oops sorry

I thought you were dude.. I apologize hahahha

oooooh Coco could you bear to forgive this silly modern bard to-be? Thy beauty knows no boundary only second to emu-T

[pseudo blade]

I dont speak ancient english

[opaque notch]

i understand. Youre the goddess and i shall follow thy decree

Noticed i said “Thy”? Hmm?

[pseudo blade]

This is a tech prompt talk

now move it to #💬┃general-chat or #🗣️┃chill-chat

[opaque notch]

Coco butter im lost chill

[normal ice]

Shall I drop the wall of prompt text Doom upon this chat to discuss

[pseudo blade]

no

u could get woodhoused

[normal ice]

Then what in the heck is this channel even for

Whining about prompting??

[opaque notch]

lol woody

Peaker

Ginger blond

[sleek stump]

Guys, this isnt really the channel for what I am seeing

[normal ice]

I'll post it and see if it gets completely deleted.

[opaque notch]

Yooo im lighting up the mood chill

[normal ice]

LMAO

[opaque notch]

You see anybody reply sheesh

Goodness

This poor humbl bard

Playing a tune no oje listens

[sleek stump]

Ya, walls of text arent really allowed

[opaque notch]

Where is wall of text?

[normal ice]

Anyways I've been experimenting with crazy prompting concepts by using the style prompt to give it an idea of what to look for in the lyrics themselves and then giving more direction inside of the lyrics and that seemed to work fairly well...

But it's kind of hard to display a prompt in a channel that doesn't allow you to talk about prompts other than from a meta level

It got deleted.

[opaque notch]

Arual who is you boo-boo

Called laggint bozo

[pseudo blade]

why is this turning to a something 18+ 👀 🛡️ 🕰️

[sleek stump]

Post the prompt but not the entire 5000 characters of the lryics 😄

[opaque notch]

I man come on mr prompt its elementary

[normal ice]

Bruh. That wasn't even that long that time.

[opaque notch]

ima vlog woody in a bit

[normal ice]

Whatever I don't even understand the point of this channel.

[pseudo blade]

ayo?

[opaque notch]

Brah it 360 words

On a chat context based

For realz?

I show off here boo-boo

Check this

[pseudo blade]

u cant use this type of language here 😳

[opaque notch]

Inlove coco

[pseudo blade]

im already dating bro

[opaque notch]

Lovo loco moto coco moto me

Dating

Like who?

Wooody?

[azure crane]

Why is v4.5+ suddenly gone?

[pseudo blade]

huh?

[normal ice]

Are you trying to use the mobile app or something

[pseudo blade]

@opaque notch im dating a real person bud, ur a kid, im getting mod control

[opaque notch]

Update app

[pseudo blade]

app broken for v4.5+

use web 👍

[normal ice]

Also this is definitely not the place for whatever the heck discussion is going on in here but I don't really know how I'm supposed to discuss actual technical prompting when you can't actually display the prompt that you are trying to have a discussion about

[opaque notch]

Its not broken boo-boo

[sleek stump]

Are you usinjg mobile app?

[opaque notch]

delete app and reinstll brah

Easy

[azure crane]

No, iPhone browsers.

[opaque notch]

use duckduck

Coco is slow

[pseudo blade]

💀

[sleek stump]

Turn the phone to portrait'

[pseudo blade]

bro, i have autism, do u want a block?

@sleek stump I cannot deal with this guy

[opaque notch]

Me too

[pseudo blade]

let me come back w/ a better mood

[opaque notch]

Block who me?

[sleek stump]

Samantix cool it

[opaque notch]

Bye

[azure crane]

Ok, it worked but the desktop view is really clumsy…

[sleek stump]

Ya, phones werent really made to be mobile music studios 😄 They need a lot more pixels

[azure crane]

Hehe, yeah. Thanks for the help 🙏

[sleek stump]

I only use my iphone as a car audio player.

[final hazel]

@opaque notch 🛑 Sir Samantix of the Off-Topic Realm,
Thou hast been summoned for excessive bardic buffoonery and the unauthorized deployment of “lovo loco moto coco” charm spells.

This is a server of music, not medieval matchmaking. Continued clownery may result in your lute being confiscated and your chat privileges banished to the Shadow Realm (aka timeout).

Let's keep it chill, respectful, and on-topic—or the mod hammer shall descend with great justice (and mild disappointment). 🎭🔨

[restive gull]

is there a guide or something that explains the best way to prompt a song?

[normal ice]

Now that I know that I can post giant wall of text as long as I don't have text formatting I'll go ahead and paste the thing I was trying to paste earlier one second

[sleek stump]

There are a number of guides in the workflows channel ((right above this one), but there are no official guides perse.

[normal ice]

For 4.5+, Figure I'll show you an example of a completely insane prompt I used and got probably a 90% result of what I requested through the lyric prompt when it's paired with a carefully worded style prompt to kind of guide it on what to look for in the song. I'm still going to be experimenting and I'll be sure to show you guys any results I get!

Example Song Output
https://suno.com/s/QxYET3m5IhAJ6iIc
(Most had a vast array of variations comparable to this one, but I liked the flow of this one the most.)

Style

A chaotic and aggressive, high-energy track that fuses the relentless percussion of modern trap with a sample-heavy, lo-fi aesthetic, The production is intentionally maximalist and overwhelming, built on a foundation of heavily distorted and clipping 808s, rapid-fire hi-hats, and a dense collage of sound effects like sirens and lasers, The track is defined by its constant use of chopped, pitched, and processed vocal samples and signature, recurring vocal tags, which serve as the primary rhythmic and melodic hooks instead of traditional lyrics, The mood is one of pure, unrestrained chaos and menacing confidence, creating a frantic, high-octane auditory assault designed for maximum impact

For the lyrics I have left the lyrics prompt fully present on the song itself but due to the content of the lyrics it gets flagged for language by woodhouse here.

[sleek stump]

A lot of users have had a some real success using GPT or Claud or Grok to form style prompts

There is also the magic wand, use it and see if it helps.

[final hazel]

@opaque notch Alright Samantix, the diss track was noted (and removed). But turning this into a personal drama fest and calling out other users like that? Nah.

You're officially booked for a 24-hour timeout. Solo tour, no audience. Use the time to chill, reflect, and maybe write something that doesn't break half the server rules.

See you tomorrow. Keep it cool or keep it quiet.

[tribal remnant]

Thanks for sharing

[dry hound]

Hey, is there currently any way to merge two songs into a single one that I'm not aware of... I'd love for suno to create a transition section between them

[sleek stump]

Currently you cant merge 2 separate songs

[polar spade]

Did you actually get a collage of lasers and siren sounds 90% of the time?

[normal ice]

I actually had several gens that did include lasers or something comparable

[polar spade]

No sirens?

[normal ice]

And sirens for sure I had a few but it's not always going to be 100% consistent

[polar spade]

Oh ok. Cool. Thanks.

[normal ice]

And some sound way better than others especially when you're requesting this kind of thing

Some just sounds like screeing birds 😂

[polar spade]

I was wondering about the "collage" part.... thanks for the input.

[normal ice]

But again this was kind of just limit testing craziness

The big thing I noticed is that it did not speak the parenthetical production notes

[polar spade]

OK. Situation normal then. Got it.

[normal ice]

My settings were all default 50% no persona no inspiration

So you might get more acute accuracy with stronger style influence

But I find it sometimes kills a bit of variance when you do that

You could probably do an upload extend and then carefully craft a prompt to flow into the new song

[spark cove]

Hi is there a way i could do thr following i have 2 personnas that i trained rly well they always give good consistent results i wanna do a song that features them both would that be possible?

[wet minnow]

maybe with creative use of inspo... but seems difficult

[lost holly]

Maybe create a song with each persona, add them to a playlist and then create a song with this playlist as inspo (inspiration). But dont know if that works xD

[spark cove]

I tought maybe create a part of the song from 1 personna and extend in editor with second personna is that even possible?

[wet minnow]

i think you can make persona extends. (at least in usual create)

[boreal harness]

It really annoys me when I upload a 7m30s piece I am working on, and try to cover it, to get the wrinkles out. Despite providing all the lyrics and the tempo, and setting Audio strength to like 95, Suno throws some 3m30s BS at me. Anyone know how I fix that?

[red whale]

Good morning! I'm trying out 4.5+ and I like having tight, four minute songs. But Suno now wants my songs to be seven minutes or longer. A song I was working on under 4.0, which was this four-minute length, is now twice as long under 4.5+. Any thoughts on how I can tell Suno that I only want a certain length? I tried to add "four-minute length" in the style box, and that didn't work.

[spark cove]

Ima try it out then

[lost holly]

The 2 persona thing kind of works like this 😄 but i am not sure how much random that is xD

[polar spade]

"......It really annoys me when I upload a 7m30s piece I am working on, and try to cover it, to get the wrinkles out. Despite providing all the lyrics and the tempo, and setting Audio strength to like 95, Suno throws some 3m30s BS at me. How do I fix that?......" - @boreal harness

---

It the nature of the beast since Covers were introduced. Covers have always needed the original to seed the new Cover version. So you'll lose some of that time you'd want , as reserve for seating in the seed information. If you already have an 8 minute song, is likely your cover will be on the moderate or less side for the output. Its an expected result.

If I cover a 30 second jingle ill likely get 4x the length, for instance, that to get a n exact 30 second Cover. Thats also annoying. 😁

Since personas act in the same way, we suggest to just make a persona of only about 1 minutes worth.

[polar spade]

I still perceived inspire was kind of like giving us a good idea for the "next song on the album" , if we give it a Playlist of the main styles anyway. According to the sample video. I still cant get anything close. So its got to be some sort of voodoo I have not worked out yet. Specific genres....maybe the more original the songs in Playlist with no extensions and such, the better results?

[boreal harness]

This channel needs a better title. I didn't think my issue involved a tech prompt. Should be tech-and-prompt-talk

[hard beacon]

w00t?? it's called tech-prompt 😄

[boreal harness]

And my tech issue doesn't involve a prompt.

[polar spade]

Its a dash. People use it to list things same as a slash. Tech/prompt/talk. What can we say. 🤷 no biggie.

