📚┃suno-school

[polar spade]

I did mention back, to throw it in the editor UI and add small gaps between each one to see what happens. I did that. I put a 4 to 8 beat gap or segment between everything, and typed "oh oh" as lyrics.... so she'd sing one part then sing "oh oh" then go to sing the next part. It was semi-terrible-trash, but good to see that it was possible to be creative and do something with adding small segments.

[echo moat]

You've been a massive helpt oday

[polar spade]

Just thoughts and a snippet where some things seemed to work. With Suno, you might want to damn a river, but it may then flood a town. So its like throwing spaghetti against a wall and seeing which noodle sticks. You never really know.

[echo moat]

i'm still a little confused on the whole persona deal.

[young geode]

Just remove the persona and generate, see if it still causes the issue: make sure the breaks/interludes, ect are metagged between the sections.

If it goes away, it was the persona.

[echo moat]

10-4

but then if the voice changes which i imagine it will, how do i get her voice back?

[polar spade]

I thought you were working off the output 1 of the original output (OG) you appended the persona to?

(OG) w/Persona---> output 1
Output 1 + edits (no persona) and Sliders ----> output 2

Are you going back and reworking the OG with Persona ? Or just reworking output 1 without any persona now?

[polar spade]

@noble galleon Sent you a DM referencing your issue that might help. I assume youll check it Monday.

@karmic egret. Reference Student ID promotion response from 3Daisy #archived-ticket-help message

[echo moat]

the guide i read basically said persona is style and cover is melody for the most part

is there a minimum length i need to capture a voice?

[wooden sleet]

Suno is really annoying me now I want to make a very short song about 1 min preferably, but it's dragging out the songs and adding unnecessary instrumentals breaks and intro & also dragging out the words, how would I fix this? also my lyrics are only 2 verses long

[polar spade]

We tend to feel 1 minute is adequate. The answer is really to test and see.

That really came out of a persona taking too much of the reserved area when making a cover or extension that ends up shortening the covers or extensions.

As far as working, 1 minute seems to be enough. If its hard to find a good minute, throw the song in edit UI and spice together a minutes worth of varied vocals.

[polar spade]

Not sure the style of song, but the word "jingle" is recognized by Suno as in these examples i whipped up. https://suno.com/playlist/f64592d8-952d-479f-b46a-5de36a01e073
Memo can also help. Certainly anything over 1 minute is more useful than less than a minute, as you can simply cut out a bit of the areas which have just music , so its down to 1 minute.

[polar spade]

The instrumentals and intros are helpful since they are fairly easy to edit down as much as you need to get it down to a minute.

[hollow hull]

Hi I want to make a person pero I don't want to make it public I am not sure how to know if I am making it public or not, which button is the NO PUBLIC , red or green? https://prnt.sc/WP4BeT9ocXNn

[sleek stump]

Dim button means it is not shared.

[hollow hull]

ty @sleek stump

[unique stone]

This seems to be backwards for some reason.
Just checked my personas and ones I have public have a dimmed button.
Ones that are not public do not have a dimmed button.

If the circle is to the left, it is not public. If the circle is to the right, it is public @hollow hull
So the green one in your screenshots is not public.

[polar spade]

Why cant it just turn pink like before? 🧐

[sleek stump]

Ya its backwards... to the left its off, to the right its on.. but the highlight is counter to what has been a standard for decades.

[young geode]

It's likely in the reverse so you aren't aware of it/ignore it after it's on.

Suno wants personas to be public, so giving it the lower contrast while on keeps it hidden from sight, but the high contrast brings attention to it.

[young geode]

I've done anywhere from 15 seconds to 45 for a voice. If you're trying to just isolate the voice, you should split the stems and then find the areas of the vocals that have the vocal qualities you're looking for, that are also clean and with little distortion.

When it comes to Personas, the longer the persona is, the stronger it's affect on the song will be. Shorter length vocal personas will introduce the 'vocalist', while moderate length personas will capture their 'vocal style'. Longer personas will start to cement the 'vocal structure' (IE when the Vocalist sings) and Complete personas (a full 1:1) will effectively cause the song to completely revolve around the vocalist.

So if you're just looking for the voice, a 10-15 second clean cut of the vocalist is sufficient as a persona. If you want their style, you ought up that to 30-45 (just long enough to capture 2-3 sections [verse-verse-chorus]), as most vocalists will switch it up in that time frame. I wouldn't use anything longer than a 45s cut for voice, unless you really enjoyed the way the vocalist parsed the song, because anything longer than 45 will start to force a narrowing/expansion of 'space' within the song: IE the vocalist is always active/too many gaps, creating instrumental filler.

[small bridge]

is anyone having duets turning country no matter what they try. I'm trying to get a two person female male contemporary R&B feel. But it's giving me country or sixteen thousand voices for every line.

[young geode]

DM me your prompt and the song

[buoyant lotus]

Not the same, but for some reason all my songs turned metal until I removed "organ" from my prompt (I wanted church vibes)

[devout sinew]

Hi everyone! I am trying to get Suno AI to follow specific choir assignments in my song prompt. I want a children’s choir singing only the pre-chorus with soft harmonies, and then a gospel-style adult choir singing the chorus and final chorus with full harmonies.

I have tried several different ways to phrase this in my prompt, but so far only the children’s choir is being executed, and the adult gospel choir is not showing up at all, no matter how I specify it.

Has anyone successfully pulled this off? If so, how did you word your prompt to make Suno follow the choir instructions correctly per section?

Thanks in advance. I would really appreciate any guidance!

[young geode]

It's likely cause the AI is confusing both as the same.

In order to get these kinds of effects, the 'vocalists' have to be distinct enough, so as to force the token prediction to treat them differently.

There are several things you can try:

1. Make the Children male/female and the Adults the opposite gender.
2. Give the children a distinct range/polyphony and the Adults the opposite.
3. Make one a choir and the other an a Capella.

This doesn't guarantee that it's going to work, but it leads to a higher probability of it occurring. Doing it all has an even stronger effect

For example, for 3 you'd likely write Vocals 1: Children's Choir, Light, Airy and soft. Vocals 2: Adult's a Capella, Annunciated and direct. Harmonies in the Final Chorus.

And then you'd strengthen this using Metatagging.

[Vocals 1: Children's Choir, Light, Airy and soft]
Lyrics```

```[Chorus: Adults]
[Vocals 2: Adult's a Capella, Annunciated and direct]
Lyrics```

```[Final Chorus: Harmonies]
[Vocals: Children Choir, Adult a Capella, Harmonies]```

[molten spire]

Hey y’all. Why is one song out of the generated two in a bad quality. Any idea how to fix it?

[pine carbon]

Hi everyone!
I have a question about the new v4.5+ model. When I'm making covers of songs (since the remaster function isn't available for v4.5+ yet), I've noticed that the generated tracks always get cut short by about 5-10 seconds. This didn’t used to happen with the classic v4.5 model.
Does anyone know if there’s a trick or workaround to prevent this from happening? Thanks for any advice!

[polar spade]

You might mention how long is the original output and the length of the covers. Are they consistently the same time or varied around the 5-10 seconds. And how many did you generate at the same close length?

Within the 8 minute time limits now, generally, there should be no reason a 4 minute song should not cover anywhere from 3 to 5 minutes.

[polar spade]

Has that been consistent over 10 or so generations (eqch 2 outputs) of different songs?

[echo moat]

i'm having a really hard time trying to create a person based on an acapella

i go to upload audio, choose my acapella, then it asks me to do a cover for it, but doesn't let me edit it to trim the 10-15 seconds

like once i upload my acapella, it asks me to either do a cover, extend or save to library, if i choose save to library, it never saves

[magic shale]

Hi, why does a song of mine show "remixes are disabled for this song" ?

[echo moat]

@young geode OMG thank you so much, i did a 10-15 second person and now she is allowing instrumental solos and interludes

so i literally had the perfect song and in the end it just ends abruptly, what should i do about that? will doing a cover upscale or remaster fix that?

[polar spade]

@pine carbon seemed to have the same issue today, though seeing a cover to do that now would be considered odd. Having two users on the same day experience the same might be a weekend glitch happening. #📚┃suno-school message

So if you have a moderate song of less than 6 minutes or so, the Cover should make a completed song of anywhere from 2:30 to 4 minutes optimally but no matter, it should end the song by button ending, stab or fade. Just cutting off somewhere is again, odd.

[echo moat]

yeah cover didn't fix it either, it ended abruptly as well

[polar spade]

Yes. We're talking about covers that end abruptly.

[echo moat]

even worse when the song is literally perfect from start until finish than bam

[hybrid vine]

I'm trying to find a way to have hints of chiptune in one verse of a song, I tried putting it in the style prompt, I tried [Verse – Chiptune] (and many variations, like chiptune hints/elements, 8-bit video game sounds, etc.), but no luck

[wind barn]

is there a way to pronpt chord progression for like verse and chorus etc? thank you

[polar spade]

Same for instrumental ..... and Custom....toggles too

[sleek stump]

They are literally backwards of how things actually work. When you turn on a circuit with an indicator, it lights up. We are trained to understand from a very early age of what on and off are. Flipping this for buttons in the UI causes confusion, and the ensuing questions.

[young geode]

You're welcome, glad it worked out.

As for your issue of it stopping: how long is the song right now; it just stopping near the end?

[echo moat]

what's the best way to edit lyrics? i click on the box on the left, and it won't let me type anything in, i create a new segment, and it seems to jack everything up

like some of the words at the very end aren't correct, and want to add an outro

[young geode]

That's a question for @polar spade

[echo moat]

idk if the editor on mobile is any better but the one on desktop is a little rough around the edges

[young geode]

Its not, use the web browser one

[azure prairie]

it says tech so ill asume im in the right area. if not sorry, but im using the website and for some reasons my songs dont publish on suno or do i have to pay for that?

[lost holly]

What is your Suno handle?

[azure prairie]

whats that?

[lost holly]

Your Suno profile name, so i can check if any songs are published.

[azure prairie]

Tensefan704, there are songs published but the ones from this month dont publish at all. it says publish but when i come back later that day its not published

[lost holly]

I see 35 songs, does this match? What is the last one from this month?

[azure prairie]

Most wanted 2012 edition and Clem. are my latest. the button goes purple and i get notification saying its published than 30 mins later when i come back its no longer published

[lost holly]

Hmm true, the last one i see is from june the 29th.
Have you tried clearing cache/cookies, relogging, logging in through another browser or inkognito tab?

[compact shuttle]

Hey! Just wanted to pop in and say I got my original issue solved — but I ran into something else.

Stripe isn’t accepting Cash App or Google Pay on my end. I had to manually enter my card info to get it to go through. When I try to pay using Cash App or G Pay directly, it says either “Cash App is experiencing an issue right now” or “Google Pay is experiencing an issue.”

Just wanted to bring it to y’all’s attention in case any mods know what’s going on or if it’s a known issue; because its been an issue since the 20th.

[azure prairie]

no not yet. ill try those now. thanks

nah nothing worked. thanks anyways @lost holly

ill maybe try the app, i doubt app is avaible on PC

[lost holly]

It sounds definitely like a weird case.
Only other thing i can think off, is that something is blocking the publishing, likely on your site, like an add blocker.
Maybe someone else has an idea.

[azure prairie]

i know for sure suno is the only webiste/app that i havent made an admin pssed at me, let me check the ad blocker after searching what ad blocker is

english please

[lost holly]

Sharing links like that works though, right?

[azure prairie]

let me test on #🐣┃suno-showcase

[azure prairie]

the answer is no. it also shakes my damn screen lol

[azure prairie]

the links share in my server and other servers im in except this server

[half dragon]

Share the link here and see if Discord embed can play the song then we'll know it's actually published
If the embed doesn't have a play button - it is still private and not published

[azure prairie]

link doesnt want to share to this server only idk why

https://suno.com/s/qui6aHnhdz7mguxk

oh but it will share here!

[half dragon]

yeah - it's not published to public

did you click Publish button on the song - gotta ask the obvious question 🙂

[half dragon]

I think it's the showcase - you can only post once per certain amount of time

[azure prairie]

once! ouchi

[azure prairie]

i did

it keep showing the purple dot as well

be back soon.

[hard beacon]

Good evening.. somebody heard something new about remaster in 4.5+ yet??

or is it in 4.5+ but displayed as 4.5?? Coz the color is purple and not white 😄

[raven turtle]

That’s interesting…. Ooooh

[hard beacon]

It was just a thought I had..... maybe they forget to change the layout in the remaster-windows, but... well... who knows 🙂

ok.. after remaster in 4.5 the color is white in the library-screen.. so... i was wrong 🙂

[raven turtle]

We live in hope my friend! 👍

[hard beacon]

oh ya 😄

sick of creating cover-versions in 4.5+ ... of course good output, but.. naaahhh.. to much "cover-style" 😄

it's disappointed if you have the "perfect" 4.5 version and you wanna do an upgrade and then.. NARF...

[junior sun]

anyone know a reliable way to instruct a voice to use for a given verse or line.
I tired this
[Verse 1 – Male Voice1 (soft, drifting)]
(verse1...)
[Verse 2 – Male Voice1]
(verse2...)

But after 2 generations... so 4 songs all same result.... first verse is a male voice... good... but then 2nd verse it switched to a female voice.... why?

[half dragon]

There isn't a realiable way to achieve this. Just keep on trying until you got it.
Other options are Extend each part sepearately or use Editor to replace with the correct gender vocal.
But it's still not reliable in any options. Be ready to spend lots of time and credits to achieve what you're after.

[wild egret]

has anyone has a clue how to replicate these kind of songs?
https://open.spotify.com/intl-es/artist/0dgJbQ0bKPyUXco8hEXN7X

have tried many things but its not there yet

maybe suno is not good enough at this point for this kind of songs??

[solemn lintel]

What do you mean by "replicate these kind of songs?"

[wild egret]

the style, the ambience, the sounds, the vibe, the reverbs, the synthwaves, etc

[solemn lintel]

There are a few techniques you could use, but to start with, have you tried experimenting with different style tags to try to get that genre?

And messing with the sliders, like upping the Weirdness?

[wild egret]

yeah, i've tried many things, tuned up the sliders too

it turns out lofi, for some reason, eveytime, even after banning "lofi"

[solemn lintel]

I've found a lot comes down to just experimentation, and... throwing tons of credits at it until you get the sound you want. ^_^
But there's also the persona approach. Though with copyrighted music, I'm not sure how ethical that is.

[wild egret]

turns into lofi or rocky
https://suno.com/s/WEVMUN11TbDjQBI6
https://suno.com/s/k6BB9x0i2CClgSp6
https://suno.com/s/PpaEAbV4XYtPoWAC

[wild egret]

havent messed with the persona thing, what is that?

[solemn lintel]

Do you have a free account or paid?

[wild egret]

paid

[solemn lintel]

OK. So a persona is basically a way to capture the "sound" of a song. Style, vocalist, that sort of thing. How it works is kind of mysterious and magical to me, but it works.

Basically everything I do now is using personas.

[wild egret]

where is that different from uploading an audio sample?

[solemn lintel]

So you get a song that sounds the way you want, then click that song to go to its page. Then click the little "..." and you should see this.

When you select "Use Persona" it will then make a persona based on that song. Then, any time you make a new song, you can click "+ Persona" to add that persona to it. I've noticed, though, that 4.5 tends to really like to reuse the actual tune of the persona rather than just the "vibe." So I often use 4.0, or just have to generate a handful of songs before it stops using the original melody.

I hope that makes sense? ^_^;

Right now the issue I'm running into is I made a song yesterday using a persona, came out AWESOME on the first try. Then I made a different song using the same persona, and ran about 20 generations and it simply would not use the vocalist. So I gave up, and ran a new song, and once again it worked perfectly on the first try. It's so weird and flaky and inconsistent, and I really wish we had some way to give more instructions to Suno on what we want it to do.

[wild egret]

hmm okay, got it, i think haha, i will try that and let you know, thank you!!!!

[solemn lintel]

Of course! Good luck, and have fun. ^_^

[wild egret]

and i wish you better luck with your productions 🙂

[solemn lintel]

LOL thanks

[hybrid vine]

I'm trying to find a way to have hints of chiptune in one verse of a song, I tried putting it in the style prompt, I tried [Verse – Chiptune] (and many variations, like chiptune hints/elements, 8-bit video game sounds, etc.), but no luck. Any help?

[half dragon]

Have you tried it in the Style as well? other option might be to upload a sample?

[hybrid vine]

I have, but when I do, it puts the hints on the whole song, not just the verse I want.

[polar spade]

Small point correction: when you originally make a new persona, you find "Make Persona" [instead of Use Persona] in the same location.

[wild egret]

I tried to use it but no generation comes close to what I need

So Persona will just make things different from what I want, my guess...

[young geode]

Dreamscape song, dark ambient, ambient, Chord progression is d minor, a minor, g major, in melodies, not the full chord, Arpegio style, Slighly imperfect, Syntpath included, reverb and delay, Introspective, low tempo, Synthwave, snowfall alike song, Atmospheric textures, Melodic fractions, Immersive audio, Slight drone

dark ambient, ambient

Lo-Fi is often a subset of Ambient, which is a Subgenre of Downtempo. Additionally, Oneheart is within the subgenres of Ambient>Lo-Fi.

What you might want to do is remove Ambient and instead use Era and Genre2020s Chillwave, 1990s Microhouse: Chillwave to move into the more relaxed area of synthwave and microhouse to reduce the amount of instrumentation, while targeting the years these sounds typically appear. Further you can weight them with percentages Subgenre: 2020s Chillwave 80%, Influenced by 1990s Microhouse 20%. Then you can add in Mood: cold, sadness, winter time, retrospection and style: quiet melodies, soft beats, light reverb, delayed syncopation

Prompt Rebuilt:
Style Prompt
2020s Era, Adagissimo Tempo, Downtempo, Subgenre: 2020s Chillwave 80%, Influenced by 1990s Microhouse 20%. Dm – Am – G Chord Progression, broken chord, Arpeggio chord, imperfect notes. Instrument: Serum Chillwave Synths, lush, dreamy, atmospheric vibes. Style: quiet melodies, soft beats, light reverb, delayed syncopation. Mood: cold, sadness, winter time, Introspection.

Exclude Prompt
1970s, 1980s, 2000s, 2010s, Rock, Metal, Rap, R&B, EDM, Techno, Hardcore, Industrial, Country

Scale Settings:
Weirdness: 70%
Style Influence: 75%

[dreamy heath]

is there a way to cause a "pause" for a few beats in Editing? Often it will run from one verse to the next like it's in the same sentence, ruining the rhythm and the meaning.

[devout sinew]

Thank you so much for your detailed suggestion. Your theory was sound and a really creative workaround. I tried it out in several variations, and while it was a great alternative in concept, unfortunately, Suno AI still did not execute the prompt correctly.

