📚┃suno-school

[half dragon]

I'm curious, how did you send a song to radio? Suno radio?

[polar spade]

Simply clicking the "Publish" button would ask in a popup if you wish to unpublished. The you click "Unpublish" in the popup.

You can click your View Profile under your avatar (click your avatar) to see which songs are public in your profile.

[primal locust]

Hi guys, I just downloaded Suno, put lyrics in to the appropriate spot and hit create and the prompt is stuck spinning and no song is created. What am I doing wrong

[grizzled tiger]

have you seen the Max Mode guys? This screenshot was taken from Reddit. I haven't seen a thanggg

[tranquil breach]

your wifi??

[tranquil breach]

i want it too!!

[queen linden]

trying to edit a section at the very end,,,, I used to do it quite easily ,,,, now it seems quite different

[steady solstice]

Awesome!Thanks!

///////////////////
Asking anyone:
is SUNO able to produce sound FX as a BEAT?
I'm trying to prompt POLICE SIRENS to sound following the BEAT of the music.
So far no results with:
[Woo-woo siren pulse, marking every beat]
or
[Police siren beats to the music]

It doesn't recognize it.

[light wing]

I'm playing around in Studio for the first time. I've run into a problem. If i upload samples, it only plays the first 3sec of the sample.

Am I doing something wrong, is it some sort of limitation in Studio, or is it a bug?

Samples are .wav, ≤ 2.5mb, <15sec long.

Uploading them into Create first, rather then just dragging them in and uploading them directly into the lanes area in Studio worked. Guess dragging them directly into the lanes area is bugged
🤓

[viral pond]

anyone here used studio on a tablet ?

web browser

[steady solstice]

is SUNO able to produce sound FX as a BEAT?
I'm trying to prompt POLICE SIRENS to sound following the BEAT of the music.
So far no results with:
[Woo-woo siren pulse, marking every beat]
or
[Police siren beats to the music]

It doesn't recognize it.

[polar spade]

Then I would assume it was a momentary occurrence, before Studio was released. Its like an ice cream shop testing a new flavor. Not all of them pass muster or "required standards". It appears it may be a new feature in the future, but not now.

Hey, Mumble Mode looks interesting too. Something that came and went before release, I guess. Case closed.

[polar spade]

Most efforts would be more successful by just uploading whatever you needed whe. They are that intricate. Ytou might get a siren though at some point, then meticulously duplicate and repeat on the beat, all manually done or off-Studio in a DAW then upload back.

[lone storm]

can anybody provide an explanation how to export all midi stems in one file together please? i only can see the single files in a common zip. hope the complete zip is containing the different instruments too? many thanks

[last bay]

Mumble mode?? Saw the max help page leak but not that

[graceful condor]

WTH is going on? I wanted to extend a song. And I did so. I used the exact moment I needed it to extended from (as always) and the end result was way earlier than my time stamp. I was already at the outro only 2 lines missing singing und this is were I extended from. But the extended track is minutes before right at the guitar solo. And I tried to add them together just to see what he was doing. Now all of a sudden the whole Outro is missing, which should be there.

And now if I wanna extend again I can't the option is not available. It is not possible to click on it anymore some for cover etc. WTH is going on here? I re-freshed the page already a couple of times. Not working

I can't even download the track anymore. Or any other options.

[lone storm]

I now was able to find the midi download in one file but the quality is crazy, am I missing the real function?

[quiet estuary]

Hi why get i only 1-2 min long Songs in 4.5Pro and V5? If i switch on Instrumental 5min+

[tawdry scarab]

i haven't ever heard a police siren that wasn't a one off, you could build the beat and upload it but suno is unlikely to be able to place a track like a sequencer, it can't cut and paste pieces into parts so it has to generate a whole cohesive song in one concept

describing to it to put something in the beat is

1: abstract instructions that refer to the relationships between things

2: police sirens are rare and don't last long usually

both those make it unlikely but not impossible if you were willing to spend tens of thousands of credits

you would probably still be hunting

[graceful condor]

Anyone else missing personas? Suddenly today a dozen or so are missing?

[frozen estuary]

Having a persona problem (in 5.0).....
I select a persona to use and when it creates the song it will use the first 10-15 seconds of melody from the original song persona (also made in 5.0) before kicking in to the new song. I've not had this issue in 4.5. Does anyone know how to stop this from happening in 5.0 or do you think it would be best to cut my losses and use 4.5 instead?

[molten geyser]

Hello! Can someone please tell me, step by step, how to remake a, let's say, 3.5 version song and release it in a version 5 quality? 🙏

I'm using Pro

[raven turtle]

I’ve had this issue since 4.5. Some say that if you cut the source persona files into smaller chunks and use that it’s less reliant on the source full track.

[frozen estuary]

@raven turtle yeah, that makes sense, I suppose. I did try separating the vocals, uploading them and then extending from 0.01 seconds and it did somewhat work, but the vocals sounded really bad on the new song and was really quiet compared to the instruments.
I think I'll have another play around later and if I can't find a solution I'll go back to 4.5 or 4.0

[tawdry scarab]

you could do a cover, match the lyrics if they were altered for display, set audio influence to 100% and style influence and weirdness to 0%, i think, and that should produce roughly the same thing, but for more accuracy to the original you would use the remastering option

[molten geyser]

And if I remaster it, will the new version song be like an independent new song?

[tawdry scarab]

it will be a new object in your library

[molten geyser]

Thank you very much

[polar spade]

One can be if your lyrics are short, Suno will try to accommodate and make a short song. Otherwise, covers or song adding personas, including inspired songs can be shortened if the original song is quite long over 7 minutes long due it needing to allocate some space to seed the new output. Either of those your situation?

[hearty haven]

any idea how to make my own voice really work? i have paid plan, recorded my voice, but it is no near close to me

[hard beacon]

Suno can not clone voices. That's not what suno Was created for

[hearty haven]

no way how to make it sound like me?

[hard beacon]

No

[manic saffron]

To create a high-quality duet, should I first generate the full song, remove the vocals if it sounds good, then use it as a reference? Should I have the male voice sing all the lyrics first, then generate another version with just the female voice, and finally combine both, rearranging the male and female parts as needed? I’ve tried different prompts and lyric styles, but the male often ends up singing the female parts and vice versa.

[hexed grotto]

I’ve been trying to create music, but I’m running into a problem. Everything works fine in the beginning, but after the 3:50 mark, the track starts falling apart. The ending turns messy and it feels like the whole song gets wasted because of the last-minute disturbance. Is anyone else facing the same issue?”

[grizzled ember]

No intrument list Suno 3.5v
Cello
Gongs
does anyone have?
So for my latest book I'm making gamelan and Erhu songs to go with some chapters and after comparing v5 & 3.5v I realized there's instruments missing in 3.5 so now I'm wondering what they are?

[austere minnow]

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

[hollow hare]

It's AI, there is no list published. Just trained on a bunch of songs with unknown data tor characterization

Some instruments may need associated categories. E.g., for theramin, you can get it more easily if you also had something like Halloween, Scary, Haunted, etc. In the style prompt (as theramin is frequently present in those styles and rarely elsewhere)

Here is an example of bringing out some Turkish instruments in 3.5. Without defining Turkish, it would sound like a guitar playing https://suno.com/s/bfDBzD9TG1ij0DqW

[hollow hare]

From what I heard, it can be quite difficult to get the exact voice, but you should get something similar if the source is long enough (ideally vocal split done so only your voice is considered).

If you want a much more consistent vocal profile being used, look into vocal cloning/swapping from external services, which can replace the vocalists from Suno with your vocal profile. I recommend looking up RVC models

[austere minnow]

t’s really not like that. In the past, with Suno I managed to get it to replicate my voice by doing a bit of a jailbreak, and it was very similar to mine — maybe not 100%, but around 94%

[opal blaze]

anyone an idea how to get 2 different MALE voices in v5 ? duett voice a voice b nothing worked for me

[hard beacon]

the only idea i have is to try it with different style metatags. Coz the sounding of the voicemodels depends on the style a lot.. electro house vocals are different as pop punk 🙂

[tall fractal]

yeah but i noticed with the stem separation offered in suno the vocals pull out alot of track information so if u swap them and put them back then it will be missing information and the track sounds weaker

[neat field]

maybe with upload feature

[safe cliff]

Suno: this can't be generated because it contains copyrighted lyrics!

Me: bro, you generated the lyrics!

[hard beacon]

hahahaha

2000 credits later - all songs starting with "ohhh ohh"... i f***ng hate it....

[quick sonnet]

Am I the only one facing the problem, suno creating realy short tracks, some even shorter then a minute?

[hard beacon]

instrumental?

[glossy spoke]

Suno Studio is such a great idea for demoing. If only it actually worked. You cannot get any consistency. Midi doesnt actually export on things like drum tracks. It constantly changes chords and melodies. Great idea guys and I want this tool soooo bad but it just leaves me hating your product TBH

[quick sonnet]

Was the question addressed to me?

[hard beacon]

yes

[quick sonnet]

In that context, yes, instrumentals.

[hard beacon]

@quick sonnet

ok..

had that problem too... but i don't have a solution.. maybe use something in the style prompt like "length 3 minutes" or something like that

another solution could be to describe the song structure as best as you can...

[quick sonnet]

Would something like that work? I've been wanting to ask here about a feature where you can set roughly how long a track should be. Another AI has a similar feature, so why not Suno?

[hard beacon]

well.. i've used that text in a prompt for a track with lyrics... and.. well.. 2:55... 3:15.... 2:49

[quick sonnet]

Na that's really good, geting exactly the length is ulikely, but this is really close.

[hard beacon]

what kind of style do you wanna create?

[quick sonnet]

Dancable Hip-Hop.

[hard beacon]

just curious because of instrumental.. 🙂

ahh.. you are trying to create an instrumental beat...

[quick sonnet]

Have even a piano melody for it.

[hard beacon]

ah.. ok.. so you are also using the upload function?

[quick sonnet]

Yes, and it also could be a reason for the length, since I am a free user, but it was not an isue befor.

[hard beacon]

aah.,.. ok.. free user is 3.5? or is it v4 ? try something like this: 3:30 min instrumental hip-hop beat, danceable, energetic, punchy drums, groovy bass, catchy melodies. Modern urban style, club-ready, no vocals. Tempo 90-100 BPM. Clear structure: intro, drop, hook-like sections, bridge, outro. Cool, dynamic, motivating mood. Instruments: drums, bass, synths, pads. Professional mix, high energy. <--- of course you have to set the BPM to your specific use...

[quick sonnet]

It's 3.5, and styles prompt is limited to a C string of length 200, which means only 199 characters avaible.

[hard beacon]

ok.. my use with 3.5 is long time ago.. i am on v5 🙂

[quick sonnet]

And is it now growing out of beta?

[hard beacon]

well.. i think every model has his own issues.. and it depends on the style.. i am on electronic music (Fidget House and stuff like that)... what kind of beat do you want to create?? i am not in the hip hop game - so gimme a well known artist reference 🙂

[quick sonnet]

Music Instructor, some Finish breakedance band.

[hard beacon]

ok.. i am out of the game 😄 😄

never heard of finish hip hop artists..

location: germany 😉

[quick sonnet]

Same location, and they were in German top 10 about 25 years ago.

