📚┃suno-school

[still holly]

Check this this song I've used this to use male and female verses https://suno.com/s/YkD1W67cdmX480Ig

[dusty tangle]

Course Suno will just do whatever it feels like.

[humble verge]

you don't use brackets?

[still holly]

Not in that car and it seemed to work 🤷

[naive ibex]

Hi all, is there a way to set a lang forced? For example I have text in English, but want to that singer with French or German accent.

[cosmic mica]

yes, you just have to translate it

Just wondering — does anyone here have the “Share as video” option when clicking on “Share”? I use custom lyrics, but I don’t see that feature yet.

[polar spade]

For the video with lyrics, we would generate the video (by clicking Generate , under download) first then we can download the video and manually share the m4p file, or share the link address. Paid subs only.

I may be wrong, but last I checked , I dont believe there is a specific "share as video" option.

[cosmic mica]

However, I saw several Suno songs on video. I looked into it and found that there was a feature on Suno but it wasn't available to everyone yet, but I was probably wrong then.

[uncut pumice]

Weird, you can't re-master with 4.5+? When will that be implemented? I now have songs that were originally 4.5+...that are going back to 4.5? 🤔 Why does that even make sense? Wasted credits !!!!

[near flume]

It's hit or miss on having English being sung and having an accent, any kind, come through.

[cosmic mica]

@near flume well what also works is to write the accent in the style of the song and in the lyrics write as we would say with the accent

[undone scaffold]

how do i keep the new regenerated outcome and get rid of the lyrical duplicate that it made?

[uncut pumice]

🤔

[radiant breach]

Dear Suno, there is no female in my persona and I have "male" selected..... so please tell me why your robots decide to put female vocals in my song. Makes me question WTF is actually going on behind the scenes.

[sleek stump]

You can nudge Suno into doing this by describing the singer type in styles...
The German male vocals sit front and center
This was from the magic wand, my entry was just "German Male Singer"

[sleek stump]

Its not an option yet. I would imagine it gets added as the model matures, its still a beta model so changes and features are still in store for it.

[radiant breach]

https://tenor.com/view/where-you-at-gif-21177622

"we fixed personas"

[sleek stump]

Theres the https://discord.com/channels/1069381916492562582/1082775596767117312 channel, and theres suno.com/feedback
This isnt really the place for complaints.

[random stag]

Is there any way to prompt a human to produce a song that uses one version with a female voice and the other with a male? If they’re going to give me 2 versions, I wish they would just default to this unless I specifics one or the other

[sleek stump]

I'm assuming you are wanting SUNO to make one male and one female vocal from the two generations each time you click Create? That can happen naturally when you do NOT specify a vocalist type, however is very rare. The way that generations work it is not possible to prompt for this.
When you hit Create, Suno will take your prompts and generate 2 arrangement variations, they will never be the same.

[main cairn]

Hi all, any reason why the advanced options, like weirdness, style influence, and audio influence would be grayed out in the suno iOS app? I am making an instrumental song from an audio upload.

[polar spade]

You can use Edit Ui and just select and remove the duplicate, is one way.

[buoyant lotus]

I saw a post on reddit about accents, the person had some results when changing the letters to letters specific to the language. Like writing in english but adding some random "Ø" for nordic accent. Maybe you can try throwing in some é, è, ê for french accent ? or β for german ?

[random stag]

Do you think it’s worth suggesting?

[sleek stump]

YOu can suggest any feature you want, the likelihood that it will go anywhere is slim, the way the engine renders would have to be changed and I dont think they want to deviate that far off their current path.

Just understand that it will never give you the same arrangement or vocal style between any generations, that is what cover is generally meant for.

You can already take a song and cover it, changing the gender of the vocalist.

[fluid estuary]

Is it random if you can create personas from audio uploads that u remix?

[plucky bloom]

Hi, is there a way to ensure that tracks 4 minutes and longer seem to be getting a lot of short tracks under 4 minutes?

[sleek stump]

Not sure I understand what you are asking here... Are you trying to get tracks under 4 minutes? What model are you using?

[plucky bloom]

@sleek stump trying to create tracks that are over 4 minutes (over 5 minutes would be great), seem to be getting a lot of instrumental tracks between 2-4 minutes, which im have to trash.

[compact shuttle]

More stuff in lyrics box x longer style equal longer track

[plucky bloom]

4.5+

[sleek stump]

What model are you using and what are the genre you are trying to use? song length is very much tied to genre

[compact shuttle]

For example

And this is 4.5+

Here's an example, a emotional blues. Clocked out the intro but that's alr

Outro*

[sleek stump]

I can easily get 6-8 minute instrumental tracks, but its usually because of the genre I am generating

[compact shuttle]

Lotta older styles = longer songs

[plucky bloom]

House Music - which kind of has to be longer to be able mix tracks together, 2-3min makes its difficult to blend tracks and the track is over to quickly.

[sleek stump]

I also have a different approach than most, I use symbols in the lyrics to influence iintrumentals

[compact shuttle]

Ooo tell me about that sometime cus I tried it and literally got like question mark emoji smiley face

Yada yada yada

[plucky bloom]

thanks anyway, mine are only instrumentals, i'll keep playing, just burning through credits 🙂

[compact shuttle]

You can use meta tags

Yo control the instrumental

[sleek stump]

Here is an example of one of my methods. https://suno.com/song/8d571a8a-857d-424a-b601-47ce0615705e

[peak delta]

Any suggestions on how to edit a single-word issue. Harod (as in King Harod) is mis-pronounced on my song. I don't want to change anything else, just that one word. I have respelled it to Harud to get closer to the proper pronunciation but the edits have changed more than just the one word.

[round crown]

I am new here and brand new to Suno. I already have a song that I really like, but it needs a few minor tweaks. Is there anyone that can help me do it?

[polar spade]

You can add extensions which can give slight change ups in the motif or go 180° , depending on the prompts and persona used, or you literally may make duplicates of different sections and add as many as you want.

[tiny rivet]

What do I have to do to make it work like this?

[Bridge - slower, with a gear vibe / spoken like on the street]

[polar spade]

My first thought is just use an extension to add whatever prompts to change the flow. Redo and tweak until something you like, then Get Whole Song.

[frank obsidian]

If the purpous of paying for the pro membership is so you songs can be PRIVATE, why tf is this the default. At least add an option to turn it off for EVERY song. this is ridiculous. @open magnet

idk if you are a developer, but please contact a higher up if anything.

and the "private" button on all the songs are GONE.

@steady kelp

(sorry but georg didnt respond)

[sleek stump]

We have already had many long discussions about the default setting for remix. This feature went in a couple months ago, most of us have just adjusted our process to just turn that flag off when we post a track.

And pinging multiple team members wont help.

And finally, the subscription has nothing to do with if your tracks are private or not, it covers ownership, commercial rights, and adds features. All songs are private whether you are free or paid, until you publish or otherwise post a link to any song. @frank obsidian

The permission for remix is found in the 3 dot menu under Visibility and permissions.

[frank obsidian]

quit ur yapping. i ping multiple cause they wouldnt answer. fortunatly, other members helped me out.

[frank obsidian]

you cannot private your song unless its a membership

[flint elm]

Any idea how to get a certain song length? No matter what I try in 90% of my songs suno stretches it to the 8 minute mark and the sond is not even finished. Even if I put stuff like "don't exceed a playtime of 5 minutes" it most of the times doesn't work. Same with "Keep it under 5 minutes". whatever. Maybe someone has a suggestion

[frank obsidian]

atleast the first year i joined

[sleek stump]

All songs are private until you publish them.

[frank obsidian]

must be new cause that was one of the main points advertised to me initially

[sleek stump]

Ive been here for over a year, and the remix permission is new as of a couple months ago.

[frank obsidian]

okay.

[sleek stump]

Even before hand, unless you posted a link to the track somewhere or published it to public, it is virtually private.

[frank obsidian]

that has been clearified. thank you though.

[shadow plaza]

What styles is best for the new model 4.5+

[sleek stump]

So far everything I have tried has turned out really good. You need to try your own and make your own opinions 🙂

[shadow plaza]

Show me examples

[sleek stump]

If you want to hear what differnt genre sound like, hit the explore page and listen to samples from a lot of genre.

[shadow plaza]

But thats there isnt 4.5+ model i ask precisely

Just show me your music on the newest model

[sleek stump]

I posted one of my Jazz Fusion tracks a couple pages up in this thread...

[shadow plaza]

Do you have metal rock

[sleek stump]

Yes

[shadow plaza]

Im not into jazz

[sleek stump]

Metal has had a vast improvement

[shadow plaza]

Let me hear it

[sleek stump]

Heres a 4.5+ remix https://suno.com/s/giBo0BuPsiFhjQvJ

[shadow plaza]

Thanks

[dense nexus]

Has anyone learned how to get Suno to create an extended length (over 4 minutes) track? What prompt does Suno recognize to set the track length?

[dense nexus]

What about Southern Rock? I just released a full album on all the streaminig services. It's titled, "Southern Crossroads", check it out.

[dense nexus]

I'll run back in the thread and take a look. Check out my recent album release titled, "Guitar Groove", that you might like.

If anyone would like to listen to some of my recent music created with Suno v4.5+, visit my website at https://www.scancotech.com/songmaster and click on the Discography link.

[plain pumice]

Anyone knows how the advanced options work when remixing using the persona feature, more specifically with regards to
"Style Influence", it says it addresses how much it adheres to style description, but is this the persona's or the current style's prompt description?

[polar spade]

Did about 5 replacements edits and all perfect on the first gens. No addit. edits needed. That was nice.

[plain pumice]

Not exactly sure, I used:
Custom Prompt: [Tempo: 120 BPM, expressive 4/4]
Style Prompt: [Tempo: 120 BPM, expressive 4/4] - To reinforce it.
It's possible that there is something else to it, like the actual musical content with regards to the composition.
I used 4.5+, all 5000 characters, and got a 7:52 length track.
..as soon as I plugged in a persona, that number got close to halved, which used a track length of 7:13 - each one came out exactly: 4:17.

[near flume]

Most of mine are over 5 minutes. It all depends on the amount of lyrics and structures you use + the tempo.

[dense nexus]

Thanks. The prompts I've been using are very similar. I've tried using a duration prompt such as [duration: > 4 minutes] or longer but it didn't seem to make a difference. Some of my tracks are greater than 4 minutes, even longer than 7 minutes, but I can't seem to figure out what's different.

[near flume]

That's been a long standing bug with Persona & Cover. Typically Persona uses the run time of the song the persona was created from and then stops.

[near flume]

time duration isn't a valid prompt, same with trying to put time spans in the lyrics.

[dense nexus]

Typically my lyrics are 2 verses (sometimes 3), chorus, bridge and outro.

[sleek stump]

Suno has no clue what time limits are, it makes songs that fit the logic of its training.

[dense nexus]

What if you instructed Suno to include an instrumental interlude between say the repeated chorus and bridge, or bridge and final chorus, etc? This is especially useful and characteristic of rock, southern rock or folk rock genres/styles.

[sleek stump]

You can insert (and I do often) instrumental sections in songs to give vocals a rest and let instrumentation shine, but which type works in which part of the song depends on the genre. Breaks, Interludes, Instrumentals, Solos, they only work properly in certain sections.

[wild egret]

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

[sleek stump]

Try Ambient Lo-Fi, Dreamscape

Maybe even add down tempo bass

[wild egret]

close but not there yet, it adds way to much lo-fi style, imma try without that

[wild egret]

look https://suno.com/s/ODBRfO8J1OcAnEu3

[sleek stump]

Check these...
https://suno.com/s/iybk8HAF05gi9LfH
https://suno.com/s/xtlqmJqBJZEAOYis

[wild egret]

beginnings are nice on both, but the it turns way too much into lo-fi, same with mines

those from the yt video are kinda darker, more reverb, and echoing, and almost no lo-fi

[sleek stump]

They are also constructed in a DAW not in AI

Digital Audio Workstation

[wild egret]

yeah i know, at first i though it was ai, but seems like ai is not there yet right?

[sleek stump]

Its what a lot use to process and master tracks

You can get some really good stuff with the AI, just need to learn the prompting to get what you want... that is not a quick or easy feat

[wild egret]

yeah but i dont know nothing about music sadly

[wild egret]

i've tried many, but it ends up going lo-fi even when not mentioning it

and cant fit everything in 200 characters

thats another issue i have

[sleek stump]

There are a lot of electronic sub genre, you just need to find teh right combination for what you want.

Oh, I see, you are using 3.5, free user?

[wild egret]

yeah

that limit is gone with pro?

[sleek stump]

Ya, 3.5 is likely not going to get you want you want here, sorry to say

[wild egret]

havent paid it yet because i thought i wouldnt be able to reproduce those songs

[wild egret]

is there that much of a difference?

[sleek stump]

3.5 is good for some genre, 4 some different ones, 4.5 different strengths again, and the new 4.5+ has just surpassed everything else.

But the new + model is still in beta and has some quirks and issues that need sorted.

3.5 uses simple style prompts, it doesn handle complex styles well at all.

[wild egret]

hmm okay

there's no trial for me to test if the songs will outcome how i want them right?

[sleek stump]

Not really... theres no guarantee any AI platform is going to give you what you are looking for.

[wild egret]

with the 4.5+ model can i upload a sample and make a whole song sound alike?

[sleek stump]

similar... something like that, but there are rules for uploading samples... you cant upoad anything that is recorded or owned by someone else..

[wild egret]

okay so i wont be able to upload a fraction of that video

[sleek stump]

nope, that would be a violation of the TOS and a copyright violation for the video content.

[wild egret]

got it

[wild egret]

no copyright free netiher?

[polar spade]

There are a few free genre analyzers out there to use. I like audioaidynamics.com for instance. I write down every genre and sub genre it notes, throw into suno to get a good close start. Then tweak and weave the styles around for a few pages of generations until I something closer that I like.

[sleek stump]

Correct, free user do not own or have commercial rightsa to the songs they generate, only private non commercial use.

[wild egret]

Thanks for the tip, will try that too

[wild egret]

has it something to envy chatgpt? because i tried that too and its decent to my understanding

[polar spade]

I use to help other do that. But most have an idea or a "feeling" of what they want.... usually an exact copy, so most songs to get exactly or really close takes pages of outputs and tweaking.

[sleek stump]

keep in mind that Suno is a creative generator, its not meant to be a copy machine

[polar spade]

Its free and easy. You drag the YouTube address into it. And hit play. It analyzes every second and have a live analysis of the vibe its hearing. I get a pencil and paper and write them all down. At the end it will typical still show the top 8 or so.... but you can also click on the waveform and find the genre at each point

[wild egret]

i dont want to be an exact copy either, that wouldnt be interesting either, listening to the same songs

[polar spade]

Then your taste still differs from mine. So id prefer you use your credits. 😁 but most times, I like the close proximity that it give as a starting point.

[wild egret]

oh okay, i uploaded a fraction and this is what it generated, seems worse than chatgpt so i will try that other functionality with the yt url now

[plain pumice]

Yes, that one is really good, though the actual genre analyzer, can be a bit confusing, as it can list quite a lot - not to mention the very broad genres as well can be kind of vague.

[polar spade]

Just let it go through the whole song. Then go back and scrub through the song. But also below is the colour's of the most common/frequent genres found.

[polar spade]

But you have it as an option. Im not selling you on an idea. Just saying the option is there. You asked, well Jony asked. 😆. So, Good luck.

[cyan light]

@wild egret You can try "ambient, dark ambient, emotional, sad, introspective, slow tempo, C minor, chill groove, meditative, soft textures, cinematic, no lo-fi, 91BPM" using Suno 3.5

[wild egret]

when using "no lo-fi" suno understands LOFI LO FI LOOO-FII 😂

made 4, first 2 turned out to be mostly classical piano, for some reason, the other 2 pure lofi

[cyan light]

@wild egret I see what you mean, even this is venturing into lofi https://suno.com/s/sy4N5TxkkEWMhtpv and https://suno.com/s/qtlv6ZhQg1SMzBzJ

[echo prawn]

Is there any way to let AI know that singing should pause at designated places?