Despite spending several tons of wasted credits testing it, the results were inconsistent. Either the children's choir was completely ignored or non-existent, or Suno would only include one choir and not both. On the rare occasion the children's choir appeared, it would not sing the correct section (Pre-Chorus) and only performed two lines before dropping out.

Still, I really appreciate your help and effort. It was a smart approach, and the problem seems to be more on Suno AI's current limitations than your method. Thanks again for trying!

[wild egret]

this is the result
https://suno.com/s/JudIw7vIC4HgWHJV
https://suno.com/s/zwyAV0SpZ3mzTTAw
the closer i've ever been. not bad in terms of tones, melody and vibes, but its way too fast and upbeat, seems like a happier song

[young geode]

Vocals are the hardest to do.

The best way to force dual vocals like this is to create completely contrasting lyrics, in pace, meterline, rhyme, texture, everything, with heavy meta tagging and targeting genres that have those vocals. Then generate until you find the found. From there you rip the vocal stems and then create a persona that captures both.

Then try to make the song you want using that persona. Its very time intensive, and frankly not worth doing under the current versions most of the time.

[young geode]

What BPm are you looking for?

[wild egret]

i've ran a few of that spotify playlist songs on a bpm recogniser and it turns out to be faster than your prompt, about 80-90bpm

[devout sinew]

Thank you again. That totally makes sense, and I really appreciate the honest breakdown. I can see now why it is such a challenge. The idea of contrasting lyrics, heavy metatagging, and leveraging genre-specific structures is smart, but yes… it does sound incredibly time-consuming for such uncertain results.

I agree, it is probably not worth pursuing in the current version of Suno AI unless they release more reliable vocal layering control. Still, your explanation gave me a much clearer understanding of the mechanics behind it. Thanks again for generously sharing your insight!

[wild egret]

what i meant with this suno result being faster is that the melodies are faster, when you listen to the spotify ones you can think that any of those would/could be used to fall asleep to, but the suno ones that i just created are more happier/faster/ not ideal to fall asleep to

[granite pawn]

Does anyone know how to get a 6 second intro put into a style that Suno will adhere to

[neat field]

just do the first 15 seconds of the song in the style you want, then use the "Extend" feature from 00:06 in whatever style you want

still a roll of the dice, not sure how often v4.5+ will adhere to the new prompt, haven't done extensive testing since v4 on that

[granite pawn]

It’s for Country songs and i trying to get is radio ready. I will try that. Thank you

[young geode]

What do you mean: do you just want an exact 6 seconds of intro or do you just want a longer intro without an immediate start to the lyrics?

[granite pawn]

Need a 6 second intro. Right now it give. Me a 12 second intro

[young geode]

Dont waste any more credits. Im generating something for you

[granite pawn]

Thank you

[opal orbit]

Putting a tilde and line under works great for me. As well as labeling the seperate sections in brackets. For example;

[VERSE 1]
Lyrics V1
Lyrics V1
~

[VERSE 2]
Lyrics V2
Lyrics V2
~

The tilde will signify there is a pause or "exageration" of some kind. Since its not connected to a word, Suno will do so with the music instead.

Adding that additional space between your verses also helps Suno identify them as seperate.

[echo moat]

with like trance and breaks, what are the best ways to coordinate and signal breakdowns and buildups?

[young geode]

https://suno.com/s/zlh1qyY95dhhQWkL

https://suno.com/s/WDGXSBaL1S7gBbam

Is this closer to what you're looking for?

I used an 80bpm beat upload from Metronomer.
And changed up the prompt to have descending arpeggios, more chillwave/less micro and a bit more sadness.

Style Prompt

Built at an adagissimo pace, with a descending arpeggio broken chord progression of G - Am - Dm. this Chillwave track floats on lush, but sad, Serum synths, airy pads, and gentle,. Microhouse touches the background in minimal way, with a syncopated, lightly reverbed beats with imperfect note placement, creating a dreamy, cold, introspective winter soundscape as the song progresses.```

[young geode]

Thats going to be difficult, as SUNO doesn't listen to length requests or tags.

What you can do is convert your song into 'Alien text' or Morse Code (this is text GPT can understand, but it's unspoken, so it won't be sung).

Then:

[Start]
[Intro: Quick Instrumental]
[Fade in]
[Instrument: type]
[Instrument: type]
[Instrument: type]

[Brief Instrumental]

Converted Song

Then use whatever the best intro you find from this process, cut it at a good point, then turn it into a persona. Then use that Persona to generate the actual Song with the starting Intro I gave you.

[young geode]

Buildups, in order of strength:

[Tension]
[Building]
[Buildup]
[Crescendo]```

Release, in order of strength:
```[Energetic]
[Drop]
[Cadence]
[Resolution]
[Release]```

[granite pawn]

Thanks MasterofSFL, Suno does not understand intros.

[late mason]

Hey does someone know how to prompt to make smthn like this ?
https://www.youtube.com/watch?v=gY_Br1pPtfo&list=RDgY_Br1pPtfo&start_radio=1

[viral pond]

-I downgraded, and had to change payment card, it paid, but now platform no longer thinks im a subscriber

prompt kawaii j-pop anime,. then push the magic button ?

[viral pond]

https://suno.com/s/cz07cBdiuWla1Kwx similar ?

or maybe this https://suno.com/s/wtblTNWyysRFkxz1

[late mason]

ohh you good

[wild egret]

woah, that is next level prompting, i understand half of what you just put there haha. its amazing
lets break it into parts so i can better explain what i feel about this:
https://suno.com/song/467c2dc9-6a7c-47af-a44a-80b3f696618d?sh=zlh1qyY95dhhQWkL
start is great melody wise, beats are not entirely what i want them to be ( i even think i want no beats whatsoever). up until 0:35 is kind of on point besides of the beat. 0:35-6:37 is really awesome to the point i will use this prompt next time for other projects, i loved the vibes of the song, but, still is not exactly what i am looking for, from 6:37 to the end it kinda is what im looking for

https://suno.com/song/c3f6629a-bce0-4d1d-a7a3-a8edaa0c30f2?sh=WDGXSBaL1S7gBbam
00:00-02:20 great, similar to previous song, nice sound and mix, but still not what im looking for (loved it tho)
2:20-2:41, this is it! these vibes, these echoing sounds + reverb, these waves, sublte melody, this is it.
4:12-4:28, very close to what im looking for.

To summarise, its even closer than before, i hope i can make myself clear as of what i'd like it to sound, if im not clear enough, this video has the style of songs im looking for https://www.youtube.com/watch?v=QqCZzViuHqA&t=432s&ab_channel=patient.

btw, is there a tutorial on how to prompt like that?

[pseudo blade]

Where can I find my saved prompts?

[half dragon]

Right there on Create page.

never mind

[pseudo blade]

ah got it, i never knew they moved it

[half dragon]

that's where you save it - but the add is somewhere.... looking..

Ah got it - So on the top there are + Persona, and + Inspo
Click on one of them then hit the "<--" back arrow to see this
And here we can access all "saved" prompts

[pseudo blade]

ahh okay, i thought they relocated it...

and they did

without me knowing

[half dragon]

Yea they moved it, there used to be a + button next to the Create ...then they move those on the top but left this one out

[pseudo blade]

no not next to create, it was at the top of version switching before

[half dragon]

It's appreantly going everywhere 🙂

oh man

It's right there infront of the Create button ..lol

I swear it was on the other side of the Create butotn 🙂

[weak talon]

New here. Is there a simple prompt format/template that works best?

[polar spade]

Simple in the sense of picking some styles that sound right to you, list them down with a comma and space between them and hit create. Thats about the simplest. Even just one genre is fine, but fun to combine things to see what happens.

[young geode]

Try taking out the Minimalist and Canceling out Jazz in the Exclusion prompt. Then add:
Style: anti-climatic, non-escalating, tranquillo.
Mood: Mellow, relaxing, laid-back

This might do the trick.

Note: I have my own method of generating instrumentals, so that also may be creating your unwanted sounds.

If you found the sounds you want within that, I would recommend using the prompts, generating a song, then clipping out the sections you liked, creating a persona, then regenerating a new one with the same prompt under that Persona.

[young geode]

I have a guide in the guide channel, and just posted how to do this.

[young geode]

Simple how?

I'm assuming you mean Simple andConsistent here. As I just told the other user, you can check my third guide in the guide channel and use that as a basis.

It's laid out in a way that walks you through what you need and is easy to use with AI if you don't know what you need.

[pine carbon]

I already sent the examples to support, but If you are interested, I may DM you links for full and than cut of songs. Today a will try to make another one so maybe the weird effect happens again. Thanks for your advices

[polar spade]

I understand the concept, so having a sample would not be needed. Thanks for the consideration though in asking.

As I mentioned, there seemed to be some small changes since about Thursday, so glitches or anomalies not common can be expected as things normalize. So I feel we go a couple of days, and test again is probably in order. If it continues, report as a bug and ask about again for any new ideas.

[pine carbon]

The exact same thing happened to me last Wednesday. That was the first time. But I am in CET time zone, so for someone it actually might be Thursday 😄 Thank you once again a have a great day

[azure prairie]

huh? when did it... why?

[restive cipher]

Hi Suno team,
Since last week I have been unable to generate usable tracks because every output suffers from:
Harsh, piercing highs – cymbals, guitars and even vocal esses create a painful 6–10 kHz whistle.
Digital clipping & crackling – audible crackles and short bursts of distortion even when the prompt asks for “clean” or “low-gain”.
Over-compressed / brick-wall sound – no dynamic range, everything is flattened and sibilant.
Steps to reproduce
Use any prompt containing “progressive rock”, “guitar lead”, or “French baritone vocals”.
Even when explicitly adding “low saturation”, “smooth high-cut at 8 kHz”, “no clipping”, the result is still over-bright and crunchy.
Expected result
Warm, balanced mix with controlled highs and no clipping, as was the case two weeks ago.
Example prompt that triggers the issue
Rock progressif français sombre, épique, 8 min, 70 BPM 6/8 → 140 BPM 4/4, moderate-gain guitars, male baritone soft vocals, smooth high-cut at 8 kHz, warm low-mid EQ, no clipping, vintage tube warmth
Audio evidence
I can upload a 30-second excerpt if needed.
Please let me know if this is a known regression or if a hot-fix is planned.
Thanks for your hard work—hope we can get back to making clean, lush tracks soon.
Best regards,

[bronze helm]

Unfortunately this problem has been prevalent for a while, and can be resolved with audio mixing. Use Audacity to reduce the harsh frequencies.

It'll likely be fixed with a future version of Suno

[bronze helm]

Experimentation! Just try different keywords and ideas

You can also do natural language with v4.5 and v4.5+

[fringe crane]

https://soundcloud.com/770rd/wont-diss-you

Any inspiration for a prompt that allows me to deliver vocals in the tempo of this song? All my attempts so far ends up a bit too melodic and singy.

[azure prairie]

@lost holly so i have been tryna get the situation sorted and i see i have a adblocker enabled, could that be cause?

[echo moat]

So i did a cover of a song but wanted to change the female vocals, well; the cover kept the original singer, so i said ok i'll create a new song. It stopped after the second chorus, completely ignored like 3 or 4 sections of my song. Using the v4.5 model

like every one i do lasts like 3 minutes and the song should be 5 or 6 minutes

[polar spade]

You having a Mandela effect? Its always been to the left of Create. 😜

[restive cipher]

OK thanks

[echo moat]

4.5 is a work of art

[restive cipher]

Here is the clip that interests you Suno music
https://youtu.be/nD2FU8bgznk?si=Y5jYi5b87qf7Bv0C

[echo moat]

i'm hearing clipping in my vocal stem, what's the best way to fix that?

[young geode]

Post the link.

[echo moat]

here's the link to the song but i extracted the vocal stem, not sure how to generate link for just a stem

it's very faint, not even sure i heard it in the full track but can def hear it when it's just the vocal track

[young geode]

Are you using a quality headset or just a regular headset/gaming headset

[echo moat]

old radioshack headphones

[young geode]

what are the timestamps you're hearing it?

[echo moat]

must be my headphones, cause i ddin't hear it over hte loud speakers

i can heawr it nevermind

1:01 - 1:02 i can hear it

it's faint but it's there, will doing a cover on it upscale it and fix those impurities?

[young geode]

If you're hearing it, likely. I cannot hear it already.

You could rip the instrumental stems, then cover that with the lyrics and new vocals.

[echo moat]

what do all these versions mean?

[polar spade]

Youve been clicking the Detect Stems 3 times.

It gives you 2 versions of tracks when you click Detect Stems. Immediately there will 6 stem tracks as one version and Version 2 will be slightly different.

So your Library should have 36 stem tracks now.

[young geode]

lawd help you

Thats a lot of credits

@echo moat
Did you rip just Vocals+Instrumental or all the stems?

[echo moat]

i clicked the ellipses on the right of the track and said get stems

is that not the right way to extract vocals and instrumentals?

[young geode]

Well, far as I'm aware, it gives you two options.

Vocal+Instruments, and then something like a 6 way Stem Split

[half dragon]

You might have missed the small link that says > See Extracted Stems (instead of Extract Again that costs 50 Credits for all stems)

[polar spade]

Sure. That way is fine.

MasterpfSFL was asking if you chose only two stems (vocals + instrumentsl) or let it detect all separate instruments/vocals/fx/backup vocals? The former is 10 credits the latter is 50.

[echo moat]

oh yeah i always do vocals + instruments

[young geode]

Ah, good.

Almost had a heart attack for you

[echo moat]

lol

[polar spade]

OK. Then thats 30 credits for 3 pairs of variations (versions).

[echo moat]

is there a way to get back to that screen or do i have to spend 10 credits everytime

[polar spade]

Once you do it once, if you click again to review all stems together , it will show "See Extracted Stems" below the side you chose to review again without extracting stems again. This seems to remain for about 4 minutes, and then reset.

Of course ALL stems are immediately loaded into your Library to review individually at any time.

[young geode]

@echo moat
Once you get your steams, are you looking for a completely new female vocalists? What are you trying to go for?

[echo moat]

nah i love the voice, just can't upload it if there is warping and clips

[radiant thorn]

Hey y’all- anyone have prompt advice for making an instrumental backing track from an audio demo I’m remixing? I upload an audio of a demo of just bass/guitar/drums, they’re supposed to be just straight up strumming and chords. I try to specify that it’s a backing track for vocals, instrumental, and no solos are deviations from the original form…but every single time, it comes back with a cheesy 90s-era lead guitar noodling through the entire thing

[echo moat]

i extracted a few instrumentals and vocals i will lsiten through them to make sure there are no issues, but is there a recommendation for fixing small anomalies in the outputs?

want to publish my track but can't if there are audio issues

i'll just have to listen to the vocals and instrumental stems and see if i can find any without any issues

[young geode]

The small glitches arent something that's easily fixed, you might have to try doing individual section regens if this doesn't work. But that's costly and I dont know how to do that one well.

[young geode]

Clarification:
You are uploading a demo, which you then want to use as a basis for generating a basic back track, that you later want to add your own vocals to: effectively just to get a longer repeating/continuation/mild deviation of the demo.

However, it's giving you solos and major deviations, specifically 90s solos.

Correct?

[radiant thorn]

Yup. I take the vocals out of the original audio first and then I upload it. Sometimes my prompts will ask for it to beef up the instruments or give them a different flavor, or adding another instrument to the backing track itself...and it's not just 90s solos, it's any instrument that isn't bass and drums that'll launch into a solo for the entire song.

[young geode]

What's the prompt you're using? Also, are you using the Instrumental tab or the custom lyric?

[radiant thorn]

I used the instrumental tab at first but I eventually started assuming that it took that as a cue to make a "traditional" instrumental where's a main instrument playing a melody for the entire song. So I switched to the custom lyric tab, left a space and didn't put any lyrics there. Didn’t work either

I can't find the prompt at the moment but it was something I made with the style-creator ChatGPT thing

[polar spade]

What's th reason to prompt anything when cloning an audio upload?

[young geode]

Will it repeat it without prompting?
Ive never just uploaded something and hit gen

[polar spade]

I try to just give suggestions if under two steps, so, ill only suggest making sure when you cover your audio upload that the audio slider is 98%, weirdness at 1% and Style slider at well 1% as well seem about right, to just clone/cover your audio as exact possible. It may not get the instruments exact, but the riffs and melody should be OK. 🤞

Everything else is blank.

Well.... at least try one.

[radiant thorn]

I'll give it a shot, thanks! What about if I want to add an extra instrument like brass horns to the arrangement? That's when it usually takes it as a cue to create an entire brass melody noodling through the entire song lol

[polar spade]

Oh now you're getting technical. Haha.

If you add any prompts you start giving Suno free reign to be generative. For now, we cant add instrument to instrument stems. Wait... hold on... let me check that...

Nope... cant do that yet.

I dont know how to blend in a brass instrumentation or ensemble exactly where you'd like among covering an audio upload, without it just noodling about. Prompts are styles, so you would need to start sliding the style slider up and open the gates for generative noodling.

[polar spade]

You might shed more light onto this. I've got to head out to work. @radiant thorn Good luck, hope you find the groove.

[shadow plaza]

Anyone get error elevated usage when hit generate

[young geode]

So the first thing you ought do is do as @polar spade said, which is place the Audio Slider to at least 90%, drop Weirdness down to it's lowest percentage and Style to it's lowest as well and see how that affects things.

Then, if you aren't getting what you're looking for, you can try and do the following:

Using the custom lyric mode
First, create a basic Intro-Verse-Verse-Chorus-Verse-Chorus-Outro format. Then as the lyrics, either one of two things:

1. write some repeating lyrics for all the sections and translate them into Braille.

2. add in a repeating pattern of morse code.

Then, using the same settings, hit generate.

If that doesnt work, you can slowly add in metatags and prompts, upping the style influence to 75% or higher.

Doing the lyrics as I instructed causes SUNO to recognize a pattern and start repeating patterns: especially with metatags like [Melody]/[Riff] ect.

The other thing to do @radiant thorn, is to utilize the Exclusion prompt area.
1970s, 1980s, 1990s, 2000s, 2010s, 2020s, 1970s solos, 1980s solos, 1990s solos, 2000s solos, 2010s solos, 2020s solos, Metal Solos, Rock Solos Ect. That will also help keep SUNO from going off the rails most of the time.

[slim elm]

Suno website is broken

Same issues on Bing, google, chrome and safari

[half dragon]

I use Chrome, both on pc and mobile - I just checked and it loaded fine for me. Maybe local issue? try to clear cache, cookies and reload?

[slim elm]

Thank you @half dragon I tried but still persists I’m on my iPhone

Never had this issues before and I never used pc

[radiant thorn]

Hot damn, this worked pretty well. Thank you! Now all I need to do is figure out a) how to get it to remix an audio backing track in a different genre/style (ie a piano backing track demo into a folky guitar backing track)...and b) how to get it to do the entire damn song without cutting it short, of course 😂 but thank you, this worked very well.

[radiant thorn]

Gotta admit- I couldn't tell if 1) and 2) were jokes or not 😂 but I’ll try this stuff out, thanks!

[slim elm]

I’ll try to shut my phone off

[radiant thorn]

“Repeat without prompting” as in making the generation a carbon-copy of the original audio upload I’m remixing?

[polar spade]

When they allow adding an instrumental to an instrumental stem, then we can stem out an instrumental and adding a brass ensemble persona or prompt for brass. But until then.... 🤷‍♂️

[slim elm]

Same nothing have been fixed. Everything works fine except Suno website

[radiant thorn]

I was talking more about "this backing track except beefed up with brass/fuzz guitar/etc"...I know that they don't allow you to churn out the exact same audio without any alterations + an extra instrument. I fully expect the generation to be different than the one I'm uploading

[young geode]

Not jokes.

it works, as they are readable, legible languages that SUNO/GPT understands, but they are non-spoken, so SUNO will not sing them. However the text does seem to impart sentiment and SUNO does try to play them, and ends up having repeating patterns within it's token prediction.

[radiant thorn]

This is pretty cool. I'm gonna try it. But if it goes under verse chorus verse etc, do I just put the original lyrics I'm gonna sing later? What if I don't have any lyrics or melodies planned yet?

[hollow hare]

Just catching the backlog on your discussion but I might have an inkling of what is going on.

Do you have access to a VPN? (E.g. Opera browser, Proton VPN)
If so, can you let me know if using that fixes the issue. Of it does, that would mean it is likely a DNS issue (one min while I get the link)

[radiant thorn]

What’s the audio/weirdness/style settings for when I try the braille thing?

[hollow hare]

Have you tried using a VPN like I asked?
Anyways, don't need to do that now, got the link on what to do for changing your DNS in case that is the issue: #🗣️┃chill-chat message

I am unsure if it is the cause but it seems likely because someone had a similar issue a few days ago and...well, they didn't tell me if it fixed it or not but I take their lack of response as being fixed

[slim elm]

Okay thanks I’ll try that 😍

[hollow hare]

Please let me know how it goes

[young geode]

If you have no prompt
Weirdness: 1%
Style: 1%
Audio:95-98%

If you have a prompt:
Weirdness: 1%
Style: 75%
Audio:95-98%

[young geode]

What do you mean?

These are just place holders to keep SUNO on track with it's generation: IE Intro Music, Verse Music, Chorus Music, ect, ect. If SUNO does end up creating the backtrack you want, you can then add the Lyrics as you want them, in whatever way you want them (though this will end up changing the song upon a new generation/cover).

Are you trying to generate your own AI singer over this backtrack or you want to just generate a backtrack that you or someone else will be singing over?

[radiant thorn]

Gotcha. And sorry, I should’ve clarified- this is a backtrack for me to sing over later (after I import it into Logic and record my vocals)

[young geode]

Ok, then don't worry about it, unless you want to introduce some varied textures into the song. The point of doing it as I gave you is to help create a pattern (Repeating lyrics and patterns, without any vocals), so that the instruments involved will have a high probability of repeating without much deviation.

[radiant thorn]

Thanks again. The braille method worked pretty well my first go-around!

[young geode]

Did it repeat without much variation?

[radiant thorn]

There were no solos, but the sound was a little too different for my liking. Lots of weird instruments added into there that I didn't request even after I set the style/weirdness to low percentage...but it's definitely a start.

The high audio percentage/lowest style and weirdness setting/no style prompts or lyrics definitely worked the best it seems

[young geode]

then stick with that. You might get milage out of metatags if you're getting results with no prompts.

[Instrument: Type/Model, play method] works.

tags like [Riff] [Melody], ect also work.

terms like Arpeggio, Broken Chord, ect also have effects.

so something like

[Intro]
[Instrument: Type/Model, play method]
[Instrument: Type/Model, play method]
[Instrument: Type/Model, play method]

[Melody Develops]

[Main Melody]
[Verse: Repeating Melodic Chord]

otherwise, good luck. Generation off only the audio upload is not something I've fully tested yet.

[tired basin]

morning / afternoon / evening everyone. looking for suggestions for "de-saturating" a track that comes off as mixed too loudly / overcooked. vocals sound like they're hitting a compressor really hard especially in the high energy moments of course. things i have tried: cover with 100% audio, 100% styles, 0% weirdness and style intructions like "desaturate vocals, smooth out vocals" - i've had some success with this in the past although it could definitely just be a fluke (suno's internals and how it was trained is mysterious and important work)

i probably could go back a couple generations and roll the dice but maybe someone's found other techniques

[azure prairie]

Mf **** i got too excited i thought my songs were published Damit!. any answers yet why my songs dont publish? i even have consent byt EA and Telltale games that i can make a song about their game

[shy heath]

Hi all. I have a question about lyrics. When you have a word that can be pronounced differently and mean different things, how do you get it to sing the way you want it to? For example, the word "tear" can be pronounced differently and have different meanings - like pronouncing it like 'tier' but meaning a tear drop, or rhyming with 'hair' to mean a rip. If I want it to sound like 'tier', do I need to change the lyric to 'tier' because I can generate a song several times and it's basically 50/50 on the pronunciation. I was just curious how other people handle this situation. Thanks!

[polar spade]

Most times I can get away with a phonetically spelling. So, specifically , I use teer and tair.

Depending on the song sometime I need Toke-yo, 2 syllables, or To-kee-yo , 3 syllables.

[tired basin]

answering my own question -- this seems to have helped quite a bit

[tired basin]

sometimes spelling phonetically will help, and then eventually backfire and come out weird! the most annoying example i got was "psychiatric" - also had some hilariously bizarre pronunciations for "loneliness"

[echo moat]

@young geode can you listen to this, it's doing it again where it's like wahing in and out, very strange, in the first minute it does it multiple times

https://suno.com/s/3gK7GQrpFUIMyufB

[quick trellis]

Pleased to report FenomenonX added this helpful video:
How to Add Instrumentals to Vocals Using Suno 4 5 Plus New Update - https://www.youtube.com/watch?v=iAYOL1O4gSE

[echo moat]

it's not doing it on the original song tho

[shy heath]

Ahhh okay, thanks. It's annoying when I have a song and everything is great but that one word is sung wrong lol

[azure prairie]

i do [british] or [english spanish male] then the word is pronounced in that certail language but english but if you asking for example like tear and tear, it will have to make sense with your standing sentence

[polar spade]

You can add vocals or instrumental to an upload though. So that could be beneficial.

[echo moat]

@polar spade you mean the 98% trick?

[polar spade]

If you upload an audio. Then go to your Library and click menu of that audio, then under remix/edit , youll find Add Vocal and Add instrumental.

[echo moat]

oooo

if i do add instrumental, doe sit regenerate the existing instrumental or does it change it?

[hard beacon]

interesting.. i don'T have these 2 buttons 🙂

[young geode]

I hear it, I don't know why it's doing that.

Is this a generation from the original song? I dont see any prompt.

[polar spade]

Did you tip the bouncer?

[echo moat]

i have no idea i've covered and remastered so many times on this one song i can't keep track if it's the original or not

just wanted the audio to extract so i could master it beforei published it

seems almost like compression clamping is what it sounds like

[young geode]

Might be time to start a fresh one, it's likely picking up on the subtle glitches here and there and then thinking it's supposed to follow that.

[echo moat]

actually nevermind this was an original cause the persona was taking over the other one

[shy heath]

Gotcha. I was hoping that the system would analyze the lyrics first, and see that "here" is the next rhyming word, so it would pronounce it 'tier' for it to make sense.

[hard beacon]

@polar spade : don't know what you mean with bouncer.. but.. nervermind.. i don't need that function 😄

[polar spade]

You have to look directly on the upload, not the cover of the upload.

[hard beacon]

aaaaaaaaaaahh

get it 🙂 my bad..

[polar spade]

Oh ok. 💁‍♂️ it should be there if you need then.

[hard beacon]

what i need is a 4.5+ remaster function.. it makes no sense to create a 4.5+ track and remaster it to 4.5 ...

[echo moat]

i get this weird feeling that when suno reaches a certain level, it will become unafforadble to the average user

[ebon sand]

that would be kinda counterproductive to make money from mass appeal

[shadow plaza]

can i make a song from my Pure instrumental trakcs ? i dont want to change the instrumental in any way just add singing on top of it to match the track ?

[half dragon]

There's a new feature - Add Vocals
Give that a try on your uploaded instrumental track.

[shadow plaza]

where is that option cant see it

[half dragon]

On your uploaded track - 3 dots menu > Remix/Edit > Add Vocals

[shadow plaza]

saw it thanks

[viral pond]

https://suno.com/persona/d2ef4cde-65e0-47bf-8a7c-915e716b1735 why my persona don't work correctly 🤣🤣

[echo moat]

is there any type of mastering done onthe back end when songs are generated?

[late imp]

I'm trying to do a Duet Song.

But Suno doesn't seem to understand I want it to have two different voices/singers

[vestal sparrow]

I'm not sure because it's been a while since I did mine, but generally adding something like "2 female vocalists" and genning a bunch of times eventually produced something. You could also try tagging it in the lyrics but it didn't work for me consistently

I could try to pull an old example

https://suno.com/s/EzjKDdBSDeJfTUVo lyrics weren't mine, and this song is NOT serious or good, but he prompts are: Three japanese Female vocalist, Alternative Rock and Hip-Hop/rap, no autotune, japanese elements

These are also not my prompts, I just did the final tweaking to everything

[late imp]

So far I have it like this.

(Willow — Lead)

(Grimm — Counter):

Grimm = dramatic mid-high tenor—ringmaster flair, vibrato on long vowels, dry pan-center.

Willow = deep, icy baritone—smooth but menacing

dark symphonic-industrial rock

Duet with theatrical circus overtones

Two different singers

Ok I'll try "Two Male Vocalists" then

[vestal sparrow]

but yeah, when I do lyrics I often add notes in them. I try not to use it too much because the ai will read the notes as the actual lyrics if you do. The other technique I use alot now is puncutation.

sorry it's not a good answer, it would be cool to select two personas though for a duet

or more

[late imp]

Yeah def

I'll have to keep trying but thank you

[raven turtle]

I’m
Wondering the best methodology at doing a proper “remix”. As in same vocal, but different instrumental vibe. Would the easiest way be to do the remix as instrumental, stem out the original and drop the vocal stem over the remix in a DaW?

[young geode]

If you have pro, you could take youre song, strip the vocal stems, make them a persona, then download the vocal stems, reupload the stems and add the instrumental through the 'add instrumental' feature, and likely it'll reproduce the vocals with new music.

Ive yet to try this, but it should work. Otherwise, you can just drop the vocal stem over an original remix.

[raven turtle]

Great idea. I have pro so I’ll give it go. I’ll report on how it fares!

[echo moat]

why does it keep saying my name is galvanizingbongo, i never agreed to that lol

can i change that on the website on pc?

went to edit profile, changed handle, but it changed it right back

[soft sparrow]

the notification/activity screen always displaying - been like this for two days - I've cleared cache - logged out and in - anyone know how to solve this? I contacted support they have'nt returned my email yet

[rough reef]

All the songs generated mix the lyrics with the prompts. When are you going to fix this all the time? It's been happening a lot lately, affecting my projects and losing credits.

[rough reef]

After much testing: Style is the one that adds style to the lyrics of the song and even cuts songs in half abruptly

[small bridge]

okay question is it something I did/put in wrong when regenerating? or is Suno just being weird right now?
Every time I try and "fix" a part of the lyrics its starting out soft, and correct, then halfway through the redo clip it goes loud and tinny, and then like autotune.
https://suno.com/s/lEWGC5R6OBQQiMRa
https://suno.com/s/Yh5JgxbyLyhcUYRV
I disliked the clips so that Suno can tell this is wrong, btu it's sad cause the first part is actually semi decent and then it goes south.

[buoyant gale]

I think Version 4.5+ loves Violin a lot. I prompt no band, classical guitar only, and I use creative boost but most of the time V4.5 Plus wants to add violin and percussion lol but it's still awesome though. So i got used to but I still prefer using the older versions.

The stem separation I don't use it, it cost lots of credits when we can use Fadr for stem separations for free

[small bridge]

ummm... k

Suno is having a melt down cool

https://tenor.com/view/everything's-fine-gif-14986607915005937648

[raven turtle]

A fair point regarding Fadr

[restive ocean]

I wonder.. is there like a way to control the vocals when generating? I generated over 50 songs and not a single female voice is close to what I need to fit the tone and theme of the song

[azure prairie]

have you tried saying [female vocal] above the section you want a song? i manage to do male and female in one song @restive ocean

if you do so i recommend the square brackets [] cause AI dont use the words in them unlike ().

[thin kite]

Just an idea. SUNO AI users who download more SUNO STEMS have a bit of a problem with downloaded files cluttering up their personal library. It would be great if all files from "STEMS" could be stored in a separate directory. The song storage has been separated song storage divided by "STEMS", best regards, MM * Len taký nápad. Užívatelia SUNO AI, ktorý viac sťahujú SUNO STEMS majú trochu problém so stiahnutými súbormi, ktoré zapratávajú osobnú knižnicu. Bolo by super, keby sa všetky súbory zo "STEMS" mohli ukladať do samostatného adresára. Oddelil sa úložisko piesní úložisko piesní rozdelené podľa "STEMS", s pozdravom, MM