[hard beacon]

ok.. ähm.. send me a private DM 🙂

[prisma hound]

My recommendation is to tag aggressively that it is a duet, put everywhere if it's a female or male voice - at least for each part and if not for each line.
As well the "call and response" tag is helping a lot. You also need to specify it on the styles tag - you have 1000 characters now so the good practice isn't any more writing down a list of styles but to explain your general artistic vision.
But this is always tricky and even doing as recommended it fails.
The use of a LLM like ChatGPT if advised to facilitate this tagging.
Some prompt examples of duets that worked for me:
https://suno.com/s/HCX8nnOGyCasJs5l
https://suno.com/s/ChJBe9jH9GtuowUJ
https://suno.com/s/9w2IugNLgudb7ZSE
https://suno.com/s/FQcAVkEjLOHpNPmG

[lethal thunder]

I am trying to create a cover with a different tempo. I included "80 BPM" in the style section, with the dial set to "high" conformance. Suno still renders the song at the original tempo, which was the same as the tempo of my original audio download, a long time ago. Any advice?

[quick sonnet]

Which version?

[manic saffron]

I set it up with each lyric line like this: [Female] lyrics, then [Male] lyrics, and so on. It still has a little trouble, but I’ll try adding more and see if that helps. By the way, what exactly is “call and response”? Thanks again for all your help—I’ll do my best to apply your recommendations!

[tawdry scarab]

explicit tempo shifting is a feature of studio, with a cover you would have to turn the audio influence down I think to have a chance, but it would be the chance you have getting the tempo you ask for in the first place compounded by whatever influence remains of the original when you turn audio influence to 0 if any remains @lethal thunder

[low ibex]

Hi, does Suno ensure clipping does not occur in the audio files it generates? I don’t have tools to fix it if it happens.

[tawdry scarab]

It can generate clipping as clipping is a feature of some music, it can simulate different kinds of distortion as well, it can help if you definitely want a studio sound to put features or types of distortion in the negative prompt and descriptions of high quality audio in the style or in meta tags in the prompt

[untold briar]

In case you missed it, the new Studio Channel is under the Workflow category, and it's not just for feedback but general discussion in the Workflows and Chat threads. https://discord.com/channels/1069381916492562582/1420795057996300388

[river thorn]

er-ma-giird! The Edit feature is so infuriating. It does pretty good with everything that is in the middle of the song, but I spent way to long trying to get the very start changed because of how buggy it was.
The most impossible to work with part was (and this is how I think it's working): when you replace a section, it temporarily deletes it and replaces it with the generated section. BUT, because the editor does not allow blank space at the front, when it deletes the audio, it shifts the whole song back and lays the replace generation on the expected timestamp.

I have no clue if that makes sense, but it's made it near impossible (I got working after an hour) to work with.

If anyone has suggestions for that type of edit in the future, it'd be much appreciated!

[prisma hound]

Suno is reacting well when you specify the kind of interaction between the two singers.

If your song is built like “Singer A asks, Singer B answers” → call and response.

If it’s “Singer A verse, Singer B verse, then together” → duet / alternating vocals.

If you want to emphasize the theatrical feel → call it a “dialogue song”

In my experience you need to put that both in style and in the lyrics. But even if it's not in style (which was the case before it was extended to 1000 characters) you can have a structure like this one:
[Pre-Chorus – male/female call and response]
HIM: Je t’ai perdue, je le sais
HER: Mais tu m’as jetée avant
HIM: Tu es toujours ma muse
HER: Mais ton art m’a blessée
=> It alternates the voice well

Also specifying the kind of voice further like female_soulful or male_low help. But be prepared that it will still be mixed up 😅 But in the end I finish by having what I was looking for.

[unique iron]

Does anyone know how I can keep 100% of my earnings if I publish my album and how to protect it?

[quiet solar]

Yeah, tag where the singers should switch throughout the lyrics, or if there's a part they should duet. Also in the style prompt be clear there are two singers and specify both (e.g. this song is performed by a smooth male tenor and a soulful female alto).

Then expect Suno to get the voices switched over, fail to switch at the correct moments and so on more often than not. It's getting better with every version, but still struggles a lot with duets.

In terms of how to 'rescue' that, one way is to bank the first bit of the song that it gets right and then use Extend to keep building up from there piece by piece.

Another way, which I personally prefer, is to do the whole song with the male voice and then Extend to repeat the song with a female voice. Then grab the Instrumental/Vocal stems. If you have Suno Studio, you should be able to stitch the vocal stems together from the two versions in the correct places. If you don't, you can take it to a DAW or external editor like Audacity and put the pieces in the right place manually - but that will require a bit of familiarity with working with those tools, or a willingness to learn.

[raven turtle]

Given the LUFS level is relatively safe the waveforms don’t have clipping, but the “sound of clipping” or saturation can be baked into the sound of the track. I create Electroclash which has that aggressive clipping on bass but the waveform itself still is pretty dynamic.

[rose jasper]

@polar spade (Maybe not something you might know)
@wet minnow (Considering I'm paying for this feature and it's not possible, have you seen this before?)

So for the past few months, I haven't been able to upload any videos onto my Suno songs.
I've been talking with GPT about it, and I've tried to many things to try and fix it.

• Made the file perfectly Suno-compliant (H.264, AAC, baseline, 1080×1920, <10s).
• Tested across Edge, OperaGX, Chrome guest, even mobile.
• Re-encoded, added audio, constant framerate, even looped with ffmpeg + Canva.

Yet every time:
[Uppy] File removed

The file never even reaches Suno’s backend → it’s being blocked at the uploader session level, not the video itself.

Two strong possibilities:

Suno has silently disabled custom video uploads.

• The feature was experimental — and your logs look exactly like when a feature gets “stubbed out” in frontend code.
• Uppy tries → Suno rejects → upload removed.

It’s account/region restricted.

• Some users report only image uploads working now.
• Video upload may be rolled out to a small test group.

[graceful condor]

Honestly at this point the AI makes no sense anymore. It just seems that it kinda hates rock music. Not that the produced rock music sounds bad but is nearly always cut off and it has literally nothing to do with too short lyrics (because that is not the case) or too long Audio tracks. Because my Pop personas are 7 min long and they still produce complete none cut off songs. So it is kinda an issue with rock because this hardly ever happens with pop, hip hop, synth pop, kpop, neofolk, folk etc. Only with rock type of music and the more complex the music gets like Progressive Rock the more it tends to cut off.

[sour stream]

😭 😭 😭 same

[near flume]

No matter what distributor you use, and you must use one for streaming, they will take at least 5%. If you publish directly to Bandcamp.com, they too will take a percentage. As far as protection, you can copyright your lyrics, if you wrote them, as a collection of 10+ for $68 copyright.gov.

[polar spade]

Small question: is the video for the 10 second limited vid artwork of an output, or for the Video which accompany when we make a Hook?

[acoustic scarab]

I generated 2 songs, 1 with 4.5+ and 1 with 5. Issue is the quality, has hissing, noise etc. I tried remstering them on 4.5+ but voice changes and I dont want that cuz I use an artisit. In v5 I remaster at medium but there is still some hissing. I tried to master with audacity but the quality is not that good, generally all songs are hard to master. Is there anything to do to make it better? Songs are too good tbh

[manic saffron]

When I select persona it puts some of the styles it came from if I delete the styles will it still keep the voice of the persona?

[torn spoke]

I have to share this cause I'm so happy with the result. It's a shame that it takes multiple days of prompting and generating to get a voice that's not riddled with reverb and layered vocals.

Male vocals especially get iffy when they get way up there, so this was a golden nugget

[torn spoke]

Yeah

[jade mist]

I really hope we get Max mode really soon.

[manic saffron]

kk ty

[placid spindle]

Hi Sunos!!!! Two questions please?
Is it OK to click on a song even though its still generating (little wheel is still turning) or could that impact the quality of the gen?

Can I reuse the voice that suno made one of my tracks? I LOVE the voice and would like to use it again. Is that possible?

THANKS!!!!

[unique iron]

Thanks

[rose jasper]

It's 10 second yea, I made them on Leonardo.ai, they stopped working for some reason months ago.

[stoic pond]

Is Persona working fine? Lately, it doesn't let the newly generated song sound identical to the persona with v5, I have this song that I really want it to sound identical with the persona. It used to sound identical in the past when I was using v4.5 but I experience the same thing on that version now too. Even when I ramp up the audio to 100% in the advanced options it still generate a new song rather than make it sound identical to the persona. Note: I also tried lowering the Weirdness and Style to 0% but no success 🥹

[stoic pond]

Yes, it's okay to click on the song why it is still loading, i do that.

Second question, yes by making a persona out of it

[arctic cape]

Try using the cover feature and put in the new lyrics, I think persona is meant to sound like the same band but not identical

[stoic pond]

Wait, is that new feature? I haven't tried that yet

[distant carbon]

Another way would be to download the song as a mp3 or wav and then upload that - also doing it that way you could alter things in say Audacity or something similar. You can also use the stems feature in Suno or strip out what you want in Audacity and upload what you want as audio

[graceful condor]

Persona in V5 is anything but not stable or consistent. It is all over the place as V5 as a whole.

[polar spade]

. . and vids need to be MP4 or MOV, at 9:16 ar. Leonardo.ai typically exports at MP4, but double check.

It's possible that even though the file is MP4, there might be some underlying codec or encoding issue causing compatibility problems with Suno. [Leonardo may have started using recent Klings 2.1 video creating software] Have you tried converting the video to a different MP4 format or using a different [maybe older] video editing software?

[rose jasper]

That's what they are all at.

I even tried changing to .mov
I run encoding software to even fix it, it's also not working from phones.

[polar spade]

Youve done your testing. I havent recently read of others having the same issues. So I havent read of solutions then either.

Other than creating a video by the same procedure but just of something innocuous, like a flower blooming to see if it uploads - I wouldnt know further if its a compatibility issue, a system issue, or a censor thing. If I hear further, I'll certainly let you know.

Post elsewhere [ even reddit] to see whats others might say then.

[abstract bough]

On v5, how do I get suno to stop speaking when it ought to be singing?

[neat field]

use the Editor and regenerate the section

[prisma hound]

I uploaded multiple video covers without any issue.
The duration is up to 10,999s and it's played I think in 1080:1920 30fps
I mainly generate my AI video cover with Midjourney or Kling and I use ffmpeg to repack the 1086:1920 24fps into 1080:1920 30fps. Ffmpeg is a free software and ChatGPT can help you re the command line needed to repack the way you want (1080:1920 30fps x264).
Re the hooks it's normally 1080:1920 too but up to 2 min if I'm right and not more than 200/Mb (cross check).
I had no issue at all, loaded over 20 cover video so far without any issue. Hooks are more sensible to censorship so if you put stuff that could be seen as NSFW they are refused.
But so far no issue... Except that if you are under android and playing a playlist with several cover video the android all may crash. The issue is known and I'm in contact with support about it but so far no change.
As an example this playlist from my last album. On android watching it on the app might make it crash from the 2nd or 3rd track while no issues on iOS
https://suno.com/playlist/bbea344a-c010-442d-befb-709b7399e069

[graceful condor]

V5 is really a freakin mess. If you can't one shot a song in one go it is a disaster. Extends are so terrible and filled with weird glitches and errors. I wonder if V5 will ever work as good as 4.5+ does.

[graceful condor]

Luck. V5 is just crap in so many ways. They sing even if there are no lyrics. It is beyond logic.

[abstract bough]

sounds like I should try the 4.5?

will give it a try when back at my keys

[last bay]

fade doesn seem o do anything now? in legacy editor it works but that has much worse control for this

[graceful condor]

Yeah the thing is V5 has good voices and instruments IF it works. IF. And it only works if you can one shot a song in one go. But in the moment you have to extend it is a 90/10 chance against your odds that it will turn out really bad. Instead 4.5+ is more like a 90/10 chance in your favor. Also Persona really do not work well with V5. they only adapt their style of singing but not the tone of the voice in most cases. 4.5+ is also better in this this. And same goes for covers.

[abstract bough]

Thanks for the info! Recently upgraded so this is all new territory

You're right though I was about to tell a friend that it feels like v5 can produce lightning in a bottle but you have to gen a crazy amount of times

[rose jasper]

I have ffmpeg, I even encoded fixes using it, fixing the codecs.