[cyan light]

@echo prawnMaybe if you tag it right—like ‘spoken word, dramatic pause, cinematic narration’? I have never tried

[wild egret]

yep, seems like cant exclude lofi with free tier

[echo prawn]

I am currently working on a cover of a song, but the AI is unable to sing along with the pauses in the original song's lyrics.

[cyan light]

@wild egret I tried " no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi, no lo-fi" Iam sure you can guess the outcome though...however it was always 4min full tracks of lo-fi...

[wild egret]

Dang it, who would have thought 🤣

[cyan light]

@wild egret If you have any credits left I have this crazy almost scorcery like thing you can try and get pretty good results maybe 1 piano in every 2 results but the other just may shine...

[wild egret]

Sure! Let’s see

[polar spade]

Understanding that Suno ignores the word "no", all it sees are the styles, thats one reason why the exclusion field was introduced.

[wild egret]

Yeah I imagined that. Like every other ai tool has the “negative prompt”

[sleek stump]

Anything you put in the Styles window is treated as a want. Only the exclude window tells suno you dont want it.

[magic shale]

how do you find the 4.5+ model ? For me is simply incredible !

[cyan light]

@magic shale The "gear" icon at the top of the screen on mobile and PC but also on PC just click three buttons to the right of the gear on the current displayed Version IE v3.5 and a drop down menu will appear

[magic shale]

Ah, thanks! I know where to find the function! I asked actually "how good is 4.5+ for you? " !

[eager compass]

I haven't found a surefire way but a few things that give ok results, depending on how long of a pause. There's the classic 3+ periods"...", dashes "----", and separating lines by hitting enter a couple times for a slightly longer pause.
" "
Sometimes it works.

[oak cobalt]

How to prompt to increase the length of Instrumentals/low character count songs? I get about 2-3 minutes songs but want some longer ones.

[hard beacon]

well.. try it with song structure meta tags

[ebon sand]

Negatives often are problematic for AI, it's like telling gpt to create an image without an elephant in it, most of the time there will be one

[brisk flint]

Don't use "NO", Suno does not have an understanding of 'no', so whenever you WANT to have a saxophone, just enter 'no saxophone' as a prompt. 😉 Think of how we humans do process information: Don't think about a pink elephant now!!! So, what just happened? 😉

[brisk flint]

It depends on the genre, but you can extend the song length with an intro, breaks, additional verses, etc. This often works, but not always. Accordingly, I prefer to use the old “extend” function to significantly extend a track and then use “get full song” to merge the individual parts.

[real hemlock]

Hello everyone, I'm a hardcore user since May 2024. More than 7000 generated songs and I've NEVER been able to create a drum solo in my rock / heavy metal songs.
As a drummer, I love when there is a drum solo inside a rock song.
If someone knows how to do it, please tell me.

I've tried [drum solo instrumental] and so many other things, written in the style "drum solo breaks, fills" etc.
Instead of drum solos, I always have guitar solos...

thanks for your help

[polar spade]

I can easily NOT think of a pink elephant🧏‍♂️. Watch me. See...
I'm now not thinking of a pink.....
...... doh! 🤦‍♂️

[kind wing]

Hi, Is there an option to stop random music from playing on our suno profile. No harm meant.

[polar spade]

The three lined "hamburger" menu in your playback banner at bottom will pop up a queue box. At bottom of the queue box is a double arrow and an infinite button. Clicking OFF the infinite button so its greyed out will turn off "play next song after queued songs play".

That can help. But honestly, i think it resets and defaults back to on in certain situations.

[hardy thunder]

@sleek stump At some point, when Suno produced guite "muddy" or extra reverb kind of sound for some genres, there were instructions to use excludes to avoid that. So I still have had a habbit of putting something in the exclude box.

I think when I started to use excludes, the song summary showed the exclusions with a hyphen (-) in front of words in the exclusion list. Like, if I wanted to exclude words "reverb" and "echo", in the end the summary there was a list like "-reverb, -echo".

But this is not the case anymore. The exclusions are added in the end of summary without the hyphen. So, for a new song if I just copy the summary as a text and paste it to style box, my exclusions are most probably treated as instructions to do the opposite I really want.

Ofc I'm not doing plain copy-paste, I use the option "Use styles & Lyrics", but anyway, it is kind of confusing and a bit misleading too.

[rich junco]

Where is the location of the "queue"?

[eager compass]

Bottom right on browser. Next to your volume, thumbs up and thumbs down options.

[rich junco]

Thank you so much

[eager compass]

You're welcome.

[mellow mural]

Can someone help me or point me in the right direction... im working on a piece.. a duet.. and I cannot get it to sing the lines in the correct voice.. Male/ female.. it's all over the place a mess.. I've tried a few different things.. but i wonder if there's a specific way to word things so it will listen

[lost holly]

There is no guaranteed way that always works. Something like https://suno.com/s/JLF4czgTsZmKNsoA can work, in my experience it works best if you define the male/female like this in the lyrics.
Otherwise you can try male and female vocal, duet... in the style field or a combination of that.

[lavish gorge]

Hi. How to use persona but only keep the vocal character? I try many times to use persona but it give me same vocal note and chord proggession.

[raven turtle]

I’ve had issues with Personas. If you can use it in v4 they seems to work more as intended IMO

[sleek stump]

Just to be clear, Persona are not vocal presets, they are intended to be influencers to maintain styles, they do not guarantee the same vocalist.

[compact shuttle]

Heya, I wanna try that new playlist thing they added, how do I make a playlist? Also how does this new feature work exactly? Has anyone experimented with it?

[compact shuttle]

Try something like this when entering the lyrics

[verse 1, male vocalist]

[Hook, female vocalist]

[Verse 2, female vocalist]

[Bridge, male vocalist]

[Verse three, male vocalist]

[Outro, female vocalist]

[half dragon]

Simple: 3 dot menu on the song > + Add to playlist
Then a popup shows - choose a playlist or create a new one

[compact shuttle]

Have you experimented with it yet, I currently can't

[half dragon]

Yes, I have many playlists for a while now - it's not new thing.
You can't create Playlist? are you on web browser or mobile?

[compact shuttle]

No I'm talking about the new feature that got launched with version 4.5 Plus

That's about playlists

[half dragon]

Oh!

+Inspo?

When you say you can't ... meaning you can't select the playlist to add when creating? or you selected but it didn't work

as it was working fine for me

It's like using the playlist to influence your new songs - and it's an awesome feature in my opinion

[half dragon]

With that new feature - you don't create a playlist - you use the playlist to inspire and create new songs
In your Create page - click +Inspo - then choose a playlist
Type in your lyric, and style as normal, then click Create

[patent sand]

anyone know how to make 4.5+ not hold a note for 5 seconds?
tried "[Verse 1: Female, short, detached vocals, no sustain]" but the vocals get too choppy
a Japanese glottal stop "worked" but frequently got ignored by the AI or accidentally pronounced
genre is gothic metal

[oak cobalt]

I suppose I could write a gibberish song and then extract the Instrumental stem.

[bleak furnace]

tech team can you restore the auto lyric version to 1 month ago it was great but now it can't write a song as good as 1 month ago

[daring lava]

S/O to @half dragon @polar spade @sleek stump for being thoughtful and helping people all the time here!

[compact shuttle]

No as of right now I'm waiting on suno to confirm my purchase which is taking forever

[polar spade]

Forever takes about 4 banking days. 😳

[compact shuttle]

Tomorrow makes 4 😳😳

Oh and just to make sure. General commercial terms means I can upload to Spotify, apple music, etc right?

[polar spade]

Thats pretty vague, but as a paid pro or premium subscriber you have ownership to use commercially ONLY those newly created (fresh out of the box or home-grown uploads) amd may monetize. Contingent on the other agreements you made with Suno and getting through the recent AI hurdles of Youtube and all distributors. But yea, basically.

[compact shuttle]

Thank you and sorry for the vagueness I'm just a little busy

[livid topaz]

Anyone have any tips for creating smooth transitions or revisions? I try the regenerate option to redo a bad transition in my song but no matter what options I choose it always changes the style significantly (using the same style as original generation) and almost always over generates into the next verse which wasn't selected. I am just trying to replace a short section of my song that has some weird artifacts and a rough transition between verses. Is there any prompting that will tell Suno to just revise and not completely change the section? hope this makes sense ive wasted like 200 credits trying to fix this one part of my song. It's right in the middle of the song so annoying

[polar spade]

This Velvet Sunrise Sundown incident screwed the pooch for the rest of us, but . . . 🤷 🤣

[compact shuttle]

Velvet Sunrise?

[polar spade]

Ummmm... errr..... Velvet Sundown. Sorry. The other is real, this one is AI. Just talking about it picks at the scab a bit more. Haha

[sleek stump]

Ya and Spotify is complicit in that debacle.

[hallow rapids]

In an effort to extract MIDI from a Suno render, I instructed Suno to isolate instruments and remove drums from a cover/remix. To my surprise, the process yielded more favorable results than anticipated.

[hallow rapids]

Thats a tough one. I have tried this on many renders with zero luck of it sounding good or natural. The best luck i've had is with using the replace feature on both IOS and the Web version. Mixed results there also

[livid topaz]

Yeah it's so frustrating. the song sounds perfect except this one part. It just can't seem to get it right

[earnest rose]

Hi Guys! I'm trying to make an EDM track where only single lines are spoken, not continuously. So I split the lines using [verse1], [verse2], etc., and insert [instrumental] between them. But Suno plays often the spoken lines one after another without breaks. Any tips how to fix this? I just wanna "intro text speech one line" then EDM sound - then a sentence - EDM sound - then next sentence - then a Drop etc.

[graceful condor]

Are you using the Editor? If so, try to rework the part you want to change, save it as new song and then remaster it, when editing things this worked for me but I had to give it a couple shots

[polar spade]

Tips: Slowly and carefully highlight the words or area replacing.

Or go to Legacy editor and manual type in the timemarks to be more exact. Keep in mind the timemarks will jump around as Suno try to wuantize to the nearest best blend point. Keep manually changing the timemarks until they stick, if needed.

CNTRL KEY will disregard clicking to a quantized point for a smooth highlight when using the waveform.

Use SMART blending option at least, as a failsafe option until youre use to Fixed and Classic options.

You can click on a regeneration or any segment, and slide the ends in and out to ajust which one overlaps the other and how much, if that helps.

Old school: Zoom in full and holding CNTRL key, surgically highlight to the top of a sin-wave where you may need to splice two sections together. I often dont waste time worrying about overlaps of too big replacements or overuns or garbled words, etc.. If they generally sound good, I manually splice them perfectly and run a remaster to blend them more, then go back and replace more within the new replacement. Having a large monitor or DAW experience makes this easy.

[polar spade]

Dont mark spoken section as a [verse] if I understood you correctly.

Just drop down a linespace and prompt:
[Verse 1]
Bla bla bla singing
<linespace>
[Spoken]: [male] [whisper] " and the crow flew into the mist to find the dragon"
<linesplace>
[Verse 2]
More blah blah singing

At least that would give a better chance.

[earnest rose]

i try , THANKS!

@polar spade thats correct the lines i wanna that he "speak" them but then a bit of sound/edm instrumental part and then next line etc. thanks

[polar spade]

Suno is getting better at just speaking lines that are in quotation marks.

[polar spade]

It may be hard to get precise changes, but take a shot. See what happens. Even 2 line spaces may be in order.

[earnest rose]

@polar spade that speaking work good/is ok (9 of 10 he speak, dont sing) but he ignore my [instrumental] between the sentences

[polar spade]

Try [interlude] ? For long solos, I use
[Interlude]
[Interlude]
[Interlude]

As Suno tends to give 20 seconds per Interlude.

[earnest rose]

allright, i try, thanks

[livid topaz]

Thank you for this detailed help. I've actually tried pretty much all of this. I have a lot of experience with this stuff. the rest of my song is literally perfect no artifacts or anything. fully ready to be mastered. To the point where using remaster makes it worse everywhere else. it's just this hyper local break between a chorus and the start of the next verse that is broken. it originally sounded fine but had a massive instrumental breAk between that I needed to delete. Unfortunately I couldn't get a smooth splice so I had to regenerate but that's just led to artifacts and weird transitions. ugh. so simple yet for some reason suno can't seem to get it right

[main cairn]

How do you do that? If I tell it the song is instrumental, it gives me no lyrics box to type into... unless you type them in the style... If you turn instrumental off and just give it metatags with no lyrics, the AI hallucinates like crazy and tries to make its own lyrics. Sounds like it dictated a baby babbling and incorporated it into a song.

Oops, I meant to reply to the message about metatags, but the question still stands.

[half dragon]

This is what I did when I want an instrumental of my song.
I covered it and removed all lyric and all instrumental tags are left.
https://suno.com/song/3e9c2520-adbd-473c-a78e-ce308aba5682
You can do the same with new instrumental songs too by giving it a structure.

[sleek stump]

You can still use the normal windows for instrumentals, all of my instrumentals all use the lyrics window for direction, and content for length.

[near sandal]

I ran out of credits and updated to premium today but now when I try to create I'm told I don't have enough credits,Can anyone help please?

[patent sand]

anyone know how to make 4.5+ not hold a note for 5 seconds?
tried "[Verse 1: Female, short, detached vocals, no sustain]" but the vocals get too choppy
a Japanese glottal stop (ッ) "worked" but frequently got ignored by the AI or accidentally pronounced
another glottal stop (international phonetic alphabet: ʔ) didn't help
genre is gothic metal

also much more basic question: if I close the stem extract window on a song, can it be opened again? not trying to dig through all these stems manually after-the-fact

[empty steeple]

thank you - i emailed suno weeks ago with no response. and there’s no number to call either when having problems. customer service is terrible with suno. what am i supposed to do?

[hollow hare]

Hmm very strange, 500 indicates an Internal Server Error, likely a glitch/bug in the code or something (e.g. add-on) interfering and breaking some sensitive function

I can think of three things you could try:

1. Delete your cache, log in and try to download again. Sometimes the cache retains some broken material.
2. Try another browser and see if you get the same error/issue
3. Try extending the song or, if you want it as-is, crop the entire song (crop tool but do not remove any parts, just crop from 00:00 until the end of the song). This will cause a new generation which should allow the download to occur (in case something broke when it was generated the previous time).

If that still does not work, DM me your song link and ping me so I can try a few other things on my end.

Regardless, after you try those, be sure to report the issue at #🪲┃bugs-web so it can be investigated. If you do not want to share your song link, you can ask the staff member(s) to DM you when investigating (and you can also report via the website). Just make sure that the person you send the link to is a staff, they will have the 🟣 Staff role in this server.

Hope this helps you with that strange situation

[rustic yacht]

Hey there !
I need some expert here ... I struggle to do something on one of my song ... After the intro I start my song with "I love you" (In french, doesn't matter much) but I want it to be whispered and glichted or inside a vinyl scratch ... And I can't get it right. Any ideas ?

[patent sand]

try putting it in parentheses and prompting [whispered] or [soft-spoken] in the brackets before it

[rustic yacht]

Like this :
[Whispered]
(I love you)

? But for it to be an FX should I try [Whispered - Scratch/Glitch] ?

[patent sand]

glitch might be more reliable

never tried it myself but I've seen it

FX has been historically unreliable to generate but 4.5+ might be able to if you're using that

[rustic yacht]

Yeah I've stopped trying to use FX it was too random ... but thx anyway I'll give it a try

[empty steeple]

thank you for the help but i already deleted my cache as well as logging in again. i also tried from a different browser. a 500 error indicates something in suno’s servers from the diagnostics i ran.

i also tried editing the song as well but it do not work.

[near umbra]

Does anyone have the prompting magic to make Suno make a beat like this?

[humble verge]

Help! Suno won't end my song 😆

[trail fulcrum]

Perfect — here’s your ready-to-go Suno Bug Report Template you can copy/paste into their report form or Discord:

---

🐞 Suno Bug Report: Repeating Chord Progression Despite Prompt Instructions

Issue:
Suno repeatedly generates the same chord progression (e.g., I–V–vi–IV or similar 4-chord loops) across multiple versions of the same song, even when my prompt explicitly forbids this behavior.