[restive ocean]

No no I do get female voices but like not what I want them i prompt it soft female voice and it gives me screaming woman 😅

[ashen basalt]

I am wondering if you make a replacement for a song but the volume of the replacement is lower then the rest of the song is it possible to fix that in editor?

[azure prairie]

[soft female vocal voice] try using that

or [slow soft female vocal] you also add any instrumental requirments inside of the brackets

[half dragon]

Different than folders but there's a filter that you could tick on and off to show Stems with other files in your Workspace or just shows Stems alone.
Essentially the same.

[half dragon]

Unfortunately not right now. Either regen or fix it offline.

[sleek stump]

Look at any song here for https://suno.com/@cinqueenmusic She has very detailed descriptions and she commonly gets very soft and whispered vocals. You may learn a lot from taking a look at her tracks.

[polar spade]

In the Edit UI of a paid subscription, you can click on the top banner of a segment then click the 3-dot menu below the segment to change volume of just that segment.

[near flume]

Try prompting like this: Female voice with soft, emotionally fragile tone, hushed and breathy delivery. Slight Tennessee/Southern accent. Intimate mic proximity, capturing every vocal tremble. Wide dynamic range—from whisper-like vulnerability to sudden emotional swells. Often uses sustained notes with a controlled vibrato and occasional vocal breaks. Emotive, confessional phrasing, like she's reading from a diary.

[restive ocean]

Ohhh wow, I didn't know we could prompt like that aaaa, thought we just put like instruments and such!

[near flume]

I've done songs only using that prompt, no prompts for instruments. Suno does a good job filling that part in.

[restive ocean]

Thank you so much! I'm trying that out now!

Also, I'm trying to figure out how personas work. Do I extract stem vocals first? And then what?

[ashen basalt]

oohhh thank you so much!

[polar spade]

You can make a persona of just about anything. In the output menus, click Create / Make personas. Thats about it. Once on the persona page , fill in all the blanks. Keep public or turn public off and click Save.

[sleek stump]

I would recommend getting familiar with Suno prompting and how it works, and what to expect before jumping straight to Persona, thats like skipping Music 101 and crashing music 301.

[near flume]

Personally I don't use Personas since they're so buggy. What I do is if I find a vocal prompt that turns out what I want for a certain style of song I add it to a collection of other vocal prompts. My collection has over 30 voices, so far.

[polar spade]

What's a vocal prompt? Is that specific to just in the lyric section to get the same vocalist every time?

[fair adder]

hello, i don't know if i rather post this into the suggestion or the bug channel, but i just noticed that when you cover an existing song, you can't use the inspo feature. it adds the playlist, but it completely removes the audio i want to cover

[lost holly]

Well those are kind of exclusive features. You either can have an inspiration or cover a song, but not both. Makes sence to me.

[jovial bough]

Anyone have a sure-fire (LOL) way to prompt our friend Suno to create a scaled back single instrument (say, piano or acoustic guitar) cover of a fully produced song? You know, like an "unplugged" version. I've had some luck, but many times it just refuses, sometimes even throwing in instruments that weren't in the original. This one has been the most stubborn. Note the previously non-existent violin. https://suno.com/s/MFDXBZ9zUgACL02Q

[sleek stump]

They (Suno) changed the workflow for +Inspo, it was never intended to be used for covers or personas, so they fixed that a couple days ago. From what I have come to understand.

[gloomy shore]

Why the ai kept repeating the chorus atleast a 2nd time? No matter how i labeled it and told it not to repeat but it just kept repeating it. 😮‍💨

[half dragon]

It's very difficult to have Suno to gen just one instrument - might be impossible to prompt for.
Maybe build the whole song per section? using Extend feature.

[neat field]

i already did Stems from the library, any way to now see those in the editor?

[sleek stump]

YOu can prompt for a single instrument, Ive done both Acoustic guitar and piano on a first gen, but I do not believe you can get that from a cover unless it is already a solo performance. Suno takes a heavy direction from the original song, even if you put the audio influence to zero. YOur best bet woul;d be to split off the vocals and then try and gen a new track with Add Instrumental...

[hard beacon]

Stems are for external DAWs

[sleek stump]

Not anymore

[neat field]

it's a shame the built in DAW isn't "there" yet for stems

[sleek stump]

Something is coming, Suno acquired WaveTrack, and are likely working on integration.

[neat field]

hopefully before my sub expires LOL

[sleek stump]

Dont expect anything until the end of the year or first of next

[hard beacon]

well.. Stem Export in FL Studio is good... but directly stem export in the tool (Suno) is better, because they can seperate it directly from the output 🙂

[sleek stump]

Suno stems are currently the best in the market.

[hard beacon]

because they are exported from the original output..

When you export them in the DAW, there are frequency overlaps... otherwise it works quite well 🙂

Suno creates the stems from the algorithm of the output... DAWs try to break down the song using frequencies... If you want Suno stems, you should use the stem function 🙂

At the end of the day, that makes sense 🙂 However, it would be far-fetched to say that Suno's stem function is the best on the market... that's like comparing apples and oranges 🙂

[sleek stump]

Its not far fetched, all the other splitters use Spectrographic algorithms, which is why they all inherently have bleed. Even lalala.ai has some bleed, and they were the best (and most expensive).

[hard beacon]

I think we're talking at cross purposes... It makes sense to use the stem separator, which separates the output directly... Of course, external tools use other functions 🙂

[polar spade]

Keep in mind Suno finds the nuances best suited for the genres but if you add genres (more than one) you increase the combinations by millions. So if you have a few prompts, adding folk, for instance, may not give you anything close to sounding folk (you hope not) but might give just a single instrument which folks music is often associated with.

The other version is to make a persona of what you have, then cover the same with just , say, folk, or in-the-round, or unplugged, acoustic introspective, improvisational, as a prompt with Audio and Style sliders about 70% , both.

Not sure-fire of course. 😊

[lilac linden]

I've NEVER been able to get to 100% of ANYTHING I want with Suno. It seems like If I start out with the right prompts and slider moves and I THINK I know what's going to come out, it does. But just sometimes! And it get further away from me with repeated remixing. If it's not happening, I stop at 4-5 tops!

[polar spade]

Thats about right. There's some method to its madness as a generative AI to constantly vary something with each output, but then that variance can be a 1% deviation or an 80% deviation. So we think some things "repeat" where others things "aren't even close" , but as for suno , they are all within the 80 some million parameters based on the prompts we gave it.

[lilac linden]

If I get something that's close that I like, I'M DONE! It's nothing like playing your own instruments and recording in a DAW.

[polar spade]

Well, now we can just play our own instruments, upload and it will give us a near copy, as well as minor variations and pitch correction for a nice similar sounding vocals. Its all good.

[lilac linden]

I'm waiting for accurate reproduction, continuity and consistency, personas based on my own voice (that do NOT effect the music!) and individual personas for particular instruments.

I'm mostly doing a "Glamor Shots" clean up of a ton of my son's songwriter demos. And when I set the audio influence to 97-100 it generally works the way I want it to, without adding too much that isn't already there.

Every once in awhile it does him ad a female voice or adds production I wouldn't put in.

[polar spade]

I like hearing my voice as if I were 25 again. Haha

[lilac linden]

That I haven't tried yet. Anything me!

[polar spade]

I should do a songs-from-the-80s retro album and find some old pic of me. Try to write in the style of the 70s or 80s. Any decade really. A period piece album. Heck im still trying to get it to harmonize like the Beach Boys. One day.

[lilac linden]

It's entirely possible now!

[polar spade]

Its hard to write badly as I did back then. Or as some songs were back then. Haha

[lilac linden]

You can do it!

[polar spade]

Its like Tchaikovsky trying to sound like a high school marching band.

Yea, its probable to get lucky using only Suno, but now its too easy to just compose and record the song old school and upload it. Going back to how we did it in the good old days, 3 years ago.

[neat field]

Tchaikovsky marching band is going in my back pocket of ridiculously hilarious prompts

[raven turtle]

Reporting back to @polar spade that the stem split, make persona and “add instrumental” to it as a remix works like a charm!
I took a Bowie Berlin era type track, stripped out the vocal, and added a more cyberpunk music bed behind the vocal! Brilliant!
I found I didn’t need to reupload the vocal. All I had to do was apply the “add instrumental “ to the source vocal stem already there.
One thing of note is it only seems to add instruments the length of the source vocal stem. So if it’s a 1:38 vocal stem, then the m music added is that length too.

[coral imp]

Hey guys. I'm trying to get the AI to have a female vocalist sing 2 choruses alone, the 3rd chorus with the crowd singing alone, then the female vocalist and crowd sing together on the 4th.

Currently I have the crowd tagged as "crowd choir". I'm getting the crowd to sing where I haven't prompted them to. I was wondering if there was a way to force it at specific points of the lyrics.

[silk axle]

Hello have you disabled 4.5+ preview for free users today?

[sleek stump]

As far as most of us know, 4.5+ was never availab le as a preview. V4 is for first time signups for 10 generations, but it has never been 4.5 or 4.5+.

If you were able to use 4.5+ as a free user, that would have been a bug that likely got fixed.

[formal gale]

so how do you get certain sections of lyrics to be sung by different voices? sometimes they either don't sound different or it speaks the text in parenthesis or even if I - beside like chorus etc

I love suno and am having so much fun but wanna advance my stuff

when I put this in suno --- Verse 4 – Ronni (soft to fierce) ---- it says ronni soft tro fierce

[magic shale]

Hmm, I had some words in my prompts which destroyed big time song after song and it took me a while to figure that out!
So watch out with some ChatGpt prompts, cause some attributes for vocals, singers or instruments are misinterpreted by the model and will change the sound or introduce hallucinations ! Attributes for the vocals only will be applied to the whole song or will bring up new unwanted instruments!
The word "amp with chorus" for the guitar would change the whole sound of song!. It just happened.
I lost 2 hours with that 🙂
It was a new prompt I designed and 2-3 words in it spoiled everything.

[formal gale]

any idea how I make is so I can do a two girl duet?

[coral imp]

It seems we all pretty much have the same question lol, just slightly different

[wispy laurel]

It would be really lovely if this thing would follow instructions 😂
Trying to create using only piano, drums and guitar, clearly stated in the style instructions. But the AI compulsively inserts a bass guitar too, even when I tried directly telling it "no bass" with MANY different variations of wording
Exclude Styles also doesn't stop this from happening.
Now. In theory I can just split the stems, throw away the bassline and combine the rest back together. But then I'm left with a noticeable absence where the bassline used to be, because the AI absolutely refuses to have the piano play that part instead of an actual bass.
Any clues how I can get this thing to not add instruments that I specifically told it not to use?

[sleek stump]

You need to put this in brackets [Verse 4 – Ronni, soft to fierce] for example.. anything in brackets are meta tags and used for direction, anything not in brackets likely will just be sung.

[formal gale]

is there a way to get suno to make a duet with two females? even when I say femal vocals only it throws a dude in there

[ashen carbon]

I'm trying to make a non-persona female vocalist track (in Spanish) and every track, no matter how I prompt it sounds exactly like one of my main Personas that I always use. Am I going crazy or is Suno defaulting to what I have generated a lot of already? And if so is there any way to reset or force a fresh take?

additional note: the lyrics I'm trying to use I already generated a whole lotta times with the Persona that I now feel that I keep getting even though I unselected that Persona...

Signed,
Frustrated

[wispy laurel]

It does seem to pick up influences from past creations, I did a lot of rock/metal stuff using a specific persona but now am doing a lot lighter stuff with either no persona or a completely different one, and I had a hell of a time getting it to not sing like the old one
Try clearing your browser cache, that seems like it helped a little bit, although I might just be imagining it

[polar spade]

Excellent. Thanks for letting me know, mate. 🤘

[polar spade]

And im not sure the 4.0 promo is still given as well. I tested by creating a free account and there was no trial given. I dont see it listed in Basic level perks. I could be very wrong. Is there still a promo popping up somewhere?

[ashen carbon]

Thanks for validating my experience.

[radiant chasm]

Heyas. I'm trying to use sticks or claves in some Australian Aboriginal inspired songs I'm writing. I can get a reasonable didgeridoo (its not great) but I can't get the sticks, no matter how I try and they really are fundamental to the genre. I even uploaded a piece of just sticks and tried remixing it but it didn't work. In fact the automatic summary said it was a triangle so I tried specifying those too but it didn't work either. Is anyone else able to get songs with claves or sticks to work?

[radiant chasm]

I thought it was just for the beta period

[ebon sand]

There is no such data sent which would adjust future generations based on past generations, it's coincidence

[shadow plaza]

What Styles 4.5+ is good at?

[agile cairn]

Is it possible to write a prompt that removes all effects on a vocal stem in 4,5+?

[radiant chasm]

I think the stems are always generated from the resulting song rather than being generated first and the result mixed from them. So, you would need to make a completely dry song. I'm not sure how you could do that.

[viral holly]

I have a song that I want to put vocals on and it is about 5 minutes long. When I try to create a song with it, it will only go to about 4 minutes in and not complete the full song. I'm using 4.5 beta+. Are there any prompts to determine song length?

[fair adder]

well that's a shame. i was hoping to finally continue a stuck/forever unfinished song by trying this feature. this kinda sucks...

[radiant chasm]

When I've experimented with it myself, I've only got 4 minute songs too. I think its a fixed limitation that they haven't improved yet. You can always extend the song though, I think. There is no way to really control the song length.

[worldly niche]

Hi, the music I produce almost always breaks. I have to extend a song every time, and I spend double the credits and time, and often the result isn't even the same as before.
I thought it was a problem with version 4.5+, as it's still in beta. However, 4.5 causes exactly the same error. Are the technicians already aware of this problem?

I never use Persona, I make a song from scratch, sometimes I do remixes. And they don't stop at 8 minutes, which would be the limit, but at any minute.

[polar spade]

An Inspire is considered a "new creation" (Create panel) and Covers are remix edits (Remix panel). So even the panel will change, depending which you choose last.

[brisk flint]

To extend the length of the song a little, prompts such as the following help me: First, definitely add an intro and an outro. Then define up to two breaks during the song. And interludes or ‘instrumental parts’ – this way, Suno has to generate more ‘content’ that, depending on the length of the lyrics, doesn't fit into the 4-minute grid 😉 – But of course, this has to fit the genre, but in my productions, I can often – not always! – artificially inflate the length this way.

[young geode]

[Start]
[Intro]

[Outro]
[End]

Every song you make needs to have this setup, as it allows SUNO to properly start and end a song, otherwise you get immediate starts and random cut offs: even if you have more within your song.

Next, you need to check your prompt. Poorly written, sparse and uninformative prompts create bland, generic songs and have a tendency to cut off/glitch: SUNO is a token based AI, and like all token based AIs it benefits from focus, IE setting the parameters it needs to think in and cutting off pathways.

This means using both the Style prompt and the Exclusion prompt areas.

Lastly, Metataging. Regardless of the song/genre/style you are going for, introducing instruments/vocals/moods/instructions, before and after each section, further helps refine SUNO's token prediction.

Like what's occurring is SUNO doesn't know how to continue the song past a certain point, and so it either forces it to end or it cuts. If you do all the above, it should heavily reduce the cut off/end phenomenon you're experiencing.

[graceful condor]

Is it normal (but what is normal) to do a cover of a song that is 7:27 just to have cut down to 4:24?

[worldly niche]

Thanks so much for the directions. I'll try following them to the letter and see what happens. Thanks again.

[worldly niche]

In fact, it happens to me even with the cover. The original song doesn't cut out, but the cover does. Furthermore, even when I do an extended version, the extended piece suddenly cuts out as well. Absurd. Unfortunately, all this also costs a lot of credits. So, to have a finished song, you'd be looking at up to 100 credits for a complete song, after all the changes due to the cutout.

[polar spade]

Yes. Its not an easy guess how long a song will be, Suno always uses the original as a "Seed" to produce the cover for the new output. In general, the length of the original determines the length of the space reserved for the seed data.

When limits were at 4 minutes, it was just hard to get a 4 minute Cover. Now with limits at 8 minutes, 4 minute covers of 4 minute songs are pretty simple. But if the song is already about 8 minutes long, a cover longer than 3 or 4 minutes might be pushing the limit.

Try a different way, with using Inspire instead.

Try creating a new Playlist and ONLY include your one song you want to cover -- then click Inspire button found in the Playlist.. Add your lyrics. Slide Audio to 99% and assuming you dont want new prompts, Style and Weird sliders to 1%. Run as a 4.5 ver.

See if you get a variety of lengths longer than 4 minutes at least.

I just test ran a 5 minutes song with just inspire and got 4:44 and 5:50. So its promising.

[graceful condor]

Thank you @polar spade I'll give this a try.

[polar spade]

Dont shoot me if it doesnt work better than a cover, for you. 😁

[young geode]

https://suno.com/s/U24KwPUFgqWKIEqx

Is that what you're trying for?

[young geode]

From what I have seen, this is a small glitch from the update.

One user has determined its linked to the style influence scale being high. Make the changes and reduce the scale to maybe sub-40%.

Otherwise you may just need to place that project on hold for now.

[young geode]

https://suno.com/s/E4E7azztLkfUMnQU

Are you looking for something like this? (ignore the style)

[soft wadi]

getting frustrated with Sjuno. I want a low female voice but suno is not listening to the prompt. HOw to I get suno to slisten to the prompt

[near flume]

can't get the simplest percussion elements to work, claves, cowbell, vibraslap, tambourine

[near flume]

vocal prompts are done in the styles box, but there are more details assigned to the voice than what most people use. Example: Male vocalist with an emotionally strained, soaring tenor voice, expressive and raw. Blend of soft, introspective tone in verses and powerful, anguished belting in the choruses. Vocal delivery is clean but urgent, avoiding growls; intense emotional vulnerability. Harmonies layered with a haunting, melancholic edge. Evokes frustration, isolation, and inner conflict.

[polar spade]

And those give you the same consistent character (artist) vocals each time to use, for example, in creating an album of songs by the same artist?

[near flume]

Closer than what I'm getting from Persona, especially if I use it cross genre. I released an album (female singer) about a month ago and the vocals are almost identical in all tracks.

[polar spade]

It may be very well what Suno does internally when making personas, but I assume its more of creating code to reproduce the vocal style.

Yea, if it does better than a persona for you, thats cool. I may play with it more, but i doubt I know which adjective is more accurate than another and fear a lot of experimenting. Kind of how it was in ver 3.5 before personas were introduced.

Thanks for the insight. Always helpful to get other POVs.

[young geode]

How long are your vocal prompts?

My standard is Vocals: Sex, Vocal Type, Range, qualities.

Then if I am targeting a specific kind of vocalist, I will add in the era, Genre and specifics to a singer: Bark-like aggression, Raspy and never sung.

[jovial bough]

I have decent luck creating a singer songwriter style track with only a single instrument accompaniment (piano/acoustic guitar) from scratch. That's not a problem. My problem is getting a cover of one of my fully produced tracks to carve it down to one instrument. Sometimes it works, most times it does not. I would consider a 50% success rate "sure fire". LOL. Instead, it's more like 0-10%.

[young geode]

Are these tracks your own, as in not made by Suno?

If you have access to the full file, can't you just enter an editor, isolate that, then upload that?

Regardless, I looked at your prompt and it's only 'Piano Only Accompaniment'. If that's all you're using as a prompt, then you'll run into this problem: as whatever is part of the track will be picked up on by SUNO, unless you gate it's token prediction.

You might need to do a prompt that's something like:

Description: Cover Song. Genre: Instrument Piece, Subgenre: Instrument Solo, Instrument Accompaniment. Instrument: Instrument Type, Model, 'Instrument' Only, scale type, play pattern/method. Vocals: sex, type, range, qualities.

Narrative: A cover song done as a single instrument piece and vocals. Ect, ect.

Then in the Exclusion Prompt:
(List of instruments that exist in the original), (list of instruments that exist in the original genre), (full band, full orchestra.

Then you'd use meta tags to further weight the token predictions. If you have the song itself, paste it into Google Docs/Word:

[Start]
[Intro]

Song

[Outro]
[End]

And then use the [Instrument: Type, model, play pattern] meta tag in conjunction with your verse, chorus, solos, ect.

This will probably yield a more consistent generation. However, this is one of the few areas of testing I have yet to do, so I can't tell you how much more consistent it'll be.

[modest juniper]

whats the use of extending a song when it goes off on some other genre and vocalist. It seems that used to keep the vocals at one time. Being my personal vocals.

[near flume]

Typically 300-400 characters.

[young geode]

What prompt, genres and settings for scales do you have?

[young geode]

And you find this is produces repeatable results, consistently?

[near flume]

yes

[modest juniper]

I use my own recorded song. its all original and I dont want it to go off on some crazy ride. Not sure what your talking about

[young geode]

You mind providing me an example, as you would write it?

I test these things from time to time and then add them into guides. I've found my standard method works fine, but if there's a way to get a higher level of fidelity, then I can fold that in.

[young geode]

Ok.

Say your song is Country, ala Johnny Cash. You upload a segment to Suno, and prompt it "country song, Guitar."

Your style of country, as a basic prompt, would be 1950s era, Country, Honky-Tonk, Acoustic Guitar, 4/4 Timing.

But the prompt "Country, Guitar" leaves Outlaw, Bluegrass, Urban, Contemporary, 1940s-2020s as possibilities: as these are all Country.

So when I ask what your prompt is, this is what I mean. If you are just uploading and hitting generate, then it's likely just hearing whatever genre you gave it with no other context.

[near flume]

Male Baritone vocal with warm, slightly gritty tone and expressive delivery. Emotional indie rock style with soulful phrasing and dynamic intensity. Verses delivered with restrained intimacy, gradually building to powerful, impassioned choruses. Subtle vocal rasp, clean pitch, and tasteful melismas. Include layered harmonies in the chorus—tight thirds and sixths—for richness. Production should feature light reverb, smooth vocal comping, and a natural, live-room feel. Mood is determined, heartfelt, and resilient.

[modest juniper]

I have tried both ways. It gives a little prompt. And it gives a description prompt. The large one is The song is a high-energy hard rock track in common time with a fast tempo, around 155 BPM, The key appears to be E minor, The instrumentation includes heavily distorted rhythm and lead electric guitars, a punchy electric bass playing root notes and simple patterns, and a drum kit featuring a strong backbeat on the snare, The male lead vocals are delivered with a gritty, sometimes aggressive tone, bordering on yelling in the choruses, The song follows a verse-chorus structure with instrumental sections, The production features a full mix with prominent guitars and vocals, There is some delay on the vocals in certain sections. The small one is hard rock, aggressive, male vocals

[young geode]

And are you targeting a 'Hard Rock' sound (Aerosmith, ACDC)?

[young geode]

Thanks.

I'll test it at some point and see if I can get consistent control.

[jovial bough]

No, Suno made. I write the lyrics, Suno does the rest. "Piano only accompaniment" was just my added style summary. This one just happened to be the best of the lot, waited until 2/3 through before addinback in the drums. I'll fiddle with some of your suggestions, see what happens. TX

[modest juniper]

no I am trying to get it to continue to sound like my vocals in extend. It used to without much effort. Now it wants to change everything.

[young geode]

If it's Suno made, I would just rip the Stems and then isolate with prompting and meta tags. You'd have much better success.

[young geode]

If that's what you're trying, then you need to describe your vocals to a high degree.

Instruments and genres are one thing, but vocals are difficult to produce and maintain. If you know your vocal type, Vocal range, vocal qualities, speech style, ect, ect, then you need to use that.

[jovial bough]

I don't want to mess with stems. Meanwhile, here's what I did. Our friend still insisted on keeping the drums in there. 2 out of 2. Other than that, it was all piano and vocals. It just sounds like it may not be trained (yet) for this scenario, simple as it seems it should be. If I knew how to DAW, I'd strip it out there, upload and try from there. Someday...

[weary glen]

hey guys, anybody here know what good term for "deep voice choir"?

[polar spade]

Have you covered the audio upload first to get your audio set in code suitable for suno to then extend?

[jovial bough]

I just noticed when I typed in the exclusion list, the "exclude styles" identifier disappears. Should probably be fixed to look just like the 4 categories underneath it.

[polar spade]

Was " .....without much effort...." during 3.5 era or now with 4.5 and sliders? That would make a difference for sure.

[polar spade]

The notes-with-the-slash-through-it icon ?

[modest juniper]

In the past, back when it used to only give you two minutes of uploaded audio I would take the song and find a spot where I could extend it. I would use Audacity to cut to the exact spot and then re upload the track. I would then put in the lyrics from that point on and though it was not perfect it would somewhat sound like me. The instruments would somewhat stay consistent. I did not put anything into style or genre. And it would extend that way. Now with all this other crap. Nothing stays consistent with the song. I have wanted to change the lyrics in certain parts of my completed works and it goes haywire and does not stay consistent or have anything that sounds like my vocals at all. That is what I am trying to get at. I did not have to describe my vocals or anything.

[polar spade]

Oh , you mean "Exclude Styles" should retain on the left. Good idea. I'd say, since there is so little visible room for text, they're just relying on the icon, but you can suggest it. https://discord.com/channels/1069381916492562582/1260263379243761776

[polar spade]

Have you now tried to cover the audio upload first to get your audio set in code suitable for suno to then extend? What i could do in the past doesnt all work as before for me as well. So, its relearning what works best with any new stuff.

[weak moss]

Phonetic sounds matter

A with the slash above it, the vowels with slashes above them,

literally change the voice sounds

use it

leverage it

[jovial bough]

Ha! Corrected a typo and changed my exclusion wording a bit. Success! However, the vocal interpretations were atrocious. I mean, godawful. Karaoke hour. So back to work. Time to eat some more credits.

[modest juniper]

When I do covers of my songs I am pleased with how a lot of them turn out. It gives me a different singer and that's all good. I guess I am just SOL. I dont like the weirdness etc. I want to be able to not use that. As i have tried a thousand different ways of putting the sliders. It never works. My songs I have written and recorded are from years ago. They are good songs. But my vocals and singing has caused me heart attacks and so i cannot change lyrics to my own songs any longer. Suno was my only way. But it is to no avail