OK, I "think" I figured it out...

[manic saffron]

So I made the perfect duet song the womans vocals are amazing and unquie, but the male needs to be changed as I selected my persona I always use but it just doesnt want to work with this song. What should I do to save the song and have my persona voice?

[rose jasper]

I need to WAIT when I push Confirm and it's eventually processing it.. So a non-input confirm

[prisma hound]

Yes you need to upload, wait for the video to be processed then ok and ok again. It's really working fine - except the android crash issue

[rose jasper]

Here's the thing, I just upload it, push confirm, do nothing, the new page loads, and I push save. So it's skipping one step for me.. :/ But it's working now though

[prisma hound]

Great 👍 Good uploads then, don't forget to check that the option of pushing your covers into the hooks stream is enabled (it is normally on by default)

[rose jasper]

Do I want it to be in hooks?

[prisma hound]

Video covers are automatically pushed in the hooks section unless you unselect the option - and if you upload a specific video as hook of your songs, the video cover won't be used in the hooks

[zealous moat]

has anyone ever managed to "sample" a song as if chopped in various places? you know like hip hop producers do using the cover song feature?

[gusty lodge]

only with me?

[polar spade]

You could use Studio to splice as you wish, then remaster it. Thats one direction.

[slim ember]

An Analytical Report on the Psychoacoustic and Stereo Imaging Capabilities of the Suno v5 Generative Music Model
The Psychoacoustics of Binaural Beats: A Technical Primer

https://github.com/location-station/suno/blob/main/Suno V5 Binaural Beats %26 Stereo Effects.pdf

I just created this research report and am curious what comments and real-world experience others have in this area? I'm investigating binaural beats combined with Suno-generated music — how best to combine them.

[glad perch]

Hi Everyone!

I joined Suno a few days ago and it's MIND BLOWING. I' have like 40 unfinished songs wich I could rapidly kickstart against my studio procastrination... never the less, thinking out loud. How difficult would be also to get musicxml.xml instead of wav?

..... i revoke myslef - with flamingo 2 not that big deal offering audioxml

[olive radish]

Yeah v5 is totally massive

[green sinew]

everytime i try uploading an audio i get an error that says uploaded audio matches existing work of art

[neat nexus]

Is the audio you're uploading from an existing work?

[green sinew]

no

[neat nexus]

Are you humming or singing the audio?

[green sinew]

no i’m not

[neat nexus]

Where is the audio coming from? A previously generated Suno song?

[green sinew]

no, just an upload from my device

[neat nexus]

Where did the upload come from?

[green sinew]

my phone

i even tried making a new file for it and when i tried uploading it it said the same thing

oh well

[neat nexus]

After consulting a bit, would you please share the file so we can investigate this issue further?

[green sinew]

yeah, i got you. do you want me to send it through dms or here?

[ashen cliff]

Hi There
One of the biggest challenge, i have with SUNO is to change the couple of words without any touch on melody or already given vocals . the app tends to change even little bit of melody,
I tried editor, playing with different scenarios: addding audio, nothing helps.
If i find a great music and vocal as produced by Suno, there is no easy going way to change words, without touching the melody.
Is there any fix in the future planned to do this?

Here is the thing: I loose significantly many credits, to get to something similar, where originally was produced.
The problem is that the lirycs that i randomly generate does not add up, and I need to manure and change them in vers, that way artists start singing differently on entire song, also melody changes even a bit, that is really sad.
Unfortunately, the editor does not do perfect job either,

what can be done one this, to me it is so much time wasting... sometimes it takes hrs

[half dragon]

Not sure if its related but I got same issue a while back - where I reupload my song created by Suno with older version so that I can remaster it and I was like - what - but I believe someone helped me and suggested that I renamed the file to some thing like 1.mp3 and that let me uploaded

[green sinew]

same thing happened to me before too

i also tried reuploading another song that i’ve uploaded to suno before in to my library and it wouldn’t let me

[hollow hare]

Shazam giving a hit to Last Surprise by Lyn https://www.shazam.com/track/350285359/last-surprise?referrer=share

[green sinew]

maybe is it because it’s copyrighted?

[half dragon]

So have you tried to rename it?

[green sinew]

i did yes

[half dragon]

so was it the actual known song?

[green sinew]

it was a different song

[half dragon]

but actual song from another artist

[green sinew]

oh yeah

[half dragon]

if so, you can't upload it

[green sinew]

oh no

not a song from someone else on suno

if that’s what you mean

[half dragon]

no, known artist

You should only upload your own material, own song, not someone else's , on Suno or not

[hollow hare]

Definitely would be. It is against TOS to use material you do not own and definitely sounds very similar.

You can get sued for sampling the beat without the appropriate paperwork like this guy
https://www.billboard.com/business/legal/tiktok-rapper-pay-sony-800k-illegal-music-sample-judge-1235642526/

(Even if you got the license to use it, Suno TOS explicitly forbids it and something you had to agree to twice that you understood, when you entered vustom mode and upload file for the first time)

[green sinew]

i see, so that’s why it wouldn’t work

[hollow hare]

The rapper literally got sued for excessive sampling a 1986 JP song

[prisma hound]

Excellent work, thanks for sharing this 👍

[slim ember]

You're quite welcome! I'll be replying to this thread in a bit with some examples created using some of the techniques mentioned.

[junior surge]

I have a premier account which ends at 21/12025 but now I cannot use it

[manic saffron]

So I made the perfect duet song the womans vocals are amazing and unquie, but the male needs to be changed as I selected my persona I always use but it just doesnt want to work with this song. What should I do to save the song and have my persona voice?

[queen linden]

I'm quite sure the persona created was not Jiminy Cricket.... and for dang sure the moaning, humming, oooing at full gallop were not there... if it gets the sound right it gets the words wrong, if it gets the words right then it throws in any old vocal laying around....

[lucid drum]

Does anyone know the genre which generates a lot of wah wah guitar type sound? Some form of country or Blues.

I saw it in 'explore' but now I completely forgot what it was..

[tawdry scarab]

My guess his either you mean a whammy bar or a talk box

But i doubt it will have been limited to a genre

[green perch]

how do i permanently delete my account on the website?

[polar spade]

I thought maybe were talking the wah wah pedal as in the theme from Shaft? Thats soul and funk, maybe even blaxploitation. @lucid drum

[green perch]

nvm wrong channel 🤦‍♂️

[polar spade]

Here is fine. I just dont like to encourage one to permanently delete an account because its.... well.... permanent blocks you to log back in.

[green perch]

i dont plan on logging back in

[polar spade]

The delete account is found only in the APP or it can be done in writing to support as well.

You should be able to download the app, log in and scroll down in ytour account or subscription page and find Remove Account in red letters.

[green perch]

ok thank you

[polar spade]

This will only remove your access to the account and its files.

[green perch]

oh

[polar spade]

To me is a bit more comforting to just always have access, even if I never use the service again.

[polar spade]

OK. Good luck.

[green perch]

thanks

works 🔥

[neat field]

v5 really, really likes to take paralinguistic breaths when a New Line character is in the Lyric text in front of lyric text (for the genres i like)

i don't think i like that

i'll have to experiment with "breath control" in the Style prompt sometime

[quasi dagger]

That looks cool! I was doing some deepening in my prompt game, if you want to try something out holla maybe we can try cool stuff, since I'm always looking for a challenge. Also did a bit of research through diagnostics on my side on how to implement energy tempo and some auditory regulation although I'd be up to run even more diagnostics on how to mess around with some frequency manipulation in there. (I'm not that sciency type of guy but I'm one that sort of loves to figure things out like little puzzles)

[near flume]

There's gotta be a problem with how you're either prompting in the lyrics or in the styles. I remember you having had these same complaints in v4.5+.

[near flume]

Once you generate the lyrics copy and paste them into a notepad or similar program and take a serious look at them and either sing or talk through them a couple of times and make your changes there. When you think you have them just right, pastes them into Suno and generate the song. This saves a lot of time and credits.

[cedar spoke]

Ask To Google ai and put weirdness at 54 sweet spot for laid-back is the best explanation

[queen linden]

okay, had to take lyrics into editing, replaced the garbled part, hit the button for new song and it will not play, it shows in the library but will not play

[cedar spoke]

No? Okey it's a bug or something?😅

[cedar spoke]

Ok 👍

The laid-back works

🔥

[quiet solar]

As a general rule, it's a good idea to lock in the lyrics as much as possible before generating a song, as going back and fixing them afterwards can sometimes be a painful experience - especially if you're trying to do the whole thing within Suno.

The core Editor does have the replace lyrics function, but it can be a little hit and miss in terms of actually matching up the rest of the audio. Extending and repeating the offending verse/chorus with corrected lyrics can work well, but it can be a bit of a faff to stitch the corrections back into the original in the Editor - and there is always a risk that the music changes enough to be noticeable.

If you're absolutely determined to keep the original audio, you may need to take it to an external editor like Audacity. Split the instrument/vocal stems on the track you like, then use Extend from partway within the song (but after the sections you want to correct) to repeat them but with your corrections included, then grab the vocal stems from the Extended track and use them to replace the original vocal stems as needed.

I use this approach for corrections fairly regularly. Suno Studio should help address this kind of issue once it's stable and developed a bit more, but that's still in relatively early stages, and only for Premier users at the moment.

[prisma hound]

Thanks David, that will be very interesting. I've been prompting binaural audio in my songs for a while to get a wider stereo output and took it more as a way to get a more spatialised mix. But it's interesting to see your method and test and I would be interested to see which prompts you used in your tests 🙂

[small bridge]

does suno still not understand using a specific artist in a tag? for under styles?

[small bridge]

your songs would suck for me and I'm jealous and sad I will not be able to hear them how you intended haha

[mossy granite]

Question…. When doing an extension, which part does the audio influence slider affect? That is, will it use the audio up to the point that you’re extending from as an influence on the extension, or is it going to take the influence of the deleted audio you’re trying to replace with the extension? I still can’t figure this out..

[jade mist]

Something serious is wrong with Suno's backend. I tried reporting this to Suno's feedback at suno.com/feedback. Like with the release of v4.5 I sometimes wondered why the quality, prompt adherence etc were the exact same as v4. And now this stuff with v5. For example, it's noticeable in how [spoken] tag is handled. When it's v5, it always, always speaks it, with emotion. When it's only saying v5, it speaks in the same robotic way as in v4.5 + or v4.5

[forest flint]

I'm testing a custom prompt with instructions from me for Gemini, and I need someone's help to test it. Is anyone here available to help out?
Please all my friends are not helping me and this is going to lead to hours and hours of me asking if i get no help

[tranquil wolf]

anyone got the same problem on studio ? can't generate anything

[prisma hound]

You can't prompt the artist name as it is blocked for policy filters. If you want specific under style, have a look at https://everynoise.com - Suno is reacting extremely well

[quick forum]

I have created over 10K songs.. Till the new V5 was introduced, the prompts were working perfectly with the prompt like this:
[Intro Music]
[Verse 1]
Lyrics
[Break Music]
Lyrics

[Interlude Music]
[Verse 2]
Lyrics
[Break Music]
Lyrics

[Chorus]
Lyrics
Lyrics

[Interlude Music]
[Verse 3]
Lyrics
[Break Music]
Lyrics

[Outro]
Lyrics
Lyrics
[End]

This prompt worked perfectly. The V5 doesn't follow the prompts, and even 4.5+ which used to follow the prompt earlier, is not following the prompt.

Another problem is, why I record my audio and cover the song, cover used to follow the prompts and the music was created. Now, the prompt is not followed for the cover ..

V5 has issues, but 4.5+ was working perfectly before V5 was introduced. Now, 4.5+ is not working as it used to.