Prompt Example:

[forbid: I–V–vi–IV, vi–IV–I–V, and any four-chord loop used in pop/rock presets]
[require: unique progression per section – intro, verse, chorus, etc.]
[final instruction: no preset structures or default templates – original harmonic content only]

Problem:
Despite this prompt, Suno returns nearly identical progressions across multiple generations. I've generated over 10 versions of the same song with the same harmonic structure, even with adjusted Style Influence, Audio Influence, and Weirdness.

Expected Behavior:
Suno should follow the prompt and vary progressions across song sections, or at least produce noticeably distinct harmonic structure between generations.

Why It Matters:
This behavior costs significant time and credits while blocking creative variation. Please escalate this to the team or tune the model to better honor structure and negative harmonic prompts.

---

Artist name: MeyVyn. Genre: alt-rock/grunge/biblical concept albums. Structure compliance is essential.

[shadow plaza]

Im doing house music.
I want to have a little prhrase or chorus that is singed once or twice before the drop. How to do it. Cuz right now it keep repearing like 10 times become annoying?

[trail fulcrum]

@shadow plaza If your vocal phrase is looping like 10 times before the drop, that’s why it’s killing the vibe 😅

Here’s a quick fix:

🎯 Use it once or twice MAX – like at the 16-bar and maybe again at 8-bar before the drop. Don’t let it just repeat endlessly.

🎚️ Automate the fade – volume or filter fade it out early so it doesn’t overstay its welcome. You can even chop the phrase short for extra punch.

🌊 FX it out – throw a little reverb or delay tail on the final word so it trails off into the drop instead of repeating.

🥁 Pro tip: If you’re feeling fancy, chop the vocal and sprinkle it rhythmically right before the drop. Like a little vocal fill – think Daft Punk style.

👉 Bottom line: build tension, don’t annoy the crowd.

[shadow plaza]

That chatgpt reaponse isnt what i looking for.

[dusty tangle]

is there a way to get suno to put in an extra instrumental in a second of the song through editor?

[fresh wave]

Any prompting tips? I'm sorta new

[ebon sand]

Nothing is true, everything is permitted

[fresh wave]

I did learn that out by making new stupid genres.

😛

Is there anybody that has a lot of experience that could give me some feedback?

I'm trying to see where I can improve next.

[cosmic pollen]

Try to write it like [slowly sung phrase]
(„Your phrase“ 2x) or (Adlib: „your Phrase“, „your phrase“)

[ebon sand]

Like in terms of composition?

[shadow plaza]

Have you done this before?

[fresh wave]

Hmmm... Possibly composition! I'm not really sure I've been doing my own thing so long but really new with suno

[ebon sand]

You could just share what you want feedback for, but someone who might do an amazing rap might not be able to do a great goregrind song, so just try and see what happens

[cosmic pollen]

I did it for a song to have some spoken words between different parts and it worked fine for me

[shadow plaza]

Asking because i do mainly instrumentals and want to have this part build up and say something like phrase and drop the beat.. Thats it no more talking and singing

Through whole song

Just single time

[fresh wave]

yeah, that's a good point. Also why it's hard to get any feedback going music because it's subjective as hell unless you have a weak point you think you have. I think I just wanna suno 4.5 pro 101 tutorial to get better

[ebon sand]

You could try to do an instrumental only and then add the phrase in the editor

Might take a few credits though

[shadow plaza]

Never learn how to work with the editor 😂

I prefer to be able to do in through metatags and promts

[ebon sand]

I've given it a try and the line only got spoken once, I can share the input, but it's nothing special

[shadow plaza]

Yes please

[ebon sand]

https://suno.com/s/cdb3j6axrq8MNwqd
I have not much idea of electronic music so sorry for that

[dapper spoke]

y i cant use those free 20 songs?

[shadow plaza]

Its okay.
What models was this
?
Can you share style promt.

[ebon sand]

It's on the page but yes, it was 4.5+
The prompt was
"a repetitive dark techno Song to worship a new god with a singular spoken like before the breakdown"

[shadow plaza]

Ok this is with insgrumental option off yes?

[ebon sand]

Yes, it occasionally repeats it a 2nd or time if there are more breakdowns later in the song but I think this could be solved with inline tags about stuff it should do or id just remove it in the editor because it's only one spot
https://suno.com/s/nf0wUCmjuOOL8TeX

[shadow plaza]

Thank you

[short fjord]

Does anyone else have the problem that the sound crackles when listening to their tracks?

[ebon sand]

On mobile browser Always
But the songs themselves also have it often in the audio

[ebon sand]

Have you looked into #1252267444773326848
There people share their test results, but take it with caution because it is always a combination of everything so their ways for example would never work for the songs I create as my whole song setup is very different and so the AI interprets it differently and thus the best practices will not work like that

[short fjord]

im on pc ( using chrome ) and it starts yesterday to crackle. old tracks who definetly not crackling crackles now.

i tried now firefox, there are no crackling, so , i think its a browser issue

[buoyant lotus]

Hi, anyone has a tip on how to avoid drums in a song ? I'm trying to make a lullaby (told him "lullaby" in style prompt) but it's always adding a lot of instrument and especially drums (who tf sing a lullaby with a full orchestra and drums seriously) Tried soft, acoustic guitar, a capella, soothing, etc

[ebon sand]

Maybe try "sparse Instrumentation, acoustic guitar", if you want no instruments at all it might require different wording though
Which model is it

[buoyant lotus]

3.5, still on a free plan

[ebon sand]

Easiest way would be to use a persona

I'd start with something like this and then exclude percussion or attempt to define more in the instructions what it should do
If you want vocals only it is also possible but needs different emphasis, 3.5 is great on this
https://suno.com/s/rVq9nHufH1bHVW3E

With persona it would be easier though, less random instruments and better quality

[somber heron]

Guys I have a question) can somebody reply to me , I have 600k followers on insta and 250k on facebook, does suno provide Cpp program for creators ?

[buoyant lotus]

thnks, I'll try this

[polar spade]

This seemed easy enough with "nursery" and " lullaby ", depends on how "infant"-like you wish to get : (ignoring the 12 days of Christmas feel) https://suno.com/s/aYw02Cs03YIUEJ7C

[polar spade]

Oh.... 3.5.... hmmm.... let me see what it does for that then.....

Even better! Just a music box. https://suno.com/s/nndYNCftsOREOlwy
https://suno.com/s/8kvaO7OGWE78plH1

Oh sorry... one is 3.0 the other 3.5 . Since we can do either.

[spiral frost]

Anyone good with prompting vocal style ? I really need help :/

i want a very high pitch voice, like soprano mail vocal, and i never get that

[ebon sand]

On which model version

[spiral frost]

4.5

but maybe there is a better one for that?

[ebon sand]

It should be fine doing that, I have some Bell resonance soprano vocals there

What is your prompt like to achieve it

[spiral frost]

@ebon sand i tried so much prompte but from : slow piano ballade, unplugged, solo airy youthful voice, light voice, soprano, meditative, acoustic, solo singer, male voice, [soprano male voice], guitare, drums //////// to : slow piano ballade, unplugged, solo airy youthful voice, light voice, soprano /// to : A slow, meditative piano ballade, unplugged, with spacious, delicate phrasing. The acoustic, solo male soprano voice floats lightly above minimal, expressive piano chords, conveying an airy, youthful presence. Sparse pedal use and lingering silences create space for vulnerability and reflection. and i very very very often have like "bassy low voices.." i want somethjing angelic like this : (this is an export from suno -> Ozone mastering)

[ebon sand]

I didn't listen completely through it but skipped through with several stops but for me there's no soprano in the song so maybe it's not the term you're looking for?

Soprano is by its nature a female vocal range, I could think of male soprano singers in context of church choires but that's not very common and not what's happening in the song

Try the terms falsetto, emotive and breathy maybe, something along those lines

[buoyant lotus]

Thanks, I'll try something like that

[polar spade]

Soprano male = contratenor/countertenor. (Whether that helps as a prompt, I dont know). Just use "boys choir" 😁

[spiral frost]

Ah… tried falcetto not working… but youre right soprano is for female

@polar spade very right ! Ill try it iut

[ebon sand]

Prompting for clean angelic female soprano and then switching the "vocalist: male" switch

As for model, I'd use 4.5 and not 4.5+ for this

4.5+ also should be able to do everything but it seems to be tuned to be overall a bit more gritty and... Energetic

[polar spade]

For two days, all my playbacks have lots of bumps, pops, cracks and hiccups. No smooth playback. Too many sunspots this week affecting the Server?

[unique stone]

Are these present in downloads or only when playing through browser?

[polar spade]

Just the browser on Android.. I have NOT downloaded anything to check. Ill check PC desktop at some point.

Its like a buffer problem. So could very well be my phone too. The pops and crackle (i.g. as if a bad connection on my headphone plug) are included too. So I assume it will clear at some point.

[unique stone]

I'd be mostly concerned if they happen at the same points, e.g. if you hear a problem at say 0:05 and then you play the song again and have the same problem at 0:05.
Also check old songs. If old songs (that used to be OK) are affected it's not an issue with the songs themselves.
Could still be a site-related issue though, e.g. something on the page causing it to misbehave.

[polar spade]

Im listening and scrubbing back to see if repeats. But nothing consistent. Just seems to lag, like tape dragging or buffer issues. I was joking, but seriously sunspots can do that too. Peak flares this year. Who knows. (Right now its quiet though). Or else I need to pay my phone bill and they're shadow-lagging me. Lol.

Hey, I woke up today and discovered that I , at 2:50pm yesterday, apparently made 1 generation ( 2 outputs) of a v 3.5 French song (in French) and didnt listened to it either.... and I dont speak French. So ...... 🤷

And this morning I tested 4.5 nursery lullabies and got 2 outputs, "12 days of christmas" and "we wish you a merry christmas". Hahaha... its just one of those days weeks. 🤣

[spiral frost]

Good id !!

[sleek stump]

@clever tinsel If you are using "Use Styles & Lyrics", the base song will be linked in the top of the prompt section. Delete that song and adjust your styles prompts and you can get a complete reimagining.

[spare root]

(I put this in the need help tab so kinda messed up on the placement but hopefully it works here! )

So just a quick question, how do you write a persona description without catching a false flag? Do you just keep it super basic? Like super, super basic. I've attempted to clean up the description to only mention vocal tone and range and its still saying theres a artist name in it. Im unsure as to where however so I just need to know if its better to keep it like a list instead of a bio type of thing.

[shell trout]

sometimes a regular word is an artist name

ive seen my original lyrics flagged as a producer tag

[spare root]

Oh gosh... is there a list of words somewhere? XD

[spiral frost]

sorry to get back to you with the same problem, i really can't get this voice out, even from female to male tranisiton in 4.5 / and 4.5+ even in in 4 its worst... if anyone have a workaround... a keyword.. i pay 1000 credit.

[pseudo blade]

I've been trying to get v3.5 to do from melodic to high scraming notes, (I've been here for a while now) @polar spade any advice here

[polar spade]

Honestly, that's pretty specific that sounds like the need to use extensions. So unless I experiment, I would not know which direction would be best.

[pseudo blade]

okay, thanks ❤️

[spiral frost]

@polar spade @ebon sand Guys i finally find a workaround.... Maybe useful for you guys. So : You have to get some other music that has a bit the same voice. Add it to a playlist. Create a song, Cover the song with as "inspo" your playlist.. the voice should adjust.. it worked for me

[livid topaz]

Has anyone else experienced issues where the WAV file of a song sounds significantly worse than the MP3 of the same song? tried downloading in multiple browsers, playing the files on multiple music players and devices, and even mastering the WAV file and no matter what the MP3 version of the song sounds significantly clearer, crisper, and more detailed. This makes absolutely no sense am I missing something? want to make sure I have the best possible file before I master my track

[spring token]

yes I just noticed this too

[livid topaz]

Really frustrating. Like honestly the MP3 sounds pretty great but if I'm paying for the uncompressed WAV file option I want it to actually provide a better track file. Defeats the purpose.

[livid topaz]

Quick update I tried splitting the song into stems and then downloading the WAV and it seems to have helped a lot. Audio quality seems at least mostly on pAr with MP3 now

[lunar kraken]

Does anyone know how to prompt to get get long songs about 6 to 7 minutes. I am not doing lyrics and I am doing edm music. These short 2 to 3 minute tracks are not cutting it. Once it a while it kicks out a 5 or 6 minute track and that is doable but I need it to kick out long tracks all the time

[livid topaz]

I would use extend on the shorter tracks you get. Extending shorter tracks will almost always result in better results than trying to make one giant long one in one go

[spiral frost]

wave files is great if you have recorded or composed the song yourself, ai is not cristal clear, so wave file has rly no importancE.

[livid topaz]

True. Gotta squeeze out every little detail tho🤓

[lunar kraken]

I try that but extend feature doesn't work great for me. Seems to always goes short and sometimes 30 seconds to a minute

[viral pond]

does it make sense to create a base instrumental song firs,t then cover with lyrics ? or is it not a good idea? never tried this '

also I use the main library view when i jsut want to listen to most of my latest music, but sometimes there are crap I dont want to listen to, can I just use dislike to hide these or will this have other consequences ?

i dont want to delete them, just keep them in a seperate playlist

[remote zodiac]

Hi,
many times on “type beat” that I try to upload on suno I get the message to say that the music is known, do you have any solutions to counter this?

[sleek stump]

You can use Dislike to hide them, but Suno uses Dislikes in its training. You can also send them to trash to hide them.

[viral pond]

was afraid of that

i never really fan of like or dislike..

[sleek stump]

Anything you send to trash is still there, until you physically delete them from it

[viral pond]

yeah i guess, hopefully suno will fix a better ui in the next update, cause the current ui is buggy and not really good

[sleek stump]

Boom Woodhouse in da house

UI has not been Suno's strong suite

[viral pond]

proper folder structure with workspace where you can put all variants of a song you work on, and one for actual playlist, an archive where you just send random stuff you might look at later

and a random play function 😄

[sleek stump]

There is a folder structure, btu it only works in create

[viral pond]

yes the folder structure is flat

it should be nested

workspace is buggy, if you move some around, sometimes you need to refresh the whole page to get bacl the one you look for

its nice they fix many things but its about time to focus on a better ui i think

[dull hedge]

Hi guys! I wanted to extraxt stems from a song I made and noticed it created 2 different versions (I guess two sets of stems?). It’s just above the tracks and says ”Version 1” and ”Version 2”, and they’re slightly different. Can anyone explain this? And how do I know which version is the original song?

Also, my laptop crashed so I had to redo the process, and suddenly I had 3 versions of stems. This happened a few times because my laptop is old, and every time it adds another version. So I now have like 8 versions of stems for the same song. 😳

Hope someone can explain! Thanks!

[remote zodiac]

Hi,
many times on “type beat” that I try to upload on suno I get the message to say that the music is known, do you have any solutions to counter this?

[viral pond]

how come the rearranging do not remember the positions in the playlists? i change the positions, but next time i go in its back to what it was

[half dragon]

Is it true that it's known music? you should only upload your own music. Was it something you created?

[half dragon]

They're both should be identical - main differences are different type of instruments are detected in besides the main instruments.
Yeah, when you go back into Stems, do not click Extract again, just click the small link underneath that button > See extracted Stems

[viral pond]

why I have the feeling that every song i make now days gets "cursed" ? I did make some cursed songs on purpose before, and now it seems everyone is the same kind of "curse" '

does suno remember previous generated music and learns from that or something ?

[remote zodiac]

No, it's a free type beat from youtube

[civic flint]

💥🤘🏻💥

Enviado por um membro da equipe do Suno. Sobre a restrição ou a letra. Ele escreveu que é para maiores de 13 anos. Por que não fazer músicas malucas e idiotas, em vez de sinalizar...

[sleek stump]

Suno does not allow uploads of music you do not own. Music from YouTube has copyrights, which could get you in trouble, and Suno if they allow it.

[last bay]

how do i add new lyrics for extension in the new editor? i though this worked befor but when i select a part i can only extend, i want to add another chorus near the end but using the section with "+" didn't work good so i wanted to extend but then i can't replace. is this now only working in legacy?

[magic depot]

Hi all- anyone know… How do i stop the garbling sound from a web browser experience? I'm connected to A UAD interface and its defaulted to 48000 hz.. would that cause an issue? I feel like it recently started with the last few updates

[celest smelt]

Suno seems not be able to read coptic language.
Example:

[half dragon]

It can't sing it?