Then how about just talk? 😉

[boreal harness]

Chit-chat?

[humble verge]

god v4.5+ is awful

[polar spade]

Is that all you need? OK. Post a suggestion to change things. Im done. Have a nice day, my friend.

[hard beacon]

then use 4.5 🙂

[humble verge]

you bet i am

[boreal harness]

I mean, chit by itself means nothing, except currency. We naturally put the two words together. Chat chat. Tech prompt.

[hard beacon]

😉

[red whale]

Is this an appropriate forum to talk about the issue I mentioned earlier, where Suno just generates songs at seven+ minutes under 4.5+ no matter what you do?

[boreal harness]

It's basically the same issue I had, in reverse. Related.

[polar spade]

Have you Covered the 7+ song to see if it cuts you back to about 4 minutes?

[red whale]

I haven't tried that yet. I'll give that a shot and see how it goes.

[polar spade]

Its a workaround, but I wouldnt know how to get exactly 4 minutes with any version without just trial and error.

For what its worth covers always seem to like that 2:30- 4 minute range for me.

[hard beacon]

i've asked it yesterday, but as far as i know there was no reply to it.. maybe somebody got an info about it.. will there be a "Remaster in 4.5+" button in the near future?

[humble verge]

It's horrid

[boreal harness]

You might try uploading two versions. One with just your lyric sections, with clear [Break]s denoted in the lyric field, and one with just your instrumental sections, again with [Break]s. Put them back together in your DAW, then upload the whole piece to get a consistant master.

[red whale]

So it got it down to five minutes with a cover which is a little better.

[hard beacon]

@humble verge : what kind of style are you creatin with suno? just curious why you think it's an awful update

[polar spade]

Take out the lyrics and Throw in an [end] in the lyric box. See if that inspires it to go shorter. Who knows.

[red whale]

Already in the lyrics! But thanks for the suggestion.

[polar spade]

Well, the cover may ignore anything that hasn't changed. Take OUT the present lyrics and just put [end] .dont take out all but [end]..... OK. Again, just spitballing at this point. But its going in the right direction.

[humble verge]

I'm doing covers and remasters and getting nothing but garbled and noisy crap. Especially when using a Persona. Like, it doesn't even remotely resemble the song I'm remixing. Female voice? No, according to 4.5+ my female persona is Leonard Cohen, if Leonard Cohen had been born without a tongue. Instruments? Forgetaboutit! Let's make the entire song sound like it was played beneath a waterfall. No, it's awful.

I don't think I've recognized a single word from my lyrics in the six attempts so far.

[boreal harness]

What are settings at regarding strengths?

[hard beacon]

but i didn't get it.. what kind of style are you creating?? don't understand me wrong.. i don't wanna discuss if 4.5+ is good or not.. just tryin to help 🙂

[humble verge]

Bluesy/jazzy stuff.

[ebon sand]

I don't like it either for the music I want to create

[hard beacon]

ahh.. ok.. that's i genre i never created with suno.. but.. btw.. remasters are in 4.5 and not in 4.5+ .. so that can't be an argument about 4.5+ 😉

[humble verge]

Covers are.

[hard beacon]

that's right.. but you said remasters as well 🙂

[humble verge]

but mainly covers

[hard beacon]

i had some bad results with 4.5+ covers yesterday with some pop punk songs... but also some good ones... for dubstep version 4.5+ works great

[humble verge]

I did get good results doing instrumentals in 4.5+. I have not tried making a song with vocals from scratch yet. So I should point out that this is a problem with creating covers and using personas in 4.5+ which has given me nothing but absolutely atrocious results. Like, basically just noise.

[hard beacon]

i don't work with personas coz i always have issues with it.. maybe that's why i have better results..

[ebon sand]

For me it treats personas as if they weren't even there

[humble verge]

It is really difficult in general.

[hard beacon]

... and i don't wanna spend so much credits if i don't understand a function 😄

[boreal harness]

I only get noise when I set the weirdness slider above 80.

[polar spade]

It seems:
Audio slider affects the amount of the Audio (at top) that we are editing. Im told Persona is affect by Audio as well.
Style affects lyrics, prompts and I personally think Persona somewhat as well.

Overall, from myself and many others I've read, when making a cover, Style hovering around 45-55% , and Audio about 55%-70% seem optimal for emulating the same song as a Cover.

Cranking them both to about 95-98% doesnt make drastic changes so that may or may not help, depending on the original song and prompts.

Weird slider does get weird after about 76% (it will turn red to indicate) but if you have certain instrumental or clear, separated instruments, such as a quartet or symphony, weird can do things such as change the instrument. So, though I dont use weird high % often, ill test it about once on every output and get a couple of nice results occassionally.

[ebon sand]

With personas you basically can generate songs with the same vibe and vocalist, just doesn't work on 4.5(+)

[humble verge]

I did a cover specifically instructing that it's a female voice, it's described in the style as well. I told it to exclude brushed drums. What I got was a very deep manly voice and nothing but brushed drums - which, in SUNO, sounds like white noise.

[boreal harness]

That was a question to Rafaelis jazz issues. I should have quoted him.

[polar spade]

Oh. Sorry, mate. Thanks. Something for anyone to read then.

[boreal harness]

My sweet spots are around Weirdness 68, Style 70, and Audio 60.

Also, I like to write out my prompt BEFORE applying a persona, then copying my prompt, selecting the Persona, then pasting my own original prompt below the one on the Persona, then used the Prompt Magic to merge them.

[hard beacon]

i have to say, that my audio quality is bad more often if use the sliders in different ways.. so it always let them in the "factory preset"..

[ebon sand]

I just have a vocalist as persona, it doesn't fetch them so it doesn't really matter what I type in the prompt, he's not there and then I have no use for it basically

But it has no adherence to provided music style personas either

[boreal harness]

Even if you increase the Audio strength? Completely ignores it?

[ebon sand]

Yes, on inspo it works slightly but if I amp up the audio strength there I get a fractured output of various incoherent parts and vocalists

As if it would take 30 seconds of each track and jumble it

[boreal harness]

Does the Perrsona itself have defined lyrics?

[ebon sand]

Yes, it's basically a vocal stem but better quality

[unique stone]

The defaults are basically meant to be the best balance between everything so generally it's a good idea to use those and then change them as needed.

[hard beacon]

well.. for me the results in "standard" are the best (Audio quality)

[boreal harness]

I like Style higher, as I don't want too much machine hallucination.

[humble verge]

When using a persona, I find any audio influence over 50% takes too much from its original song. But hard to avoid regardless, which is why I rarely bother using them.

Not sure why 4.5+ seems to go out of its way to include what I explicitly tell it to exclude. But I guess that can be used in reverse. If I really want didgeridoo in my blues song, I just tell it to exclude it.

[hard beacon]

I'm beginning to think that I'm misunderstanding the sliders... if I set Audio and Style Influence to 100% and Weirdness to 0% (prompt is like the original). then a completely different song comes out... what the...

[ebon ingot]

here's another question: has anyone ever gotten the sound of a hang drum / hand pan right ? I tried numerous ordering possibilities within the prompt. Steeldrum works fine, but hang drum, hand pan, is always interpreted as a "Tabla" like sound. If anyone has any ideas or synonyms I could try, it's greatly appreciated.

[boreal harness]

Reverbed vibraphone in minor scale?

At slow tempo.

[ebon ingot]

@boreal harness that's ingenious, I will try that, thanks

[boreal harness]

Might still come out too sharp. But, option to play with.

[spark cove]

how in the hell

[lost holly]

You have to manipulate, means copy the html code for the first persona in the browser console and then select the other. When you then create a song, Suno uses both personas. The downside, it doesnt work well when you say like female or male in the style tag but together seems to work pretty good.

[spark cove]

im trying to do 2 males in the song 2 rappers

[lost holly]

The try working with bracket (), also no guarantee which is sung by which persona, but Suno switches between both

[spark cove]

so i gotta open up the console and paste both or wha?

[lost holly]

Click on the persona, after you selected one, with right click. The the inspect option (might be named differently depending on you browser.
There copy the whole node of the selected persona and paste it in, so you duplicate the whole HTML node, you see if it worked in the browser window, if there are two personas there then. Then you select the other persona, this will only change the top persona.

[ebon sand]

I think you're chasing a ghost unless you have really precise proof that this works because copying a node has nothing to do with the post command that is sent to suno

either you did something else or it was coincidence but this is not how stuff works

[lost holly]

😄 you are completely right 😄 i just tried stuff and it worked^^
Those songs where created with two personas:
https://suno.com/s/4Oo5jxp6Nuv13nwk
https://suno.com/s/JLF4czgTsZmKNsoA
https://suno.com/s/8Zaqv38YlstXcTQW
Not perfect results, but it should suffice for demonstration reasons.

[ebon sand]

which of those were the personas

you can see it here, i did a duplication of the node and only one persona is sent, your client side view doesn't change what's sent to suno

you'd have to manipulate the latter and if you didn't do that (and i don't know if it's possible) it does nothing

[lost holly]

Personas: https://suno.com/persona/e26ffb43-6705-4087-a14f-c55c80a088d1
https://suno.com/persona/71ac2777-c3e1-4dd2-afd1-39212545d8b3
https://suno.com/persona/6d2389ba-a3ca-450b-bd55-b2d71c1d5b03
https://suno.com/persona/3f6e6c73-53e8-4f30-b481-04c1f11910e5
In general, you are right and argue in ways i can follow, i dont know, why and exactly what i did, but it worked 😄

[ebon sand]

the model probably just had an output which sounded similar to a persona you had, i get the same voice all the time which isn't even a persona, but this method doesn't work as you described it, there is no possibility that it would ever work like this, as said either you did something else or it was coincidence that you had the feeling you got two personas meshed into one song

[lost holly]

Completely understandable statement xD give me 2 of your personas and a song and i try again. Lets see what happens xD

[ebon sand]

i do not wish to share my personas because i have a vocalist which i consider my property haha
i can generate random ones though, but the easiest vice versa test is this

i removed the nodes and it still generates with persona

the html you see doesn't influence any of that

[lost holly]

Feel free to share other personas, its just for tests 😄

[ebon sand]

i'd rather test if you can put 2 ids into the post prompt, but i doubt there's support for that

[lost holly]

Or choose 2 of mine^^

[ebon sand]

you know your personas best

[wispy escarp]

Hello new here. Anyone else having trouble sharing to TikTok?