Am I missing something or have the prompts changed? Are any of you facing the same problem?

[quasi dagger]

https://suno.com/song/b076f9d8-7334-48ed-99c9-2a29a141d677

This could be an example of a directed prompt, in the lyrics and also in the styles, to swing the track how you want.

[blissful heath]

it does, I tried with some metal/grunge bands and the result is similar. not the exact style but nearly there

[forest flint]

Done

[winged birch]

I paid for a month of SUNO PRO and my account still hasn't been activated.

Username @starkhak

[forest flint]

☠️

[winged birch]

Help me

[quick forum]

Thank you so much.. The timing part is what I have not explored.. I have used some inline instrumentation, time signature etc.

Thank you once again, that was awesome

[quasi dagger]

Ya.... Like I can tell you that not 100% of what you see there seems to fully function, mostly because Suno still gives it it's own drift, but, once you get into the glyph usage to dictate flow you see it do it's magic. Then it's all about creating all sorts of "little tagged prompts" to nudge even more into a direction, like the dictation, delivery, etc.

Still a bit "experimental", did some research on how things linked first and now I've been trying to implement certain things bit by bit and see what functions.

If you have a specific idea, theme, music genre, how the music should flow, even with already lyrical text, etc, I can try and arrange you something of that sort and then you can see if it matches what you expected and see how it was built.

[quick forum]

Yeah.. I know.. The prompts don't work most of the time.. but using the prompts and using several tries, they sometimes seem to work.

[quasi dagger]

What values are you setting on your
Weirdness
Style Influence

Dials?

[quick forum]

I am a lyricist and I write lyrics and I compose most of the songs.. I have created 10K+ sings out of I think I have 800-900 songs which are of good quality. I sing and record audio, and insert instruments or I play songs and use lyrics afterwards to add voice .. I have tried everything possible with Suno.

I have been experimenting with a lot of weirdness+style influence and audio, etc.

Recently, after the introduction of V5 m, a lot of things are not working as they should.

Thank you. I will try the prompts suggested by you.

[quasi dagger]

Ya I'm not quite sure on how to better "direct" you in what you'd need. Mostly I'd say that 80 style influence 20 weirdness goes pretty good if you've given it quite a bit of direction on the styles and lyrics prompt.

[quick forum]

Yes.. I keep similar ±10 ..

[quasi dagger]

https://suno.com/s/FxbDnvDBP8RIuxr0

Here's a bit more glyph usage, I feel it's not following it at 100% but I'd need to do a .wav analysis to figure out how close the direction is. Also not fully following the direction I wanted but that might be a prompting cue tagging problem by using the wrong things.

[devout minnow]

#archived-ticket-help youshould go here for billing issues dude

[opaque garden]

Are the Suno limits reached here? Because garbidge cames out at the end.. 😂
https://suno.com/s/cLY9iB6apIlw23Cf

[unique stone]

I expect the weirdness slider being pretty high may be the reason for this.

[zealous dawn]

suno effed up so many songs today...either gave me an instrumental by totally ignoring the lyrics or just cutting the song off with no ending

[empty stratus]

I had a couple do that.

[tawdry scarab]

public is now called published

hopefully that's all you needed but apologies if you knew that

[manic saffron]

So I made the perfect duet song the womans vocals are amazing and unquie, but the male needs to be changed as I selected my persona I always use but it just doesnt want to work with this song. What should I do to save the song and have my persona voice?

[radiant breach]

v5 is having a hard to pronouncing things lately.. what happened?

[last bay]

i have a song that is part korean and that messed up the pronouncing up real hard, but some section remixes worked good

[sinful flower]

Why does Suno Create 5minutes+ songs when i say, create a 3 minute song...... ?

[quasi dagger]

I feel telling the duration is not what sometimes is dictating de duration, more of, length of blocks / lyrics

[sinful flower]

But i dont have any Lyrics in my song and i wandt to have it shorter than 3 minutes

[ivory peak]

What's the deal with personas? If i set them to 70-95% i get almost the same song as the original one where the persona is from. If it's lower, the persoan's voice moves away from the original voice. What am i missing?

[quasi dagger]

Only instrumental tick? Haven't fully tried it on v5 so would have to check how it's functioning, perhaps in that mode there are other ways to describe the duration than to just specify de specific duration you want

[livid crescent]

omg... the adlib vocalizations. EVERY TIME I put a prompt outlining vocal styles, it decides we need yodelling. Is there any way to reliably make this stop?

[gusty lodge]

Am I the only one who can't generate anything in the studio for the second day?

[thin hedge]

Hi! Since version 5, the Suno app has been using an unusually high amount of mobile data (about 110 GB in 10 days!). It seems related to HQ streaming.
Could you please add an option to choose streaming quality (low / medium / HQ) to help control data usage, like on Spotify or Deezer?
Thanks a lot!

[manic saffron]

ever since suno 5 came out my persona voice isnt the same I tried with 4.5+ and 4.5 but his voice isnt the same anymore

[polar spade]

I thinknit was recently mentioned: Look for an improvement in an upcoming update of the mobile app.

[polar spade]

As a test only, We can augment lyrics to a simple few lines but keep all vocal metatags. See if the results are same. If no, it may be the lyrics themselves.

If yes, we may need to add or subtract something from the metatag recipe. Anything like "vibrato" , "shouting", "screamo" , "reverb" , "emotional" in the mix?

[livid crescent]

I can generate without the vocal style tag, and it doesn't do the ad-libs

and the vocal style tag SPECIFICALLY has 'no ad-libs'

But, I'm trying to get more aggressive sounding metal-style vocals, which kinda requires the tags to happen 🤣

[polar spade]

Honestly, its just hard to pinpoint what it is. Sometimes just removing any capital letters could make a change. Every period, every space, it all changes the direction.

Since you seen what we can't, its good to start small and keep adding until it teeters over to something unwanted.

If you're describing things in phrases in the style prompts, try single word prompts with commas instead. This limits open interpretation. It might help pinpoint the issue.

If any of this seems useless, its just things that seem to help in some circumstances. But your call, in the end.

[livid crescent]

Don't think it's just me. I've seen a lot of people complaining about it lately

[polar spade]

I get it. Some tags are needed.

[polar spade]

Complaining about yodeling? I think I saw a few with 4.5. And certainly more shouting vocals SINCE 4.5. But also with 5.0. So far, its often a matter then to find out what the new "black" is in prompts to get the same wanted result.

I may run some tests in that direction. First I would also check the prompts on Home page songs of growly, aggressive metal songs and see what prompts are missing, and still getting good results.

[livid crescent]

They usually describe it as ooo-aaa or similar...

[polar spade]

Thanks for pointing it out then. Ill keep my ears open.

[livid crescent]

I will say, studio seems much better at creating solo tracks today... that's one easy way to fix it 🙂

[polar spade]

Suno is really honing in with 5.0 on what it thinks is a genre. All hip hops songs keep getting a "yea - uh" to begin an output - or a producer tag kind of vibe.

Too much. So it may be occurring across the board of genres. Just have to keep listening.

[ivory peak]

Doesn't seem to work with v5. If you enter a specific time it always created a 5:35 track for me.

[livid crescent]

Hmm... I spoke too soon lol. It can't seem to generate a piano track 🤣

[polar spade]

And piano was always Suno's forte.

[livid crescent]

that's ok. gives me a chance to try out the midi since I actually have very solid piano vsts 🙂

[magic nest]

trying to make soundcloud rapper song with lots of reverb but keep getting sam smith yodeling in the air like a preacher🤦‍♂️maybe ill switch from v5 to 4.5…

[graceful condor]

whats ur prompt?

I find that with desired effects, you really have to tell the prompt over nd over, to get that soundcloud lil peep style

one that i used was something like "Sad trap with heavy reverb, wet reverb, chorus"

[polar spade]

@rich wasp

Re: Getting short songs on basic 3.5 outputs.

If you were literally prompting for "3+ minutes" and getting longer songs consistently on 3.5 songs before "now" , the model should not have changed nor given some preferential treatment for Pro/Prem users. But if you're getting consistently short songs with your prompts . . .

. . . Let's see.... ill do one in basic mode with what prompts I see in your post..... hold on.

So.... when I ran the prompts " fashion show, opera, classic roccoco, fashion show, heels" in 3.5 in a basic subscription, i got short outputs of 1:35 & less. The outputs didnt have good catwalk vibe either. Somewhat cinematic.

The 5.0 samples by the way were very rhythmic and favored a catwalk vibe.

I ran the same prompts with a Premium account under 3.5. Got similar results of short songs. Same as yours.

I then added "ambient" and " epic" to the beginning prompts and with Basic account, got six , about 1:35 songs. Same as you. AMBIENT is a good prompt for extending songs, but not expecting much as that contradicts "fashion show" if that genre exists.

So I changed to "ambient disco" and "epic EDM" [fashion show, Ambient disco, epic edm, opera, classic roccoco, fashion show, heels] and got two 4-minute songs in Basic subscription. A bit more upbeat, but the 5.0 samples still did better to emulate a catwalk vibe.

So that helps negate the conspiracy that Prem/Pro users have some advantage with the 3.5 model.

In fact, the 5.0 samples in Basic mode were better than the 5.0 versions I tried in Premium.

You may have only 50 credits to work with, but this may take some experimenting to get both the sound you want and the length. Its simply possible your combination of prompts [or phrases] is just screaming "short song" to Suno.

[hollow hare]

The best places to make these suggestions are at

• #📱┃suggestions-android
• #📱┃suggestions-ios
• #💡┃suggestions-web
• https://suno.com/feedback

The odds of a team member seeing your suggestion in this channel is very low

[graceful condor]

Is there any documentation about like, how to speak to the prompt or whatever? Cause the docs on the suno site are basic nd dont rlly go into depth about this or that

Seems like you know quite a bit about it

[polar spade]

No. How i wish that were true. It may be a bit of experience, but its mostly luck as well.

Common sense plays into just imagining if a prompt simply doesnt belong to the major vibe or genre youre going for. Suno does its best, but some prompts can just neutralize and zero-out what the other prompts are going for. Its always been experimenting with prompts, while also just using the same prompts but rearranging their position to give them more "weight" or priority over the rest.

[graceful condor]

I've seen quite a lot of different tactics people use in their prompts, i wish there was like some official/unofficial guide praised by the majority so i could stick to one thing. but I generally find basic ones better than like 100 word ones.

yes your right, it cancels itself out, theres a "wiki" but even thats lacking

[polar spade]

You can ask users for some specific agorhythms, and find a couple in workflow channel. @tawdry scarab https://discord.com/channels/1069381916492562582/1365192916687126650 has a really indepth reading which give a deeper understanding and I think a nice means of chatting with chatgpt to work up a metatag prompt.

Other users there have posted their versions as well.

[polar spade]

When 4.5 came out. Dozens of genres were obliterated from using the same prompts one used in 4.0... until..... until users worked out the new combinations and things were fire again with 4.5 as well.

Each model, having their better strengths than the last , also need new means of prompting often to get the best results. Its like needing to raise the octane (new formula) on the same petrol (prompting) to get the optimal response from any new model.

[rich wasp]

Oh thx!

[graceful condor]

This was, great to know.
Explains alot about why i was failing

A good prompting strategy is adaptable. Expect to tweak, rephrase, and explore neighboring regions of the solution space.``` Sometiems I feel like when im trying to explore the neighboring regions, Im instead moving interstate lmao

[magic nest]

gothic emo kid screaming in a wishing well…in the style of soundcloud

[graceful condor]

You know what really confused me though,
The suggestions it gives are like 1 word.
Where as the AI-touch up thing for the prompt, writes a whole novel

so i was so confused, like do I write a book or just "guide" it

I guess its all trial and error between models

[polar spade]

By the way, respell to "Rococo" and it should help pull more of that post-Baroque vibe in.