[young geode]

What's the issue with stems, again?

If you got the track, then you should easily be able to take that specific stem. Convert it into a persona, then using both the Stem as the audio and the persona to guide Suno, you just add your lyrics back in, using the same prompting technique.

[jovial bough]

Cost. A 50 credit rabbit hole that I don't want to fall into.

[eager kettle]

Is there a contact for proposals for any collaborative or cooperative ventures>

[young geode]

Thought you would say that.

Here's the thing: 50 credits, once, gets you the sound you're looking for (Vocals or Instrument). That's 5 generations, 10 song attempts.

Unless you think you can get what you're looking for in less than 5 gens, it's worth doing it, cause you get the Stems and you can turn the Stems into persona.

You can then take the Piano Stem and use it as the basis, while turning the vocals into a persona, then isolate the instrument and vocals through Prompt and meta tagging, and likely get your song produced, in the way you want it.

It sounds costly, but the increased probability of hitting the mark does make it worth it if you're trying to do something very specific.

[jovial bough]

I was just looking for an excuse to make that cute pic. I'll look into it. Thanks for the good advice.

[spiral arrow]

有

[hearty laurel]

Hey guys. Stupid question. What are the differences between the Suno-created stems V1 and V2?

[trim glade]

Can anyone tell me why Suno always gets too creative and usually doesn't follow the song that is supposed to be covered in the chorus?

[young geode]

Explain

[trim glade]

when i cover a song , suno is to creative , they doesnt follow in the refrain the cover song melody

[half dragon]

What did you have Audio slider at?
Raise that as high would tell Suno to follow the audio that you cover and retain the melody. Maybe as high as 90+%??

[half dragon]

They're slightly different in some of the instruments it detected.

[trim glade]

I have try all.. i make it in 100% or 70% allways the same , they doesnt follow the original track in refrain.. in a atmospheric song they play happy powerful music in refrain v.v

[hearty laurel]

Even if Suno came up with the instruments to begin with?

[half dragon]

At least the melody it should be as close.
As for style, maybe the style slider would affect it? Have weirdness slider to zero.

[trim glade]

i have set the Weirdness to 0% and Style isnt a think i doesnt have write , happy or powerful rock think

i have try to make an Anno song with the original Anno theme , but when the refrain plays it comes an extrem happy music that doesnt match to the original cover song

[half dragon]

So you didn't have anything in the Style box?

[trim glade]

No, there was enough in there. But nothing in there expresses itself with "Happy, powerful chorus." I wrote down everything, including instruments and style.

[half dragon]

You mean, you copied and pasted in the Style box all the styles from the original song - but it still ignores it?
My covers been pretty success with Audio at 85% - 90% and Style at 85%-90%

[trim glade]

Yes, so to speak. I even wrote for the test that he should not be so creative and should follow exactly the melody of the cover song, but he did that just as little.

[half dragon]

Yeah. As long as in the cover, if you want the same style - it should be the same as the audio; then play with the slider to guide Suno to stay close or be more creative.

[trim glade]

That doesn't help at all, as I said I even set everything to 100% for audio

[trim glade]

That is my Style: An Orchestral song with a Nostalgic Flair , a Clear Male Lead singer follows the Instrumentals of an Piano , Acoustic Guitar , Snare Drums , Trumpet brass , Violin Quartet plays in background , Woodwinds and Cembalo

[echo moat]

does suno perform any mastering during song generation?

[jovial bough]

OK, I took the dive. Downloaded a copy of Audacity and my wife - who uses it for her "legitimate" songs - taught me how to use it. Took stems from the original, uploaded them to Audacity, chopped it up and reassembled just the piano and vocals, uploaded it to Suno, and started runnning covers. Gonna stick with this one for now. Adele will have to work out whatever kinks are left. 😉 Phew. Lotta work.
https://suno.com/s/mAZrREWNfo116WCV

[small bridge]

okay genuine question did they remove the new editor and change the legacy one because for me when i hit edit it says go to legacy editor (fine by me the new one never worked for me anyway) but the legacy editor now looks like it's malfuncitoning to me(if this is the new vibes, I don't love it, to be sure)

but i just wanna know if my suno is weirding out or if this is just what it looks like now

[half dragon]

That does look like the legacy editor ... let me open on mine - I also notice it's taking much longer to load the new editor - sitting here looking at the loading screen forever....
Serious question: did you accidentally click Open in Legacy Editor in the loading screen?

[radiant chasm]

No but its close. Its using a rim shot sound instead of sticks/claves.

[mortal crow]

This is just horrible! My Suno page has been down for several days! No one responds to my messages! The working area of the screen is constantly blocked by a notification screen saver that does not disappear or turn off! There is no response to my complaints!

[radiant chasm]

No, me either

[mortal crow]

This is just horrible! My Suno page has been down for several days! No one responds to my messages! The working area of the screen is constantly blocked by a notification screen saver that does not disappear or turn off! There is no response to my complaints!

[half dragon]

Even if you clicked the notification menu item on the left? it just doesn't go away?
What browser are you using?

[small bridge]

nah i sat there for a good five min nothing pulled up so i clicked the legacy button

[half dragon]

Yeah the new editor isn't coming up for me either... yesterday it was working but they now have the option to go to legacy editor from that loading screen.

[small bridge]

yep okay i'm not crazy woop

just had to check

well i mean i'm not loosing it ovr Suno at least. XD

[mortal crow]

No, it doesn't disappear and I didn't turn it on! It loads itself when you open the site

[half dragon]

What browser is that? Have you tried another browser? Just to see if it's browser related?

[mortal crow]

Chrom

[half dragon]

I'm using Chrome - no issue here. Maybe some extension got conflict and causing issue?

[mortal crow]

This block has been going on for three days now and it won't go away.

I haven't installed anything on the browser in days.

This screen saver seems to be stuck and won't go away

[hollow hare]

Lately a few people experienced DNS issues which caused something like that to happen. To make sure if that is the cause:

1. Try using another browser, does it have the same issue? If no, try clearing the cache of the first browser. If yes:
2. Do you have VPN (e.g. Proton VPN, Opera's built-in VPN)? If yes, does the site load properly with the VPN enabled? If yes, that likely indicates an issue with the DNS
3. For flushing or changing your DNS, see: #🗣️┃chill-chat message

[mortal crow]

I cleared the cache yesterday and it didn't change anything. I'll try other browsers but they need to be installed.

[hollow hare]

Hmm don't bother wasting your time with that, just do step 3 and see if that helps.
If it does not work, then you can try another browse, like WaterFox

[mortal crow]

I am not an expert in such matters.

[hollow hare]

Are you using Windows or MacOS?

[mortal crow]

МАС

[hollow hare]

You know how to access your terminal?

[mortal crow]

No

[hollow hare]

No worries, I will help in a moment (leaving this chat convo here to refer again in the future).
What version is your MacOS? Should be in Help or Apple icon > About This Mac

[mortal crow]

10.12.6 sierra

[raven turtle]

Just tried to replace an intro that has the arp not playing consistently and yep… sent me to Legacy Editor. I assume there’s some back end tinkering going on?

[hollow hare]

Okay, so search for the app Terminal (or execute terminal.app), which will open up the command-line interface (CLI), similar to Command Prompt in Win

According to https://www.freecodecamp.org/news/how-to-flush-dns-on-mac-macos-clear-dns-cache/, for that OS version, run (copy/paste) the following command to flush your DNS:
sudo killall -HUP mDNSResponder

Sudo tells the computer to run the command in admin mode, so be careful when you see sudo commands. Pretty clear what this one does but you can always do killall --help to see the help property (if like Linux, you can do man killall for the manual)

When running sudo, it will ask for your computer password

[hollow hare]

If flushing your DNS worked, which basically is like clearing a cache of the IP addresses stored, great.

If not, follow the following on how to change your DNS, then flush again after it is changed:
https://support.apple.com/en-mide/guide/mac-help/mh14127/10.13/mac/10.13

In case you were unaware the Domain Name Search (DNS) looks up the IP address of the website address you enter and goes to that site. Sometimes the server IP can change which is when you run into these DNS issues if your DNS didn't get flushed (usually happens every hour to a day or so automatically iirc)

The problem can actually step from your default DNS lookup, which is typically assigned by your ISP. This is why you change the DNS Lookup address in this situation - you want to get the correct IP looked up instead of the wrong one by your ISP. Well, that's the gist of it

Please ping me if you need any further help, clarification, or letting me know of the outcome 🙂
You can DM me but ping me after you do so I can allow the message from the filter

[mortal crow]

I don't know my computer's password(( I've tried every password I know and Terminal won't accept it

[hollow hare]

Ah, please see https://support.apple.com/en-us/102367

[raven turtle]

I’m generally getting about -13LUFS on generations from Suno. Is this fairly consistent for more people, and as such are you then pushing it up in this party software?
I know streaming services like -14 as a rule of thumb, but we all know artists push up around -8 as the loudness wars are still a thing. I’m lifting to around -9 or -10 ATM finding it that the WAVs don’t like being pushed too much from Suno, even with tricks like multi band compression. Maybe the dynamic range just isn’t quite there yet

[radiant chasm]

There seems to me to be some kind of gain calculation applied to the songs to make them around -13 but its annoying because most of the time the peak is around -6db!

[raven turtle]

Indeed. And I’d hate to try and use too much of an expander on the TP

[radiant chasm]

I wish the peak was the 98% like its supposed to be and that they embedded the replay gain info instead of applying it but this wouldn't work on a Web browser playback

I'm using Ozone on my tracks when I have "remastered" them by using the stems. It seems to work alright

[raven turtle]

That’s seems a fair method. I’m using Bx clipper if needed, Shadow Hills for some of the heavy lifting, and some Ozone on Imager, Maximiser etc if needed. But just sounds a little too forced maybe as you say the true peak is being crushed before we download the WAVs…. Hmmmm. It’s like… almost there for dynamic premaster quality. Frustratingly so!

[radiant chasm]

yeah I find it very frustrating. There is definitely some kind of "post processing"

[raven turtle]

I need to analyse the waveform more. I’d tried Abbey Road TG Mastering suite and Waves IDX to try and uplift it, but it’s just too harsh as the dynamics sound like they almost cut off. I wish glee could get a full range dynamic WAV

[radiant chasm]

Yeah. I've noticed strange "drone/saw type" noise in my tracks that I think it very likely from watermarking.

[raven turtle]

Had the same thing today. Oddly only on a section from an insert edit.

[radiant chasm]

strange

There might be a difference from version 4 to version 4.5 in this regard

They did sign on with some new thing for it I think

[raven turtle]

That could be a factor. I’ve found remastering in app down to v4 evens out the sound range and it seems to low cut the sub to a more manageable level for mastering out in Logic.

[radiant chasm]

hmm.. I think I can see how that would work

[granite pawn]

Is sno editor down

Suno

[raven turtle]

Some further tests and yep… it seems everything I’m Getting out of Suno has a really throttled true peak at about -4.5 or more for some. Pushing it up to -9LUFS is inflating the true peak a bit but it’s still quite subdued. Overly cautious. But cleaning up the mid range mud drops the life further out the waveform.
I’m assuming there’s some backend algorithm that’s rendering their masters out to be more streaming compliant and an overly safe true peak, but as the platform doesn’t seem quite there with the fidelity (yet) it’s just not sounding “really good “ when pushed to CD volume if around -8 to -9.

[radiant chasm]

Yeah, I'm sure its for playing in a browser. It seems to me to be much the same that YouTube does when you upload a video.

[raven turtle]

Frustrating as the dynamic range is “almost there” at times

[radiant chasm]

yes...

[raven turtle]

Guess we play around and hope for 16 or 24bit in v5 then! LOL

[soft wadi]

Could someone please tell me what all the weirds sounds are that suno is generating. there are scratches, glitches. etc. what is going on? I am trying to regenerate an intor and it all sounds like craPPPPPPP

[plain pumice]

I'm not sure, but the new Song Editor isn't loading for me, legacy works as usual, or so it seems.

[granite pawn]

I’m getting neither, legacy is freezing

[plain pumice]

Legacy seems to be working for me, at least for now.
..but some changes don't seems to take effect - like matching lyrics to the song.

[soft wadi]

it appears to be down

legacy works but not generating anything good

[small bridge]

are peoples gens going from quiet to loud on their own?

[plain pumice]

Definitely not following my style instructions.

[modest crow]

Yes

[soft wadi]

replacements are trash.. glitches, sounds that don't below

[torpid thicket]

There a reason why we can't download our 'Contest' songs anymore???

[small bridge]

it's really bad this week, can't get gens on much of anything that I like without wasting a huge amount fo credits. I literally spent 2000 credits on one song. Worth it cause the song is really really good imo but still man, it's bad this week.

[torpid thicket]

What prompt are you using?

[small bridge]

it's all good i got help with it already from a staff, it was just being really finicky the problem is most of the parts (i was going in and doing line by line or full verse redos) were breaking with like the voice going wonky, or like starting soft, then going loud and mechanical.

it's a bot issue for sure not user error.

or it ignoring the lyrics thats another favorite error i get -.-

[raven turtle]

Yes. And also some left To right panning of whole audio. It’s subtle but there.

[small bridge]

great, sigh.

I swear this month I am only gonna be able to work on 2 songs cause it's that bad. Like all those credits on just 2 songs. Used to be able to get at least 10-12 done in a month to a decent ending. Now it's so bad. I have to be really picky.

[soft wadi]

has anyone tried Udio?

Suno is getting ridiculous

I am trying to replace one sentence and it keeps grabbing the entire verse. I don't want to change the entire verse UGH UGH UGH

[small bridge]

this was happening to me all day wasted over 100 credits. it's legit so annoying I feel your pain. pat pat

[soft wadi]

I just wants to make me go somewhere else

[small bridge]

yeah I feel you. I did try udio free and it was utter crud. So I dunno what their paid is like but their free was so bad. and you really can't do heavy lyrics like I can on here. For example it seems to be glitchy and it has a space for your song title but I for the life of me cannot get that box to click so I can put anything in it so it gens it's own title as it's blank when it goes through. The genre's and everything I tried to mix just came out really glitchy. I can't say paid is better but free just didn't have the autonomy that Suno does. However Suno is very broken so. haha Take that as you will. XD

[half lintel]

Hi I’ve written the lyrics to a song , how do I post it to my socials ?

[torpid thicket]

[sleek stump]

Cant dload someone else song, even a remix of it.

[sleek stump]

Go try it yourself, you will be back to Suno in about 5 minutes 😄

[small bridge]

Suno's like my third child I pay 10 dollars a month to have ignore me. I already had two. But the third one at least is cheaper than the other two. XD

[sleek stump]

The other music AI platforms all suffer similar issues, because they all work the same way.

[torpid thicket]

We use to be able to, bruh

[sleek stump]

🤷‍♀️

[half lintel]

How do I upload the song I wrote to my socials ?

[sleek stump]

You can link back to Suno for an embed... in your Suno library, click the 3 dots, then share, then link, that will copy the link to your Suno song and you can paste in in your socials.

[half lintel]

Oh thanks ! Will try that

I’m missing something I think

Share , then link - is that coply link? Can’t seem to find how to paste it to socials

[lost galleon]

I played around with Persona for the first time today. It seemed to work pretty great, then I noticed it would inject a beat from the original song most of the time and it was very noticeable. Is there a strategy to focusing a persona on the voice rather than the origin song?

Should I get stems and then remove the instrumentation?

[half lintel]

I did press copy link

[sleek stump]

Now you can paste that link in your socials

[half lintel]

So on TikTok say , where do I paste it ? Or is it easier to create a video with that as the audio and upload that way?

[sleek stump]

I dont use TikTok, I have no idea there... sorry

[half lintel]

Okay so instagram ?

[half dragon]

You'd need to download the video for tiktok or instagram

[sleek stump]

download is only available to subscribers

[half lintel]

I haven’t made a video yet . I am a subscriber

[half dragon]

yea - if you're a subscriber

[sleek stump]

Ah, ok.. You can generate a lyric video, and then download that.

[half lintel]

So if I made a video on my phone with the audio I should be able to just upload that ?

[sleek stump]

Pretty much

[half lintel]

Okay do I need to add the video via Suno or can I just download my song and then create a video with it?

[half dragon]

You can create your own video with your audio downloaded from suno. OR
you can generate lyric video like @sleek stump said above then download it

[half lintel]

Thanks for the clarification it’s much appreciated

[formal gale]

dude seriously I am trying to use names of my band members to describe their sound and it's telling me I can't use artists names

[sleek stump]

Then don't? Use nicknames or somethinbg

[echo prawn]

Why does extending a song always generate something unacceptable?

[stoic vine]

there are some words that it randomly will say you can tuse because its an artist name but it will be like a normal word lol

[cunning kelp]

Who can help me? I put 4 minute length song at the beginning of the song prompt and it gives me results of 1 minute or seconds. thx

[stoic vine]

well this is definitely not the right channel for that

[cunning kelp]

sorry, im not that familiar with the server

[stoic vine]

i mean you can just read the channel names

an example json prompt that seems to have produce multiple good generations at least for me:```json
{
"positive_prompt": {
"styles": [
"instrumental progressive metal",
"math rock",
"jazz fusion",
"djent"
],
"elements": [
"complex polyrhythms",
"clean arpeggiated guitar passages",
"occasional djent-style palm-muted riffs",
"rhythmic tapping and sweeping",
"jazz-inspired lead phrasing",
"tight syncopated drumming",
"ambient background textures",
"dynamic shifts between soft and heavy",
"rich clean tones blended with mid-gain crunch",
"high technical precision"
],
"bpm_range": "100–140"
}
}

negative prompt with it:```json
{
"negative_prompt": {
"exclude_styles": [
"death metal",
"metalcore",
"synthwave",
"trap",
"pop rock"
],
"exclude_elements": [
"screamed vocals",
"overly distorted walls of sound",
"808 sub bass",
"autotuned vocals",
"generic power chord progressions",
"heavy reverb on all elements",
"four-on-the-floor beats",
"EDM-style synth leads",
"simple song structures",
"lo-fi sampling artifacts"
]
}
}

example result: https://suno.com/s/O2BdbPNKvo1d63K8

[young geode]

Is this all within the style section or is the negative in the Exclusion?

[stoic vine]

negative fits in the exclude styles box believe it or not

[lusty rune]

You're using straight .json in you prompts? What is the stability of that? Verses common language prompts.

[stoic vine]

No idea, could be placebo but feels a bit more consistent

[lusty rune]

Cool so how long have you been using that type of prompt science? My methods are simular in a way.

[stoic vine]

Just started using json today, I should note I set weirdness to 10 and style influence to 90 also

[lusty rune]

Ahh okay so I'm curious if weirdness was set to zero would it strip away anything. We should chat?

[stoic vine]

I just got into bed actually lol but I'll check back here tomorrow

[lusty rune]

Let me send you a friend real quick

[young geode]

Just need to know.

I like to test these things. I've seen proper Json code work for other LLM AI prompts, and it's strength is in it's degree of specificity. However it's extremely character heavy, which typically means losing out on certain context you could otherwise input.

[vapid bronze]

Hello all - please forgive me any mis-steps as I'm new here... I'm open to any correction on how to correctly interact on this channel.
I've been struggling a bit on a couple of songs with how to get consistent production to drive consistency at certain parts of the song. One example is the voice whispers a lyric at the end of the first line in a couplet and I'd like it to repeat that same performance at the end of the second line - but I'm struggling to figure out how to do that in an already generated song - I'm trying to replace the lyric but that doesn't seem to work - any ideas?

[young geode]

Give the link the song you're trying to work with

[young geode]

So: The woman says "Goes pop" in a whispy manner and you're looking to recreate that whispered/whispy quality in other portions?

Which areas exactly? I noticed you don't have a verse or chorus structure built in, is that on purpose?

[vapid bronze]

Sort of on this one it's on purpose, but I really don't know what I'm doing smh I've spent almost 200 credits just trying to learn over the past couple of days 😅

[vapid bronze]

But, yes, that's what I'm trying to replicate - is that whispy "goes pop" at the "cheap red wine" lyric.

[vapid bronze]

I'm working now on adding the Verse, Chorus and Bridge headings now - does that make it easier to deal with in the Editor?

[young geode]

Well you don't need [Verse] or [Chorus], but it does help keep context, but this is more like a poem so it may not be necessary.

-Lyrics-
/
[Vocals: Female, Whispy, Breathy]

• Goes "pop" *
  [Vocals: Female, Whispy, Breathy]
  /
  -lyrics-

And

-lyrics-
/
[Vocals: Female, Whispy, Breathy]

• Cheap red wine *
  [Vocals: Female, Whispy, Breathy]
  /
  -lyrics-

If you add the verse and chorus tags,, you could add to your prompt "Whispy Vocals at end of sentences".

Control of lyrics in this way is rather difficult, as it's all about isolating the lyrics and the structure of the lyrics and the punctuations.

Asterisks lyric Asterisks has a tendency to give a lower, more whispy vocal. Sandwiched between these meta tags, you will likely get it, especially if you place space between the lyrics.

[vapid bronze]

Super cool! I will play with that... Do I add everything literally and exactly as you typed it including the "/" or do I just start at the square brackets "["

[polar spade]

@graceful condor (response moved here from "need-help" Billing / Support guide channel.

No.

The spinning wheel you see before the song's length is displayed is just your computer processing the file's information (metadata). It's getting the song's length, artist, album, and other details ready to show you. It's not a sign that the file is in a fragile or "unfinished" state. Listening to the song while this is happening doesn't affect the file at all.

" Because yesterday, so much went wrong......"

That then still leave you with "well if not that, then what is it?" Theres no great single answer, and experience would say , its only a guess to try saying exactly. We do our own testing and report what appears to be a bug or glitch. One test is to always try a simple creation - one prompt, a few lyrics, click Create. If THAT doesn "cut off" (and often those generate long songs), then it may be more a combination of things were trying that just dont work well together. Then we try to talk it out here.

Having chats with other in the tech-prompt-talk or even general chat channels here will help other users to give you their ideas as well.

For now, I hear clicks and pops when I listen to created songs. I've reported it and several others in the past week have the same. So far, it seems a playback issue.

As for "cut offs", its common that trying to cover an already long song can create "cut offs". But I've noticed more users talking about cut offs recently, so it can literally be a backend issue as well, as its clear that there are some changes in the formats and layouts of the program lately.

If you're getting 100% constant issues, all similar, no matter what test you try, then definitely start an email to [e-mail] to work it out.

[young geode]

The slash is just to indicate space in the lyric structure. You do need the brackets and asterisks

[vapid bronze]

Thanks for your patience with my noob-ness and your help! Is there an article with all of the symbols and what they do somewhere that I can use as a reference?

[young geode]

Don't worry about it, I write guides for people like you.

There is no instruction manual for Suno or AIs, so you're the norm.

Check the workflow channel and find my thread. You'll find everything I've learned and tested up to this point.

[vapid bronze]

Is this what this verse should look like using your guidance? I think Discord modified one of your astericks into a bullet point...

[young geode]

Make sure you have spaces between the lyrics you're isolating.

Otherwise you have a higher chance of the Metatag bleeding over

[vapid bronze]

Do you mean an line break? above and below the bracketed instructions? to isolate them or just spaces on either side of the lyric on the same line?

[vapid bronze]

Good grief ... okay this should be the new image with the line breaks - is this what you mean?

[young geode]

[Vocals: Female, Whispy, Breathy]
Goes "Pop"
[Vocals: Female, Whispy, Breathy]

Little marble falls in like a guiltless shot—Cherry bomb bleeding into

[Vocals: Female, Whispy, Breathy]

• cheap red wine *
  [Vocals: Female, Whispy, Breathy]

Kawaii Kitty whisperin' "girl, you're divine."

Discord uses markdown so I can't just add in the asterisks properly, but they need to enclose and connect with the text in question.

[polar spade]

@soft wadi transferred from need-help channel: "....Could someone please tell me what all the weirds sounds are that suno is generating. there are scratches, glitches. etc. what is going on? I am trying to regenerate an intor and it all sounds like craPPPPPPP...."

Before using more credits, have you downloaded one or two of the outputs to see if they are clear and its not just the playback that's having issues? Of counrse, if you can note the scratches and clicks are in the exact same spot each time, then thats more to the output itself, right.

Also, does it crackle simply everytime you click a spot along the playback line. It has for me ,which is definitely not an output issue, but a playback problem.

[vapid bronze]

Like this? I'm reading both your "SUNO Guide_Research...." and your "Using Style Prompt..." right now - there's a lot there! I appreciate all the work you put in and I'm gonna study it - thank you! ❤️

[young geode]

Yes.

That isolates the lyric. Try that out and see if it gives you what you're looking for.

[vapid bronze]

How does Suno know that the 2nd set of brackets is the closing set? In other languages, html for example we need to specify with a closing operator like <html> and </html>

fyi... you can use This is inline code. to enter exact code into the Discord markdown... for example,
[Vocals: Female, Whispy, Breathy] Goes *"pop"* [Vocals: Female, Whispy, Breathy]

[young geode]

At this point I don't know if SUNO uses operators, or even parses a form of them.

It recognizes Meta Tags and lyric structure and punctuation. So if it sees a meta tag and text different from the rest, it'll tend to treat it differently.

Also I forgot about the vide box, thank you.

[vapid bronze]

Using this article for reference and backticks https://support.discord.com/hc/en-us/articles/210298617-Markdown-Text-101-Chat-Formatting-Bold-Italic-Underline

[unborn ermine]

Hey folks! I’m trying to make an instrumental with multiple sections (intro, build-up, climax, outro), but SUNO seems to struggle to follow that structure. Am I hitting SUNO’s limits here, or is there a better way to guide it?

[sly badger]

Hey everyone, I've been having an issue with editing sections of songs. Sometimes it will repeat a part twice or not connect well back in. I feel like I am throwing away credits trying to fix one or two lines that came out badly on otherwise perfect songs and don't know how to fix it

[graceful condor]

If I upload a recorded piano piece to create a song out of I can't use the vocalist Suno creates as a Persona, why? What if I really love the voice how can I save it aka re-create it?

[young geode]

You can control SUNO instrumentals by using the Lyric box and proper prompting.

First,
Take some basic lyrics, could be new ones or ones you already made, convert them into grade 2 Braille or Morse code. Structure your Intro, Verses, Chorus, ect, as you want.

Then use heavy meta tagging to guide SUNO

[Verse: Elements]/[Chorus: Elements] 

[Instrument: Type, Model, Scale, play method]

[Mood: Emotion, Environment, Concept, ect]

[Instrument Solo: type, instructions]/[Instrument Piece: type instructions]

[Drop: Instructions]/[Interlude: Instructions]/[Riff: Instructions]```


And make sure to sandwich the tags, instructions and lyric structures:
```[Instrument: Type, Model, Scale, play method]
[Verse: Elements]
[Mood: Emotion, Environment, Concept, ect]
Translated Lyrics
Translated Lyrics
[Riff: Instructions]
Translated Lyrics
Translated Lyrics