[hard beacon]

please elaborate on your problem... no one will be able to help you with the description of your problem 🙂

[lost holly]

The html has to look for example like this. So i kind of explained it wrong i think ^^ but i am not fluent in html.
The href links there you can replace with other personas.
Last time i had duplicated a bit more, but at the end it had the same effect.
This works for me getting Suno to use 2 personas.

[ebon sand]

The html doesn't matter to it, it's only what you see

it doesn't change anything about anything the servers receive

[unique stone]

Yeah I checked one of the song links given above and there's only one persona associated with it.

[ebon sand]

yeah and the persona entry isn't an array so you also cannot manipulate it to carry multiple entries

the server would likely reject it with bad request or trim it to the first or internal server error

changes to the html are purely cosmetic

[lost holly]

I only can say that my song this way uses both selected personas. I cannot explain why or how it worked.

[gloomy shore]

I accidentally created a song with 3 vocals in it. I don't know how i did it. So can anyone help me on which type of prompt should i use to create 3 vocals?
Here's the song.
https://suno.com/s/uAC1Kxl2KdTepbC4

[unique stone]

Song 1 used persona 1 only
Song 2 used persona 2 only
Song 3 used persona 1 only

[ebon sand]

There is no foolproof way but high success for me was for example to provide the lyrical themes in a way that they only make sense with several speakers, and the inline tag [narrator] usually gives me a 2nd voice, the only foolproof way is using edit, there you can add any voices you might want but for the initial prompt it can work or not, there are only likelihoods

[lost holly]

If you say so, i just like trying stupid stuff and see what the stupid results are, dont want to argue who is right and who is wrong 😄 having a good time with suno is my goal, thats all 😄

[ebon sand]

i usually get 3 different vocals by having for example chanting cultists, a hierophant preaching and a narrator, but that won't be the desired result for everyone haha

[gloomy shore]

I'll try your and see if it works lol

[ebon sand]

if you liked the 3 vocals of your track you could try to persona them, maybe it picks them up in future songs

[boreal harness]

Anyone had a problem with Remasters turning your horn sections into female vocalizations?

[gloomy shore]

How do you persona them?

[ebon sand]

basically rightclick on the song or on the 3 dots and then create -> make persona or so

[gloomy shore]

Oh reuse the song style might not work 100% isn't it?

[gloomy shore]

Chanting cultist, hierophant preaching, narrator... didn't work lol

[ebon sand]

i don't put them into the styles, you kind of need a song setup which even would support these topics, i just use inline tags, i have no singer clarification at all in the styles but you have to find something that works for your case

when i want a duet i prompt for a priestess or a succubus because that fits the song topic and will to a high chance generate the right delivery

[gloomy shore]

Oh so add them into the lyrics on the part that you wanted to have a different vocal?

[ebon sand]

Yeah like here, but it's always very unstable, if it switches you can edit it later though, my serious songs rarely offer a different option than using various vocalists
https://suno.com/s/gvo5kCjWZHuwlAYB

[lunar anvil]

I have had 11+ voices in a song, but there was so much background noise that it wasn't release quality.

[ebon sand]

you can add as many as you want with edit and patience, to me it's the better option than trying to prompt it initially, if i can get it i'll take it though

[lunar anvil]

This was before the editor.

[primal juniper]

With 4.5+ model, i get alot of yelling instead of singing. Doing Thrash metal and techno atm

[ebon sand]

Having the same problem, in my case country where it doesn't fit at all

[dark linden]

Does anyone know how to get the vocalist to not be a screamer/screetcher but keep the like nu-metal/post grunge sound?

[boreal harness]

Share one of mine with you:
"Modern, clean production, Male voice that blends a deep Southern crooner's drawl with raw intensity, Early southern preacher-style vocal phrasing: sultry, bluesy, full of swagger; merges with the tortured growls of some wandering demon in an alleyway, Verses begin smooth and sinister, like a gospel-voiced devil whispering sweet doom, Breaks and bridges erupt with unholy distortion: guttural, ragged, with demonic clarity, Voice should shift between clean, honeyed rockabilly vocals and feral growls, Each line is performative, possessed, ritualistic, and tinged with theatrical southern gothic fury"

[zenith depot]

Having GenAI add Sound Effects and Vocals to a Video (any video) needs to be AI MLed.

Should even be possible to take out the audios that are already within the video through the ‘stem’ method and such.

[gloomy shore]

So you have to specify which verse to have the vocal sings it.... 🤔 why did you specified the Vocalists at the beginning of the lyric? Is it to let the AI knows that there are 2 vocalists will be singing?

[ebon sand]

yeah this often had good results for me, i don't play around with it enough to call it a method but i did that for a while to have some sort of instruction block at the beginning, for example also to include what the song is even about so that the AI can get the delivery right, to me it worked well, it did something with the output but not enough data to call it a way to do it haha

[gloomy shore]

If it works it works 😁 thanks

[analog sparrow]

I'm coming back to Suno after spending months mastering it about a year ago. Back then the way I got the song I wanted was to keep adding and extending stubs and importing it into my DAW. I tried the new editor but it really changes the sound after saving changes. What is the way you all edit and add to segments now?

[humble verge]

is it just me or are 'S' sounds more shrill in 4.5+? It's like getting stabbed in the eardrums with an icepick.

[brazen yew]

Any ideas how to get longer songs without crowding the lyrics page. I'd just like a long track with a few bits of spoken word scattered in amongst it. At the moment if I don't add anything into the lyrics box I'm getting long tracks but as soon as I add lyrics, takes two mins off and desperately tries to get all the lyrics out. Will just filling spaces with randomness help increase the song

[delicate oriole]

Can someone plz help me? 🙏

[analog sparrow]

I think I may have answered my own question. It looks like Remix is what I would've loved a year ago and Editor (in beta) will at some point replace it. So to avoid the Editor really changing messing the song up on saving changes, I'll use Remix? Also Persona will probably help. Stems and the DAW it is!

[delicate oriole]

I'm just trying to figure out how do a parody in suno

[daring lava]

Hey there! How can I help?

[delicate oriole]

Thank you, I'm try to make a parody of a song, specifically the Bezos song but idk how to

[daring lava]

With Suno you can't use another artist's work as a reference if that's what you mean, but you could make songs that have a similar style, vibe and feel! If you'd like help with that, please let me know!

[analog sparrow]

David, I'll take advantage of you being here. I used to create music by generating short stubs, then repeatedly using the basic extend function to build full songs section by section before importing to my DAW. Now should I stick with the traditional extend method, use the new Remix panel's Extend tab, or try a different approach to maintain sound quality and avoid the editor's audio changes?

[delicate oriole]

So I can't follow the same melody as the original song but with different lyrics?

[daring lava]

Hey there! Thanks for reaching out!

That's a tough call, the extend approach is a classic. Plus it definitely has a lot of cool perks when it comes to being able to build out bit by bit while retaining consistency.

Personally, I've been trying to one-shot everything recently. 4.5+ is a beast, there's an 8min limit, and well... it's pretty fun laying the whole foundation through prompting and seeing what comes out.

I guess it's dealer's choice. Interested to hear what you come up with though!

[daring lava]

Great question! As long as what you are using to create belongs to you, you can do it. But if it doesn't then it's away from Suno's ToS. I attached a link to the Terms of Service in case it's helpful!

If you have any other questions, please let me know!

[analog sparrow]

Yep, one shot is definitely the best sound. Agreed. The model is great! Excellent job on those improvements. But that is why I'm torn on how to proceed. I need to add a longer intro to an absolutely beautiful composition. The only way I've figured out to preserve it is to import the stems into my DAW, Then create the intro in Song Editor, which fuses it into the opening lyrics nicely, import those stems into my DAW and put it together manually. Any other ideas?

[daring lava]

To do what your saying, the simplest approach that comes to mind would require stiching the song in post. Sounds like you already have the body and tail of the song, you just want to extend the head.

• go ahead an drag that off to your DAW and clip the intro out gently

• after that, extend from the part of the intro that holds best and add complimentary guidance so that it builds out how you want it to

• once you have that, download it, throw it in your DAW and edit the tail off where you want and stitch it to the beginning of the initial track in post 🙂

Good luck with your song, feel free to ping me to it if you share it here in server! Always looking to check out cool jams!

[neat field]

v4.5+ seems to have the Remaster function built in, at least seems that way to me

i should test V4.5+ Cover/Remix a song a second time with same setting and max Audio Influence 100% to see if that does anything at all

if Remaster is built in, announcements should be made

[mild egret]

For @winter patio = Curious... just vocals https://suno.com/song/55948936-4d62-4a3e-86c0-c154517dd6aa only 1:17 but Style was "spoken word only, one male, ‑music" . 2nd song was 7:35 but instruments came in: https://suno.com/song/83f86512-b3f7-4c8c-8421-aeaf4fe83394

[misty oyster]

Is there a problem with uploads right now? Im trying to upload a 14 second audio and it never ends up uploading

[untold briar]

One of the biggest changes since you last used Suno is that the maximum song length is so much longer now, that you can make a whole song without using Extend. This opens up a whole new direction for workflows where instead of Extending (or Replacing) in small chunks, you iterate on the entire song at once using iteratiev Covers (Cover of a Cover and so on). Note that you can also iterate on Personas, and use any particular Persona and a Cover at the same time. All of these features don't even use the new Editor!

Conversely, features like Replace can be used in the opposite way as well, to generate entirely new music and not just "fill in the gaps" or make a small change in the lyrics. That is less well supported in Suno, but it can be quite useful. You can think of Replace as having two audio prompts, a left and a right, and just using the middle gap as the product. This lets you not be constrained by the left and right audio having to be continuous and make sense as a whole song.