[polar spade]

Yes. Im biased to single prompts. But there should be no difference, generally as I understand. Im a less-is-more type user. Easier to de-bug that way.

[graceful condor]

since your here, Can I ask,
Are the newly generated tracks influenced by previously generated tracks in the playlist?

[plain pumice]

I'm curious if anyone knows, but I was trying to get the midi of a track, from within the studio.

1. Click "Get Midi".
2. It does something.
3. Then Nothing.
   .. I'm not seeing any popup, doesn't show as downloaded, there is nothing.
   Trying to use "Get Midi" again gives some kind of error, anyone who has experienced this and know what to do?
   Downloading the Stems does not download the Midis.

[polar spade]

Personally, it feels that way sometimes. So I have a habit of refreshing before any change in my Create. Its more rare that I think it does that these days than wth 3.5, I just feel better to start fresh each time to remove doubt that something is in a buffer somewhere. I could just be mental. 😁

From the developers: "Thats not possible"

In theory: Suno uses the best possible combination of melodies, so that means avoiding just random melodies as well, , at the same time theres got to be a bit of algorhythms to keep the melodies randomized so we all dont sound like Dolly Parton or Judas Priest every time we use a Country or Metal prompt. But there are less "optimal" chord progressions , and only so many millions of good melody lines that work well within a genre. So its inevitable, even within our own outputs that things just start sounding similar occassionally.... probably.

I could say, the more "common" we are with the metering of our lyrics, the similar combination of prompts, attempting to get a certain sound and so on, we can certainly limit Sunos creative flow and song number 12 through 16 sound alot like songs 1. 3 and 7.

If anything , using Suno or other Generative AI helps to remove that trend of humans to subconsciously pull a melody or "vibe" of some other artist we heard 5 years ago. Nonetheless, in the end, if it sound too familiar, best to try to augment it.

[whole cobalt]

guys how do i find my old suno song generated on this discord app

[raven turtle]

I’m finding the vocal hums at the beginning of each generation too much! Every single damn time! I’ve switched back to v4.5 for initial generations now and then remaster into v5 . Seems to be an OK workaround for now until any update can stop its constant intro warbling

[plain pumice]

Honestly, remasters from any version into v.5, has butchered them.
..and any previous version I've wanted as v.5, I've had to rebuild.
This pattern was also true in many cases from v.3.5 to v.4 and v.4.5.

[quasi dagger]

I've been trying to find ways to tag it out but this feels stronger on some styles than others

[tall fractal]

i agree somewhat, but if u keep trying you can get a better version

just gotta hit the right prompts with the right luck of generation

once you found the prompts giving you basically what u want then just pop out a lot of them

[raven turtle]

Absolutely!

[raven turtle]

I’m Only using the “subtle” setting and it seems OK. It’s not brick walling or anything crazy.

[quasi dagger]

I feel that some keywords set in certain tonality in the styles can also make suno drift there, its hard to detail what exactly unless you start extensively testing and that costs quite alot of tokens

[torn badge]

One message removed from a suspended account.

[tall fractal]

where do you find a setting called "subtle" I dont see it

[tall fractal]

yeah just do it

[nova turtle]

anyone have any hints on finding a GOOD singer. I assume if I find one I like I can then make a "persona"?
In all the song I've made, I don't really like the singers much. Had better results with udio a year and half ago
I'm and old man - love classic rock and prog - think yes - genesis. All the singers sound recent and artificial

[half dragon]

The settings on Remaster tool.

[tall fractal]

ahh i see ty ty

[half dragon]

Maybe do a search for it? depends on the term you might need to scroll through many reault pages

[tall fractal]

it has to be a song generated in suno as far as i can tell. so generate until u find a voice u like

hopefully there will be an option to upload an A cappella in the future

[polar spade]

Suno can "sing" uploaded spoken words, if you also provide the lyrics/text. It doesnt have a means of "auto-tuning" though.

But the spoken words couldnt be something already recorded and publicly available, no different than any other form. Even if a parody or public domain. Technically, it likely would not sound like the voice/person/artist on the audio either.

[polar spade]

Thats a wide spectrum between Jon Andersen and Peter/Phil. 😁

Prompts such as: anthemic, prog rock, ethereal, deep cuts, barbershop, 70s synth.... maybe avante guard, Improvisational, free interpretation.

Throwing in those prompts got a natural sounding vocals some harmonies with some energy and vocal fry and a cool improv-ed 6 minute song on the first gen.

https://suno.com/s/LvlOfsaHi1OoQKLg

[iron timber]

I find that [tags in lyrics] tend to be more controllable then the actual prompts .. still briefly touch in prompt but high detailed in the lyrics and the word dynamic is huge for me at leastt

[karmic oxide]

Can anyone help. I signed in on Suno AI using discord profile. I was able to sign into discord and see the suno server, but the Suno website itself seems to be having trouble loading my discord profile. it seems stuck.

[short arch]

hi , theres some trick to make the extend version with more time , is giving me only 15 sec more

[surreal citrus]

I've been working on this piece for a 𝐰𝐡𝐢𝐥𝐞

[last bay]

not klicking on that, why is it so small and why in tech prompt?

[abstract bough]

Finally had the opportunity to try 4.5 but unfortunately it just sounds like a shoddier version of v5, and it similarly ignored the instructions I left in brackets. I skimmed some of the convo above and it seems like 4.5 may have also changed: someone else mentioned that they used to have success commanding it with brackets, and now brackets are doing nothing as is with my situation

So I'm still stuck with Suno talking instead of singing :/

Out of the ~20 gens I tried, only one seemed to obey and sing almost the whole thing, but I can't tell if that was simply sheer luck

[last bay]

huh i only get spoken word when asking for it and even then it sometimes sings. you can also reinforce in style prompt aside the instructions in brackets in the lyrics

maybe share one of the songs to see lyric and style prompt

[abstract bough]

my suspicion is that it doesn't like to sing things that don't look songy (no rhymes and inconsistent length between lines)

[last bay]

even without rhymes i should sing most times 🤔 what is the genre?

[abstract bough]

choral, prayer

I've tried a bunch of diff prompts that worked fine on other texts

my most common is - choir, solemn, reverent, al capella, dark strings, deep tones

[last bay]

ok prayer might push it some, maybe just change or leave that out

[abstract bough]

sorry to be clear I did not make that part of the prompt

I wasn't sure what you meant by genre

[north plinth]

I have noticed that you only have a certain number of downloads per month on the free plan. Is this the same for the sub?

[quiet solar]

Ultimately, prompts are just a way to feed in reference points for Suno to look up in it's database of learned music to try to steer it towards what you want.

What tends to get people frustrated is they think of it as being a command prompt where the AI is given strict instructions to follow - it's perhaps more helpful to think of it as a set of thoughts on a page given to a session performer with "...but ultimately, do what you think makes sense" added to the end.

When it comes to approach, there's not much functional difference between, say, "pop punk, upbeat, melodic" and "a pop punk track with an upbeat tempo and melodic guitar riffs" as the AI will still interpret it from keywords. The benefit of being more narrative only really kicks in if you're giving guidance around specific parts, e.g. "instruments build up slowly towards the chorus" or "guitars have a crunchy, post-punk distortion".

As a general rule, prompts are more effective the easier Suno can look them up - so things like [Verse] [Chorus] tags in the lyrics it finds very easy, and you'll generally do better with more musical references, e.g. "fingerpicking arpeggio guitar" versus "singer yowls like a raccoon that just burned its paws"

Similarly with lyric prompts, that can help direct the AI to do something specific at a specific time, but it's not a guarantee. Even in most examples people give of "Look what I told Suno to do!" you can see that the AI disregarded a substantial number of their prompts.

If you want to yell louder in Suno's ear and direct it more closely, it's not just a case of prompting, but things like uploading pre-existing audio where you've laid out the general song and then turning up the audio influence and style influence sliders to the 70%+ range, and the weirdness slider down.

[summer garnet]

When extracting stems, are they suppose to sound louder than the original track. They always seem to sound louder and some of the instruments (e.g., guitar) sound so different. I wanted to change just one part of the vocal but leave the rest of the instruments alone, but it sounds so different with the steams, I am almost hestitant to even do it.

[quiet solar]

They're not supposed to, but sometimes it happens.

Splitting out all recognisable instruments is more likely to have something go a bit off in the mix than a simple vocal/instrument split. You can always regenerate the stems to try to avoid the issue.

[smoky igloo]

anyone has issue in upgrading plan to use Studio

[flint haven]

You are misinformed @lavish shard. If you have < 50 credits, it will be refilled to 50 at 00:00 UTC+0 (unless you have a rare glitch that prevents it from happening)

[graceful condor]

Send me a message of your Song prompt maybe I can help.

[humble verge]

why do covers sound like absolute 💩 in both v5 and v4.5+ ? Audio quality is in the gutter.

[prisma hound]

I've been exploring daw command prompting recently and it's really working well. It's completely undocumented but it seems the training data and the engine are made for that purpose.

[ivory peak]

Is it me or Edit tool is useless? It doesn't continue the style of the song and if the section is at the end, it cuts off lyrics. What's the point of this?

[livid crescent]

Is there a way for Studio to use v4.5 instead of v5? Instruments in v5 seem fine, but I can't get aggressive vocals from v5 to save my life.

[devout minnow]

I've found that the edit tool can be hit n miss, some days it has been torturous to use and other days, like yesterday it was on point,

[outer bough]

What is your approach for recreating drum track in Suno? I have a song in Suno Studio and would like to change the sound of drums according to prompt. But I get only complete arrangements and sometimes totally unrelated to prompt, even when I choose "Drums", prompt drum style and play wirth sliders. The same goes with any other track - it is extremely difficult to change sound of existing track to other instrument or vocal on this point. Is there any way that could make it easier?

[livid crescent]

I use [Drums Only][No Other Instruments] and it works about 50% of the time

[outer bough]

thanks, I will try 🙂

[hard beacon]

sure.. just change the model 😄

[quiet solar]

It's definitely a really efficient way to give Suno the information. I'm not sure if it's a more effective method of getting Suno to 'listen' to the prompt, but I've seen some really good results with it. Ultimately, the trick is to find what works best for you. 🙂

[livid crescent]

Where exactly can you do that in studio??

[hard beacon]

ahh... well.. Studio is a different one... afaik it depends on the output...

so maybe you have to create a cover in 4.5 first

[livid crescent]

yeah, but you can't generate vocal only tracks outside of studio... at least I haven't found a way. When you extract stems it gets all artifacty... so you try to generate again, then extract stems, then generate again, and extract stems... you get the idea. I'm pulling from a 4.5 cover, but trying to remake in studio is impossible. which is really sad, the main thing I liked about suno before was that it could do aggressive vocals better than the alternatives. Without that I don't really need Suno, I can do the instrument stuff without it lol.

[hard beacon]

what happens if you create a 4.5+ cover (aggressive vocal model) and then remaster it into 5?

[random pecan]

Hello guys! Creating songs is now impossible via discord?

[livid crescent]

you get an aggressive metal track with an R&B singer. sounds really weird.

[hard beacon]

i know what you are talking about.. but i had some aggressive nu metal v4.5+ generations in the past and a remaster in v5 can work well 🙂

[prisma hound]

Yes it's very efficient. I built some tools to investigate and made deep analysis in the literature to get back the good information. I'm still learning the way but this kind of prompting really leverages the quality and the 5000 character limit of the lyrics field is well optimised 😅

[livid crescent]

I'm using 2 singers, trying to get the layering like I want it. So I've been trying to make a vocal track for each one that I can mix/match... but can't get the raw vocal tracks to do the layering from.

since every time I try to recreate the stem in studio it uses v5 and loses the style completely.