[Drop: Instructions]```


Then for your style prompt,

Description
```XXXX Era,Rhythm: Italian Tempo,Time Signature,Beat,Genre,Subgenre,XXXX Era Subgenre: Influenced by/combining (Sub)Genre{If requested},Instruments: Type/Model, Effect/Play Method;Type/Model,Effect/Play Method,(...),Vocals: Male/Female,Italian Vocal description,Ranges,Vocal style,Vocal Effect(s),Mood: Tone,Theme,Context. Mix: Raw/Wet/Ect. Scales: Intro in E Major, Verse in E Minor, ect```

Narrative
``` The Song begins as a slow emotional Adigo piano piece. As the piece progresses, war drums fade in with a soft thumping beat, the song stops, then drops into a Symphonic performance.```

Then whatever you don't want in the song as the negative prompt.


This is an example: https://suno.com/s/WDGXSBaL1S7gBbam

[polar spade]

You can separate the vocal stem of your creation and Make Persona of the vocal stem, if you wish.

[near flume]

The asterisk doesn't necessarily give a lower, whispy vocal it all depends on the genre you're using. Typically the asterisk will signify importance to that line. I've used it in darkwave & black metal and it produced more of a shouted growl.

[near flume]

You can place the vocal controls directly into the Verse tag. Use them with individual lines if you are changing the vocals from what is being used as the main vocal. Normally you'd declare the main vocals in the Styles box, and in the lyrics box id you want to tweak or change a portion.

[young geode]

I've only ever heard it as whispers/under sung.

[vapid bronze]

I'm so impressed by Suno - Even though I don't know anything, I think this song is a banger - I'm just too detail orientated and want some slight changes. For example, I want her to pause in Verse 5 between the word "Pretty" and "fable"

1. do I need to completely regenerate the song or is there a way to make that happen in the Suno editor
2. Or, should I just pull the stems and send them to a sound engineer to have them do it?

[Verse 5 - beat stutter - whispered reply]
…that was one time.
Okay, five.
But this time I’m pacing, I’m zen, I’ll survive.
Besides, it’s artisan! Look at the label—
she’s winkin’ at me like a Hello Pretty

[pause]
fable.
[pause]

[jovial bough]

@young geode Is there advantage to using the meta tags in the lyric box vs just relying on the Styles box? I've seen here and there (including you) that the metas in the lyric box help refine what you want, but no real good in-one-place discusssion of it anywhere. You got a link to something like that? (What you said to MilesJoe above leaves a lot of things undefined that I don't understand at all)

[young geode]

Yes, because it gives additional context and instruction.

In the Style Box, you can add in both descriptions (what's in the song) and narrative (how the song is played), but within the Lyric section, you can be more specific.

For example: https://suno.com/s/rYowuAgzrrC1dfsf
V
https://suno.com/s/3LqurQbYDmUijpV5

I added scales to the Prompt, but I also added them to the lyrics, and it changed the entire speed and sound of the song to be more like Metallica, and the switch between the scales was consistent.

But if I just added the scales in the styles box, then you occasionally might get them, or get them at improper times.

So the rule of thumb you should be working with is:

Style Box, to introduce and focus SUNO on a sound.

Metatags, to guide and order SUNO's generation of a sound.

If you have the style prompt, but no meta tags, you'll likely get the sound, but without control.

With the meta tags and no style, you might get the sound, but only if if works with what SUNO is already doing.

If you have both the Style and the Metatag, you have a high likelihood of getting the sound and getting the sound at the proper place.

[jovial bough]

Good stuff. I'm a word guy, so I will never achieve the kind of granularity in Suno that you can get. (With a 50% success rate, maybe? 😉 ) But I will definitely start spending more time up there in the lyric box. I've generated a ton of songs off my lyrics relying just on the Styles box, but this is good info for someone who is basically still a neophyte.

What did you mean by "Translated Lyrics" up above?

[young geode]

I'm not even a musically knowledgeable individual, a lot of this is because AI makes knowledge access highly easy.

The easiest way to do what I do is to simply work from the basics to the intricate.

Lyric Structure (Intro-Verse-Chorus, ect)>Lyric Configuration>Lyric Punctuation>Basic Metatags (Instrument, Breaks, Vocals)>Intermediate Metatags (Play patterns, scales, qualities)>Intricate Metatags (Instructions, Chords/Notes, Ect)

[young geode]

Turning lyrics into Nonspoken print language.

IE, Grade 2nd Braille, Morse Code, Cipher, ect.

[jovial bough]

Uh-oh. The jet scalped me. (Over my head)

[young geode]

What do you mean?

[jovial bough]

Don't understand a bit of that or its relevance

Again, may be over my head and beyond the granularity I seek.

I've seen otheres here doing soething like that though. Talking straight to the software .

[young geode]

SUNO uses GPT as it's translator, so GPT picks up on all kinds of languages, which it then translates into semantic notions, textures, ect: not just lyrics.

IE sad and melancholy lyrics will produce sad, melancholy sounds (unless over ridden), ect.

So if you're trying to make an instrumental, instead of relying on the instrumental tab, you can instead use the lyric box to introduce 'lyrics' that won't be spoken, but will provide additional semantics and mood, as GPT will still understand the language.

It helps both slightly lengthen the instrumental, and also give its sound added complexity.

[brittle apex]

Interesting idea. I'm going to have to try this

[brisk flint]

Do you use the 'lyrics'-section? If not, ignore that 'instrumental'-toggle-button, deactivate it, use the lyrics-sections and just enter the instructions within the lyrics and you should still get an instrumental without any voice, any singing, any spoken word but following more or less the structure you provided.

[stoic vine]

If you are using any AI to generate your lyrics I highly recommend adding in your prompt "don't use cliche rhyme schemes or word combinations such as "neon glow" and things that "hum", use more interesting vocabulary" of course this can depend on your style but a lot of blanket lyric generation uses the same sort of phrasing unless you really work to get it unique

the amount of neon glows, something with the word hum or shadows is in like every low effort ai song

ive found grok actually is pretty good at getting it to better lyrics with some nudges

chatgpt is not creative, claude can be at times, but grok is easier to massage

[fast patrol]

that actually works?

[stoic vine]

sometimes, thats just like some tips, of course you may need to massage it a bit but definitely a place to start

i really need to make a like a list of all the very common ai phrases that are used too often

[fast patrol]

i really need help getting my songs less ai lyricy fr, cuz they aways be using the "in this game" and stuff like that

[stoic vine]

yep, you could even try to add the stuff i mentioned about and then a just an "avoidance list"

[fast patrol]

thanks

[stoic vine]

"where the shadows play."
"the neon-lit night"

[fast patrol]

Hush now, stop your brag,
keep yapping, I’ll put you in a swag,

bro who says that

i was legit cringing

[livid trout]

Hello, when i'm editing a segment (in the edit song feature) and you get the 2 examples, how can you dismiss them ? There seems to be no UI to do this. Even when resetting the whole track, the 2 replacements keep appearing in the segment I want to edit. How do you 1. remove these 2 replacement and 2; regenerate more ? Or cancel replacement altogether without having to reset the entire song ?

[jovial bough]

Are stems saved somewhere? I did a full song stem, clicked outside the box, and now I can't find them anywhere.

Before I could export them, I might add. Poof.

[terse edge]

Hello! Is there anyone here that is able to use the editor decently?

I feel like it does everything except whatever i'm actually trying to do lol

[jovial bough]

LOL. Just found that at same time. Very odd. Like you give up and go to do it again, and there they are!

Not the best of programming.

[terse edge]

I'm trying to do something apparently very simple, add a couple lines in the middle of a verse.

[jovial bough]

Theoretically, open in Editor, highlight a couple lines in the lyric box, click replace lyrics, insert the two lines with your added words. I also typically click Fix Alignment. Oh - and cross your fingers. If it's add a couple lines, highlight lines before and after, and then in the new lyrics box, insert the new 4 lines

I said it a little better in that llast sentence.

[stoic vine]

day 2 of testing json prompting, really seems like its producing results closer to what i'm desiring https://suno.com/s/CdRwy93ALWyuA6sq

[young geode]

A good way to de-AIify lyric generations is to force the Ai to deal with odd patterns, rhyming schemes and phonetic asks, and if you can, uploading documents that contain much of the wordage you would never want to appear in your lyrics.

From there, it's just making minor adjustments yourself

[graceful condor]

How?

Okay most of today was great considering creating songs but now with my current project it gets massively annoying. Yet again like yesterday the same issue came up where my songs are cut off in the end. And even me having a Pro Account with max. 8 min length. Most of the created songs are exactly cut off at 5:03 which is really a weird number. Why? It is frustrating. Closing the Browser etc. didn't help at all.

[graceful condor]

Now it is 4:51. Honestly this is just frustrating and no fun anymore.

Why do I even pay for 8 min and not get more than 5 as it seems.

I get the feeling the Generator runs worse and worse the closer the night time comes.

[graceful condor]

And the extend thing is pointless if the extended part is completely different in sound and all to the part which came before. This is so dumb.