[celest smelt]

It says "Ni na na la la la" the usual mumbling sounds.

[half dragon]

@flint haven did not like that 🙂

[half dragon]

I did a small test - chatgpt gave me the transliteration version of the verse in the language to use and Suno sings that ok
https://suno.com/s/3h3owbxrqO9H9UQ7
Note: I used a generated Coptic verse - not from yours as I can't copy it 🙂

[celest smelt]

I also tried to romanzied it. But every time, chatgpt or grok gives me arabic romanzied. It seems that old egyptian coptic is not possible because is not source enough like old latin to make a music.

However, Old latin works perfectly. As counter example.

[celest smelt]

Cant copy you the text or the bot will delete for "spam" reasons.

[half dragon]

Yea, I tried to post it too but it didn't like it

[celest smelt]

Yeah, i wanted something ambient as an homage to the egyptian mythology gods. But is not possible. 😦

[half dragon]

I think you can - I think Suno is capable of creating that kind of music. It might be challenging but it's not impossible.

[lunar anvil]

Kind of weird the Suno profile shows up sometimes and other times the browser says that profile cannot be found.

[sleek stump]

Might try using styles like the following... @celest smelt
Shaabi
Classical Arabic (Tarab)
Nubian
Sufi/Dhikr
Mahraganat

Additionally, Suno relies on its training models for everything from understanding text and script to what music styles should sound like. Old Coptic is not something that Suno is likely to have a lot of training on, and many LLMs like GPT have problems with it, but not transliterated text. I would suggest sticking to transliterated lyrics at this point.

[lucid turtle]

how to do accapella? im a user since V1 and suno has always seemed to struggle with this. is there a way to prompt it that im not aware of ? thanks if you know

[polar spade]

What end result are you looking for can decide your prompting. These days "a cappella" is more expected to sound like Pentatonix than a chapel choir so that can sneak in neatboxing, percussions and an instrument or two and lots of layers.

[quaint whale]

Does the Suno LLM use it's own user inputs/output for its training data? Some potential evidence:

1. I crafted a song that randomly included the lyrics (not listed in the lyrics section though) of another song I crafted.
2. I got a song that used hard swear words despite not prompting it to.

This will be beneficial in the beginning, (lots of data), but it will create a feedback loop, just like what happened to Stack Overflow, and Suno will become super generic and may even collapse if they are not careful.

[sleek stump]

We do know that Suno uses User outputs and dislikes for training, but to what extend is something they have not disclosed. We also know that Suno uses ChatGPT for a number of agents (Content checks and moderation for sure), but we do not know what they are using for their language parser. It could be and likely is an internal parser, because all the big LLMs really dont know what works 100% in Suno and never have.

[quaint whale]

Appreciate it. They may be playing with fire then.

[humble verge]

every time i open the editor i think to myself, here goes a thousand credits down the toilet 🙄

[azure crane]

V4.5+ adds way too much reverb. Is there a tag that can reduce it?

[raven turtle]

I’ve done minimal tracks that includes some instruments,
Then stem out the vocal and cover it as vocal only. This has worked for me

[versed harness]

hey not sure if this is the right place but does anyone know a good female voice prompt that dosnt sound like a 18 year old girl, trying to create a rap battle song my credits will thank you for any help 🤣

[narrow citrus]

Is there already a plan to finally improve the overmodulation, especially in purely instrumental pieces, but also with vocals?

[daring wharf]

You could specify Contralto Vocals, and/or add a Vintage Vocals tag, and other specifics trying to call up particular voices. Another trick is to use a male persona and then tag everything for female -- this can push things pretty far away from the usual ska pop range.

[narrow citrus]

Thanks... but how is that supposed to change the sound overmodulation? Basically, it's too loud and some notes sound distorted. This is particularly noticeable with piano instrumentals. Mastering doesn't help either. Once the notes are overmodulated, you can't effectively correct them. The overall volume shouldn't exceed 80% so that you can do proper post-processing.

[polar spade]

Nowadays you may want to build a persona thats can then consistently gets your results. That would be the simplest solution once you have a few in your pocket. Even a deathmetal screamo woman persona can be smoothed out by balancing with new prompts and adjusting sliders to get a warm subtle tone. It just depends on all the other factors, so its a crap shoot, but will occur at times.

Just throwing it out there, genre wise, folk and country, tend to have more mature sounding vocals overall.

Otherwise, yea, someone might suggest charateristic prompts such as "mature, contemporary, smokey, warm rich, 30s to 40s" as well as exclude bubblegum, anime, vocoder, teen, kids songs, and others to help weed out teen sounding vocals, and credits be damned, experiment until you find your voice.

[versed harness]

https://tenor.com/view/happy-tears-gif-4478570828550002515

[azure crane]

V4.5+ adds way too much reverb. Is there a tag that can reduce it? I create instrumentalsz

[quaint ginkgo]

I have 20 songs v4 BUT I cant use it - WHY??

[polar spade]

As in? You see them, but clicking on them does nothing?

[quaint ginkgo]

yes

[polar spade]

Are these 20 the only songs in your catalogue? Or you cant click on 4.0 songs only, but other songs are fine.

[quaint ginkgo]

other verthions ok but tup on v4 and nothing

[quaint ginkgo]

other verthions ok but tup on v4 and nothing

[polar spade]

That is strange. There should be no reason to keep one format from playing over another, no matter which subscription level you presently have.

Have you before published any of the 4.0 songs that I could find one and see if it plays for me?

[quaint ginkgo]

no I didnt use anyone __
v4
Pro
20 songs left

[polar spade]

Obviously its like specific to your account, but Can you play 4.0 songs found on the home page of other users songs?

[quaint ginkgo]

they dont let to choose v4

[quaint ginkgo]

yes

[polar spade]

20 songs left? What can you tell me that that means? I dont understand "20 songs left". That sounds like something Suno would post about your remaining credits.

[polar spade]

You have a Pro account, you say?

[quaint ginkgo]

when you have new account you have 20 songs v4 but they dont let to use them

[polar spade]

Ahhhh... when you have a new Basic free account, they give you 100 credits to use for making some v4.0 version outputs.

[quaint ginkgo]

I see it 20 songs left - but cant use it

[polar spade]

Thats equivalent to about 20 outputs, good or bad.

Once those 100 credits are used up we, can not use 4.0 further to make songs.

[polar spade]

You started a Basic account, and you see you have 100 credits, but you dont have access to make 4.0 songs? I see.

[quaint ginkgo]

and I see 20 songs v4 left

[polar spade]

You may have more than one Basic account associated with the same email? I dont know. Its my guess, that ALL accounts with the same email are considered.

[quaint ginkgo]

no not 1 mail

[polar spade]

Beyond that, if they say you have the credits, but you cant choose 4.0 , then yes, that unusual.

[quaint ginkgo]

no I see 20 songs v4 left - BUT I CANT USE IT

[polar spade]

If you have more than one basic account, they also might monitor your IP address as the same. It is just a guess.

You may SEE IT, but you might also be blocked from using it.

At that point, I can only suggest to contact [e-mail] to see why it is not working.

[quaint ginkgo]

we havnt suport here?

[polar spade]

Suno also monitors IP addresses. While using Suno, it constsntly pings your computer to monitor usage and verify user.

So if the same logins of different Basic accounts happen within the same IP address, it could be a security issue.

[quaint ginkgo]

I tryed from my frean home and also not worcking

[polar spade]

Discord is not a support format, merely channels where users can help each other. There is the feedback channels so you can give feedback or report as a bug https://discord.com/channels/1069381916492562582/1082775596767117312 but those are still not a tech-support to "fix your problem".

Only [e-mail] will get a direct response from Support to discuss the issue.

[polar spade]

Since we seem to have eliminated many possible reasons, I am out of suggestions.

I will mention, you are only allowed ONE free account per person (not per email, location , etc.) So again, there is the possiblility that you could be blocked from making more free accounts just to use the 4.0 trial promotions.

Who knows, but its always a possibility. 🤷

[polar spade]

Good luck then. You are not out of any extra money as they are free accounts. So thats a good thing.

[sacred sorrel]

hey

[sleek stump]

See this article in Workflows... https://discord.com/channels/1069381916492562582/1367549634536542389

[sleek stump]

Showcase or General-suno might be a better place to post songs..

[flint haven]

✅ Warning logged for hannah068793. I couldn't DM them. || Do not spam your music

[brittle ivy]

Hey Suno team! I'm really interested in subscribing, but I’ve been running into a couple of issues that are holding me back. First, I’m not able to paste copied text into the input—I've tried different browsers and it still doesn’t work. Second, I can’t change the length of a song after selecting it; the option just doesn’t respond. Not sure if it’s a bug or something on my side, but I’d really appreciate any help! Thanks a lot

[narrow citrus]

The copy/paste problem must definitely be local. Have you installed any internet security tools? Often, there are all-in-one solutions that install antivirus, spam check, and browser filters. That's probably where the problem lies.
Tip... a decent external firewall and Defender is sufficient for Windows, and you won't have any real problems. Malwarebytes is definitely a good idea as additional protection.

[half dragon]

What do you mean that you can't change the length of the song after selecting it? in Editor?
Are you trying to make it shorter or longer? crop or extend?

[foggy phoenix]

Anybody experience a lot more shouting in their songs?

It sounds so weird rn

The whole structure i was doing is now falling a apart

The webplayer didnt work earlier and then after a fix my outputs sounds so shout like

Oh okay the vocal selector now works my issue was typing vocals in the genre

[weak moss]

How do I hard prompt to ensure I get black voices speaking?

[wise kindle]

you can't really ensure anything, but you could try "urban" perhaps

[polar spade]

Yay!! The Explore button has been moved two spaces down below Search and now Radio. My fat thumbs thank you Suno!! 🥰🥳

[wild egret]

copy paste issue is related to the limit of characters, if you have +200 it wont paste anything, unlike other systems that paste until the limit, suno does not paste anything.

[rigid pasture]

I am making a sad song but all the songs are sung in the same tone. Can you suggest any sad songs in ar style?

[polar spade]

So we kind of have 2 radios now?

[polar spade]

Im still asleep maybe. What's "ar" ?

[rigid pasture]

I don't understand

[half dragon]

New one is just Instrumental Radio - kinda neat

[polar spade]

@rigid pasture "......suggest anything in AR style....? " you said.....

I asked.... what is AR style?

[rigid pasture]

sad,lofi ,

[polar spade]

Its hands-on though. That's a tough one to focus on very long it seems. Wonder if it goes through ALL the genres before coming back around.

[rigid pasture]

@polar spade If you have to say a word or shayari before singing a song, what kind of promo should you give?

[half dragon]

Yeah - it's fun to vote - but I'd give up on that quick and just let whatever plays

[polar spade]

After three hours does it eventually vote on deathmetal instrumental haha... maybe it resets every 15 minutes after 5 songs?

[half dragon]

Maybe lol

[polar spade]

I always LOVED Microsofts mood radio back in about 1999. I played it all the time. Choose 3-5 artists you like, or choose a mood and every song was perfectly curated.

[polar spade]

That seems more of a personal thing. You choose your own shayari, or i guess some "producer tag" phrase.

[novel narwhal]

is there a prompt to get my house music over 4 minutes long?

[zealous summit]

how do you extend longer than around 20 seconds?

Like, my gens are around 2min for some reason

[polar spade]

Most outputs around 2 to 4 minutes i feel are reasonable, but if they are down to 20 seconds, it can get frustrating.

I responded previously to Smasherdash on how to get a single run output over 4 minutes, and the same would apply to extensions -- since extension rely on about 2 minutes of the original to "seed" off of, the combination or how long is the original, what prompts used, lyrics and how lyrics , or even just the metatags, are arranged can make a difference and thus vary extremely from output to output.

I try not to answer questions I can not give a solution to, but for me, the only solution is experimenting and trial and error with each new output.

Spacing the metatags in an arrangement so visually it appears about a page long, so Suno "feels" it should be longer might help.

Using [interlude] in triplicate stacked can add up to a minute of a solo piece sometimes.

It may help to also use Get Whole Song after each extensions, so that the outputs stay more consistent and thus might add more length.

And using prompts such as trance, dance, French house , disco, EDM and such will indicate to Suno this should be a longer song.

[zealous summit]

Thanks

[polar spade]

Ill add , certainly be selective on where you set the timemark of the extension. For longer ones you can find a solo instrumental and extension 7/8 through that so Suno thinks you still want it to go on. Same applies for extending one the third line of a chorus or verse, suno doesnt feel its wanting to end it, and finishes the verse and keeps singing or playing more. This rather than just extending from the end of or 5 seconds from end of the previous piece.

[novel narwhal]

Hi Dasvon, I didnt see your previous response. So just to clarify are there any specific metatags I can use to get a song longer then 4 minutes? Most house music is up to 8 minutes long, some longer, but id be happy with 7 to 8 minutes long. At the moment the some songs say 7 minutes long but past 4 minutes there is no music just silence

[polar spade]

I was referring to my response to the same question by you in the need-help channel where I mentioned there is generally no magic bullet. Every piece of a prompt or metatag is considered in combination when an AI generates a new output. So , what works for me may simply fail for you.

But the suggestion to Toast might help as well. These are things that help me (sometimes -- and sometimes are just trash) or that I've seen have been suggestions by some seasoned users here.

[polar spade]

@novel narwhal @zealous summit

Im so used to blending several extensions (and including personas with each to vary the output) to get the length I want or adding segments post production within the edit UI, that I don't think in terms of a one-shot 5 to 8 minute song anymore - so its harder to answer a question like that now.

[plucky bloom]

This works for me, not always, but most of the time, put it in the first line - [Length: >8 minutes]

[topaz narwhal]

I keep getting static using persona. Are there prompts to help reduce songs that could lead to this lesser quality

[novel narwhal]

A little frustrating but ill have to drag the stems into a DAW and manipulate it from there i think

[novel narwhal]

Your a star let me give this a go

[opal orbit]

Don't know how this would affect your output, but I've noticed that many of my longer generations get tagged as "cinematic" or occasionally "ambient" in the short style summary.

You could try adding "cinematic length composition" or something and see if that works.

[patent sand]

What are the persona version, prompted version, and genre?

[polar spade]

As you see, there are many suggestions to try, depending on your original prompting. I do alot of new age ambient ethereal stuff, so long songs come easy with those. As @opal orbit mentions, just popping "ambient" in the prompts or even metatags might nudge Suno to push the length further without actually vibing into an ambient sound.

Bottom line, suno tends to remain forever experimental for a random but often cool outcome where DAW is more elbow grease and sweat for a specific outcome.

I try to keep at least 20,000 credits in my pocket and have at it a few hours until I get what I want.

[topaz narwhal]

The persona version is a bit old but that song is very clear by comparison, the style was bedroom pop hip hop the new song will be a totally different genre it’s soft rock metal but I have no other way to keep her voice. And generally it works but the static this time is so bad Also a way to shorten long notes she held a note for like legit 9 seconds and she does it too often during the song

[patent sand]

yeah on 4.5+ the notes get held way too long. I tried asking before and got ignored

[topaz narwhal]

Ya it’s every genre too

[opal orbit]

In case the static is because of different versions, have you tried making a song with the persona in the same version, then remixing into the updated ones?

Then you could make a new persona based off the updated song version.

[patent sand]

it also kind of ignores the persona

[topaz narwhal]

Well the genes I’m doing and I have a broadish genre range I can’t say they all do

[patent sand]

4.5 (not +) generally keeps the song close with persona but it's not great for metal

might be a good start though. also found this for exclude that might help: #💬┃general-chat message

[fresh wave]

Yeah, I was surprised how much persona covers the song because I thought it was just going to be the voice of the singer.

[patent sand]

I've not tried persona outside of using it to "upgrade" songs so I don't know how well it applies to different songs

[topaz narwhal]

Yea I just want the voice instead I’m always fighting with the music making its way into persona unless you drastically change the genre which was working for me until after 4.0 I think 4.0 was the last time that worked well

I’ll try some things mentioned here

Thanks

[patent sand]

you can stem split the voice out and try remastering/editing it to just get it as the persona

so the instrumental doesn't carry over

[opal orbit]

If all I want is the voice for a totally different style, I always separate the voice stem out, then make a persona of just that.