[hard beacon]

coz it seperates "sounds" and not each layer

it's not a real multi-steam extraction tool... that's why vocals (male and female) are on one layer

[livid crescent]

I know... That's why I'm trying to make separate tracks. But it makes them using v5, I just wanted a way to use 4.5 so that they wouldn't change styles. I've tried 100s of generations and while it makes separate male and female tracks, they sing like it's a pop song no matter what I tell it to do.

[hard beacon]

i know exactly what you are talking about..

[livid crescent]

A guy in another thread earlier said suno hates rock and metal... He's probably right lol

[hard beacon]

well... with 4.5+ all vocals very screaming ones 😉

well.. of coz no ballad was screaming.. but.. even dubstep or EDM was with aggressive vocals

[livid crescent]

Agreed...4.5 does screaming very well. I just wish I could use it in studio

[hard beacon]

aaaaaaaaaaaaaaaaahhhhhh... now i get it... you load a 4.5 version in studio - recreating a stem - and then it's in v5 sounds... correct?

[livid crescent]

Yes

[hard beacon]

tricky one 🙂

[livid crescent]

Recreating to get rid of the artifacts from stem splits

[hard beacon]

you also have some artifacts in v5 stems... coz.. well.. it's not a real time multi-stem-export tool... sometimes you have guitar elements on the fx channel

[livid crescent]

Yeah I know. I'm trying to get a vocal only track to generate... No method to do this using 4.5

[hard beacon]

tbh... i never get an vocal only track within my last 100.000 credits 😄

[livid crescent]

I can get vocal only about 60% of the time... Just always whiny

And it thinks everyone wants auto tune

[hard beacon]

well.. at the end of the day it's AI and not a DAW 🙂

[livid crescent]

Yeah guess I'll keep waiting. If I get too tired of waiting can always try other software

[hard beacon]

that what i am thinking since v2 😄

[livid crescent]

Lol

[hard beacon]

but i have to say it depends on the style you want to create...

genres with sounds, that are also in the industry created with software", are much better.. an electronic guitar is harder to immitate as an EDM Sound 🙂

[livid crescent]

Well, for me I'm just trying to get vocals to go with tracks I've already recorded. Released an industrial metal album thru a label a few years back, but the singer moved on... Had a lot of material still needing vocals

[hard beacon]

ahh.. ok.. so you are more on cover function..

[livid crescent]

Everything I work on starts as a cover... I'm bringing at least a guitar riff for it to use as a guide. I don't mind the changes to the music most of the time, but my primary use case is as a vocal generator

[hard beacon]

well.. there are are some gooooooood ai vocal generators for different DAWs out there...

or are you "just" a musician and not a producer?

[livid crescent]

Well, the one I released was 100% produced by me. I've looked at other generators, but same problem.... They don't usually do metal well

There are a couple I haven't tried due to pay walls

[hard beacon]

mabye it's easier to hire a new vocalist? 🙂

[livid crescent]

The first year I looked non stop. Pretty much stuck hiring as a gig... That gets expensive real fast

Singer for one song would pay for a year of premium here lol

[hard beacon]

but.. what is your personal girl?? Own written with ai generated vocals?

girl = goal

hahahahaha

sry bout that

[livid crescent]

Honestly don't care about whether the vox are AI or not... Just would be nice to be able to finish the tracks I've spent years working on

Yeah I knew what you meant 🤣

[hard beacon]

sometimes the german autocorrection kicks in 🙂

because it's a lame genre??

sorry.. just kidding 🙂 What prompts are you using for the style?

try to write the word "chords" more often in your prompt.. and play with the sliders

what model are you using?

[livid crescent]

make your prompts more compact and less wordy. can use chatgpt to shorten it if you want, that helps

[hard beacon]

naaarf

[livid crescent]

I found it helps with both. one sec

[hard beacon]

@unreal sleet : send you a private link 😉

[livid crescent]

[Trap-soul, 70 BPM, moody, hypnotic, driving, chord-driven, prominent hard-hitting minor chords on Rhodes piano, bright bell pluck motif, heavy 808 bass with glides, sharp hi-hats, crisp snare, subtle vinyl crackle, hook with emotional female vocals, verse with confident male melodic rap]

use that as a starting point, elaborate as needed

[quasi dagger]

Can you give me a highly descriptive take (genre you want to hit, instruments you want in, etc etc, mood....) of what you're looking for, maybe I can roll some example for you through my workflow see you notice something different from what you're doing.

[livid crescent]

I've had these too 🤣 . absolutely no place in the metal songs I got them on

[quasi dagger]

Give me what you throw at chatgpt to try and get what you're looking for, and, if you want, some trial lyrics so I dont have to get some my own way which could muddy some things.

Since the more descriptive that I can be to what I want to achieve the better the result will be in getting what you want

(you can do it DM if you dont want to polute here with lots of info)

Them lyrics, ye?

[graceful condor]

I am so sick that SUNO is butchering words in lyrics either leaving them out entirely or mumbling some dumb sh.t or making up words. I waste so many credits because it messes up one word one FREAKIN WORD and all the rest is great. I am so sick of it! What is the worth of better voices in V5 if they sing absolute braindead bullcrap.

[quasi dagger]

Alright I'm still processing some diagnostics I was doing before. So this might take a while until i can throw something back at ya but once I do I'll check see if it's running towards what you want.

[livid crescent]

personally I feel like their idea of better is a bit flawed, but yeah it screws up A LOT

[graceful condor]

And it seems they do not care at all. They release some new stuff totally flawed und bugged and that's it for them. At least it seems like it. Because there is no real communication. We give feedback over feedback but I do not see any progress of things getting better. In fact issues just get more and more. I mean even Persona are worthless now. They do not sound at all like the persona only using they phrasing and pronunciation if you are lucky. I mean what was the reason to give us V5 if it is just a mess?

[livid crescent]

I will admit, compared to a week ago I've noticed significant improvement in the single instrument generation for studio... no mention was made about it but I can definitely tell a difference.

It still struggles something ridiculous on vocals though, with both isolated tracks and general song creation. As I mentioned earlier I've made 100s of tries to get 4.5 vocals into a studio track. Finally just gave up

[graceful condor]

OK that one I noticed too. Instruments somehow got better. Not all types of music but overall.

Yeah and it is even more frustrating if you already got a nearly perfect song but regardless how many extends or re-tries you do the AI says "NO".

[livid crescent]

the one I'm working on it studio. I've got maybe 3 tries on all the instrument tracks. then I've got like 5 vocal tracks where each one has nearly 100 tries and still is nowhere close.

[graceful condor]

I feel your frustration. I really hope this gets better soon overall. Because it is really ruining the fun.

[livid crescent]

Like I said. I gave up

[slender rapids]

Hello, can the [spoken] tag work?

[calm apex]

I am having problems using a persona from someone else, is a public persona , i have it favorite but when i click it nothing happens? is this a new problem?

[half dragon]

[Spoken Words] works pretty reliable

[slender rapids]

okay thanks

when there is a [child] tag that works ?

[coarse river]

I have been unable to log into my Suno pro account for several weeks. When I enter my phone number it goes to an account called TokyoDesigns911 which is not mine. When I check the account it shows my phone number but of course none of my library is there. I have contacted Suno support several times but have not received any help.

[polar spade]

Less likely for a [Child] tag to speak as a child. Children vocals and speech is more commonly heard in children's song, so if you're writing a kids song, maybe. And for kids songs, often those are sung by adults as well for the kids.

[quick sonnet]

At least they should be aware, that isue was already mentioned many times.

[hollow hare]

#📚┃suno-school message

[foggy inlet]

Hi,

I’ve completed listening to songs and rating them on Sonu as instructed, but I haven’t received any points for these activities, even though I finished the required

I have tried refreshing the page, clearing cache, and logging in/out, but the points still do not appear.

Please check and help credit the missi

Th

[quick sonnet]

Why the f... even mp3 downloads are not avaible for free users? Befor this only vaw downloads were limited to pro users, but mp3, not the best quality, was avaible for everybody.

[remote skiff]

Anyone know how to produce a good song ?

[short arch]

what do you wanna produce?

[remote skiff]

Like kpop song

[short arch]

wow good selection ,i dont know if its possible to make kpop

let me check something

[short arch]

Professional K-pop style with energetic and polished vocals.
A mix of electronic pop, dance, and hip-hop elements with modern percussion, bright synths, and powerful bass.
Full commercial structure: short intro, verse 1, pre-chorus, catchy chorus, verse 2, bridge, and explosive final chorus.
Expressive lead vocals with ad-libs and layered harmonies.
Lyrics half in Korean and half in English, about confidence, ambition, and youthful energy.
Clean, dynamic, stadium-level production with a clear and punchy mix.

[raven turtle]

I’m currently throwing credits down the drain here trying in Studio to get it to fix a mis pronounced word “tastic” should be “plastic”.
How much credit cash I’ve chucked at this!!! You’d think it relatively simple fix… agghhh.

[tawdry scarab]

there's only loose control over the voices with direction, so the thing to do is get the right next line, then extend after it and get the next line right, etc., to piece the work together, though there are various conventions people use to try to get it to work better from the get-go. that involves putting "duet" in the style and using tags with brackets [] to do things like [male] or trying to give them names and then refer to them

@unreal sleet

[quick grove]

How do I share my songs in #🐣┃suno-showcase ? It seems to keep getting deleted. My songs don't have anything inappropriate in them.

[livid crescent]

I tend to generate one male, then another female, extract the vocal stems, then piece it together the way I want in a daw. then reupload and cover it. It's not perfect, but better chances

less credits than constantly regenerating...

[quick sonnet]

Not for suno.

Write it in the style and lyrics, this increases the cances to get the right result.

[summer garnet]

Right next to the lyrics or is it part of the tag for the section in the song like [Verse 1] or [Chorus]? I generally will describe where I want the female or male lead voice to sing. [Chorus - Male Lead Voice the first 2 lines and Female Lead Voice the last 2 lines]. It seemed to work for me for the most part or throwing in a [duet] tag when I want them to sing together.

[quick sonnet]

Like I said, it increases the chances, but gives no guarantee.

[livid crescent]

well, extracting vocals costs the same as one generation. do it for the male and female separately, then piece them together. run it back through as a cover, and you're at 30 credits. Same as 3 failed generations...

[safe chasm]

I have a weird thing going on, making v5 now completely useless unless I create in v4.5+ and then remaster in v5 . It is using my style prompt for lyrics constantly. is there a general bug for that or do I have a weird disease? 🙂

[summer garnet]

Are you doing that through Studio? Extracting the stems with the vocals. How are you generating the male and female separately unless they are already there from generation? Are you doing the entire song with male and then with female? From what I recall, you can do a seperate track and have it be vocals female and then do a cover on the lyrics and then mash them together?

[livid crescent]

studio doesn't let you specify male or female. I generate in basic, extract stems, mix them like I want in ableton, then reupload

[north plinth]

i saw they limited the amount of downloads per month or free users dont know if the same for subs, i asked but had no reply. 😩 this seems to be a recent thing

[north plinth]

mine arrived the following day

[reef tree]

Will personas be incorporated into the app? Or will this onky ever be on desktop?

[summer garnet]

Never used abelton before. Sounds interesting. I've only ever used Audacity in the past. Not much of an audiofile or music producer until recently getting my feet wet in it. I believe you can do it in studio, but you have to drag the section you want to change into the custom area on top left depending on the layout where it has Lyrics, Styles, etc. It will let you redo that section as a cover and you can select female and play around with the advanced settings to get it where it needs to be go along with the track.

[tiny girder]

hey does anyone know if i can use suno to sing on a beat i already made myself

[summer garnet]

I believe you can use the upload audio with the beat and then go from there and generate the style and lyrics into it via generation. I haven't tried to do this myself yet though, just saw videos that suggest or show it as an example.