[neat field]

instrumentals on v4.5 are something special, never heard a cello treated in such a manner

https://suno.com/s/Wfb0g5nbgLzCsTPH

[viral pond]

aaaaaaahh... why does the first generation always become the best 😄 and you mess up the vocals, is it even possible to recreate it with the correct vocals using a mix of persona and demuxed instruments ?

[radiant breach]

Best tips on maintaining a persona? Seems like I'm still getting gens that sound off. Not as consistent as 3.5 was. When I get a good song I make a persona, then any song after that sounds nothing like the original.

[hasty bridge]

Hello, are you also having trouble with Suno at the moment?

He no longer wants to give me a female voice, and he's no longer taking into account my recorded personalities.

[radiant breach]

Above the audio sliders now you can choose male or female

[hasty bridge]

I saw yes .. but despite the fact that I put women's voices .... I have women's voices once every 20 songs and they are not the ones that I have recorded even though I select it

[spark cove]

Hello, so i just generated a track i rly liked the voice on it i tried making a personna off it but it keeps giving me voices that are very different from the track

I rly liked that voice as it sounded cool and had a cool accent too

[humble verge]

I get an error when trying to create a persona.

[analog sparrow]

Thank you! Yes that is very helpful. I just spent five hours since I chatted with David trying to extend a 30 second intro by 17 seconds. My conclusion is that it's not possible using any method aside from an act of God. I'm trying to time it for a video you see. Both the "Legacy" editor and the new one seem deceptively tailor made for this purpose - but I fear it is actually something far more sinister created by a product manager and head of marketing.

[near flume]

You may be able to use the method you described as a way to Remaster but it's not something the average user would know about compared to a Remaster v4.5+ link in the menu.

yes. The app and website have different access.

[tawny kelp]

Question, I put in lyrics that should generate a 30 second max clip, but Suno keeps repeating the lyics and tries to make the song 2+ minutes long. How can I get it to only follow the lyrics I have written and stop repeating them?

[sleek stump]

If you recently signed up for a Suno account, the initial free tier gets a limited trial of v4, you can gen 5 times on the web, and 5 times on the mobile app for a total of 10 free trial generations. Currently on mobile devices, to see the model selection drop down you need to turn your phone to portrait view.

[sleek stump]

This is incorrect. access levels do not change between devices, what does change are the features in the mobile apps, they are limited due to the platform. iOS has more features than the Android app, but it still has fewer features than the web version.

Normally you get 5 and 5, but I have heard of all 10 being used on either platform. regardless, your access to v4 will be done very soon\

Keep in mind that 10 songs is 5 generations, each gen creates 2 tracks.

10 total generations for free v4, that is a total of 20 songs. Once you use them you are then limited to v3.5 on a free user tier.

I cant do voice, I have no microphone on my PC

[humble verge]

Is there some kind of descriptive tag that might make 4.5+ actually use the persona? I think I've tried every setting by now, and haven't had any results that used it.

[naive ibex]

Regardin 4.5+, as I understand that 4.5 (without +) is now fixed, solid, and (finally) nobody will mess aroud it, no model background changes - stable and reliable to work with no hidden changes later? (4.5+ messed up some stable style presets as usual...)..... I honesly would add 4.50 from middle of the May, it was the best results of all time.

[sleek stump]

4.5+ is now the beta model, its going to be tweaked for a while.

[naive ibex]

I mean 4.5 without plus. Just "4,5" looks like not in "red" color. And the question is - will they mess with update it, or it supposed to be rock solid 💎

[sleek stump]

It's no longer a Beta model, so I would assume that for the most part, they wont be making any more big changes to it, outside of bug fixes and the like.

[naive ibex]

What could be good keywords to make sound level lower? A lot of times composition is perfect, but sound level like clamped (like above 100% and start to sounds dribble/noisy). This just major and the only problem with 4.5x..... so might someone have a good presets to avoid that "overloading"?

Especially extremely sad when you get perfect composition, musically ideal, but that overloading noise is ruin all song, and not fixable in post production

[polar spade]

Not sure of your "extend" methods you've tried, but If you already have an intro, it should be a workable option to go in Edit UI and add a segment as long or short as you wish. 💁‍♂️

[raven turtle]

Just tried remaster 4.5 on a 4 track. The results are now excellent. Not CD quality but close enough that with a little izotope Ozone and Shadow Hills Class A I can get it sounding damn close. Hats off to Suno for continuing to push the envelope at pace. If this is now the standard, v5 will be incredible.

[bronze helm]

Oh, we have Remaster 4.5+ now?

Last I checked there was Remaster v4 and Remaster v4.5, didn't see v4.5+

Exciting if true, let me see

Did you perhaps mean you covered your track?

[raven turtle]

My apologies I means 4.5pro on a 4 track! Oops!

[bronze helm]

All good

It's totally harmless

I won't lose sleep over it 🤣

[raven turtle]

I’m finding the remaster really upping the quality since yesterday for some reason. Although the base tracks are sounding clearer overall anyway, but I having to do very little post production work

[bronze helm]

Maybe they improved it as well

Suno models do get tweaks after they are released

[raven turtle]

The EQ curves and general waveform compression still allowed for a lot of dynamics. Really impressive

Now if V5 would let me dial in the LUFS I’d be set 😉😂

[bronze helm]

You seem to understand post-production edits. Can you tell me if you've figured out a way to "increase the resolution" of vocals, for the lack of a better term?
For instance, if I have me some vocals and I want to edit them in Audacity, they sound a little bit like their bitrate is lower, which is enough to noticeably decrease the quality.

[bronze helm]

Wouldn't be surprised, Suno's kinda starting to feel like a DAW

[raven turtle]

Are you stem splitting to get these vox out like that or just noticing that in the flat WAV?

[bronze helm]

Yeah I think I used stem splitting

vocals & instrumentals, so I just took the vocals

Let me send an example if I find one, even though I edited it slightly, but you will still probably hear what I'm talking about

Oh I found the unedited vocals

They start at about 0:19

[raven turtle]

I’ve noticed a drop in the fidelity on stem split. Perhaps try doing a remaster of the split vox first? TBH sounds like you may need a little clipping or de easing?

[bronze helm]

Hmm let me try that, will get back to you

[raven turtle]

I’ve listened. Trust me I’ve had to work with many 3rd party real vox a lot worse than that! 😂

[bronze helm]

A little busy this morning so I'll tag you later @raven turtle with results

[gilded mortar]

Hi guys I have a song that I want it to be extended in the same style. What should I do . I'm new

[untold briar]

Ah, if you're trying to extend something from 30 seconds to exactly 17 seconds, and that's a hard limit, then that's a very challenging edge case. It's not impossible in Suno but I would have really think about how to pull that off. You might need to do something very fancy, like insert a 17 second block of silence into the track, and use the Add Instrumental/Vocals feature.

[raven turtle]

Go to the extend function, add in what you want in the lyrics section and you should be good to hit create

[gilded mortar]

It's instrumental. So I don't want any lyrics within the track

[raven turtle]

There is an extend option and it listens to your source track. Or if it’s just repeating what’s go before you could go to edit section and do some copy paste and regenerating of sections. But sounds like you just want to hit the extend option in create

[polar spade]

We try for 17 seconds and suno might give us 1 minute, so It would be similar to doing a jingle: As long as the insert goes over 17 sec (by only a few seconds) and not under, its likely snippets can be edited back out to get an exact time and still sound fine. cc @analog sparrow

[gilded mortar]

Sorry for my ridiculous annoying questions . But how to make it listens to the source track ?

[bronze helm]

Is it just me, or does it kinda feel like v4.5 and v4.5+ kind of go overboard with the background instruments and noises...?

Like, it feels less like background music and more like an eardrum assault

v3.5 and v4 were less liberal with the background instrumentation

I just appreciate that v3.5 and v4 try to be a little more modest with the background instruments and effects

Where v4.5 might go overboard at times but not always and is overall the best

But v4.5+ tends to go overboard most of the time

[raven turtle]

Whe you extend it has the source track lined up to extend from it👍

[raven turtle]

What I’ve done to combat this is remaster Down to v4. It cleans it up allowing me to Then to remaster back up to 4.5.

[ebon sand]

It's not just you, I appreciate v3.5 much because it uses silence, simplicity and repetition effectively and the newer models always try to create a hit

When I include that it's desert vibe then 4.5 gives me a hans Zimmerman bombastic movie performance and 3.5 adds a guitar playing a canyon sound

[eager axle]

Has anyone noticed (it may have been like this for a while) - that even though 4.5(and 4.5+) supposedly can create tracks up to 8 mins - they always seem to cut off and end it at exactly 4.54, for some reason - anyone else having this issue?

[winged peak]

Yes, happened to me yesterday night for quite a bunch of tracks as well ;/

[unique stone]

Are you using persona/cover or other things beside just a text prompt?
Anything that needs to reference audio will result in songs shorter than the maximum length.
If it abruptly cuts off, you can extend the song to continue it.

[winged peak]

@unique stone Partially I have been using a persona, but it happened as well to 'new' creations, i.e. no remix, no cover, no persona. Someone at reddit complained about that a few days ago as well and I guessed back then that simply extending the lyrics would help, but I was wrong.

[unique stone]

Hmm. Are you using the advanced options sliders? Changing from the defaults can make things behave unpredictably.

[winged peak]

@unique stone Yes, I did and I do, nevertheless seems to be a bug 😉

[eager axle]

Yes -that would definately be the reason - i didnt actually think of that - thank you

[modest patrol]

Yo, how do I actually to this from the announcement? "🧑‍🎤 Add Vocals"

I have a bunch of instrumentals that I usually use extend to transform into songs, but this new feature, how do I use it? Where is it?

ah i guess this just works with covering now? it seems to pick up my lyrics

[solemn heath]

hello how can i put music on discord when i create it thank you very much

[lethal ravine]

Has anyone addressed the center channel collapse that happens to most songs when played on 5.1 or higher home systems?

[sleek stump]

Not really? Tracks are stereo waveforms, not multichannel surround encoded.