[topaz narwhal]

Wait can you make a vocal only stem a persona

I feel stupid for never trying that

[opal orbit]

Yeah I do it all the time

[topaz narwhal]

Jeez thanks

[patent sand]

you also might have luck on 4.5+ getting the lines to not hold so long using this glottal stop ッ

[topaz narwhal]

You rock guys thank you

[patent sand]

best of luck

[near flume]

Anybody having a problem with songs abruptly ending at 4:52 when using a Persona created in v4?

[ebon sand]

I havent checked if this is the case anymore with the newer models but with the older ones when you had a long persona the outputs were shorter because of the max context limit, so maybe if that v4 persona is a long song cropping it would help

[near flume]

The Persona's were created from songs under 4 minutes.

[ebon sand]

try to crop it to 1 minute and try again

[near flume]

System isn't allowing me to do anything to the song the Persona was created from.

[ebon sand]

then there's nothing you can do

[lunar kraken]

I am trying this now but its very frustrating when it constantly kicks out 2 or 3 minute songs and sometimes less. I pay for the highest tier and feel like this should be something we can control.

[lunar kraken]

I am putting this prompt right at the beginning and its not working. Where are you saying to put it?

[junior phoenix]

Hey, I'm having a problem with version 4.5+. The songs I play and the ones I listen to from other users are making noise.
I tested my phone, but that's not the problem. Anyone else having similar problems after the last update?

[patent sand]

what noise? static?

[lunar kraken]

Ok it seems to be helping. Thank you

[lunar kraken]

I think using this long song persona is helping too

[urban gyro]

where are the workspaces located?)

[cyan light]

Under the Create area

[hollow hare]

Not there anymore for mobile users due to the new UI update today

[hollow hare]

At the moment, the only method I can think of at the moment is to either use Desktop view or go into landscape while in mobile view (which causes the layout to switch into the desktop layout due to the page width), both of which you can access the workspaces through Create (desktop view only atm)

It currently is Friday Evening in the east coast so it may be some time before that issue is corrected...

[cyan light]

I had no idea about mobile, thanks for that

[hollow hare]

No problem. If you were curious, for mobile users, going to create and closing the creation popup would list the Workspaces in the past.

Now, performing the same action will pull up a second creation screen. The popup is the old creation (picture 1), the screen beneath is the latest creation UI (picture 2)

[patent sand]

every single 4.5+ cover comes out to 4:44 if the song is longer than that

bug? feature?

[lunar gyro]

so when you extend a song, and then you want to cut a part from the "part 2".. how can you still "attach" it to the first part since its not listed as part 2 anymore

[patent sand]

attach it and then cut the part

like

"Get Whole Song" and then cut from that

[lunar gyro]

yeah the other way around would be easier in that particular case 🙂

[sleek stump]

Are you using a persona?

[patent sand]

no persona

clean cover

[polar spade]

When you Get Whole Song it blends literally about, well, from nothing to about .03 sec of the two. What part are you cutting that would still get a perfect blend of the two?

If you cut first then "connect" there's no quantized points nor blending available. Even do that in a DAW would be more work than needed. You would cut and paste, then likely have to splice in the Editor again to get a better good splice point. Then likely remaster to blend.

I may miss the logic of @patent sand suggestion being harder. I would think its better.

[sleek stump]

Running a test on one of my 6:30 tracks...

[sleek stump]

OK, first gen with sliders at their default positions, I got a 4:24 and a 4:55... Turning up teh style and audio sliders to 90 got me a 4:19 and a 4:24... YA somethings borked

[fresh wave]

How do you get it from starting dance songs in the middle of the song? It will literally start the song in the middle of an arpeggio. It makes everything sound confusing for a couple seconds.

I'm also not sure how to extend a song before it even starts, so that you can make it actually start.

Most of my regenerations today have been from songs that just didn't start at the start.

[patent sand]

how are you prompting the intro?
[Instrumental Intro - build-up]? something like that?

[unborn star]

i would remove the first -

[Instrumental Intro build-up]
like this

[fresh wave]

It may have just been [instrumental] for lyrics

[unborn star]

try avoiding any sign aka ,:-, in any []

[patent sand]

why? works fine for me since 3.5

[unborn star]

It could happened that any sign will stop suno for following the rest, just try it yourself, it's not always the case but for me often it happened it got ignored

For me it improved my songs

It looks like 4.5, if you use on any sung word with * Word *, he will stretch the word a lot. Not sure if know but it often happened to me

Can you useful

[patent sand]

giving it a go rn

[unborn star]

Just tell me your results and which version you used

[patent sand]

have to gen again

I left something in exclude and it muted my instruments entirely

v4.5
"[Instrumental Intro buildup]" : both seemed to build up nicely
"[Instrumental Intro build-up]" : one built-up ok, the other took longer but eventually did
"[Instrumental Intro - build-up]" : one built-up ok, the other didn't really build up

it seems more consistent

[haughty epoch]

Is there a reliable way to keep a voice male or female? I'm specifically (at least I think) to make the vocal male or female and it just randomly decides to change it mid verse

[patent sand]

clicking "Male" or "Female" in Advanced Options might help
see this also: #📚┃suno-school message

sometimes it just flips because of genre and you have to gen again

[haughty epoch]

Well for this song it's a back and forth male and female, I'm specifying verse 1 female only and about three lines in it switches to the male voice

for like one line then it goes back to female, it's really odd. I'm burning through credits trying to get it right :/

[patent sand]

as in you're prompting it to go back and forth? or it simply is?

the other thing you can try is replace section in the editor, but that's very taxing to get something decent for anything longer than a few seconds

[haughty epoch]

Yeah like [Verse Female] then it'll randomly switch to male durning line 2 or 3 of the verse

[patent sand]

check out the prompt link

I sent

it's more reliable to define the actual vocal instead of just "female"

[half dragon]

It's never reliable when it comes to duet - You'd just have to keep trying and the other option is to use editor to change the voice but it's also not guarantee.
Just need to be patience and ready to spend credits to achieve what you want.

[patent sand]

it looks like a lot but you can describe it to chatgpt and have it fill it out based on the template

[haughty epoch]

yeah I went into that editor but I couldn't figure where I could change the type of voice

I was trying Grok maybe I'll have to go to chatgpt lol

[vapid nymph]

Anyone else having issues with the mobile site not losing?

[prisma charm]

u can also

[Intro]
[Intro - Instrumental builds to a powerful and anthemic hook]
(...)

[Hook]
[Hook - Do That and that, you can also (Vocals: Do this) (Instrumental: Do that)]
Ald So On~~
—And So On!—

[eager compass]

Anyone else having issues extending? All but 2 of my extends today have started singing the end of the song and never touch anything else. Like, I'll have a verse, pre-chorus, chorus and the next verse or bridge and it will start and repeat the last 3 lines at the end.

[half dragon]

I've been working on my songs all day - lots of extensions... no issue for me.

[eager compass]

At least it's not site wide I guess. Think I'll just cover the song and try to extend from there. Maybe this one is bugged. It's 4.5+ and can't remaster it. Already tried everything else from changing the language, persona's, browsers, clearing cache etc.

[placid tendon]

How do I access the advanced menu on my mobile browser? I only get there by chance.

[half dragon]

Make sure you select 4.5 model and up.

[placid tendon]

Yes I have, but when I click on the create button below I only get into the without the advanced mode

[half dragon]

it looks like you still have the old UI.. strange! Clear you cache and relogin?

[eager compass]

Try tapping the "X" at the top right.

That worked for me earlier.

[placid tendon]

If I press x I have the workspace to select, so I am out of the create menu

[azure crane]

The reverb in v4.5+ that makes every track sound like you're in a large church makes me nauseous... Why is there no way to reduce it?

[eager compass]

Maybe you have to select a workspace first? This is what happens with me. I probably should have chosen a shorter song for the example though.

[half dragon]

Have you tried this? Heard it helps some users
#💬┃general-chat message

[azure crane]

I tried a some other “styles” in the exclude area. Gonna give this a try. Thanks 🙏.

[junior phoenix]

@patent sand Like turning on an old record

[steel pecan]

Don't use long intros, otherwise he won't interpret them very well.

As far as possible, separate the message in verse 1, you will take it as a dialogue

[eager compass]

Not sure I fully understand.

[steel pecan]

It doesn't recognize intros well, it usually cuts off words.

Most of the time intros work very well, but when they are too short

[eager compass]

Got it. That's not an issue I'm having though.

[steel pecan]

I was just giving you a recommendation, you could take it with good character.

It's always good to learn something new

[eager compass]

I wasn't trying to come across rude. I was honestly confused. I do appreciate the info and will try to remember it if I come across it. I also did not want you to try to fix an issue mistakenly.

I'm sorry if I came across as rude and disrespectful.

[steel pecan]

Don't worry, I hope they help you solve it.

[eager compass]

Thank you.

[steel pecan]

A pleasure

[placid tendon]

Thank you, now the old UI still comes when I go to create, when I click on x the new UI comes, but somehow I can't get to the workspace overview now 🤔

[unborn star]

Like I said =)

[azure crane]

I tried it and it didn't work at all :-(.

[ebon sand]

Do you have an example song where this happens with the reverb?

[azure crane]

Send it to you in DM

[ebon sand]

I'll have a look, brb in a bit

[shadow plaza]

i have a friend who have his lyrics and want to sing the song so i just need to make him diiferent version of the music any idea how to do that with SUNO ?

[ebon sand]

you mean you need the instrumental track? maybe getting the stem

[shadow plaza]

well i need to use his voice into different styles of music

and of course the singing style is adjustes based on the music

is that even possible ?

[ebon sand]

i don't understand what the goal is, so the friend doesn't want to sing the song, he wants to upload the vocals or something?

[shadow plaza]

to use his voice basicaly

[ebon sand]

so you want to upload it to suno and generate songs with his voice?

[shadow plaza]

i tried that i get different voice

[ebon sand]

if this works use a different model eg v3.5 or v4 and generate until it fits

[graceful meadow]

Is there anyone else who has a problem with high-pitched female vocals? I have been trying to make another lullaby, but since the update, I only get high-pitched female vocals that sing as if they are performing a powerful hymn in a church, rather than a soft and gentle lullaby. No matter what styles of vocals I put in.
Does anyone know what to do?

[desert wadi]

For some reason when i try a cover a song it does not go beyond 4.22 minutes? Why is that? i am using version 4.5+

[polar spade]

Covers and Personas need a bit of reserved space to include a "seed" info from the original. Its somewhat proportional likely.

When 4 minute limits were covered we could get only 3:14 songs , but those could even get cut off and not finish.

With 8 minute limits that give enough to cover a 5 or 6 minute song, reserving a bit for the code to seed and leaving enough to complete about a 4 minute cover. Im estimating everything.

If you covered a 1 minute song, the possibility of getting a couple of 7 minute songs out of a cover may be higher.

So, what happens if you take a 8 minute song, just splice out everything thats repetitive, so you're down to 1 minute, save, then cover that with the complete lyrics (if lyrical)?....... a weekend project for you to find out. 🏋‍♀️

[shadow plaza]

so i mark 10 songs from my Workplace
create new Workplace Move them there
but they are now in Both Workplaces ? Is this normal i wanted to organize my main Workplace litte bit

[polar spade]

Workspace and Library are the same thing basically, just a placeholder for all things created. So you cant "move" anything away from them, that would be the same as deleting the information.

The Filters added to Workspace was to give it more of a workhorse feel. But enough users want that for Library too, I guess, so you got basically the same thing now in both (we really no longer need the main Workspace. ) so just the separate workspace folders are a convenience and nice to have.

Nonetheless, its still just one file/output. Then you put peacemakers where ever and how many ever you wish. Delete from one and it disappears from all.

So a new Workspace folder just helps you temporarily organize your outputs. But the files always remain in Library/Workspace (main)

Since they clogged up the Create page now for mobile with two (2) Create panels at the moment, i cant even see my Workplace . So, its good that Library covers the same thing with filters.

[tardy coral]

dear god please create a button that ensures NO DRUMS or PERCUSSION are generated.

it shouldnt be so damn hard..

[polar spade]

If they reverse stem so you can integrate (reconstitute) all tracks back as one but excluding those you dont want, then that will get you what you want -- and get others what they want when they dont want piano, for instance. Guess maybe in ver. 5.0?

[tardy coral]

yeh for sure can create stems... but I want minimum artifacts for professional use, and if its generated without drums-percussion in the first place then the chances of better quality audio parts is much higher

[plucky bloom]

Hey, im looking for a good AI checker, found this one - https://www.submithub.com/ai-song-checker

[polar spade]

To do that, the model would have to be extensively trained to recognize drums, and then percussion (distinctive from a xylophone or handpan or guitar tapping) , and then trained to exclude piano, and then sax and then......

The cost would be extensive.

Presently all it can likely do behind the scenes is to create the generated output based on genre and prompts, then , now that its limited down to one output, distinguish the exact moments and frequencies of drums and percussions, then separate out the drums and regenerate all other frequencies to reconstitute a final output. This all done behind the scenes. They would like then charge Pro members $45 a month to cover the additional server load expense just to convenience us.

Better to just pass the charge to the user. If thats what we want, we pay for it, individually. Once it generates new separate tracks, and then re-generates a new reconstituted track without drums, it should never have artifacts other than any normal issues with overtones and dissonance anyway, which have nothing to do with artifacting.

Again, just a thought. Your way could be better. 🤷

[tardy coral]

I´d pay any price to reduce artifacts.. $1000/month

[polar spade]

For $1000 a month I think most songs we could just hire good musicians. Haha. Well, I do alot of folk. I just need a good guitarist. 😁

[tardy coral]

yeh $1000 pushing it.. was just saying would be happy to pay for better quality if its simply a matter of price and processing power etc..

[wise kindle]

seems useless. it's pretty telling if a song is AI generated or not without the assistance of a tool

easier than image detection anyway

[patent sand]

I put 3 songs in and the server crashed

notably I'd also used this before when trying to find an alternative to sonoteller (good when it works, painful to get it to work) and it got the genre completely wrong lol

for a site that requires a sign-up to use, it's poor at best

[prisma tendon]

Now even on PC i can't play songs wth is going on?

[patent sand]

I would say "that's just you" but someone last night had the same issue

actually his issue was that he was turning off his phone and wondering why the music stopped playing from the browser

[patent sand]

check if the browser or an extension got updated

or clear cache for suno

[prisma tendon]

Already updated... Sometimes it starts to work normally, some time need to wait like about 5min ( i have 1000mb speed and it works well, i did check ) sometimes it just doesn't work at all... Like servers are half dead

[patent sand]

no I mean if there was an extension update, it might have broken something

usually that's a UI problem though

[hollow hare]

Have you tried using another browser to see if the same issue can be replicated? It seems to work fine for me in Firefox Beta

[dawn pilot]

Does anyone have detailed style instructions. Earlier it was limited to 200 characters, so it was pretty clear that you could not specify genre and a few things. But now it can be much longer. The knowledge base on suno website is not that descriptive. Does any one have a good resource that I could use?

Also, I am new to discord, so let me know if this is not the right place to ask this question.

[patent sand]

this is the correct place

what I have started doing is

Genre(s): Genre1 (XX), Genre2 (XX), etc.
Subgenre(s): Subgenre1 (XX), Subgenre2 (XX), etc.
Lyrics: detailed description of lyrics
Mood(s): Mood1 (XX), Mood2 (XX), etc.
Instrument(s): Instrument1, Instrument 2, etc.
BPM: i.e., 109
Key: i.e., A Major
Vocal Pitch: i.e., Low/Mid

where "XX" is an influence factor generally between 50 (least influential) and 90 (most influential) and ordered in the list such that Genre1 is the most influential, Genre2 is the second-most influential, etc.

I developed this after using sonoteller for a song awhile back and it pulled this information for it, so I've been adjusting it per song now based on my template. sonoteller itself is relatively reliable if you can get it to cooperate, but it's more for pulling genre/etc. information from existing music. it's good for if you have YT uploads that you want to check information on

there are other style instructions if you scroll through here also

[graceful condor]

I don't know if this is already discussed but do you guys know if Suno plans to include some kind of ability to work with impulse responses like those in convolution reverb plugins?