[buoyant lotus]

Hi anyone with a tip to force male/female voice (free plan, so no persona and stuff)
I've set my lyrics like this

Prompt like this

And it STILL sings the first verse with a feminine voice

Second verse usually male, occasionally female too

I used all my attempts for today, and it happened for all 10 tries

[median obsidian]

One thing I am learning is that you can't force the AI. You can play with the Weirdness and Style Influence sliders but it won't be 100% accurate in any cases

[half dragon]

Duet is tricky and not reliable at all. It's going to take a while to get what you want and costly.
One option is you know it, Persona and Extend - another option is to use editor to replace the voice but even then, it's going to take a few tries.

[median obsidian]

Another option is to edit yourself a little bit. Meaning you generate all verse and chorus for your male voice only, then generate another one for female voice and you blend//edit them in the same track yourself

Software like Audacity allow you to do that

Then you give the full track to Suno and it'll smooth it

[buoyant lotus]

Guess I'll try again tomorrow 🤣
Maybe if I tell it in another 5th way...
Or change the lyrics to fit female 1st, male 2nd (but I'm pretty sure THAT'S the best way to have male first)

[median obsidian]

For me it's the opposite. I tried to generate female-only verses and male-only verses, Suno was obsessed by mixing both inside same verse ...

[buoyant lotus]

Happened to me once, was quite unexpected but sounded cool so I was happy 🙂

[granite pawn]

Hi, need help with cropping the intro of a song.

[raven turtle]

Getting all my tracks cut off abruptly at the moment around 4-5min mark. Just stops. It didn’t used to do this. Extending as I thought, extends too long even if I just add [end] to the extension prompt. Feels like something has gone backwards with 4.5+ last couple of days.

[raven turtle]

Having same issues!

[stoic vine]

i cant get over how well json prompting (at least for me) is working out https://suno.com/s/JMNFBzkF5XVPefRS

big for consistancy so far, just set weirdness to 0% and style influence to 100% + json prompting

[neat field]

i wonder how many parameters Suno v4.5+ is... guessing it's like 12B or 30B3A like qwen

[stoic vine]

prompt suno to write a song about how many parameters it uses lol

[half dragon]

Does it need to be json though? Can it be :
style: ....
instrument:....
mood: ....
..etc..
or it works better with json? have you tried to compare?

[neat field]

this is clever, might have to look up some text-to-music HF datasets that are a few months old and check the parquet formating

probably if it's on HF it's in the Dataset

[stoic vine]

it could very much be placebo, but iive done standard prompting quite well but results seem to vary more than this way has in the past 2 days of testing

[raven turtle]

Honestly, for synthpop, was Suno only trained on 16th repeating note bassline? Jeez... It's getting so I listen to the first 2 seconds, and hit delete. Zzzzz. It's getting too predictable on basslines for electronic pop. Type in melodic bass... get a 16th repeating arp bass. Every time. Maybe I'll try again in a few hours...

[stoic vine]

there definitely seems to be some genres suno really shines at and some that no matter what you do are pretty mediocre

certain types of metal produce very rough noise to a point its just not listenable

then others were its like spot on

was trying to do some Com Truise type mid-fi stuff and it always pushes too much into generic synthwave

[terse wave]

Can we share the song we made here?

[stoic vine]

plenty of channels, do you just want someone to hear it or you want to talk about the prompting for it?

https://discord.com/channels/1069381916492562582/1083129341870878780

[terse wave]

I want someone to rate my song objectively.

[stoic vine]

yea showcase or somewhere else, this is channel is more for prompting talk as it says in the channel name

just skimmed your track through, feel like ive heard that done a million times at this point, sound like generic ai written lyrics and the style has been done countless times

[terse wave]

What types of songs do you recommend?

[polar spade]

You cant use phrasing in 3.5 model as you now can in 4.5. Single word prompting is whats recognized.

You may wish to search "duet" just to what others have dont in the past and their sample. Various ways to prompt for duet, which means its pretty versatile or pretty useless.

Any prompts in line over three will tend to have Suno
ignore all or parts of it, or Suno will often voice ( sing) the fourth , and sometimes third prompt.

Section headings, like Verse, Chorus are section headings and not so much prompting, though they help Suno give character appropriate for the sections. Put them on a separate line.

Style prompt box can reference one or both, male and/or female. No brackets. No "is" (though Suno likely will just ignore it anyway)

Personally, if xxxx vocals are in style prompts , then just [male] or [female] is enough to signal to Suno where to sing the parts (20% of the time 😉)

Extensions are just an easier way to "force" a gender vocal onto a single stanza.

I consider vocals as NOT a genre, and , as a style, has no gender, just frequency and characteristics, no different that picking your version of a guitar or sax for Suno -- without the other prompts to set the mood and groove, with 3.5, Suno will just give you any various style of a guitar or sax. So , while everyone has their way of doing duets, its still my opinion that the rest of the prompts, and the "meaning" and vibe of the lyrics have more to do with Suno making the decision of which voices goes where, unless more is done, like, within extensions.

[stoic vine]

thats all up to you man lol ive been a musician for 15 years as well as a programmer and i just mess around here

like what are you trying to accomplish? me personally i don't like the deception i see with ai music where people try to pass it off as music they wrote or whatever (caveats of course, maybe you purely wrote the lyrics but had the ai do the music, or maybe you sampled stuff you generated, etc...)

its cringe to just generate music and then try and distribute that music as if was music you organically made

[terse wave]

I was using Suno because my voice was bad

[stoic vine]

hey if you write the lyrics thats cool 🙂

ive actually used rvc to train a model on my voice to make it seem like i can sing 😛

[polar spade]

Suno calls it "remove". Try the Legacy editor UI. Go to edit UI and upper left click Legacy.

Then click Remove. Click-hold on the point to splice and mouse left to cover the intro segment youre removing, and release mouse.

Click play to listen to make sure you dont select more than you wished. Adjust by sliding the right side handle of selection bracket.

With Legacy you can micro- select your endpoints by typing in the timemarks exactly . (This cant be done with new Edit UI)

Click OK and youre done. Your remaining bit will be found in Library.

[stoic vine]

look into applio.org

[granite pawn]

Thank you so much

[olive quest]

Any recommendations for extending portions of a song? I am trying to lengthen an intro instrumental but the gens never carry the same atmosphere as the original. It's very jarring

[pseudo blade]

can i see what u got?

[olive quest]

like share the song?

[pseudo blade]

yes

[olive quest]

https://suno.com/s/GPvixRbEt9FgWFvP

trying to extend the piano part at the beginning so it plays more before the vocals come in

[pseudo blade]

[Instrumental Intro]

[Spoken Pre-Lude]

thats what i suggest

[olive quest]

Thanks. Will give it a try

[boreal zephyr]

I could use some advice. Still rather new here. I created song today and one version was pretty close but had an issue. So I did remix. Instead of Suno remixing the whole song, it remixed the first half and just bluntly cuts off the second half. I tried to do an extend to finish the second half, and now the part 2 cuts off bluntly too. I've made other songs and remixes before without an issue, so I'm confused as to why Suno is doing this.

[olive quest]

I know Suno typically cuts off remixes, covers, and songs made with personas for some reason. Its rather frustrating. Have you tried fixing the issue in the editor?

[boreal zephyr]

Problem solved.

Thank you for the tip. Still learning here! I didn't use a persona, but it was a custom generation.

First, I was able to combine the two halves within Suno, but it still left an abrupt ending.

But then I just extended it one more time with a prompt for concluding the song.

That did the trick. A bit convoluted, but I got what I wanted.

[sterile yacht]

someone knows how i can make it so it doesn’t sing the Notes: section from the first chorus, https://suno.com/song/ddc548a5-c5c6-40bd-8ec5-766be8b14b69

no matter how i prompt it i cannot get it to just use it as stylization, like it does for every other section.

[half dragon]

A song writer and a programmer 🙂
Maybe it doesn't look at it as structured json? if it doesn't know what to do with it - either it ignores it or tries to sing it.
This is not usual way to write it - while it might seem working ..I'm not sure if it actually did.

[boreal zephyr]

Although just when I thought I was in the clear, I discovered a broadband hiss introduced by Suno at one point. Fortunately, I was able to do a spectral analysis in Adobe Audition and remove it.

[young geode]

Have you tried a proper Json format with lyrics?

Cause the moment I tried this, it actually started quoting the style prompt:

Gen 1: https://suno.com/s/kVp8AsCz9uOI7wmI

Gen 2: https://suno.com/s/pwOzVN51YWYtTN6K

But when I removed the JSON and just kept the prompt, this didn't happen:

Gen 1: https://suno.com/s/Q9lnoyLQFdfHSXFr

Gen 2: https://suno.com/s/pNtuyxUoVpeiSXWf

[young geode]

Remove the Json and use Metatags

IE:

[Intro: Fade In]
[Mood: Silent introspection]
[Instrument: Synths, haunting wail]

[Verse: Quiet]
[Mood: haunting, internal dissonance]
You wake up and you smile,
the cameras all adore
But there’s a voice inside that’s asking
"What are you doing this for?"
n' You try to drown it out with every cheer (and every shout)
But the doubt you buried deep is finally creeping out

[stoic vine]

havent done json with lyrics

[young geode]

And youre just using the instrumental tab, correct?

[stoic vine]

yep, Instrumental, Styles (positive prompt json), Exclude Styles (negative prompt) - although from more testing, json for this doesn't seem necessary at all so for those im just doing simple comma list, weirdness 0%, Style Influence 100%, sometimes I play around with weird at 10% and style influence at 90%

[vivid wasp]

Curious what would be the best way to get these words drawn out? "So" and "know"

[young geode]

Additional letters, phonetic extensions, [Vocals: Held Notes at end of Sentence], - Enjambments, (Parentheses) [So (ooooo)]

There isn't a 'best way', so to speak. A lot of it has to do with the genre(s) you're in and the prompt you're using.

[young geode]

More testing with the Instrumental?

[young geode]

Also, there may be a way to prompt directly to have a certain section held, given what I found the JSON prompt I used either was doing.

I havent tested it, but within the style box, you mayt be able to use something like Vocalist draws out the words "So" and "Know" within the verse/chorus/bridge"

[sterile yacht]

seems to work thanks: https://suno.com/song/7bc57383-0a8e-4bd6-a7ce-45d9903b4073

saved me a week of using all my free credits figuring this out.

[main gulch]

Who comes in the country trendinglist and who not? Many creators told me: It's chaos!

[twin bough]

I'm not sure if this is where I would ask this.

But if I want the first first to be sung by the female I know to put the prompt like this.

[Chorus]
[Female Vocalist]

But if I want the chorus to be sung by both a male and a female together, I'm using the same square bracket but I'm putting "duet male and female vocalist" and it's randomly picking one of the two. How do I get suno to recognize that I want a man and a woman singing together?

I'm not pro btw. Can't afford it. So I'm using 3.5. 🥺

[twin bough]

Ughness. It seems whatever gender I start with, it makes the majority of not all of the song that. This is so frustrating.

[young geode]

You will have high difficult then.

Getting multiple vocals is hard enough, much less timing within the song on 4.0 and 4.5: and that's with the benefit of more prompt and lyric characters.

I wouldn't wrack yourself over this, just practice with 3.5 and when you have some dosh, then try.

[sterile yacht]

had some success with using [layered vocals: male & female vocals] but yea it is rolling a dice when doing it mid-song.

[ebon sand]

i had success with
[siren]
text
[odysseus]
text
[both]
text

But wouldnt touch it seriously without access to the editor

[twin bough]

Thank you for replaying. I'll skip it then. Thanks. 😇

[twin bough]

What do you mean by editor? 😳

[ebon sand]

the replace section editor

this thing here

[polar spade]

@Lunacy is showing the Legacy or older editor which has a more clear selection box and is a good learner editor to work with. Look for "Legacy Editor" in the upper left of the new Edit UI page.

Or we can click here just below "Loading Edit Session" on the secret hidden button to go the the Legacy page.

Oh... and to get to the New Editor UI click the Edit button.

[twin bough]

Oh yeah. I don't even know what that is. I'm not that advanced into it. Between my full-time job, my part-time content creating on twitch, single parenting, I don't have time to dive into all that stuff. 🤣. Maybe one day. 🤔

Thank you though gusu. I appreciate all the help and suggestions. 😇

[pliant seal]

Has anyone had good luck in making a duet song where Suno obeys when the male and female voices are singing? I’ve tried a lot of prompt and lyric variations but it seems to want to use the gender from the audio it’s covering (or randomly mix it up).

[pliant seal]

So far I’ve tried [FEMALE VOCAL] in front of the female lines and [MALE VOCAL] in front of the male lines. I’m covering another version of my song which has alternating male/female lines and it really wants to use the gender from the source audio even when I turn the style to 90-100 and audio influence down lower to 20%.

[half dragon]

Most asked question in here - I suggest you search the channel for past discussions for more details.
In short, there's no reliable way to do duet. Good luck.

[soft wadi]

i included samples

[pliant seal]

I thought for sure I had a good solution and separated out the stems to drop out the vocals, and then did a cover on the instrumental version of the song. And it still just sings with sometimes male and sometimes female voices randomly. Tricky! 138 remixes in but not giving up yet.

[half dragon]

Sometimes, finding a way is the fun part with Suno - don't give up your song!
It's possible, it's just taking more time and of course, more credits 🙂
Accidental duet are nice too and some case fits the lyric and usable.

[neat quail]

this is a completely random idea. try...

making a single gender vocal version of the song. cut out the parts that are intended to be for the female and leave in the gap (audacity or smn just cut that out)
upload the audio into suno
"Cover" the song in suno and add for each empty section [female vocals]

might even help if you negative prompt the male vocals

[hallow mason]

I have a question sometimes when I make song it acts like it makes them the they aren't there

[polar spade]

Sometimes the progress spinner will run a while and then nothing. Things don't always complete and they time out. This is common with processing or download/uploads since 1985. Occassionally things timeout or stop working midway.

So there should be no charges to credits and if it repeats a couple of times, then better to find a slower time of the day or just try back again in an hour.

You can also do some cleaning on your end, as it can mis-function if you have corrupted cookies or client files on your end. Opening more than one browser can affect things, leaving a browser open for days, having a loaded processor on your end, an electromagnetic storm somewhere, who knows. Close things, clear things, refresh things, restart things, try again.

[hallow mason]

Thank you it was frustrating me lol

[granite pawn]

Anyone having trouble downloading stems. It just keeps kicking me out

[tidal scarab]

Q: Can I downgrade from monthly to free? I currently only plan to use Suno for 1-2 months. (Testing and I want to edit a few songs.) If I end up using it more often, then I'll upgrade to annual.

[half dragon]

Can't you just cancel your subscription? once the monthly sub expires, you should be back to free plan; that's what I think should happen.
Important thing is to cancel it, not removing the pay method and let it expires - you'll get your account locked

[tidal scarab]

Wonderful -- so there is an option somewhere to cancel the subscription. That's the part I didn't see mentioned.

[half dragon]

yeah, click on your Upgrade button --> takes you to the screen where you'd see Manage button next to Buy More credits .
Click Manage --> Stripe page --> where you see the Cancel Subscription button.

[tidal scarab]

Perfect. Thank you!

[polar spade]

....or use the Cancel button in the Basic panel. Its NOT canceling all subscriptions or canceling out of Basic into nothing. It really means: To become Basic level from your present Pro status, please click this Cancel Button.

See the screenshot( either below) . I am Premium now. So if I click "Downgrade" I am NOT downgrading out of Pro, I would downgrade out of Premium INTO Pro. If I click Cancel, i would just cancel my Premium straight into the Basic account as of the next renewal date.

[half dragon]

That's another option.
After clicking Upgrade, on this screen you'll see your current plan, and there are other buttons on other plans.
If you want to go back to Basic - just click on Cancel plan button under Basic option. @tidal scarab

[polar spade]

Once canceled, at the Plan end date (screenshot 1) , you will forfeit all your subscription credits. Keep that in mind. See also the Renew Subscription button for later, in 2 months.

And check to make sure you have a Plan end date, not a next Billing date (screenshot 2) .

[lime canyon]

Yo, I need some help with my "Summer Love" remix. If there's anyone that's familiar with the edit function, that'd be quite nice. The issue that I'm struggling with is that when the original 1:10 song ends and then my extension starts, it doesn't sound natural. I'm trying to use the editor to be able to transition into my verse. Any help would be appreciated.

[wide hatch]

https://youtube.com/shorts/3478-HmZx4w?si=17kk6rpFT08OVrwl

[polar spade]

Though its exciting to see ones creative juices oozing, the tech-prompt-talk channel is really not for only displaying YouTube videos, unless were having a discussion about some technical aspect to deal with Suno.

If you wish, there is the https://discord.com/channels/1069381916492562582/1083129341870878780. SUNO-Showcase for showing off your creation.

Further you might pop in General channel to ask ones opinion of your song. But again, participation is wanted and preferred. Join in the chat. Use a translator if you need.

But as this is twice I've seen youre just posting a YouTube video, just thought I would let you know what this room is best suited for.

창의력이 샘솟는 모습을 보는 건 신나는 일이지만, 기술 프롬프트 토크 채널은 단순히 유튜브 영상을 보여주는 용도가 아닙니다. Suno 관련 기술적인 부분에 대해 논의하는 경우는 예외입니다.

원하신다면 https://discord.com/channels/1069381916492562582/1083129341870878780 에서 SUNO 쇼케이스를 통해 여러분의 창작물을 자랑해 보세요.

또한, 일반 채널에 접속하여 여러분의 곡에 대한 의견을 물어보실 수도 있습니다. 하지만 다시 한번 말씀드리지만, 참여를 원하고 선호합니다. 채팅에 참여하세요. 필요하시면 통역을 사용하세요.

하지만 유튜브 영상을 올리시는 걸 두 번이나 봤는데, 이 채널이 어떤 용도로 가장 적합한지 알려드리고 싶습니다.

[polar spade]

Its sort of an avant guarde, free expression, hip hop groove, so personally I think a fantastic song, just a hard one likely for an extension to follow. So you're correct there, as it may be hard to pin down the vibe. I dont know of any suggestions ...ummm.. easy suggestions... to nudge Suno into behaving with extensions for that one. Making a persona and extend AND add persona perhaps?

I've thought of trying a remix myself, but im think of a chillpop dance and just use different lines as topline structure. Just get away from the whole total vibe hes got but find a small melody lines for a motif to permeate the tune.

[Anyone else like that idea, have at it]

OK. Its that time of day where I seem to be talking to myself on here. Lol. I think ill have a chat with my washing machine now.