[lethal ravine]

I agree they come out in stereo but after many tests I have found many times that voices push more towards center/mono range. which causes surround systems to collapse into center channels

[sleek stump]

You can link your Suno songs from the right side menu on each song, clicking this icon will copy a link to your clipboard which you can then paste into a Discord channel. Additionaly, there is an option in the 3 dot menu for sharing

[echo prawn]

Why do I exclude male voices in the settings, but the generated songs still contain male voices?

Suno still hasn't fixed this damn bug

https://suno.com/s/ltKdkNfpLjxsymK5

[polar spade]

You might have to wait. It seems you can add vocals/instrumental to uploaded audio and to instrumental stems [and add instrumentals to vocal stems]. But actual created "instrumentals" does not bring up the option to add "vocals" in the menu. [ADD features found in the remix/edit of the menu]

And these options may change in any way or disappear, but for now, you can spend 10 credits on stemming out your instrumental into another instrumental and "Add a Vocal" to that if you wish. How exactly this would differ from just doing a Cover of your instrumental and adding lyrics, I'm not quite sure.

[polar spade]

"......Dears,no matter how I write the prompt, I keep asking for a song with a 16-bar verse and a chorus that repeats twice. However, Suno always generates a full track where the chorus section is only 8 bars long and doesn’t repeat. Has anyone encountered a similar issue? I really need some help—please!....." --- from @zealous bay (moved from need-help channel)

@zealous bay Im not sure of your process, but do you literally have a 16-bar verse and two 8-meter duplicate choruses in your lyric box?

[zealous bay]

“I didn’t enter any text in the lyrics section. I only wrote all my requirements in the style section, including a request for the song structure to have a 16-bar verse and chorus.

[polar spade]

Ahhh.... might it be easier to ask chatgpt (or your choice) to write up lyrics to match those meters then paste into Suno. It would be the same results as you'd want.

[polar spade]

You can also ask Suno to write the lyrics in the Write Lyrics part of the Lyric box. "Cherry Bloom Dreams" on the right matches your criteria.

[polar spade]

If we've answered your question here -- then please delete your question from the need-help channel (its there for billing/ account support help). Thanks. https://discord.com/channels/1069381916492562582/1372994553233276970

[tranquil thistle]

Does anyone have the glitched stem results?

[humble verge]

SUNO added this style to an uploaded vocal stem I'm using for a persona - which means it's also the style for the persona. I did not write that and my persona is female. The style has little to nothing to do with the song/persona. For one thing, it's a vocal stem and has no instrumentation. It's a FEMALE vocal stem, but according to this SUNO-generated style, it's a 'male vocalist with a deep, resonant baritone'.

There's no way for me to edit it or get rid of it, and every result using this persona comes off male. I've even opened it in the editor, changed the style and saved it as a new song. But still the SUNO style remains. I've tried re-uploading the stem and SUNO just goes ahead and adds the same style to it. What's going on here?

[final rose]

Thank you! Sounds like the answer is no for now, at least for metal.

If anyone would care to share their 4.5+ model mp3's for non-metal songs, especially if they feel the production quality is good, I'd appreciate it.

[near flume]

You need to declare "Male Vocals, Male Voice, Duet" in the exclude box. You need to use both.

[near flume]

It's still glitching, I wouldn't expect to see this fixed until sometime on Monday, I doubt they work weekends.

[tranquil thistle]

Thank you for the confirmation on the issue, sir 👍

[cloud chasm]

If you’re referring to the new 4.5+ “Advanced Options”, then the interface might be confusing you.

“Exclude Styles” looks like it could be a heading, but it’s actually a text input box. You can type text in there to guide Suno away from any unwanted music styles that happen to be creeping into your song generations.

Below that, “Vocal Gender”, is an entirely separate option independent of the “Exclude Styles”. Therefore, the “Vocal Gender” is an option that is asking you the type of voice you want to be used in your song.

So, in “Vocal Gender” select “Female”, and your songs should then have Female singers.

[buoyant gale]

I used up many credits in the latest version 4.5+ instrumental music, it's great but what I don't like is the Overwhelming Melody riff. I prompt pop music, instrumental acoustic guitar, Relaxing Music then use Creative Boost. The result is the Melody is so overwhelming, it's like a flamenco or jazz guitarist is playing. It's not natural anymore. This is just my honest review, we would like the music to be as humanize as possible

[hot trout]

Does anyone know why it is suddenly impossible for Suno to pronounce certain words properly? The song I am currently working on includes the word "flag"... which really means (according to Suno) "flaygh". Or "farg". Or insert slur directed at homosexual people here. But so far, out of more than four dozen attempts, not a single generation that uses "flag" as written.

[verbal knoll]

Some of Suno AI music glitches at times...with low quality at times at +4.5 when going towards the end of the music

[neat field]

RE: v4.5+ Remaster

covering a v4.5 song with v4.5+ and setting "Style Strength" to 0% with a high 95%++ "Audio Strength" is the closest i've gotten to v4.5+ Remaster equivalent -- and leave Style tags empty

[shadow plaza]

is it worth remaster my songs from 4.5 to 4,5+ ?

or should just go straight to gen new ones

[humble verge]

God I hope SUNO improves female vocals in the future

[polar spade]

Do you mean making a cover of a 4.5 to a 4.5+? There are only 4.0 and 4.5 Remasters.

[shadow plaza]

Ao i cant remaater my old aonga to 4.5+

[polar spade]

There is not an 4.5+ remaster option. Not yet.

[steep river]

hello, i am wondering, when i go on my phone to get to my account, i enter my number, and i get the code to enter, but it seems like someone else created the account somehow, i cant get to my library, it shows nothing

[shadow plaza]

Thanks for the info disnt know that

[polar spade]

If there are issues such as different volumes or clicks, at least 4.5 remaster could help. But if youre just hoping to up the fidelity of 4.5 songs, just the 4.5 remaster is the only choice.

[steep river]

it shows up correctly on my computer though

is there someone here today that can help?

[shadow plaza]

So what is better in 4.5+ from 4.5?

Which style perform best with newest model?

[polar spade]

You may have created a new account by mistake which you have associated your phone number.

Check here for some suggestions. https://discord.com/channels/1069381916492562582/1374814090282930246

[polar spade]

A master is intended to improve the sound quality in general , but everything is subjective.

4.5+ is also different in how it handles certain instruments and of course may push or pull sounds more forward or back in the mix.

I would say its is just anyone's opinion. Many prefer using 3.5, for instance, for its more analogue, authentic and organic sound.

[near flume]

The female voices are fine in Suno if you prompt correctly
What type/style are you trying for?

[humble verge]

got nothing to do with prompts.

[near flume]

so what's the problem you're having?

[humble verge]

They degrade over time - usually around the 2:30 - 3:00 mins mark. They become increasingly shrill and tinny beyond that. Consonants harsher - K's, T's, S's especially. I've spent hours in DAWs softening them. It's a general issue with AI vocals.

[near flume]

what genre are you working in?

[humble verge]

i've worked in lots of genres. Folk, blues, rock n' roll, blues rock, metal, renaissance ballads. It's the same and doesn't have anything to do with styles or prompts or whether or not I use personas.

[steep river]

Is there someone from Suno that can hlp me with a account situation?

helpp*

help*

[near flume]

Gotta say I haven't had that problem in v4 and above.

[humble verge]

https://discord.com/channels/1069381916492562582/1372994553233276970

[humble verge]

Do you listen to your music in headphones?

It very much depends on the length of the song. If I work with more than 3-4 verses of 4 lines each, the voice degrades fast after 2:30 - 3 mins

[steep river]

Rafaelis, i went there but im not receiving a response

[humble verge]

I don't think they work on weekends.

[polar spade]

If your computer is logging in write down the phone number found under Manage Account.

Go back to phone and use that number instead so that the phone logs in.

[steep river]

ok thank you

[polar spade]

If the good account on computer does not have a phone number associated, then write the email and which SSO you use to log in on computer.

If you keep using the URL in memory, you should make sure you know which SSO (single sign on) authenticator you used to sign up your original account. That information is ALSO in your manage account section.

Mine , has a phone as Primary and two sources (with the same email) to log in with my email

[steep river]

now they have my account locked, so im stuck until i can contact someone

[polar spade]

Multiple attempts may lock the account. You can try again in a couple of hours just to see if it temporary.

Do you know the correct SSO used ? Many will just use the same URL to "auto-log" in while the browser still has it in the cache, and then after awhile, they lose track of actual log-in info.

So check you email for a Suno Receipt to help know which email you signed up with and hopefully which SSO you used.

[steep river]

ok, maybe i cant do anything until monday but im trying, im still able to create songs

[cloud juniper]

uhhhhh pardon me suno?

[raven turtle]

I’m trying out Inspo with a 5 track playlist. Finding new generations a little too similar to the source playlist. Anyone know if adding more tracks to the playlist would help “dilute” the sound and give a more generalised take on the playlist? 🤔

[polar spade]

https://discord.com/channels/1069381916492562582/1374816730546962462. You payment may not have been received and posted yet to Suno. This occurs on new paid accounts and manual payments.

[polar spade]

You may be more instrumental than lyrical , but:

The idea is it should take the general ideals, nuances, vibe and somehow create a song. In other words as if making a persona of each track then add all those codes together as one big prompt to make a song that sounds like the next on on the album.

So, far, I do agree it seems to pull alot of riffs, intro, and even melody lines and almost veers south of the instrumentation sometimes.

There may, as you say, a way around it to "dilute" down the melodies and such and just focus on the vibe.

Ill be trying to crop 1 minute snippets that have less melody and no intros and make a Playlist to see what that does. If its good then form it back into a song.

I'll let y'all know if its a decent option or bust.

[cloud juniper]

hmmmm soo it locks you out of using the stuff you have over it?

[near flume]

I use 3 different types of headphones, Bose Noise cancelling, AKG studio headphones, & cheap earbuds. All my songs use no more than 3 verses with pre-chorus and riffs between chorus and next verse, so maybe that does some type of reset or clearing of the vocals.

[polar spade]

"Stuff you have" ? Im not sure what that means.... but... Often people just never delete their cache and keep using a shortcut or a URL to log in. When something happens and they need to log in, they forget how/email/etc., try many times and then get locked for multiple attempts.