[half dragon]

I've never heard they ever announced their plans - so no one really knows.

[winged jewel]

Anyone can help me to get different time signatures? I tried months now to get anything but 4/4 meters. Any help would be appreciated.

[shadow escarp]

I’m trying to have a spoken word dedication prior to song. Tried [spoken word] ( this song is dedicated to….) as well as other prompts but nothing works. Suggestions?

[cold citrus]

I've found doing just spoken word stuff a but difficult, I use eleven labs for simple spoken word, make he song with suno, and blend them with audacity

[shadow escarp]

I would like the same voice talking as well as doing the song. Not sure I can do that

[lost forge]

Generally
[Intro]
[Spoken]
This is spoken text

works for me. The simpler the better for Suno to follow it from my experience.
Example: https://suno.com/s/qcDRn9oPJrv2lRZf

[lost forge]

And yes, that works for me normally as well.

[graceful meadow]

For me to get Spoken word to work, I need to both put [Spoken word] & [Spoken verse] in the style prompts, then put [Spoken Verse] over every verse, sometimes in the middle as well if I don't want it to turn into song. So far (knock on wood), it has worked well.

[shadow escarp]

I’ll keep trying thanks for all responses

[cursive lava]

so what you all putting in excluded styles? think it matters much?

[lost forge]

From my perspective since 4.5 I‘ ve not used exclusions.

[cursive lava]

i just put rock and pop hope it does stuff

you guys using claude for advance prompt making ? - gpt 4.0 keeps fukking it upp, you know why claude so good at these complex abilton prompts?

[empty bobcat]

I'd love to listen suno radio in my music player. are there any direct stream URL?

[patent sand]

"Instruments: Synthesizer"

in my case specifically the synthesizer keeps fuzzing out the output after a few minutes

I mean I like the synth, but not that much

[olive crystal]

Is there a limit to the number of instruments that can be used in generation? I'm trying to figure out how to prompt the use of different cultural instruments for some sections, going from Sulmerian into Slavic into Celtic into Japanese. Can't seem to figure out if this will work or not.

[fringe crane]

When saving a song that had its lyrics edited in the editor, i always get BAD sound degradation on the WAV download of that song, but it still sounds good when played in browser on suno webpage. Is there a way around that, where u still can use the editor and then download the wav or am i forced to cover/remaster the edited song?

[raven turtle]

I usually put guitar, as I make electropop. It tends to work most of the time

[glacial moat]

HELP 🙂 How do I extend a song from the end. Its made a song I like but its only 1:28 I want it a fair bit longer . I've clicked edit extend, set the red line where I want it to carry on from, moved the white slider as far to the right it will go, but it just seems to made another song 4 seconds long then refunds me.

[glacial moat]

what Celtic instruments ?

[half dragon]

Try not to extend from the end - move the marker back a bit into the song - let it remake the last verse and continue on...
On the lyric box - make sure you include that last verse and add new verse to continue
Most of the time for me, a last verse part should be good enough - but some song it needs more depends on the structure it ended on.

[glacial moat]

aghh it worked once, added a min on now its back to doing the new sonf 4 seconds long thing

[polar spade]

Maybe not necessarily as you create the song with extensions, but if you use remaster it may blend a few of the instruments to sound like one another. But why should I be a wet rag, just have fun and come up with the answer yourself.

Putting them all in at once though isnt necessary a productive thing as you'll just get a song with maybe a couple of the instrument spread out as fill-in elements.

Youd be better to use extension with each elemental change.

[glacial moat]

ahh me fault I forgot to get whole song 🙂

[lyric snow]

anyone have a working app producing great voice mine are pretty robotic

[polar spade]

I will add, Suno says it will work to phrase prompt "first this, then into this , then into this..." I cant say that I completely understand what it means though. But you can try phrase prompting in that manner too. Someone might help here who does it that way so it doesn't get the instruments of one section to bleed over into the other sections.

[brazen trail]

Hey everyone,
I’ve been using Suno AI with a Premier yearly subscription for a couple of months now, but I’ve been facing persistent issues. I’ve tried the V4, V4.5, and V4.5+ models, but the results are consistently poor — distorted vocals, audio glitches, incomplete songs, and lyrics that don’t follow the instructions. The overall quality is so bad that the generated tracks are completely unusable and not listenable.

Honestly, it feels like Suno AI has always been plagued with bugs, and no matter what approach I take or style I try, the issues persist. At this point, it feels like a waste of both time and money.

I reached out to support a month ago requesting a refund, but I haven’t received any response. As expected, Suno’s customer support is as bad as their product or song quality.

Does anyone know how I can get a refund or escalate this issue? Any help would be greatly appreciated.

[wise kindle]

I found 4.5 pretty clean actually. these days I make the song in 4.5+ because the arrangement is better, but quality sucks, then feed it back into the original 4.5 model

as for refunds, payments are generally final and non-refundable. you can email them and try your luck if you wish, but I wouldn't give my hopes up: https://discord.com/channels/1069381916492562582/1373016771103166575

[sleek stump]

A couple things to note about your issue here.. First, you need to have some specific reasons for a refund, they are outlined here https://help.suno.com/en/articles/2550209 . Secondly, Its more likely a prompting and guidance issue you are having. There were significant changes to the way models are prompted starting with v4, and then in 4.5 it all changed again and got significantly more complex. 4.5 and 4.5+ need very verbose and descriptive style prompts and meta tag guidance in the lyrics to achieve what was relatively easy in 3.5 and older models.
Don't get me wrong, all the models have issues, and likely always will as this is an experimental platform, as are all AI platforms and anyone using any AI platform should be wary and set expectations accordingly.

[dawn oasis]

Has anyone had any luck with getting Suno to accent phrases the way you want them to? For example, a song of mine contains the word “music.” Suno rendered it as, “mu-ZIK.” Obviously I would prefer that it be pronounced “MU-zik.” It fits the rhythm of the song just as easily to pronounce the word naturally — better really — but I don’t seem to be able to get it to pronounce it the way I want. I’d rather not spend too many credits editing, and getting the same results.

I have another song where it insists on pronouncing “thirteen” as “thirrrrrrten.” I’m not sure how to communicate with it what I want better.

[sleek stump]

Phonetic spelling will help you a lot here.

[dawn oasis]

Thanks for the suggestion. To do this, would I use accent marks or just capitalize the syllables I want emphasized? I don’t want it to shout the syllables, but just stress them.

I also don’t want it misinterpreting the capital letters as an abbreviation

[sleek stump]

You can try quotations, exclamation points, Caps.. theres a few ways to do that

hyphens work also sometimes

Cap letters hyphen rest of the phonetic word.

[gloomy shore]

Does the ai recognize the command, [Backing] singing vocalist?

[hollow hare]

Yes, you need to use parenthesis to prompt for it

[Female] Echo, echo ([echo, male] echo)```Depends on the flow of your song but it usually works.

[gloomy shore]

I want to make a backup singer to be singing one single line together with the main singer and while the main singing a different line. It is similar to duet but each one sings different line but together. Is it possible?

[half dragon]

I don't think you can control that. It coud happen but not possible to instruct.

[polar spade]

That's "counterpoint" or informally a "medley" or "singing in unison" .

You can try the simplest prompt of putting the back up lines in parentheses like :

Mary had a little lamb (Baa baa black sheep)

Typically it will sing the second just after the first
But you might get the illusion that its singing somewhat in unison.

What have you tried?

[rough reef]

"MAKE PERSON" is completely broken. It repeats the same vocals from the previous song in the new cover.

[daring lava]

Hey there! Great question!

The model being used would determine the best approach. The earlier models respond best to genre tags in the style of music that typically move in a 3/4 structure such as waltz.

With the newer models (4.5/4.5+), you’ll find that including things such as 3/4 time signature, triple time, three-quarter time can be helpful within the style of music field. Would be mindful that the weight (amount of) tags included that suggest against this rhythm and pacing may reduce the likeliness of achieving the desired result.

Additionally, emphasizing musical direction contained within brackets inside the lyrical field can be useful to hammer in guidance. This too, has a higher response rate with the newer models.

Lastly, and my favorite approach, inject the rhythm via upload and use it as a base to build from with one of the Remix tools. This can function consistently with all models pretty well.

Hope this helps!

[midnight plume]

Type it in as mew sick.

[polar spade]

Slight correction?: mew zick

[magic shale]

I want to make an identical cover of a song. Anybody who noticed what effect the "weirdness" slider has on this ?
For the other 2 sliders I know the effect.
And also, what happens if I dont place any promp in the style window ?

[sleek portal]

where is the option to reuse prompt from old songs hiding? 🤔

[young geode]

It's possible, but as I've told users in the past, it's difficult.

In order to get a second singer, you first need to have it within your style prompt, something like: (Main Singer: Male, Baritone, Range, Style. Second Singer: Female, type, range, style)

Then you need an adequate lyric 'texture' within the structure of your lyrics. This means creating opposing/complimentary meterlines, call and answers/echos/filler:

Ex.
I step through the veil/fear and doubt
u/uu/ - u/u
Accentual/Amphimacer

Next, your metatags and instructional punctuations need to be folded in:

[Verse: Call and Answer]
[Main Vocals: Male, Baritone, Range, Monotone]
[Second Vocals: Female, Type, Range, Airy]
I step through the veil / (fear and doubt)

Additionally, giving enough space within the lyrics, literally, can also help Suno identify what you're asking.

That's how you get a second singer in.

Then, you'd want to ensure Vocal Harmonies are in your style prompt and layered into the areas you'd like. In addition to this, you'd want to include a genre/subgenre that regularly uses backup singers and harmonies:

[Verse: Call and Answer]
[Main Vocals: Male, Baritone, Range, Monotone]
[Second Vocals: Female, Type, Range, Airy]
I step through the veil / (fear and doubt)

[Vocal Harmony] A blinding void, surreal, yet without sensation. [Vocal Harmony]

Note, this is not guaranteed, but this weights all the probabilities within that direction. Once you finally find the Vocal harmony your looking for, rip the vocal stems, then edit them to have the kind of vocal delivery you're looking for and turn it into a persona.

From there you should be able to continue.

[young geode]

First and foremost: research your genres. If you're using genres that almost only have 4/4 timing, you'll almost never get anything but that timing. If you want different timing signatures, you need to be using genres with those timing signatures.

Second, look up the different kinds of beats, such as Train Beat, Ballad 6/8 Beat, Shuffle/Swing beat. Suno responds stronger to the (technical) names of beats, than simply the timing signatures.

Third, focus on the kind of Drums/Bassbeat you want to hear and search for the named kits/models/machines that are used: Tama Kits, for example. If you write your prompts with just (Drums, 6/8), it far less likely to work than (Drums, Tama Kit, 6/8 Train Beat).

Lastly, as the other user pointed out, You can goto sites like Metronomer and download 10-30 second beats under the timing signatures, and BPMs, that you'd like and upload them. That will guarantee that the song follows that pattern.

[neat prism]

Bonjour a tous....un peu d'aide please ! Utiliser sa propre voix sur Suno est vraiment une galère, il y a t-il une vraie solution une méthode qui fonctionne presque a 80% ou 100% ?

Hello everyone... a little help please! Using your own voice on Suno is really a pain, is there a real solution, a method that works almost 80% or 100%?

[spring spindle]

Unfortunately, 4.5+ always inserts strange words at the end for me. How can I fix this?

[vivid cedar]

@spring spindle I've had the same problems a few times. I'm thinking Maybe in a early seed og the tune, I might havde writen som credits (for the video version), which I dont do anymore. Anybody else experience strange word at the very exit of the son?

[polar spade]

Above Create button:
Save prompt / Clear All

Then click "+" beside the Create button to Add Prompt (use a saved prompt)

[neat prism]

Mais c'est quoi ce nouveau Suno...je reviens 6 mois plus tard depuis la version 3.5 avec laquelle je pouvais enregistrer ma voix et cela fonctionnait plutôt bien et maintenant sur la version 4.5 , je crée una persona pour ma voix certaines fois ça fonctionne mais il ajoute toujours une 2eme voix sur les chansons ???!. Ensuite Suno a supprimé "Add song" donc impossible d'utiliser cette option pour ajouter ma propre voix ????. De plus je constate que des centaines de personnes veulent poser leur propres voix et demande comme moi comment faire...et chose bizarre personne n'as de réponse ou de solution !!!

quelqu'un pour m'aidé ?? ou pour me répondre et me dire si oui ou non on peux ajouter sa propre voix de maniere fiable ?? merci

[weary glen]

guys what prompt do i need to make the voice is maintain

like i want it on high vocal only

[polar spade]

You were responded in need_help channel as well.

I guess the answer is "no", there is no means with Suno to "reliably" clone a voice. If any one of "they" say it can be done 98% of the time for every upload, then ask "them" directly to show proof in real time. For me, I upload my Audio of my singing, slide Weird to "0" , Audio to 78%-98% and Style "0" or it varies depending on the previous outcome.

Je suppose que la réponse est « non ». Suno ne permet pas de cloner une voix de manière fiable. Si l'un d'entre eux affirme que c'est possible dans 98 % des cas pour chaque téléchargement, demandez-lui directement de fournir une preuve en temps réel. Personnellement, je télécharge l'audio de mon chant, je règle Bizarre sur « 0 », Audio sur 78 %-98 % et Style sur « 0 », ou cela varie selon le résultat précédent.

[sleek stump]

If you are asking about the Reuse Styles and Lyrics its now in the Remix section of the 3 dot menu.

[oak osprey]

Hi all, I have a question, maybe someone has encountered this please advise me how do I make a song exactly 45 seconds long? Tried to prescribe promt to create a song for 45 seconds but still ignores and creates a minute or more. Tried to shorten the text of the song the same story. Maybe someone has encountered this situation, please tell me how to solve it.

[sleek stump]

You can not prompt for a specific length/time. The best you can do is ask for things that are short, Jingle, Demo, Intro, ect. What youy will likely have to do is gen a short track, then crop it.

[oak osprey]

Ok, thank you

[weary glen]

do you have more example that sound similar?

[sleek stump]

I have tested this before, using Jingle in your styles will get you anywhere from 30-60 seconds, its usually right at 60 ish

[echo moat]

Hey guys, been using suno for about 2 weeks now, read through the wiki, but still seem to be struggling to force instrumental and solo sections in already completed works. What's the best strategy for this? I Tried
[Epic Guitar solo] and [Instrumental][Epic guitar solo} but she keeps singing

[polar spade]

@oak osprey its been easy for me to do this way Ginger suggests. You can at least try "Jingle" and let it give you a 1 minute output then splice 15 seconds out.

[young geode]

To maintain a high vocal pitch?

Male, Countertenor, Countertenor E3-E5
Female, Soprano, Soprano C4-C6

The inclusion of genres that have these voices will also help. Be aware that voices are the hardest to control for, because most genres have a 'voice' that's associated with them, so the AI is only trained on those voices.

[polar spade]

I just noticed the "activity feed" . I know my comments aren't private, but they're intended for just the song and user I send to, not for everyone to see. Sheesh.

So we can leave a long spam comment on our own songs, so it shows publicly and not get blocked for it. Cool.

[half dragon]

I do this usually after a Chorus and it works 99% of the time
[Melodic Instrumental]
[Guitar Solo]
If I wanted longer solo section - I would add again right after
[Break]
[Melodic Instrumental]
[Guitar Solo]
...
[Epic Guitar Solo] works too but epic or not it's an opinion 🙂

[karmic pewter]

Some accounts are bugged, if you get literally zero good results, that may be it

[gusty chasm]

Hi fam! Quick question, I can´t find the answer in the docs or the server. Can I distribute and monetize a song I create with vocals in Suno? I mean, I wrote the lyrics and produce half of the instrumental but the vocals are AI. Thanks in advance!

[viral pond]

how can I create a style prompt to sound like dragon Ball Z ? 😄

[weak moss]

what are some prompts that yall have used to affect the timing of instrumentalization in your songs? Like minimalistic instrument playing solo, then another one layering at this time?

[viral pond]

ia there no way to create a cover song + a persona of own voice ?