[tiny girder]

can you send me the link whether here or personal message

[north plinth]

can anyone tell me why Suno have limited the number of mp3 downloads per month? it is not said on their tiered plans that this is the case. I don't now if its for all tiers

[polar spade]

How much of the area around the word are you including? Suno still needs the typical 3 seconds of blend room, left and right of the area youre making changes to. The larger area you chose, the easier wiggle room can be made to fit the letter "p" back into that space.

Overall, I try to at least select up to two lines, but go as far as there is a gap eventually somewhere on both sides of the word you're changing. Typically, theres a pause at the beginning and end of a sentence.

The "l" in lastic can easily become a "liason" with the word before or will blend the sounds together. So even just adding a "P" or "pl" sound in plastic needs about 3x as much space as however it is pronouncing it now.

Rewriting the whole line slightly but still attempt to write a line with at least one syllable less than what you have, can help.

What's the whole verse the word "plastic" is in? Is it a lone line or a whole standard 4 line stanza?

Where do your lyrics come from: an audio upload or just written text?

[polar spade]

What do you feel is the limit? 50? 100? 1000? Daily or just monthly?

[polar spade]

@tiny girder Thats the idea behind uploading an audio then clicking "add instrumental". It will either be on the beat or syncopated one beat off, more often than not. But ultimately it can also depend on your prompts whether this is a bouncy pop song, a rap, a free flowing improvisation jazz piece, etc. as to how Suno is going to match your melody/beat.

As most things go, just try it and see.

[raven turtle]

I’m trying to replace two lines so will try to open it up. They are my own lyrics.
I’ll try it. Cheers

[north plinth]

it says monthly. I don't even download 50 a month. I noticed it yesterday when I went to download and it said I had x amount left this month

[summer garnet]

I have had issues when editing stems in suno and exporting to a song. It would just spin and nothing would get created. Not sure if this is just a bug with this workspace and song or is affecting anytime I want to do one again at this point. It was a bit aggrevating to say the least. I ended up just exporting the stems and using Audacity to make the edit I wanted and exported it from there.

[dapper fog]

Hi all, new here and I guess this is the channel I should turn to if not then please point me in the right direction.

I'm trying to figure out "what means what" in the lyrics from Sunos perspective.
Ex.
[Chorus]
(Whispered/Shouted)
Lyrics... LYRICS, lyrics lyrics lyrics

If I want to give "notes" in the actual lyrics I have yet to find an easy way to do it. Using parentheses sometimes works, but other times those words are sung. Is there an easy way to write things to give notes because sometimes it works so there must be some better type of syntax that I can use to get a certain part whispered, or shouted, or make the singer shout a certain word such as with capital letters, which sometimes does work but...

[livid crescent]

in my experience you never want to use parenthesis, use brackets [] instead.

[dapper fog]

So essentially [Chorus - Whispered] ?

Or [Chorus]
[Whispered] Lyrics...

[livid crescent]

or [Chorus][Whispered]

[dapper fog]

So even in the middle of a section just?

"these are the lyrics [shout] sing with me [whisper] sing with me"

[livid crescent]

theoretically, yes...

Keep in mind, the Intelligence part of AI comes into play too. Sometimes it'll just do whatever it wants no matter what you type lol

[dapper fog]

Yeah I've noticed xD

[torn spoke]

has anyone here been able to get the new "Recreate" function to work in studio? For vocals?

[livid crescent]

so it's always a diceroll, but that will improve your chances of getting what you want

[torn spoke]

It has never once worked for me, legit, never. If the audio contains vocals, it will ruin them, just mumble them, say them wrong, or create random new ones

It's so frustrating

[livid crescent]

there's one called recreate... but haven't had much luck with it

[torn spoke]

Its the recreate function that isn't working for me

the old replace actually did work

I edited the original msg

[livid crescent]

I've had better luck just doing a fresh track, then i turn off all tracks but the vocal I want it to recreate

[torn spoke]

I'm just trying to fix a tiny section, like, 1000ms

[livid crescent]

haha better to make the whole part and edit it in

[dapper fog]

That's so smart... I feel so dumb right now

[torn spoke]

My problem is the entire song hiccups in a place

I just need to replace that small section

I've kind of hacksawed a solution in audition

and cleaned up the static

I'll make a personality from it and simply try to get a more polished version

I think

[livid crescent]

don't get too excited... even with that I have made over 100 generations on a single track, trying to get the vocals to do what I want

[torn spoke]

those are rookie numbers

[livid crescent]

Did I mention there were 5 of those tracks on one song? Only reason there weren't more is I just gave up

[dapper fog]

I'm more worried in general of the songs just coming out sounding like modern pop music so I over complicate the prompts xD

[torn spoke]

O shi, the actual normal old replace is working

let's... GO

shame studio's vocal generations are so bad

[livid crescent]

OH? Are you doing metal too?

[dapper fog]

Hmm, I'm doing... Grim storytelling?... Essentially creating a library of music for my RPG world. So songs are "in world" but generally themed as "rock" or alternative at the very least. Grim musical-eqsue. Wry troubadours and gods of plagues.

[livid crescent]

Very nice 🙂

[dapper fog]

Can I share stuff here? (Like a link to one of the songs if you want a sample)

[livid crescent]

Hmm.. Dunno. Hold on

[torn spoke]

yeah file uploads work here

links too

it's just in #💬┃general-chat that there's policing

[livid crescent]

I try not to here, but feel free to DM if you want

[dapper fog]

Yeah seemed to remember that from reading it

[crude mountain]

Anyone else getting any use out of suno studio? I have been playing with it for tracks and I am just getting garbage generations? Like I ask it for backing vocals over 4 bars, it generates drums or guitar strums, etc.?

[gritty narwhal]

Right there with you on this one. Its super frustrating and has pitiful documentation or tutorials on how to use it properly (at least that ive seen). all backtracks I try to get includes the melody that the singer is singing which is... not ideal.

[queen linden]

if it gets the lyrics right the vocals are off. get the vocals straight and the words get fudged at the ends. if you remix you get cooing, moaning, humming ,,,,, maybe its me. Im not a professional but i think THIS MUSIC THESE Lyrics IN THIS Persona would be easier .... sorry for the caps,.... but man

[weak wagon]

Hey all. Trying to get some vocals similar to this track ivan b - like the way. The first female vocals. What prompts shall I use? Cant seem to get it right.

[stoic vine]

Pro tip: use something like Spectral Layers to separate a drum track then use a drum replacer like MDrumReplacer, found this has made an big improvement for final results. Makes cleaning up the suno noise so much easier

Plus more control over better sounding fuller drums

[stoic vine]

Meta tags might be your friend

[queen linden]

I appreciate the Spectral Layes and Kits programs ... but really not another nickel is going in . I wish they would allow In The Style of .. type of prompt. female vocals should include those that have heart Janis Joplin, Debbie Harry, Tina Turner. I can add and subtract all I want but you end up with breathless, whispering teen queen gak

[short arch]

theres's some way to control the time of the extend versions?

[rose folio]

I tried to make a quick song with just a violin and male bard / singer but it keeps putting a band in and since the subject matter is sad it plays it slow. Spent so many tickets to get this right. Tried 5 and 4.5 Anyone got a foolproof way to get what I want? 😭

[queen linden]

you can use advanced options. use the word troubador, or solo . In style less is more. and the 5v is making me feel its more like a Slot machine I put credits in and maybe i get something good maybe I should try again and again. But do try the advanced options.

[short arch]

can somebody help me pls

https://tenor.com/view/kevin-hart-help-me-gif-18351511120546678838

[queen linden]

I have had no luck with extended music .

[short arch]

is difficult im trying in song editor and the song wants to finish now , dont want to extend , make lil extends with a critical cut end

[velvet jungle]

So I noticed that's when I do the extension on Versions 5 it's having a very difficult time pronouncing the words, it's mostly all gibberish especially towered mis to end sentences gets even worse, no matter where I set the extension, no matter what style genre, the problem is I just can no longer keep up and sing , rap, speak once you extend, now on Verizon 4.5 I had the same issue but got better especially these last 2norn3 months, now it seems like we're back to square one again, we need a update here and address this issue I can't be the only one like Highlander!?!?!?!? 🤘🏻

[summer garnet]

Has anyone used the video feature on song details section for suno? I see it works fine in suno, but I tried to export it as a mp3 for say Spotify it won't even let me play it via local file when I do that. It works fine if I just do an image.

[karmic oxide]

It's been 2 days, still no profile establishment on the Suno app side. It's still stuck in a bootloop tryng to establish my profile from the discord login...

tried loging in from phone browser and get a "404" undefined @undefined undefined "something went wrong" also if "you" are leaving me on read, waiting for me to wade through all the FAQ's hoping I will get it fixed myself... already done that. nothing is working...

[summer garnet]

That's what my export song does from suno studio lately. It just never creates a song and just has the cycling wheel next to it and never completes

[polar spade]

So, no more than 50 before getting a limit notice? I'll test on my end as well to see. Of course all tiers have un-promoted limitations such as content censor moderating. So there may be a limiter, if I were to guess, if possible to help keep scrapers from mass downloading others file -- thus we get equally limited. I dont know.

Even still, it seems downloading 10 or so a day seems reasonable. Thanks for pointing it out. I'll check out whether I can reach a limit.

[wise saddle]

if i sing my own version of a song and upload the cover, is it allowed? not use the lyrics but the basic melody etc.?

[rose folio]

Style: 100, weirdness 0 and style of solo troubadour with a fiddle, fast song. Did the trick.

[north plinth]

thanks. I can't download other people's files, I'm not interested in that anyway, just my own. I only noticed it yesterday so not sure what the limit is, i've just got one free download for the month left, so I'm assuming it is less than 50. I'm on free tier but thinking of subbing. I don't know if the download limit applies to paid tiers or not.

[humble verge]

seriously though, why is 4.5+ worse after the v5 update? They shouldn't be connected.

[half dragon]

That's a gray area but I wouldn't recommend using a copyrighted melody though.
Stick with your own melody or generate new melody.

[dapper fog]

I have downloaded over 200 tracks of mine so far (with pro) no limits

[dapper fog]

If you intend to publish it and monetize it you will be slapped due to copyright laws. Covers of 'real' songs are a very tricky area to navigate.

[raven turtle]

Same. I’ve been working in Studio and the simplest prompt on a new layer for synth pad gives me arps. A simple drums IDM style prompt I get whole instrumental tracks. Also that old problem of degeneration over time is happening . The chorus after 2nd verses get washy. Yeah it a beta and early, but it’s based on v5, but results are very patchy. And credit costly too! It’s a great concept that’ll be superb by the time suno v7 hits!

[woeful lava]

Can SUNO generate drone zone textures. I've been at it for days but it always plops in a rhythm and percussion even melody no matter what I do?

[midnight copper]

Hi

Does anyone have an issue of cloning voices for suno

Everytime I use my voice it sounds like a whole different person and I even payed premium

^^

[devout minnow]

Never used my own voice so cant comment too much, have you gone into advanced options and put audio to 100%

But afaik you arent going to get a perfect copy of your own voice

[midnight copper]

I did but I literally sound a whole difffernet race

[midnight copper]

That’s no problem but am not even same person at all

[devout minnow]

https://discord.com/channels/1069381916492562582/1416189341486678128

Have a check on this in workflows, theres a bit of guidance in the comments from the op, hopefully it helps...

[midnight copper]

I am trying

[raven turtle]

I’ve found It never replicates the exact voice uploaded. It’s a new iteration of it.
I assume the sliders have a bearing on how much

[frigid willow]

When I extend a song, it's only extending for about 45 seconds to 1 minute. How can I extend my instrumental tracks to the full 8 min? Are there prompts to help?