@lunar gyro

To make a hallucinated song into a "real" one you have to match the lyrics to the melody. To do that, you have to write words which make closely in syllables and metering to the melody.

The most precise way is to transcribe the gibberish into actual text. At as many points possible try to use words that sound English or your language of choice. This will give you anchor points to meet up between phrases as yours transcribing.

I would run a Cover at this point to check your transposition for accuracy.

Then the harder part, to write lyrics to match those meters and syllables.... as exact possible. You can try to do it in one shot or use Cover or Replace to do it in segments.

Thats about it except for all the tweaking with prompts and arrangement or some may come out as planned on the first shot.

[polar spade]

Some may post a shortcut version if they wish.

[humble verge]

I can send you examples if you want. I mean, it's not like degradation of audio quality isn't a well-known issue with AI music. It's not just vocals either. Drums tend to get more 'tinny' the longer a song runs. A song that starts out really nice may end up sounding like it was recorded inside a metal bucket by the end of it. I've had to scrap lots of songs due to issues like these. Currently I'm at my 15th remaster and fifth remix of one song, in the hope that one will come through ok, but so far no luck. 4.5+ is much better than 4.5 at maintaining instrumental quality, but vocals still degrade a lot once a song runs past 2-3 minutes.

[lunar gyro]

cool i'll try 🙂

problem the song is (i dont know how) 8.5 minutes long and it has 2 singers , and a choir part 😄

[lofty scroll]

Hi—this is my third week waiting for billing activation. I’ve emailed [e-mail] on July 5, July 16, and July 19 with proof of payment, but still no reply. Requesting urgent assistance, please.

[neat nexus]

The correct place for these types of inquiries is #archived-ticket-help but since it's the weekend that channel is on slow-mode. So given the length of time you've been seeking help I'll waive that and ask you to please DM me the email address you used in each of those emails and I will flag the situation internally. You will still need to wait for the Team assigned member to act on that message during business hours but it is something I can do to bring attention to this issue.

Please do not attempt to post it in here. Woodhouse doesn't like it and will automatically remove it.

[polar spade]

OK. You got the idea. Ill remove my samples.

[sleek stump]

You have been making a pretty concerted stink about this, what exactly is your goal here? Have you used the feedback form to send Suno your feedback and issues you are perceiving?

https://suno.com/feedback is a direct way to send them feedback, and the most effective.

[mint bramble]

I don't see "Inspo" where can I find it?

[north plinth]

is anyone having issues signing in via google acct? I get a JSON script when I try to

[sleek stump]

At the top of the create window, above the lyrics window..

[sleek stump]

Try clearing your cookies and your browser cache.

[north plinth]

it's now working thanks was very odd

[remote kelp]

What are people's impressions of v4.5+?

[remote kelp]

It is virtually IMPOSSIBLE to get Suno to generate an Acapella that doesn't have some sort of percussion rolling in... whether it's Acapella sounding percussion or like, actual drums... ughhhh

[tranquil thistle]

Does anyone know if the stem stutter issue has been fixed?

[ebon sand]

It works with v3.5 and V4 when you have a persona without instruments
It also works with personas that have music but then it's less reliable

[ebon sand]

I think it's better than v4.5 because more creative but ultimately it doesn't generate what I like thus I don't like the model the outputs somehow are very 'annoying' to me in the way it assembles a song and the delivery

[polar spade]

I spent a couple of hours yesterday look at your issue. I want to try a few more things, to see if i can get good stems out of it.

What did you need the stems for ? That may help me a little.

[polar spade]

Last year it may have been an issue. This year, just separate the vocal tracks now is the easiest route. Then edit off the vocal track to get what you like.

Since we have melody AND "vocal" instruments (beat boxing) as a feature of the current pop style "a cappella" genre, (definitely NOT chapel singing anymore) then Suno doesnt always sort out the difference of percussion AND beat boxing and throws them both in.

Unknown what your parameters are (and dont want to know 😉) certainly you can add in things like >> Spoken, narratives, audiobook << and as many voice-without-music prompts or tags are available, even choir, boys choir, Gregorian chant, chant, native American chant, exclamation points, quotation marks, etc. Again, depending what your goal was.

You'll use more credits experimenting as there's no right way of doing this. Our final product will be different every time, than if you just separate out the vocal tracks of one you already like, but it's good experience and you may get something on the first try.

Also, try spelling it with the musical Italian form of "a cappella" [as the proper "Italian" music term] if that helps, which may give a different results -- as spellings like "acapella" and others are informal but common usage and generally accepted, but worth trying the proper spelling as well.

[polar spade]

@lyric scarab. From need-help channel.

Apparently you only have Basic subscription? Popups that suggesting we upgrade are on buttons and toggles that are not available in Basic.

[lyric scarab]

Thanks

[polar spade]

You have an introductory trial of 100 credits for use of 4.0 version. After that, its only Basic tools.

[neat field]

i tried 1x experiment
Style Strength: 0%
Audio Strength: 100%

using Add Vocals + Inspiration setting Styles to empty elements

output is ok with "pure" audio input guidance + lyrics

it's worse to my subjective ear for my genres than Style tag + Style Strength 100%

[boreal harness]

Remastering not allowed? The Hell?

Am Pro.

[sleek stump]

Try logging out and back in, if that doesnt help, clear your cookies and try again.

[boreal harness]

Same thing happened.

[sleek stump]

Are you trying to use the remaster back down to 4.5?

[boreal harness]

Used 300 credits to finally get it to pronounce the vocals correctly, and now I can't get a consistant master. Bleh.

Nope. This track is pure 4.5+.

[sleek stump]

Not sure what you are doing here, 4.5+ does not yet have a remaster function.

[boreal harness]

The track is built in 4.5+, not Remastered. Trying 4.5 for Remaster. 4.0 is greyed out.

[sleek stump]

Still not understanding what you are trying to do... You edited a track, got the full track and now you are trying to do what?

[remote kelp]

I actually did the "a capella" thing you mentioned, but interesting advice.

[remote kelp]

You're saying V4 is better for this?

[boreal harness]

Remaster it. Get all the audio levels from the pieces at consistant EQ.

[sleek stump]

And how exactly are you trying to "remaster" the track? Are you trying to use the remaster 4.5?

[unique stone]

Does it offer you v4 Remaster for any other song or is it always greyed out? (Just wondering if this is a song-specific issue)

[boreal harness]

All of these versions can be Remastered.

[sleek stump]

I just did a remaster of one of my 4.5+ songs and it worked fine. It may be you have a borked track.. try making a copy in the editor and then remaster that one

[ebon sand]

v4.5 also showed some potential but doesn't respect personas much, v4.5+ did whatever

[unique stone]

Ohh. It might be too long for remaster.

[boreal harness]

Is there a limit posted somewhere?

[sleek stump]

Oh my, those are all very long

8 minutes

[boreal harness]

8 minutes is for track creation.

[unique stone]

Not a documented one as far as I'm aware but it's probably meant to just grey it out like it does for v4. What used to happen is that it'd let you do it but it would just spin for ages and then error about the clip being stuck.

[boreal harness]

documentation oversight, then.

[sleek stump]

documentation has been one of the weaker parts of Suno...

[boreal harness]

I bet the limit is 15 minutes.

[sleek stump]

HAve you been able to remaster anything over 8 minutes?

[boreal harness]

Yup. Everything in the last screenshot.

[sleek stump]

interesting

[remote kelp]

V4 has generated some amazing things for me, but the shimmer sucks and it is not consistent AT ALL

V4.5 is much more consistent

[unique stone]

Just trying this by adjusting speed on one of my songs and it seems to be 16 minutes

[boreal harness]

I guess I will have to cut it in half and hope the two halves get a similar sound.

[ebon sand]

I don't have issues with it anymore really, but i only use it with personas

[boreal harness]

Confirmed. Trimming to 15:47, it Remasters just fine.

[sleek stump]

Theres another aproach you might try, adjust the speed to under 16, remaster, then adjust that speed back to the proper length

[boreal harness]

I'd worry about frequency and pitch corruption. But then again, I did speed it up a little already. The longest version was 22 minutes.

[remote kelp]

With Personas it is much more consistent?

[frail epoch]

Well I can’t log into the app
Logging in does nothing

Pressing log in with Google does nothing but load then stop and nothing happens

[hollow hare]

Try changing the default browser on your phone. When that issue used to happen to me, defaulting to the default Browser worked.
Once you log in, you can change your browser back
(The app uses whatever your current default browser is set to for logging in)

[frail epoch]

It’s the App that’s not working

I’ve always use safari, I can log into it on safari but not the app

That shouldn’t effect the log in of the App?

[fluid estuary]

Anyone here good with FL Studio that's working with edison denoise to remove the noise? I've been trying to fix my drums, but I get clipping noise all the time.

Thought I'd ask here since we love working with noise in Suno 🙂

[hollow hare]

It does

[frail epoch]

idk why it would, I dont feel like downloading another browser just to fix a paid service thats another app

[hollow hare]

I was able to log out and back in with no issues on my side

[frail epoch]

I can show you a video, it loads the google thing, goes back, loads a bit then doesnt bother leaving the log in page

[hollow hare]

So you only have the default browser installed, no4 like Chrome or Firefox?

[frail epoch]

then yuou click the google log in and it repeats

never logging in

[frail epoch]

iOS has Safari, Safari isnt ram hungry and terrible like Chrome

and it has an AI get rid of ads feature which chrome lacks

or anything really

get rid of any part of a website you dont like

Safari is just better

[hollow hare]

Ah, forgot about iOS users. Hmm, verify that there are no settings enabled which would affect Javascript or third party cookies eince it sounds like to me one of those is causing the issue (likely the cookie).
Regardless, you should report that to #📱┃bugs-ios

[frail epoch]

I use Chrome on my Macbook Pro because I use two or three browsers at once but Safari is still my default

[hollow hare]

You may want to clear the browser cache in case something in there is causing the issue (happens on the rare occasion)

[neat field]

every time i renew my sub (not every month) i can consistently run out of the 2500 credit plan.... fun times

[near flume]

You use Remaster to upgrade the sound not to downgrade, that's why v4 Remaster is greyed out for v4.5 songs. If you want to EQ levels download the wav stems and adjust in Audacity or Cakewalk.