[thin kite]

Hello, I have a question: Can two independent songs from SUNO AI be combined into one SUNO AI song? Second question: For some reason I can't upload "VIDEOLOPS". for about 3 months now. Until then there was no problem, since then it doesn't work. Can someone take a look at it, I'm getting desperate!!!

[weak moss]

and, how do we incorporate samples in songs? for examaple, I want to preface this as a foreword: https://www.youtube.com/shorts/puTEFDXvHyE

[opal orbit]

If you have a pro or premium account, yes. Anything you create in Suno is considered yours to distribute.

If it was created with a free plan, then no.

[dawn oasis]

Oy. Seems like everything I try just makes it worse. I’ve spent more than 100 credits trying to fix one word. I looked up phonetic spell for “music” and went with “MYOO-zik.” Suno apparently reads that as “my-oo-zik” or just gets completely confused and says what. (I didn’t try to replace just the one word — I did the phrases around the word, changing only the word “music.”)

[sleek stump]

Its against Suno's TOS to upload any audio or samples that you do not own and have complete rights to. If you wanted to do that you would have to do it outside Suno, and then risk a copyright violation on what ever platform you put it on.

[opal orbit]

What has worked for me is phonetic spelling it out with actual words instead of sounds, with connecting hyphens. Then add instructions to sing consistently at the beginning of your sentence so it doesn't 'stylize' the phrase.

Try putting something like this;
[Sing smoothly and consistently]
"Your lyrics 'Muze-ick'"

Or whatever works best for the pronunciation you want.

[sleek stump]

There are multiple ways to phonetically spell things. I would personally try Muuzic if yo uare trying to accent the first syllable.

[sleek stump]

On your questions... First one, no, Suno does not have the capability to combine 2 completely separate tracks, only through extending can Suno combine clips.

On your second, what is "VIDEOLOPS"?

[weak moss]

I'll risk it. Sampling is creative.

sampling is common throughout raps

thus, I'm preserving the spirit

not doing so, would in effect

not give flowers to John Witherspoon

who passed away

the meaning is in the sample

preserving the spirit of the artform

nostalgia about this point in time

[sleek stump]

That is your perogative, but Suno won't allow it on its platform, you would need to add that outside Suno

[weak moss]

I provided context

I'm an artist

IYKYK

[sleek stump]

Still doesn't change the facts 😛

[weak moss]

i'll share with you lol

its a rough but here goes,

Say man,
Chips ain't worth a dollar know more;
Daddy tolled about burgers when they were 5 cents,

Acting my wage, I dont know nothin,
Tryna enjoy the days to mind from suffrin'

And family divided over inheritance, and I learned and lot of gems from my service,
I'm still making sense of the Suffrage!

Machine taking its toll, stratified roles from professional, there's some lessons,
How much is your time worth compared to mine; You get the message,

Stratification here feels so divisive,

I hope this provides extra context, from the sample, in the same spirit.

[sleek stump]

You dont need to justify anything you want to do, I am only informing you of Suno's stance and terms. If you attempt to upload that clip the content filter will likely not allow it from the start, and if you do get it through, it would be nuked later as a copyright violation.

[weak moss]

You're ignoring context willfully

heres a solution.

leaving the sample out, how do I preserve the spirit using prompts.

focussing on what I wrote instead

[sleek stump]

Context matters not, its the Terms of Servicxe you agree to.

[weak moss]

I'm just an artist expressing myself

and the constitution ranks above TOS

freedom of speech

lets talk about prompt talk

[sleek stump]

If you want to include what the artist said, you can type it in and a spoken intro to a track you are generating...

[weak moss]

Why type what he said, when I can simple sample the original

its a double bind

catch-22

thats why I'm asking

I provided context

[sleek stump]

Its still copyrighted, and you would be on the hook for that if someone that owns those rights wanted to push it. If teh content filter ctaches it, it also wont gen the track with those lines in it.

[weak moss]

Ginger. I'm willing to take the risk.

I want a solution, not a conversation about the NOT solution

[sleek stump]

Go ahead, I'm not stopping you, I'm just giving info.

[weak moss]

You are, but I'm asking for help.

[sleek stump]

YOu are asking other people to be complicit in a legal violation...

[weak moss]

Is it illegal what I wrote after the sample?

"Say man,
Chips ain't worth a dollar know more;
Daddy tolled about burgers when they were 5 cents,

Acting my wage, I dont know nothin,
Tryna enjoy the days to mind from suffrin'

And family divided over inheritance, and I learned and lot of gems from my service,
I'm still making sense of the Suffrage!

Machine taking its toll, stratified roles from professional, there's some lessons,
How much is your time worth compared to mine; You get the message,

Stratification here feels so divisive,

[weak moss]

You are obstructing, respectfully I want to ask someone else

[sleek stump]

Maybe @wet minnow or @final hazel can weigh in here.

[weak moss]

I will clarify, please read the context

I'm not trying to detract from prompt talk

[final hazel]

Hey, I hear your passion and the intent behind your art. Sampling is a big part of music history, but Suno’s platform doesn’t permit uploading copyrighted material, even if the use is artistic or respectful. That’s not to block expression, but to protect both users and the platform legally. If the sample is important, consider paraphrasing the essence of what was said and build that into your lyrics or prompt instead.

[weak moss]

I will use another software to add the sample ma'am

May you help me with this? Say man,
Chips ain't worth a dollar know more;
Daddy tolled about burgers when they were 5 cents,

Acting my wage, I dont know nothin,
Tryna enjoy the days to mind from suffrin'

And family divided over inheritance, and I learned and lot of gems from my service,
I'm still making sense of the Suffrage!

Machine taking its toll, stratified roles from professional, there's some lessons,
How much is your time worth compared to mine; You get the message,

Stratification here feels so divisive,

how to preserve the spirit of the inspiration?

like art directing the pacing and structure

and timing

[final hazel]

Hey, totally get wanting to preserve the vibe. The best way to do that in Suno is to rewrite the sample in your own words and guide the tone with your prompt.

[weak moss]

I did

what I wrote above

I'm asking here because I need help with how

you all are more experienced than I

I'm just some guy

Moreover, does our first ammendment rights not apply over TOS?

considering the context

and the expression

[sleek stump]

No they dont, actually

[weak moss]

Please give me guidance

[final hazel]

Hey, totally hear where you’re coming from. But First Amendment rights apply to government action, not private platforms like Suno, their Terms of Service still apply here.

[weak moss]

I will take full liability. Not suno. Its in the TOS

I just want to make progress, I'm being patient, but I just want to make progress

[final hazel]

I get wanting to push boundaries creatively, but if you're saying outright that you plan to ignore the TOS, we can’t support that here. This community exists to help each other within the rules, not around them.

[weak moss]

Omg

[wet minnow]

You called?

[weak moss]

I need help with this: Say man,
Chips ain't worth a dollar know more;
Daddy tolled about burgers when they were 5 cents,

Acting my wage, I dont know nothin,
Tryna enjoy the days to mind from suffrin'

And family divided over inheritance, and I learned and lot of gems from my service,
I'm still making sense of the Suffrage!

Machine taking its toll, stratified roles from professional, there's some lessons,
How much is your time worth compared to mine; You get the message,

Stratification here feels so divisive,

[wet minnow]

i think you have posted that already like 5 times or so... ok

[weak moss]

Because I don't want us to lose track of context

my intentions

vs what these people are focussing on

I proposed the solution

[wet minnow]

just answer people. it makes chains of the things.

[sleek stump]

You dont need to keep pasting teh same section of text over and over...
WHat are you trying to do with it, do you want it spoken at the start of your track? Sung? You havent really detailed what your actual goal is, or what model you are using or your prompt attempts so far.

[weak moss]

Jesus f christ

[wet minnow]

even woody isnt amused. yay

[weak moss]

I never wanted to talk about it over and over

I wanted to talk about a solution

[wet minnow]

Is the content copyrighted?

[errant silo]

yo, anyone down for vc? Been looking for people to talk shop with.

[weak moss]

NO

[final hazel]

#📚┃suno-school message

[wet minnow]

lol

just lol

a lightly sad lol tbf

[weak moss]

I'm going to be honest. and as a foreword I mean this respectfully

[wet minnow]

in all respect. you dont own that sample.

[weak moss]

Figuratively, I went to a 7/11 to get some water, and came back with none,

[sleek stump]

Have you thought about using a AI chat model to help you form prompts to get what you are trying to do?

[weak moss]

I don't want to.

Its improv from me

Time isn't on my side

thats why im frustrated with you

your obstructivism

[sleek stump]

Yet you want one of us to tell you the same thing...

[weak moss]

sh$$

YOU

brought them here

because you felt some kind of way

you didn't help me

you didn't listen to context provided

[wet minnow]

if she told ya its a tos thing that samples (you dont own the rights to) are not allowed. thats correct. nothing more to discuss. no support to give.

[sleek stump]

I am just another user here, that spends a substantial amount of my own time helping people, I dont get anything from this, and neither do any of the moderators.

[final hazel]

No need to get disrespectful to other users. You openly said you’d use a copyrighted song sample. Whatever your reason, Suno’s TOS clearly says not to use copyrighted material in music creation. Once you go against that, you're on your own. As you said yourself: you'll be liable. So don’t pull us—or this platform—into that.

[errant silo]

facts. what you can do is make your own sample pallette by genning a bunch of different types of stuff. I set weirdness high, 80% or higher, then when i get something cool i drop it in a daw. i take other stuff, match the bpm, and slam it into suno. have had really cool results. @weak moss

[weak moss]

I apologize, I got censored for 10 minutes for censoring cuss words

[half dragon]

Want to talk solution? many times the prompt solution has been provided - have you tried it? even using chatgpt to see if the prompt could describe what you need - but you refused to.
Also been warned about using the sample is against TOS - but if you really want to on your own - try it then. No one is stopping you. See if that helps and see what other roadblocks you could run into from that. I don't understand the resistance. If Suno couldn't provide what you need - there are other solution - sample it externally in our DAW.

[weak moss]

I am being respectful Ma'am. I will adress Ginger with a suno song

lets not talkn about the sample

Im not going to use it

I'm going to use what I wrote: Say man,
Chips ain't worth a dollar know more;
Daddy tolled about burgers when they were 5 cents,

Acting my wage, I dont know nothin,
Tryna enjoy the days to mind from suffrin'

And family divided over inheritance, and I learned and lot of gems from my service,
I'm still making sense of the Suffrage!

Machine taking its toll, stratified roles from professional, there's some lessons,
How much is your time worth compared to mine; You get the message,

Stratification here feels so divisive,

I make It CLEAR

the solution

is to NOT

talk about the sample

WHY

are we still talking about the sample?

ask

why do I keep pasting what I wrote

over the sample link?

I'm here,

because I want guidance on prompt talk

I want to talk to humans

[wet minnow]

dont istruct users how to use samples they dont own the rights to.. that would be nice.

[weak moss]

Did I do that, sir?

quote me.

[wet minnow]

sounds like it

[weak moss]

quote me

I didnt edit anything I said, just quote me.

[errant silo]

@weak moss add me back, bored and looking for the exact same thing.

[young geode]

Just DM me and I'll help you mate, let's keep the tech chat free from the back and forth.

[final hazel]

It's simple. If we are aware you are making a song that has copyrighted material in it, we will not longer assist. I like to repeat: the root is that you will add copyrighted material. We don't want any liability here.

[sleek stump]

Maybe a different approach would have garnered a different vibe and more public help... I asked earlier what your goal was, what you wanted the intro to be and what your prompting was already like, you didnt elaborate on any of that.

[final hazel]

I wouldn't go there. That sounds like you're planning a diss track, which crosses into harassment. Please be civil and keep things respectful.

[wet minnow]

but... i will remove the screenshot in favor of actual prompt talk to continue here... hopefully

[final hazel]

Moving on

😄

[errant silo]

So any insights on how i can get a hybrid theory-esque sound? That's been a real nightmare for me. Both me and Gemini are stumped.

[young geode]

You mean Linkin Park, Hybrid Theory?

[sleek stump]

As in like Linkin Park?

Myself and others get a very LP esk sound using Nu Metal as part of the styles prompting..

[final hazel]

nu metal, rap rock, alternative metal, hard rock, electronic rock, aggressive, emotional, anthemic, glitchy beats Is what I would try

[honest compass]

would be really good if you could just give it one keyword to say "look whatever genre... no beat, no percussion, no drums, just no freaking beat at all"

[sleek stump]

You can, you can just use a base genre in the styles window and let Suno grab somethign random from that genre.

[honest compass]

just doesnt seem to understand that concept at all, even thought it can have intros and outros without beats in

[sleek stump]

Unless you are trying to do somethign specific?

[honest compass]

just have something chilled where it doesnt advance into any kind of percussion.

unless you go with "acoustic" which then gives it that kind of style, you always have drums in there

[sleek stump]

You can specify like a Solo Piano Melody, or Solo Guitar folk melody...

I have had some good success with just an acoustic guitar

It also depends on teh model you are tryign to use.

[honest compass]

Solo could work, but then its just one instrument

[sleek stump]

You can also prompt for a 2 or 3 piece band and specify the band instruments. Genre will have a huge influence on whether you can exclude percussion or not.

[brazen trail]

You mentioned that "all the models have issues and likely always will," and described Suno as an experimental platform. If that’s the case, I find it very troubling — and honestly unethical — that users are being charged, especially at Premier yearly subscription levels, for a product that is openly acknowledged to be unreliable and not production-ready.

If the platform is still experimental, why is it being marketed and priced like a stable, commercial-grade tool? There’s a clear mismatch between the expectations set by Suno's presentation (including its UI, subscription tiers, and promotional language) and the actual experience users are having.

To charge users premium prices for a tool that "likely always will" have major issues — and then dismiss refund concerns with that justification — feels unfair and misleading. I signed up in good faith, expecting a usable creative product, not a beta test disguised as a polished service.

Is that really fair or ethical?

[young geode]

Then prompt it X as drums, X replaces drums, X keeps the pace of the song. And exclude: Drum, Drum beat, percussion, percussive beat, metronome, Bass, Bass beat.

And focus on genres that use no drums, like a capella. Classical, folk, Bluegrass.

[native minnow]

Hi there, I'm sorry if maybe this question was made before. Any expert "suggestion" to try to avoid a kinda metallic voice?
For now my only way is: try try try again and again (95% of the time I get a metallic voice), but maybe there is some tips or "special style" to use.
Like: High Audio Quality, Professional Mastering, Clean Mix, Minimal Background Noise, Less Distortion, Balanced EQ, No Heavy Autotune, No Robotic Tone
Btw I'm keep using 4.5, I found 4.5+ "too complex" for my taste.
Thank you so much in advanced.

[final hazel]

Your concern is understandable, but it is important to recognize how AI actually works. Unlike normal software that produces the same result every time from the same input, AI models generate outcomes based on learned patterns. Developers shape how these models learn, but they do not fully know nor control every behavior directly. Even the people who build these models are often still learning how to communicate with them effectively.

This is not unique to Suno. Every AI company faces the same limitations, because the technology is not a finished product. It is constantly evolving. Suno is upfront about this, and that honesty reflects a commitment to improvement rather than a guarantee of perfection.

If your expectations come from thinking of Suno as a traditional software product, that might explain the frustration. But if you approach it as a creative tool with potential and unpredictability, the experience can be much more rewarding.

[quick trellis]

Any official information on the "“Add Instrumental” feature? Perahps I'm looking in the wrong place and I can find little on youtube.

[native minnow]

101% agreed with what you said

[young geode]

What model are you using?

Is the metallic ting always there, or are you getting it after extentions?

What genres are you relying on and what instruments are you using?

[native minnow]

First of thank you for your very quick reply. As I mentioned I'm using only 4.5. I'm not really happy/entusiastic about the 4.5+
No, I don't use the editor at all. Only get problem from there. It "cost" me less, redo the song.
As a geners, I'm going: Nu Metal, Rock, Pop... and in all 3 cases (with many alternative) I have pretty much always the same issue.
Or metallic voice or really noise background. And unfortunately there are no 3rd party software to clean it.
I'm pretty happy anyway because I'm using Suno for my playlist and listen "my lyrics" during my work day, but in the end, you have to deal with it.

[young geode]

So you're getting the metallic ting at the beginning of the song.

Send me a DM with a link to it and I'll check it out

[native minnow]

No, usually I start to have problem after the middle. Sure, thank you so much!