[hard beacon]

whazz up 😉

[pliant seal]

That was a really good idea. I made the female version of the whole song — even using the female tag, adding it to the lyrics and the style it still took a lot of takes to get one! Then I took that into logic, split into stems and sung all the male parts. Then I took that back into Suno for a remix (to hopefully improve my voice) and guess what: it still swaps the genders on random lines! Male/female parts that vary thru a song seems to be cryptonite to Suno. 167 songs later I’m not yet given up but getting worn down. 🙂

[lime canyon]

Oh, I never actually thought about the persona route. Might give that a go, and then I'll let you know how it goes, I guess.

[half dragon]

I don't think you can create Persona from someone else song; unless there's a way I'm not aware of.

[neat field]

private studio uncut acoustic session, post-emo
0% Weirdness
Style Influence 100%
Audio Influence 77%

this produced a nice acoustic conversion to my ear

original:
https://suno.com/song/7d353a4b-a1dc-431f-8bdc-4bd9de470103

hymn #5 2025 (B) (Uncut Acoustic Session):
https://suno.com/s/YEEDEv2YVJ6YTUlk

[polar spade]

Hadn't thought of that. I forget those things. I dont remix peoples stuff for the most part.

[hard beacon]

nah come on.. what's going on?!?!?!... go fu** yourself

[near flume]

That is correct.

[devout sinew]

I have a question about prompt formatting. Since the styles description area has limited character space, is it ideal (or at least acceptable) to include detailed vocal instructions directly in the lyrics area? I am referring to things like [vocal cry], [glottal stop], [whisper tone], or [breathy delivery], using square brackets [] to make it clear they are not meant to be sung.

If using the lyrics area for this purpose is okay, how much detail is too much? Can I provide vocal phrasing notes for nearly every line (e.g., dynamics, inflections, tone changes), or would that overload the model or interfere with performance?

I want to make sure I am helping the model—not confusing it.

Thanks in advance for any advice or clarification!

[lost holly]

Yes, you can add Style instructions in the lyrics. Personally i prefer to use it like [Verse 1, female whispering vocal, slow]. You also can add [Male/Female vocal] and other instructions before every line. My recommendation here is though to limit it, as much as possible, for example 1-2 instructions in the [Verse/chorus] part.
In my experience it does work best, when you use as few style instructions as possible. The more you use, the more tries you need for a good result in my experience.
The hard part in all of this is learning what style instructions Suno actually knows and thus needs and which are irrelevant.

[devout sinew]

Thanks so much for the helpful reply!

That makes a lot of sense, especially the part about limiting vocal instructions to 1 or 2 per section for better consistency. I was wondering, when using more expressive or emotional phrasing like [vocal cry], [glottal stop], or [crack in voice], do those types of instructions actually get recognized by Suno? Or are they considered too detailed to be reliably processed?

Also, has anyone experimented with embedding vocal tone or delivery per line in a subtle way, like:

[female vocal, head voice, soft rasp] I tried to hold on, but my hands let go
or
[vocal fry] Why did you leave me in the dark?

I am curious where the line is between what is helpful and what might be ignored. Any insights from your experience or clarification from the dev team would be really appreciated

[lost holly]

soft rasp is one of those instructions that Suno knows but not necessarily follows. That what i meant with you have to figure out what and what not to use 😄
I would recommend you to start out with common phrases, that often appear and then go into more detail. As well i would recommend to limit the style instruction to 3 words max and not more then that. In doubt look up special words for what you want to achieve instead of using more words.
Its hard to generalise an answer here, since some things might work better in the way i prompt and some better in your prompting style.

For me it works best when i use as few instructions as possible in the most specific ways to describe of what i want to achieve, so that there is as little room for interpretation as possible for Suno.
Other then that you have to learn certain style connections, for example that country often contains an acoustic guitar, so if you don't want a guitar here, you should use other styles.

[livid trout]

Something weird is happening in the editor. When I want to try generate a better intro that is in tune with the rest of the song, the 8 bar structure of the intro is lost and replaced with an intro that overwrites the 2 first bars of the first verse,

I think its because the intro "flam" before first measure of the intro. Does the replacement algorythm gets confused with that ?

[slim elm]

Hey guys I hope y’all doing good

There’s probably a bug generating videos on iOS

It was working fine until this morning

Videos generated previously I can still dl them

New songs or old songs if I haven’t generated the videos I can’t do it now anymore

[half dragon]

Videos need to be generated manually before they can be downloaded (for private songs)
Published songs - videos will be generated automatically.

[slim elm]

I know otherwise I couldn’t have made my previous videos

The issues is now

It’s broken

After creating new songs and all adjustments are done we must manually generate videos

Once the video is generated we can dl the video and its appears like this below when it’s working correctly

But currently it’s not working and it’s showing like this below so I can’t dl my videos

[half dragon]

Report it in #🪲┃bugs-web if you haven't.

[devout sinew]

Hello again. I am running into a problem when generating cover songs. The audio often cuts off abruptly near the end of the final chorus or halfway through the outro, even though the full lyrics and timing are supposed to be complete.

For context, I do not enter any style or structure information in the prompt when creating the cover. I just select the original and generate the cover version.

Is that why it is getting cut off? Should I be adding a prompt or instructions to help Suno complete the full song? If so, what kind of prompt is recommended for a cover to avoid it stopping early?

Thanks in advance for any help!

[pseudo blade]

I got a special workflow but Woodhouse deleted it :(

[slim elm]

Thanks @half dragon

[shut ferry]

Best way to create a progressive rock/metal song without Suno having the singer scream almost every final word of each line? Literally put 'no screaming or no head voice' in the style and yet the sustained notes continue.

[half dragon]

Extend one of the best versions to finish it.
Set the extension point after a line in that final chorus, let it remakes and finishes the outtro.

[young geode]

https://discord.com/channels/1069381916492562582/1382019029266600067

In there is a series of guides to help you with this.

The [Bracketed Instructions] are referred to as meta tags. Be aware that not just any instruction or reference can, and/or will be picked up.

From my testing, the best practice is to name the musical quality you're targeting, then provide as direct a description/instruction you can.

For instance:

[Guitar] may be picked up, but [instrument: guitar] has a far higher chance of being recognized, while [instrument: Guitar, Stratocaster] will be recognized and give a specific sound, with [Instrument: Guitar, Stratocaster, in E Major, hammer-on arpeggio] being the best.

[young geode]

When you say scream, do you mean the power metal scream?

You'd need to target power metal, speed metal, Symphonic Metal in the Exclusion prompt box, and add in vocal holds, vocal sustain, vocal fermata as well.

You'd also want to add to your style prompt Vocals: Male, Baritone, Staccato, Bark-like style

The reason for this is that power/speed metal is where you get that high note sustain from, while you're looking for a crisp, enunciated vocal style, which is called Staccato.

This should help fix it. If it doesn't, you may need to employ some meta tagging and lyric restructuring.

[shut ferry]

Thanks, I will give it a try

[young geode]

@polar spade

Using the editor, am I able to splice in stems, and other gens, that have been generated together off the same base?

I noticed that it was giving me options from other songs when in there and ok not sure how it works and the degree of control I have with it.

[half dragon]

I don't think it's possible yet but I'm curious of the options you mentioned as I don't see it.

[devout sinew]

Hello. I have a question about Suno’s public sharing settings. Is there a way to disable or delete the original prompt after generating a song, especially once it has been shared publicly?

Right now, the only options I see are remixing or editing, but those do not fully solve the issue. For example, remixing sometimes cuts the song awkwardly, such as in the middle of the final chorus or during the outro, which disrupts the flow.

If there is any way to turn off or remove access to the original prompt after generating, I would really appreciate any guidance. Thank you!

[near flume]

Once a song is generated you can remove the prompt. Your only option is to do a Cover with the Style box empty. This "should" generate a very close match to the original but there maybe slight changes. Publish the Cover version, only the lyrics will be visible.

[trail fulcrum]

I am attempting to surgically replace an instruction I missed in lyrics, but I love the tone shift so am rplacing lyrics to keep it.... NOT WORKING. I have chewed up 1000 credits in attempts.

[trail fulcrum]

To avoid MUCH frustration, I always have ChatGPT open while creating and run my issues through it as an advisor of sorts. I believe it has access to most info on the web which would include forums like this. I find it super helpful..... Good luck!

[pearl arrow]

Ho un problema con la Libreria SUNO, nel senso che mi dicono in giro che se voglio depositare musica AI presso una società di protezione diritti, per dimostrare la mia attività di "arrangiamento" musicale dovrei dimostrare il lavoro svolto. Allora ci sarebbe bisogno di stampare l'elenco di tutti i brani e quante volte sono stati provati e ripetuti per pretendere royalties. Si può fare una stampata dei file audio della libreria? grazie a quanti vorranno rispondere.

[sterile moon]

Is there currently a way to add choir? i dont want lyrics added though

[polar spade]

Thats something many ask for, just splicing together the best bits from different outputs of a similar song or edits of the same original, but nothing like that yet. I haven't seen.

[sterile moon]

also anyone else having issues with the ai not following style instructions?

im wasting a ton of credits when its not following my prompts at all

[polar spade]

Choir can mean a few things so you might detail what your trying to do. Just putting choir as a prompt without any lyrics on a first creation? It might hallucinate, or you can put "ahhhhh" and you may get a lot of ahhhhhs, which may be what you're looking for.

[sterile moon]

good to know, i havent tried adding choir yet i didnt want to waste my credits on it

but is there a way to get credit refunds? because i dont think its fair to use my credits when it didnt even follow the prompt...

[polar spade]

But what specifically is your goal. Im not testing it either for you unless I know a bit more. 😄

[sterile moon]

im just trying to get it to use a composers style of arrangment, but some some of the flair in her songs have choir vocals in them.

[polar spade]

A cappella? Boys choir? Gospel choirs? Backup choir? A choir sounding epic metal ambience?

[sterile moon]

sorry, yeah guess i can be more specific. More like church choir

[polar spade]

Such as Enya or many New Age styles?

[sterile moon]

erm, not sure what those are.

[polar spade]

im sure nothing as Gothic as a Gregorian chanting .... but you say composer... as in classical style or more just a modern piece. Atmospheric?

[sterile moon]

modern ppiece, atmospheric definitely

like everything its outputting has sad music piano tones, and thats not even in my prompt

i have 'epic' repeated twice

e.e

and the

'exclude' styles option doesnt even work

and it still does

[polar spade]

Ok. Got it. I think i know what your trying. Let's see. Gmmm.... and atmospheric" would be one prompts. But instrumental. Probably using Custom setting, not Instrumental.

[sterile moon]

mb woodhouse

[sterile moon]

but what about the AI not following prompts?

[polar spade]

After near 2 years, I dont know what "follow prompts" means. Suno looks for similar attribute which prompts have and using more than one prompt can mean you open up possibilities to be random by 100 fold.

So, back to the atmospheric piece with choir. Let me check on something I think I did a month back.

[sterile moon]

.... the prompt you put into the description?

[young geode]

Why am I getting 'alternatives' then? Have you seen this?

[polar spade]

Description prompts, yes , is where Im indicating. But even in lyrics we can tag prompt for a few things as well.

You said a composer, who is it, if I may ask? Jazz, classical, avante guard?

[sterile moon]

Yuki Kajiura

[polar spade]

Intros are good to start with ahhh and oooos without lyrics.... ill grab one with African djembe and kalimba sounds... and add choirs prompts, see what happens.

[sterile moon]

She's made some beautiful combat scores for some movies that i want to tap into

[polar spade]

Oh cool. Ill check her out. She's not the jazz sax player right. I dont think so. Anyway, brb.

[sterile moon]

no she doesnt do jazz

shes mainly an anime composer

[polar spade]

Score composer... got it.

Cinematic is a good prompt for that too.

[sterile moon]

it is hey? kk ill try that

[polar spade]

Yea. Darned, I wish I could remember I had something just like what you need. Grrrrr. But its been a couple of weeks.

Ok. Ill make one up. Let's see. Ill burn my creds. Hold on.

I had an X named Yuki. So I already hate this. Hahaha.

[sterile moon]

no need man, theyre precious

[sterile moon]

oh god lol

[polar spade]

Its what we do. I've got 30000 at the moment... im good.

[sterile moon]

lol

[pseudo blade]

@polar spade Hey I have a brand new workflow, would u mind reviewing and looking at it?

[polar spade]

Ok. This one isnt bad on a first generation. https://suno.com/s/VdLQbYH5NoawwabH. Then chorus comes in about a minute and the ending is nice.

The second output was OK. Similar. But not as driving. So its trash.

Anyway, I think you can see we can get a choir to sing ooooos and ahhhhhs without much effort.

[sterile moon]

ill use this as a template, thank you

[polar spade]

Workflow? As in.....? Well nevermind... is it posted. Ill go see what youve done. Sure. Anything anyone does is usually helpful, right.

[pseudo blade]

Yeah it’s posted

[polar spade]

As you can see, I may try some random prompts to nudge Suno one way or another. For instance, when speaking of an epic orchestration, the drive of a metal guitar can be an influence, even if we never hear a metal guitar in the piece.

[polar spade]

Is it Easter? Or do you have a link.... lol

[pseudo blade]

https://discord.com/channels/1069381916492562582/1401551088930259084

[polar spade]

Wow! Pretty extensive. Ill be using that to help me think up some new ideas. Very nice , Coco. You must have spent some time on that. Good focus and laid out well, also. Youre a genius!

[pseudo blade]

5 hours spent, but hey at least my summer school finished

[polar spade]

5 hours. That would take me a month. We'll, frankly, I never finish what I start. OK. Im hiring you for all my projects now. Haha.

[pseudo blade]

🤣

[placid aspen]

Hi. I'd like to play around making cover versions of some tracks by other mainstream artists, but I'm guessing that just isn't possible due to copyright issues etc?

[polar spade]

Moreso due to it being material that is available to the public so not exclusively created and controled by you. Copyrighted work is included, but this includes pretty much ANYTHING publicly available. The ABC song, a weather report, poetry from the 3rd century, mother goose rhymes, producer tags, quotes, and so on.

[placid aspen]

Thanks

[cursive island]

can suno do car engine sounds?

[sleek stump]

I don't know about 4.5+, but every previous model I have not been able to get any automotive sound effects. Suno isnt really trained on SFX, so unless the songs it is trained on have engine sounds, its going to be really difficult to get them.

[weary glen]

hey guys how do i use the same kind of music i make and just change the lyrics of it?

i want the same instrument and everythings

[near flume]

Have these all been tested in Suno or is it just a list of genres/sub genres?

[near flume]

Suno is horrible at SFX, better to add them in post.

[pseudo blade]

just the list, i accept genres that are can be used as examples

[weary glen]

someone?

[polar spade]

Changing the lyrics will make a new meter and melody, so suno would sing it differently.

But.... Paid subs can disect the instrumental stem (Get Stems) and add new lyrics to the instrumental though if they wish. Remix/edit Add Lyrics.

[winged peak]

There is no 100% guarantee, but here is what I would do: 1. Create a new playlist. 2. Add only ONE song to the playlist. 3. Click on ‘create’ and create a new song. Click on ‘Inspo’ and add the playlist. Now enter Lyrics / Style Prompt and move the ‘audio influence’ slider to 100% in the new settings. 4. Cross your fingers. Best, Tia!

[weary glen]

the same style right?

[winged peak]

Yes, but you'll figure out that this way Suno most likely even ignores the style-prompt 😉
Not always, but sometimes.
So, the 100% audio-influence-slider somehow overrides the influence of the style prompt, which still is been used.

[weary glen]

its sound so different for me 🙁

[winged peak]

@weary glen Do you have both links for me, I can offer to have at least a quick look and listen.

[polar spade]

I like when a good idea makes it around. 😉

[polar spade]

OK. Same style? Separate out the instrumental track. Add vocals. #📚┃suno-school message

Are we assuming you want the melody to be exact, but wth different lyrics? 🫣

[sleek stump]

I've done a fair amount of experimenting with adding different lyrics/vocals to a instrumental stem, WHat I have found is that the backing track is mostly the same, but changes happen to accommodate the new lyrics and their structure. I can tell its the same general arrangement with the same band makeup, but its just not "the same". I'm not willing at this point to just burn credits experimenting with this limited use feature with all the myriad slider settings.

That would have to be left to the user wanting to make that feature work for them.

[polar spade]

Yep. It physically won't work as AI re-tokenizes the new lyrics. Its about an 8 step process to get something close , if we want the same melody. And even then..... 🤷

[sleek stump]

It does a really good job keeping the same-ish melody and composition, but I feel its just not going to be able to keep a backing track the exact same unless every stanza has the exact same sylable make up and flow. And in my experience thats going to be a really hard thing to do unless you are doing a parody, which historically has that exact thing, same syllable and flow as the original. Think Weird Al and all his parody tracks.

[graceful acorn]

Has anyone found a way to make a persona without the system saying you're using some random artists name? Every word is a band apparently. Making that feature completely useless.

[sleek stump]

just give it a non descript description, and just a s ingle name.

[graceful acorn]

I've put in random words and still someone has aband named Trash Dirt Feature

[sleek stump]

Its GPT, the assistant they use is really squinchy

[graceful acorn]

And in any case if I can't organize with it so I know what is what in my own persona list, then I don't know what I'm paying for anymore.

[sleek stump]

Persona is just a very micro part of what you "pay for"

[graceful acorn]

Oh it felt like an older non-ai system like every word that is tied to a band is not allowed.

[sleek stump]

But if thats the one thing you cant do without, then it is what it is.

[graceful acorn]

Yeah it's just frustrating. Straw and the camel back I guess.

[sleek stump]

I have half a dozen persona, and I never use any of them lol

Ya, I dont quite care for their use of GPT for moderation, its frequently just flat wrong. But it is what it is, we learn to work with or around it 😉

[ebon sand]

i have around 400 personas because i also use them for in between steps in generations but i just favorite the few i repeatedly use, i'd rely on a numbering system though, also have one persona which reminded me of led zeppelin so i did that

[graceful acorn]

Hmm that might work. Thanks.

[sleek stump]

Thats the work around part 😄

[graceful acorn]

Something tells me the AI will figure out what we're doing and ban those too. But until then that'll work.

[ebon sand]

pretty sure it's just a manually added word blacklist and they forgot to set it up correctly because it also hits symbol combinations within a word and that's not common practice, likely a mistake by the ones who built the filter

[sleek stump]

Ya, pretty sure there not a lot of maintenance happening on the assistant.

[grizzled ember]

Yeah I'm going crazy over too high pitch. the hack I'm trying add soulful voice & after you make song add where it's too high add that on lyric line & add negative prompt, no inhuman high pitch, add sombre if it's sad

[torpid thicket]

How the hell is this a 'Producer tag'????

[sleek stump]

saying from Hide & seek... it appears all over the planet for 50 years 😛

[torpid thicket]

Over 50 years, and not copyrighted

[sleek stump]

The GPT filters are super sensitive, Suno doesnt want to risk any possibility to be pulled into any more litigation.

suno.com/feedback

[devout sinew]

jschires — 5:59 AM
Once a song is generated you can remove the prompt. I can? How?

[near flume]

read the next line.