[polar spade]

Have you ever logged out of and back into suno from your browsers or at least one? It seemed before you may have been "logging in" by just keeping it always open and never logging out. Maybe I assumed.

From @tame moth : (in need-help channel)
Hi Suno team!

I’m using Suno v4.5+ and I’ve uploaded a WAV file to try out the “My Voice” feature.
However, I only see the Persona option, and there’s no My Voice tab in my interface.
I would love to try singing with my own voice model.

---

Response: I have not heard of a "My Voice" feature. Are you seeing videos or seeing somewhere where they mention a tool called '"My Voice"?

We can upload out own song of us singing. Then Covering it is a fair way to try to emulate our voices, but its often not that close and hard to store as a persona for making future songs.

[tame moth]

In the mail i got it said that we can use our own voice for all song. Maybe I read that wrong? But the problem is that I can't hear my own voice in Suno. It creates a song with a different voice.

[polar spade]

If its short, can you DM me the sentence or the little advert. I'd be interested in seeing how it's worded.

The usual way, so far, is to click upload, upload our audio.... we can be singing with instruments or even just a vocal. Then choose the uploaded result file and Cover it. Make sure Cover tab is chosen. Slide Audio slider (if any) to 99% and you can leave the others alone.

Dont add anything to the style prompt box. Leave blank.
Leave lyric box blank of you wish. It might auto fill in with a transcription of your singing. If so you can check it for errors, then use it for one test or copy and remove it for another.

Click Create and see what it does.

[neat field]

part of the challenge may be a dialect prompting improvement, you may have a regional dialect or a Standard X English Dialect and it may be good to prompt any Extension or Cover to contain the name of your spoken dialect and the emotions you want Suno to convey in your voice

it's part science, part linguistics, part psychiatry, part art, and lots of luck

[cursive island]

im trying to saxophonize a certain word in the chorus whats the best way to do that?

nevermind this worked very well https://suno.com/s/I1NgoAGszeXGH89M

completly impossible without AI

[inner coral]

I wish we can go back to the old way to edit. The new update just does not work for me

[sleek stump]

Use the legacy editor

[fresh wave]

Is anyone here able to confirm this?

[humble verge]

what's the deal with the editor? It's super bugged

[polar spade]

It was doing some strange things yesterday, like making an entire song as one of the two of the sample regenerations on every chick. Weird stuff like that for me.

[humble verge]

I'm getting errors in the new editor and when using legacy, the Styles in Extend is bugged when typing something in or pasting it.

also having the usual issues with Extend not generating more than 10-20 seconds. I think I'm just gonna drop the whole Extend feature. It's useless.

[polar spade]

I'd say, personally I have no idea. No reference,no date, no authorship, no idea.

But......Seems pretty harmless. You can't break anything by doing that. Why not just follow that and see what comparative results you get.

[neat field]

crosspost with General because maybe this is a better place to ask:

has anyone tested v4.5 or v4.5+ crows or [crows] with "birds" style tag? i am out of credits but curious

could be potential for a "Crow Storm" song that has a lot of literal crow noises in it

[fresh wave]

It seems like it works for taking a vocal part from one song and putting it with the instrumentals from another song. I think just the interface doesn't, uh, it replaces the first upload, so it doesn't appear like it's referencing it. I definitely was able to combine two of my songs the way I was trying.

[polar spade]

I've always copied what i want to keep and reuse, then, clicked "clear all" and refresh the page before simply just doing a new generation from an empty Create or Remix. I have no idea about the rest of that balderdash.

[fresh wave]

It blocked me describing a vocalist as 'black' without telling me it was banned?

[polar spade]

Everyone is privvy to the same censoring. Its not as if we have a list. Maybe re-word it or re-arrange the phrase. Nowadays, we all have a 15% chance of being blocked for something, no doubt.

[fresh wave]

Yeah, I mean, I managed to get it through. It's funny because it doesn't exclude it when I mention a vocalist is black. It just didn't seem to like it for the persona description.

Kind of ridiculous, because half of all popular music is by black artists. I didn't know there was a way to make it problematic.

[cursive tartan]

I can give it a try, if you want...

[polar spade]

Stop clicking on a greyed out button.

Scroll up! See that shiny bright "Renew Subscription" button? That with stop the pending scheduled cancelation of your Pro account and put things back the way they were before you canceled -- if that's what you want.

Dont think too much into it. Its logical but not clear, I know.

[humble verge]

creating personas seems to be broken as well. Lots of work for these guys come monday

[polar spade]

I just stopped and had a nice weekend. 😂

[humble verge]

lol

good advice

[pseudo sapphire]

Has anyone had luck with duet prompting?

I'm trying to do verse 1 female pre chorus female so forth

[near flume]

birds works, crows doesn't. Sound effects are hit or miss in Suno.

[neat field]

Can confirm spent 20 credits making nightmare fuel that didn't even sound remotely like crows or a crow storm. Not making those public

[neat field]

Duet works. Specific person prompting sometimes works and the model will do as it pleases

[near flume]

In the styles box do this "Duet, Female vocal declaration.., Male vocal declaration..."
In the lyrics declare the voice like this [Verse 1: Female Vocals]
[Pre-Chorus: Female Vocal
If you want to switch per line do this;
[Female Vocals] the line to sing
[Male Vocals] next line to sing

[pseudo sapphire]

I'm trying, should I go to 4.5?

[near flume]

yes, it might take a couple of generations to catch on.

[pseudo sapphire]

I got it, thank you so much

[remote kelp]

I'm sure this has been pointed out already, but anyone notice that for v4.5+ it's "Styles" again and not "Style Description" like it still is for v4.5 regular?

[junior sun]

prompt: "Epic,Dubstep,Power Metal,Rap. Got Raided in ARK Survival Evolved. My poor dinos are all dead. Meat ran by that giga. My metal ingots & Element are all gone. Back to the beach. Lets tame a dodo."

Take 2.... i change "dinos" for dinosaurs.... now you say im using artist name "gigas" ..... wtf

your ai needs to get smarter and need to understand we are not refering to an aritst in ALL contexts... im out of character limit... i cant change giga to giganotosaurus

[brazen yew]

Is there a way to keep a strict vocal structure? I have 4 lines of spoken word throughout the song but singular lines will repeat intermittently aswell as the vocals basically slurring resulting in something that sounds inaudible

[hard beacon]

sometimes metatags like [spoken] and [sing] are working good for me

[urban canyon]

try this: describe it in general in the style section under verse and chorys and in lyrics put it part of the header tag [Chorus - male only] or [Verse duet male female]...it works sometimes after a couple of generations for me ...let me know if you had luck..

[bronze helm]

I have never heard of "giga" or "dinos" lol.

As artists I mean.

[junior sun]

lool yeah....

[dark linden]

That's weird because it let me use "gigas" in my prompt with no problem. I have a streamer friend that I wrote a song for that is an Ark OG lol

[novel scarab]

is there a way to force male/female vocals] eg [male voice] text [female vocals] text [duo vocals]

[viral pond]

okay, so i might found a way to recreate a song pretty close to original, i was just going to make a test song, but it turned out too cool... but the lyrics had too much glitch and neon in it... Two songs to compare :
https://suno.com/s/UzzlYLfHA1iJugZn original
https://suno.com/s/2NWYbRND085ndOZl remake
am I onto something here? I did not edit the song, i downloaded it and stemmed the vocals away, but used the vocals as persona, set weirdness to 0%, style 50% and audio strength to 100% ??? any better methods or is this a good way to do it ?

[brazen yew]

That's not what I meant but I think I've figured it out. You can't have the ideal vocal structure in a long song. You either have many lyrics or short spoken sections will just repeat even if you slot them in amongst your prompts. Best to just add vocals to the music retrospectively. Vocals are definitely suno's weakest link

[flint elm]

Yo guys! Love that the 4.5+ version finally makes metal sound rich (sound-wise) but what can be done about the volume that keeps increasing throughout a song?

[brisk flint]

Yes, in my opinion, you can almost ignore the two new buttons. It is more important that you include this in the style prompt AND in the lyrics as prompts in square brackets. Then you can even define the timbre (whispered, soft, harsh, loud...) and this works quite well with 4.5+.
Since I've been working with this method more since last week, you're welcome to take a look at my prompts if you go to my profile on Suno (this isn't meant to be a sneaky advertisement): @tia
This will give you a good impression of how I manage to shape the voice colour pretty well according to my wishes. It often works – but not always, of course 😉

[magic shale]

What methods of labeling and rating the songs or parts of it do you use? Considering we only have a thumbs up and one down - Suno did not want to help us here, all sugestions were ignored 1,5 yeasr), only text places in the song title could be used. Placing songs in different playlists or workspaces it is too time consuming.
I used special short strings placed in the song title to mark a good intro, a good voice or a song which I want to redo/cover/finish later and also for rate a song 1-3 stars.
This way, using the search function I can find all these song fast and easy again.
Of course, further selection can be used by throwing them in a separate folder later.
Who has such a thing and is interested in developing together this further can write me a direct message

[humble verge]

I've had some really good results lately. Telling SUNO to use female voice in styles and brackets - and then adding a male name to the styles. Like "Think [insert name]" or "in the style of [insert name]. I've used male singers who aren't caught by SUNO's filter, but I wonder if any male name might work. I've had about a 50% success rate so far.

[lunar anvil]

I tried doing that and the AI did its own thing with the same success rate or less than your 50%. edit: to add success rate

[humble verge]

I'm going to experiment with it more. I just happened across the method yesterday when adding a particular male singer to my Styles, but keeping everything else female - including persona (though I haven't had any luck with personas at all since 4.5+). I'll have to try and see if I can do duets with two women or two guys - like soprano/alto or tenor/bass.

[lusty jungle]

if ya'll are going to continue lengthening the ability to make songs, now its up to 8 minutes, ya'll need to put in advanced options a selector for song length. All i keep getting is 6-7 minute renders and all i want is a 4 minute song. Very frustrating. I was already struggling on 4.5 but 4.5+ is making me mental. Everything else is good, i just want a choice on song length..