[polar spade]

You start with a vocal stem. Then Add Instrumental [beta] is under the Remix/Edit section (at the bottom of the dropdown menu) of your output menu. Maybe you knew that, so Im not sure of any "Official" instructions on its use. Simple that you start with an audio vocal from your stem, include your text lyrics in the lyrics box , or new lyrics as is your choice to get a more weird outcome -- add any style prompts you wish, and click Creat/Remix to let Suno add an instrumental backing to it to output a new output

[brazen trail]

Thanks for the response, but I need to be honest — I find it extremely frustrating and unfair that a platform, which you now admit is experimental and unpredictable, is being charged at Premier-level pricing.

I subscribed with the intention of using Suno AI for creative and commercial work. However, despite trying every available model (V4, V4.5, V4.5+), various styles, prompts, and even "negative styles", the results have consistently been unusable — distorted vocals, audio glitches, broken lyrics, and incomplete tracks. It’s not just inconsistent; it’s often completely non-functional.

Meanwhile, instead of addressing these core problems, the focus seems to be on releasing flashy new features while ignoring long-standing bugs and user-reported issues. I’ve submitted multiple feedback forms and bug reports, but there’s been no visible improvement or acknowledgment, and it genuinely feels like feedback from paying users is being disregarded.

If Suno AI is still in a highly experimental phase, then charging commercial-level subscription fees without clearly communicating that upfront feels incredibly misleading. As a paying customer, I expected a functional creative tool — not a constant stream of trial-and-error that ends in frustration.

So I have to ask: if Suno isn’t a finished or reliable product, why is it being priced and marketed as one?

From my experience, Suno is not working at all, and I’m tired of wasting time and energy on a platform that delivers random, broken results and offers no clear roadmap for improvement.

[sleek stump]

Yes, I did say that its an experimental platform, I also said that all AI platforms are experimental, because their very nature of being under constant heavy development would preclude them from being a finished and released professional product.
Do professional people use any of these tools? Sure, a lot do actually. Its more about realistic expectations and how you manage them. In regards specifically to Suno, I can find nowhere in any of their pages or marketing that states the platform is a fully functional professional tool or platform. Again, expectations being what they are.

Is it misleadeding with the terms they use for their subscription levels? Maybe, I'll give you that one. But again, what are you expecting? What we pay for is processor time, which is paid for by using credits. We are not paying for the service, the service can be used by all for free, they are just limited in features and processor time.

I get it, you are not satisfied, Suno didnt meet your expectations. By all means contact support and ask for a refund, but only they can say one way or the other if they will or not. All any of us here can do is point you to published FAQs from Suno on how they address things like this.
I hope you can get a satisfactory outcome.

[echo moat]

how do i add vocal pauses in between verses?

[native minnow]

Honestly I totally get your frustration and sometime (dipends of the mood) I'm thinking the same.
But then... as mentioned before, you can't have too much expectation from the AI.
Different topic could be the "quality" of the interface (can I complain about it? ahaha)
The interface and the "quality of life" could be improved A LOT, just an example: in 2025 don't have a "sort by" it's a bit... eheheheh

[final hazel]

We hear your frustration, but it's important to recognize that AI-based tools (whether from Suno, OpenAI, or others) are not deterministic like traditional software. Every generation is the result of a complex statistical process, not fixed logic.

Even ChatGPT, (which you seem to be using here to answer me), occasionally gives incorrect or misleading outputs despite OpenAI's best efforts. That’s the nature of AI: it is powerful but inherently unpredictable. No AI tool, especially in music generation, can yet promise consistent commercial-grade results every time.

[polar spade]

what is a vocal pause.... a pause between the vocals? with or without music?

That might be harder than we think. Suno likes to fill in the spaces. Its often just a combination of the style of Genre and delivery, for instance, if we prompts for a spoken rendition of a long poem or monologue.

The one trick ive used is to create a blank space [using the Edit UI page] on your output then duplicate and insert (paste) as many and whereever you wish.

[echo moat]

my wife commented how it seems unrealistic cause she never stops to breathe lol

[sleek stump]

You can use musical pauses like Instrumental or Interlude or Break or Solo something... if you just want the whole thing to pause you can use a [Pause] tag in the lyrics.

[echo moat]

i have [pause] tags but i didn't capitolize the P so maybe that's what hapened

i just want the vocals to pause

[sleek stump]

Sounds like you need stuff between verse and chorus, like a Break or a Interlude or even a Guitar Solo...

[echo moat]

my wife said ai music would never be a thing, then i played v4.5 for her

[young geode]

I subscribed with the intention of using Suno AI for creative and commercial work. However, despite trying every available model (V4, V4.5, V4.5+), various styles, prompts, and even "negative styles", the results have consistently been unusable — distorted vocals, audio glitches, broken lyrics, and incomplete tracks. It’s not just inconsistent; it’s often completely non-functional.

Well. There's your problem: it's not meant to be used for Commercial Grade works. Maybe at some point in the future when new controls and weights and standard tagging recognition is implemented, but that's at least 2 years off: minimum.

However, all the complaints you have are a lack of capability and competence with the technology, not the technology itself. That's not your fault, cause there is no guide and everything is self-taught and communal discovered: but none the less it's a competence and capability problem on your end.

You can probably look through any of the Tech helpers here and find perfectly fine, near commercial grade work: and I guarantee we've all spent weeks fiddling and testing and trial and error.

[quick trellis]

Thanks. It just seemed strange that the Suno help contained details for "Add Vocal" and no mention at all for "Add Intrumental".

[polar spade]

That much is true... but on the faster pieces , the breath is likely more indistinguishable than it not being there at all.

Keep in mind (1) ALL noises are calculated to fit within the space Suno works out. and (2) Suno , since 4.0 usually always puts beathes where they should go , as if by a human. But yes, again, we typically cant sing as fast nor take a breath as fast as Suno can.

I choose genres and use sparse lyrics in order to both, sound more realistic and be singable by most talented humans. So it makes a difference.

putting in a blank space of non-breathing would sound just as strange.

[sleek stump]

Can you link one of the songs you generated so we can see what you are running into?

[echo moat]

well i think my main problem was originally before i read the wikis and guides, i just had everythign clumped together.

what does [break] instruct the ai to do?

[polar spade]

oh? I didnt know there are only references to one and not the other. OK. Well, they are about the same thing, just using the other stem vocal or instrumental.

[echo moat]

https://suno.com/s/AKqm9XxgkQ2wTOui @sleek stump

[sleek stump]

a Break will put a musical pause between stanza.

[polar spade]

@quick trellis Add though, the more clear and more melody you can add in the vocal, the better the instrumental may follow, but there is an issue on where exactly the instrumental may start, so you get different outputs.

[brazen trail]

Thanks for your input, but my concern remains the same — the platform is simply not working for me, regardless of how others may be using it. I’ve spent enough time experimenting, trying different approaches, and yet the results are consistently disappointing. I’m not interested in whether it works for someone else or if others have spent weeks fiddling with it — my frustration comes from the fact that I’ve given this platform a fair try and it has failed to deliver usable, consistent results.

At the end of the day, I'm paying for a product that doesn’t meet its promises, and I’m tired of wasting my time on a platform that doesn’t work for my needs. The lack of reliable output and the constant trial-and-error process isn't acceptable, and I shouldn’t have to become an expert in the system just to get basic functionality.

If the platform isn’t ready for commercial use, then why is it being changed as such?

I’ve made my concerns clear, and I’m not interested in justifying its issues with "it works for others." The reality is, for me, it doesn’t work — and I’m done with this failed platform.

[polar spade]

I sang along, I sang loudly and with force the Chorus in two breaths just like the singer without problem. The chorus seemed to have the longest phrases between breaths. Of course, I've had some vocal training and very good VOX capacity even at my age now.

[lost holly]

Is your goal something like this? https://suno.com/s/uhgYlIGNWQWSWQ5G Also like the song 😄

[echo moat]

ther'es also this one that i tried to make, but kinda gave up on cause i would try to create sections for instrumental solos and it would always put in vocals - https://suno.com/s/m7AxCS3erqQPa58h

[sleek stump]

Ok, some things to note about the structure, the genre you are using doesnt really delay between Verse and Chorus, so putting apause after every stanza will only confuse teh model and it will just ignore it and do what it thinks is correct from its training. What I would suggest here, and what I often do in some of the specific genre I write music in, is to put a Instrumental or break between the chorus and the next verse, with an instrumental and solo in the middle. and then a break after the Bridge.

[echo moat]

the lyrics are my wifes poetry, i do all the editing so it works out

[sleek stump]

Cool song btw

[young geode]

Then stop paying for it.

This is Tech and Prompt Help, not a forum.

[echo moat]

Thank you

[sleek stump]

YOu might also describe how you want the whole song to transition in teh styles section, which will help guide your tags in teh lyrics.

[echo moat]

thanks, this is all great information

is it better to write styles in narrative style or block formatting?

[sleek stump]

You are using 4.5+... 4.5 and + actually benefit from really descriptive and verbose descriptions of what you want in Styles, which is why the styles window was increased to 2000 characters.

You can also try using teh magic wand and see what the AI translates all you have there

[echo moat]

i haven't tried that yet, been using google gemini as of now

i'm stuck in my old ways i guess

if i want to go back and add pauses to a song, do i do a cover, extend, or reuse?

[sleek stump]

Using an LLM for prompting guidance can really help, but keep ion mind that none of then know exactly what Suno expects and uses, none of them have access to Suno training.

[echo moat]

i used the civitai 4.5 guide as a system prompt

it does well, could be better i suppose

[sleek stump]

I would suggest moving away from using [Pause], use more standard block tags instead, like a Break or Interlude...

[echo moat]

so just [Break]?

[sleek stump]

And yes, you can do a cover and change things around

[echo moat]

i'll try that, thanks

[shut ferry]

Is there a list of useable prompts anywhere?

[sleek stump]

Not really, there are users who have posted stuff, but there is no official list.

[lost holly]

Ok, if i publish a remixed version of your ellipse song? https://suno.com/s/bsEcm02FDwJG1L7S Really like it and want to be able to listen to it^^

[sleek stump]

And prompting does change from model version to version

[echo moat]

sure thing, i'd love that

What's the proper tagging for an instrumental solo?

[sleek stump]

The one you used worked as expected...

[echo moat]

i tried [Interlude] and it ignored it

in between verse 2 and chorus

pre-chorus my bad

[sleek stump]

If interlude did work, try Break, certain genre need them to be in certain spots

[echo moat]

it ignored my instrumental tag as well, wth

tried [Instrumental]
[Epic Violin Solo]
[Break]

[polar spade]

Just to point out, Suno sees something titled "pre chorus" then its a ramp up to the chorus and suno knows an interlude is not usually there. Even changing that title to Verse or something else can help.

[echo moat]

thanks

i put the instrumental block after the chorus

[polar spade]

You may also need to back down on the Audio Slider in order to get some influence on the Styles changes you are making.

[echo moat]

the audio influence?

[polar spade]

Are you just starting over? or Covering your original in order to make these lyrics/pause changes? Either way, actually, they both should have Audio and Style sliders near bottom for 4.5 Create/Remix panel

[echo moat]

just covering the original to make the pause changes

wanted to add a solo as well

some reason my prompts always seem to adhere better when it's a new song versus editing an existing one

[young geode]

Well these will create sound filler.

You can attempt use [Fade out] and [Fade in] and sandwich pause/silence

[Fade out]
....
[Silence]
...
[Silence]
...
[Silence]
...
[Fade In]

And within the prompt add in something like "Periods of silence between Fade outs and Fade Ins."

That might work if you're looking for actual silence. Hard cuts to silence are difficult for Suno

[echo moat]

i don't even need silence just the woman to quit singing but she's ignoring every directive to do so

[young geode]

Then Riff, Melody, break, interlude, instrumental, solo will all work, depending on how long you want that to be

[echo moat]

i have instrumental, solo, and break tags and she's ignoring all of them

i have interlude as well

audio influence is set to 20%

[young geode]

And place Sustaining notes, vocal sustain, vocal legato, vocal hold, note hold in the excluded.

[young geode]

Aree you using a persona?

[echo moat]

i think so yes

well the original song was, i didn't add one for the cover but

[young geode]

Is it your persona?

If so, find the original song, listen to it. Find the area where the vocalist immediately starts to do a note hold. Cut it at a good moment a line or two before that happens. Generate a short instrumental off that. Then recut and create a new persona.

Use that persona.

Likely your persona is forcing it.

[echo moat]

that's next level, i'll give that a shot

[young geode]

Remember that, for personas, you dont actually need the whole song either. Just capture the sounds/vocals you like and cut the rest out. Itll stop this from happening in the future

so even in the original song, if you find an area shes not doing this, you can use that as a clean cut

[alpine sand]

Hello, are the models which were used to create the text to music AI somewhere available for custom training?

Background: I realized that some vocals are not available and I would like to train the model on certain songs.

[young geode]

no

[echo moat]

the original song i did had no break tags or interludes, it was all one big paragrfaph basically, and i created a persona based on that. Are you telling me now that persona is trained to basically never stop singing since i chose the whole song as the persona?

so basically creating a persona is "Training" it based on what data you want it to know

[hollow hare]

The vocal system is a derivative from Suno's previous project, Bark, which they did make available on GitHub and huggingface.co:
https://github.com/suno-ai/bark
https://huggingface.co/suno/bark

Jonathan Fly also made a fork called Bark Infinity:
https://github.com/JonathanFly/bark

There are other Text-To-Speech (TTS) programs out there so be sure to look up which ones actually fits your need!

Be aware that support for Bark is not available, so if you run into an error, not much can be done to help you and you will likely need to solve it on your own.

[young geode]

If the Base song has that quality: yes.

Personas are effectively placing constraints on the models predictive token generation, trying to keep it down one specific path.

So if your original song had this run on quality where the singer was just going on constantly, than anything generated off of it will also have that qualite: especially if you made a persona out of the entire song.

[azure crane]

Why is it that some genres benefit from v4.5+ in terms of sound and production while others sound terribly bad, especially reverb and noise distortion completely ruining the songs?

[young geode]

probably how the token generation is confusing multiple genres/sounds.

If your prompts are somewhat sparse, and your lyrics without metatags, and you're not using the exclusion prompt area, it's probably trying to pull from too many places at once

[azure crane]

Ok, I just created a hard rock song that I really don't like at all, just to check it out, and it sounded really good with no loss of quality throughout the song. Really clean sound with perfect sounding vocals and balanced effects.

[polar spade]

I dont like suggesting much unless I can give examples, so heres a snippet of a couple of tricks:

divided the lines of the verse to help distinguish between verses and chorus, so it doesnt just keep hitting the downbeat on all the lines. It retokens the verse but keeps the chorus as it was, hopefully.

Adding some Ohs, and ahs, and mmms as part of the stanza to break up the 4 line trap. This , for me, tends to "reset" Suno and hopefully give a pause. I also added an extra "i" to "miine" to extend it.

Add a spoken part of anything... a whisper, silence or in this case i put a gasp

oh, and whether true or not, I did kick Weirdness up about 20% to again, try to get it to improvise more and hope that leans towards the vocals being more expressive than on the beat.

of course the rest of the song is a mix of trying things that didnt pan out as well, so at that point, i pesonally would throw it into the editor and start extending each stanza and to control the pauses better. For me, I dont mind going through 5 pages of outputs and credits to get the final result.

So I do as you've done, try to get it in one shot, then cover or make changes , llisten, then add changes again, and try a few times. Finally getting the best introductory ones where the first 1 or 2 minutes is best, then go back in and do extensions on the rest of the stanza. Sometime, thats just whats needed just so I can go eat lunch sometime that day.
https://suno.com/s/tSCISFRa8EVodwJh

[polar spade]

Its my opinion too that the song started much too strong and gave the vocalist nohwere to go by the end. Changing up the arrangement, along with luck, I got that lower energy one, which I think works well when later we build to the chorus.

It also, as one of the outputs. gave me a slower one, about 5-10 bpms slower as more of a ballad piece which seemed to lose energy which maybe wasnt then for the best.....but at least I could see that changing the arrangement and not just prompting, at least, Suno reads the room and makes some changes to fit.