[lunar anvil]

The longest song I have gotten is 5:07 in one shot. Is your lyrics or prompt contributing to the length?

[shrewd willow]

Thats crazy, my musics neves pass 4 minutes

[humble verge]

I often have the opposite problem. It may depend on genre.

[lusty jungle]

nope. in fact i put in the tags 4 minute song. prior to that i didnt have a tag in for length. Check it out. even tried using the "ai prompt for styles" thing... hoping it would help. i do not have a genre picked out either.

[humble verge]

I bet it's the word 'epic' in your prompt.

[shrewd willow]

I use the rule of thumb of never passing 4 phrases per verse (except chorus where I use 6 to 8)

[lusty jungle]

epic has never effected it, only gives it a bold trailer style feel

[shrewd willow]

2 verses, 2 pre chorus, 2 chorus, 1 post chorus.

[humble verge]

'Slow' might also have an effect. Not sure how much you use it. But when I've used it, I get longer songs in general.

[lusty jungle]

i rarely use pre chorus. AND these are the same lyrics i used on a previous version of suno and it had no issues making a 3:30 minute song out of them so im not sure whats up. i'm usin gthe same lyrics just trying to get a better sound

i use it a lot

I jus want a song length slider. thats all

[humble verge]

That would be interesting. Input a 10 verse song and ask SUNO to make it 30 seconds.

[shrewd willow]

It would probably skip some phrases. It already does 🤭

[humble verge]

Or make it a Smurf song

[lusty jungle]

lol hey i've seen weirder. one night, suno would only make metal songs out of anything i put in there.... it was like stuck... so i made a song and the only lyric is "suno's going to turn this into a metal song" lol

[shrewd willow]

4.5+ is indeed a bit crazy for me. Its going crazy on effects and over the top singing even on 0% weirdness. I went back to 4.5 even admiting instrumental on 4.5+ was so much better

[lusty jungle]

i've been a long time supporter here, since 2.0 and i jus wish... that some things we've been askin for had been implemented by now, like a song length choice. this has been asked for for months. since before advanced options was a thing. and alsooooooo advanced options... vocal gender... needs a dang DUET option

[shrewd willow]

Intro on 4.5+ is top notch for my style, but the singing....uhg....

[lunar anvil]

I haven't been following any structure of late. My last song had an Intro, 5 verses 1 chorus 1 bridge and an outro.

[lusty jungle]

felt, quite a few of my renders feel almost... showtune-y. not jus on this one but on other songs i've rendered recently too till i jus scrapped the project and will save my lyrics for another day

[shrewd willow]

I ended up making a set structure due to Suno repeating lyrics or hallucinating mixing chorus or verses. I got used to it by now and my credits thank me

[humble verge]

It's still hit or miss. My biggest beef with SUNO is the gradual deterioration of vocal quality during a song - mainly when they run longer than 2½ - 3 mins. Sibilance and hyperarticulation of hard consonants and S's goes through the roof. As does the intake of breath. I've had to scrap so many nice songs due to this. 4.5+ is slightly better, but it's far from perfect. That whistling 'S' sound is like an icepick straight into the brain.

[shrewd willow]

I think Personas make a huge difference, I admit I`ve not been having a lot of song deterioration on 4.5

4.5+ is a train wreck

Sometimes it says the word Ribbon like a russian accent word, for whatever reason

[humble verge]

So far, 4.5+ has just ignored my personas. Hasn't worked once.

[shrewd willow]

And it has been constantly repeating the beat that I took my persona from. 🤭 , I used around 2000 credits trying to figure it out just to realize its still not in a good shape for my style...

Its going in the right direction though. Intros were top notch, way better than the ones in 4.5

[lusty jungle]

ok so same lyrics, same tags.... different version of suno... this is what im talkin about... they need a damn time slider

[humble verge]

Yeah I'm almost certain it's something in the prompt/genre that determines it.

mine average out at around 3½ to 4½ mins

[shrewd willow]

If a song of mine ever reaches 4 minutes

something is wrong

It barely ever passes 3 minutes 🤭

[humble verge]

i've had several 7+ mins with 4.5+, but then I wrote a fairly long song to push this new 8 mins time limit. And with extensions I now have a few 10-11 minute songs.

[lusty jungle]

so do mine, except on 4.5+

[shrewd willow]

on 4.5+ its averaging 3:00 to 3:30 for me

on 4.5 its 2:30 to 3:00

[humble verge]

No doubt there's some tweaking they need to do as is often the case with new features. The Beta version will over-emphasize that feature.

[shrewd willow]

There is a lot of small fixing to do. Not sure how is the experience for you guys, but the "new" (not so new anymore) song editor is completely busted for me. The legacy editor is still much better

[humble verge]

Oh absolutely. I don't use the new editor at all. It's buggy and, let's face it, SUNO is not great at doing UI. Sadly they messed with the legacy editor as well.

[rose jasper]

Not possible to Extend?
An error occurred.
Invalid request. Inconsistent task and generation params

Ok, addin Inspo breaks it

[novel torrent]

What is this server?

[wild egret]

how can i crack the code of this songs? https://www.youtube.com/watch?v=QqCZzViuHqA&t=3797s&ab_channel=patient. even suno remix/cover cant do the job
https://suno.com/s/XwkqrjpUJapaSiVK
https://suno.com/s/JPpDM8ZuFer6Sibu

[near flume]

It's Tempo and what you put in the lyrics box that mainly determines how along a song will be.

[humble verge]

It can vary a lot too. I'm currently working on a song and despite using the same settings/lyrics/prompts - I get anywhere between 2:50 mins and 7+ mins. It's cool. Keeps things interesting.

[radiant breach]

Any tips on having a persona make a new melody? Seems like any setting I do my gens always start sounding like the persona.. same melody tweaked slightly, way too similar. Nothing original...

[sleek stump]

Persona are meant to be influencers, not bases...

[radiant breach]

Ok?? And? In 3.5 I used a persona and was able to get 12 unique songs with the same instrument and vocals. With 4.5 I can't get new songs. They start like the persona.

[sleek stump]

Are you adjusting the sliders?

[radiant breach]

I do get unique songs but their the ones that sound different them the persona. They said there was a fix but it seems like it just doesn't work like it's supposed to.

[sleek stump]

3.5 is totally different from 4 and 4.5

[humble verge]

Weird. I have the opposite problem. 4.5+ completely ignores my personas. Even at 100% influence.

[radiant breach]

I've spent enough credits to know I've tried the sliders I've tried inspo I've tried everything. Right now it just all sounds the same it's frustrating. And the persona doesn't hold up.

[sleek stump]

When 4.5 beta dropped, they recommended to make new persona for 4.5

[radiant breach]

Add fade to an old gen and use that, yeah I did it

Also a persona I'm using is from a 4.5 song

[sleek stump]

Persona have never been stable from my experience.. I dont use them at all because of all the issues.

[radiant breach]

3.5 worked perfectly for them so idk

[sleek stump]

Some folks have great luck with them, others, not so much

[radiant breach]

It was your "band", it let you get different sounding songs with the same general sound. In 4.5 it's just taking too much from it but still doesn't sound the same. The vocals are different.

[sleek stump]

I really dont have much interest in persona, what I want are vocal presets.

[radiant breach]

I don't want every song to start with the same melody

[humble verge]

i'd pay twice for SUNO if it could reliably do harmony, double tracking, background vox and duets under full control. And triple if I could retroactively edit songs to use the same voice throughout.

[sleek stump]

We might get closer to those kinds of features when they start implementing Wavetrack...

[humble verge]

what's Wavetrack?

[sleek stump]

Its a DAW that Suno acquired last month

Thyey integrated their entire dev team

[humble verge]

Oh, nice. It will be interesting to see how it's integrated

[sleek stump]

indeed

[quaint whale]

Did anyone else break Suno by adding too many styles?

[sleek stump]

My fear though is that its going to be expensive to use...

[sleek stump]

At some point you can just have too many styles and it just creates what it wants

[humble verge]

it would have to make sense to use it instead of just exporting stems to an external DAW

[sleek stump]

Especially when I already own 3 of them...

[quaint whale]

Like it leaked either a partial access point to the Suno template system or the Suno LLM construction scaffold

[sleek stump]

sounds like a classic hallucination

[quaint whale]

yea, what I was thinking

[sleek stump]

'it does that when it doesnt know what to do

[quaint whale]

testing

Someone is trying to prevent me from talking rn...

[humble verge]

Have you tried RipX?

[quaint whale]

it cascades

How is that politics?

They don't want ppl finding out. W/e

I'm trying to help by bringing it up.

Thx for the credits. I'll keep my mouth shut.

[midnight steeple]

Can someone tell me how can I write a duet song and control what lines are sung by female and what lines are sung by male.... I've tried everything but cannot get SUNO to allow a female to sing all 4 lines of a song even with the tag [verse 1 - female vocals][female] and (female) at the beginning of each line... the male singer still takes over.

[zenith depot]

Suno is like Leagues better than Udio for the Instrumentals.

[humble verge]

extend must be the single most useless feature in SUNO

[sleek stump]

Nope, I use FL, ACE and Audicity for the majority of what I do.

[sleek stump]

I have never had any issues with it, but I havent used it in the last month so I dont know if I would. I had my own process for getting what I needed and it was 90% success.

[humble verge]

Here's the two types of extensions I get 90% of the time: 10-20 seconds of basically the same as it replaced (not extended) or 5 minutes of repeating the vocals from earlier in the song. That's about it. I have wasted thousands of credits on this.

[gleaming locust]

Hi can someone help - the ‘replace lyrics’ function is suddenly not working on any of my songs. I can’t click on the button. Any idea why?

[humble verge]

the editors have been extremely buggy lately

[gleaming locust]

So you think it’s a system issue rather than specific to me?

[humble verge]

yes, most likely

i've had lots of problems over the weekend

[dusty tangle]

Trying to have a duet between a female singer and a goblin kinda creature.. How can I get the vocals to sound angry guy punch wall and more evil creature thing?

[still holly]

You need to go into edit mode