[midnight copper]

back then it use to take my voice and make a song now i came back from four months its a different person

[devout minnow]

Ive only done small exgensions for when i wanted to add a little extra to songs, have you tried putting a time in [] or anything like that? I don't know if it would work, hopefully someone else can give a definitive answer

[willow ocean]

V5 keeps adding all these weird hums and shouts in the songs — is there any way to make it stop? It’s honestly kinda bad.

[frigid willow]

I haven't yet. Will try this, thanks

[midnight copper]

^

[quasi dagger]

It's all some tags we add that hit Suno with some chanting, I was trying to dismystify it a while back, somehow someway my GPT figured it out but I sitll don't know exactly what it is 😐

One of the ones I noticed was when I added a tag like "conversational" on the delivery of Verse 1, then Suno just decided it should put chanting everywhere, was odd, one tag, full flow of chanting

[near flume]

Use the Exclude Styles box and enter what you don't want (ex: noise, distorted voices, hums, garbage).

[devout minnow]

I suppose something else you could try is to extend it, then grab full song, extend again, grab full song, rinse and repeat until you get the length that works for you

[willow ocean]

Honestly, I think the issue isn’t about the prompts — V5 itself just feels kinda broken. I’m probably gonna stick with 4.5 for now and wait till they drop a better update.

[quasi dagger]

Nah nah, i know what you mean, I've had this issue as well. I think, for example, I poluted a Workspace with these tracks then all of them came through the same spot. Try swapping Workspaces, try not to add many vocal leaning tags, etc. It also matters what kind of music, some lean more on these kind of pick ups.

For example, right now I'm just rolling lots of gens and I've had 0 issues so far.

[quiet solar]

Typically no. Any version of 'yes' requires wading deep into the absolute minefield that is music copyright law.

As a general rule, it's a straight 'no' to uploading original copyrighted tracks (music or audio) that you don't hold the copyright for to Suno as that's going to be a clear breach of the terms of service. Indeed, Suno's own copyright detector will probably block you.

Covers may fall into a grey area in some countries - provided you've puchased a mechanical license for the composition, and have uploaded your own cover performance of the original music and lyrics, that may be okay. If you're in that boat, I would still recommend keeping your cover private on Suno rather than publishing it, and distribute it separately wherever you'd planned to in line with the requirements of the mechnical licence.

If your cover includes video in any way, you also need a Sync licence or to comply with things like YouTube's Content ID if you intend to monetize the performance in any way (note: monetization of the cover may not be allowed by the copyright holder under Content ID or similar systems).

It's a bit of a messy system at the best of times, and the introduction of AI performers only makes it more complicated. So if you're not sure that you definitely have the legal right to do it, it's best to assume that you don't. And if in doubt, consult with someone who knows the details of copyright law in your country.

[willow ocean]

Sounds like a solid approach, I’ll test it out and see how it goes.

[quasi dagger]

https://suno.com/s/7OaR3jZG08urh7ju

This is one of the last gens, can you see the style prompt and lyrics prompt? It's not an exact 1:1 example since I fuse the dumbest stuff together with the most goofus thematics but I can set almost anything together and hit no humming unless I go for exact Ritual/Chant/Choir/Gospel genres.

[quasi dagger]

One thing you can do is try and give me the exact thing you want to do with as much info as you can and I'll try and generate it on my side and see the difference between your style and mine and see if you can notice differences.

[quick sonnet]

Asked whom?

[distant abyss]

Has anyone else noticed that v5 seems to REALLY like to add 20 - 45 seconds of instrumental intro at the start of songs? I'm getting frustrated that so many generations are tacking on a super long instrumental intro when there's not even a [intro] header in the lyrics.

its not hapenign every time , but it feels like 85% of generations have this huge unwatned instrumental section at the start of them

anyone else experiencing this and/or have suggestions on how to prompt to avoid it? its getting frustrating to get a decent generation but then have to spend an hour fighting with the terrible 'Editor' interface to try to make a shorter start that actually blends in.

like here https://suno.com/s/ebAKiimp9dvC73Io the first line of lyrics doesnt start until like 55 seconds in

[lilac gulch]

I'm trying something with these styles: Spooky reverbed Whistle , Nightmare, rap Hardcore ,Loud Bass ( soft), Catchy, and the result is coming soon

[tropic topaz]

I try to export my song from Studio, but getting a failure. The console shows me this

[cursive kiln]

v4.5 lite, is it a new model for free (because it doesn't have a pro tag)? Just not publicly released yet.

maybe this is the one they announced?

[autumn sable]

anyone else having problems saving a song after editing it?

[autumn sable]

@timid marten because #archived-ticket-help has a 2 hours cooldown on messages i guessed i answer here:
Hello, i am not using Studio, i am using the normal song editor on the website.

Basically the same problem perewilliams described in their messages in #archived-ticket-help

[neat nexus]

Please try again now. Slow mode has been turned off.

[radiant lake]

Hello! Has anyone else experienced this? I was composing an instrumental song and I asked it to create the second part. I got two songs, one 3 minutes 07 seconds long, and the second 2 minutes and 47 seconds long. I played the 2 minutes and 47 seconds one, and I noticed that it didn't stop at what was shown on the screen. It kept going, sounding very good and nice, and reached 8 minutes and 30 seconds before finishing, even though the screen only showed 2 minutes and 47 seconds. Where is the song that I listened to for 8 minutes and 30 seconds? Was it cloned by someone, or what should I make of this? I also have proof; I filmed it with my phone from minute 5:24 until the end of the second part of the song, which in the end shows 8 minutes and 30 seconds.

[quick sonnet]

Any idea how to let suno sing in the same chords progression, as some other song?

[quick sonnet]

Where you got that?

[polar spade]

It's not uncommon for suno to vary an instrumental length.

[polar spade]

If you are familiar with "add vocal" to the instrumental, that would at least be a melody within the same chord progression. What exactly are you asking? More of an a cappella vocal?

[summer garnet]

Quick question. If you say generate a few versions of a song and like parts of one song vocally than the other. Are you able to extract those into another song version and make it fit the beat, etc? There are parts of one song I absolutely love and others part of the same song, not so much. I want to combine all the best elements of the song variations.

[polar spade]

@cerulean nebula

If we make an extension, we can "get whole song" by the menu of our extension / Create/ get whole song.

If an edited song is not appearing in Library from an Edit UI (or Legacy) edit, we need to click "Save new song" at the top of the screen to put the edited song in our Library

[graceful condor]

There is no possibility to stop Suno from kicking out my prompts if I change the Audio (Cover, Extend), right? It is so annoying because each time a change the Audio it overwrites and kicks out my Style and Lyric Prompts.

[autumn sable]

Thats the problem, exactly this “Save new Song” isnt working.

[polar spade]

@cerulean nebula I could suggest to wait after clicking "Save new song" and let it refresh the edit page, then check Library. Also, check in the workgroups.

But note: some outputs simply can take a couple of minutes to show up, for whatever reason.

Are no edited songs showing yet above youre edit-output since posting 2 hours ago? Have you clicked "Save" more than once and what errors are ytou getting? And does the green "new song saved" banner even pop up in the edit UI?

[polar spade]

Since saving prompts is more difficult, im often using the "Use styles and Prompts" on a previous output or even copy and paste. That seems something they must focus on soon. I agree.

[autumn sable]

It doesn’t show any new song in any kind of library after clicking on “Save new song”.

It says that the song is save but I can’t find it anywhere but clicking on the “Go to song” after clicking “Save new song”. The problem is that the “Saved song” doesn’t work. I can’t listen to it, I can’t download it, nothing.

I clicked it in the timespan of 24 hours at least 10 times (the first try still doesnt work).

Thats what i have in my chrome console when i tried to play the “saved” song

Note: I don’t have any kind of plugins/add-ons in my chrome browser

I can play normal generated songs, i can listen to music others did but my edited songs don’t work

[icy valve]

have you tried refreshing after you save it?

[autumn sable]

Yes

I tried different browsers, PCs, flushed dns etc

[polar spade]

A 403 error means we make a request that cant be executed by Suno. This can include a server issue. If we use a VPN, this can interrupt Suno from executing a request as well.

The answer is often to just wait, but knowing a "final" process may not have happened amd we lose a song that never was.

The point that you get a green popup saying there should be a output is interesting, but I wouldnt know more.

So if two users have it happen within the day, amd at different parts of the day, then its worth reporting in bugs to have them check on their end.

The better test is to make new edited and saved song over a period of a couple hours apart rather than saving the same song.

Ill stick it on my TrSh list and test a couple myself later today.

[icy valve]

yeah it's probably just a bug. CORS issues are typically easy to solve on their side

they'll probably fix it soon

[autumn sable]

In #archived-ticket-help another person had also the same problem

At least similar problem with saving stuff

[polar spade]

"So if two people have it happen in the same day....."

Yes. Thats why in include @cerulean nebula in this thread.

I should have said "So AS two people have had it happen in one day...." 😉

[cerulean nebula]

This is what i get. If i click on "Go to song" it open the song in a new tab but it doesn't start. It's stuck on 0:00 whit the loading circle.

[quasi dagger]

Anyone else suffer from different dialect (e.g. PT-PT and PT-BR) deliveries and how to prompt onto one than the other? :V

[summer garnet]

I've also had this issue with the "Save new song" not working. It worked when I first started using it but the last song I did it would never let me save a new song from studio. It would just spin like it was going save one but in infinite loop

[hollow vector]

Hey Suno team 👋
I’ve reached out twice by email but haven’t heard back yet, so I’m hoping someone here can help. I haven’t been able to create any new songs — every time I click “create,” the screen just goes blank.

My account shows 2,360 credits and 472 songs, so I definitely still have credits available, but it just won’t let me make anything new. Been trying since yesterday.

Could someone please take a look or let me know what’s going on? 🙏

See attached image

[quiet solar]

Not really within Suno, no - at least not across separate versions of the song. You'd need to take the extracts into an external editor and stitch them back together. And unless you have a very smart piece of kit, making manual corrections if the tempo or (worse) the pitch needs changing to match the original.

[quick sonnet]

It's some kind of parody song, with similar melody as original, but complete diferent lyrics which are sung the same way.

[quick sonnet]

You should adres DAViD from suno team directly.

[hollow vector]

@stuck junco hi David can you please assist me having issues with creating songs.
Hey Suno team 👋
I’ve reached out twice by email but haven’t heard back yet, so I’m hoping someone here can help. I haven’t been able to create any new songs — every time I click “create,” the screen just goes blank.

My account shows 2,360 credits and 472 songs, so I definitely still have credits available, but it just won’t let me make anything new. Been trying since yesterday.

Could someone please take a look or let me know what’s going on? 🙏

[polar spade]

We can write a song,
separate the instrumental,
Write new lyrics to match the old (lines, syllables, metering....)
Click "add vocals" to the instrumental stem
Paste in new lyrics and click create

[polar spade]

Have you tried literally to turn off and on (cold boot) your computer? That could reset enough to recognize all client files, if that would be the problem.

[quick sonnet]

It's not about instrumental part, but the vocals. I have uploaded an acapela version of the song and then asked suno let sing the complete new lyrics the same way, but the result had 0 similarity.

[hollow vector]

Im using through the phone app

[crude onyx]

Hey there! 👋
I’m on the Pro plan and noticed that when I upload a custom image for my song, the final video still adds a Suno background and the “Created by Suno” label.

I looked everywhere (Manage Account, Edit Profile, etc.) but can’t find a way to turn that off.
Is there a setting for Pro users to remove the branding, or is that built into all videos?

Thanks so much — just want to make sure I’m using the Pro plan features correctly! 💛

[polar spade]

Right. Many chord variations can go with melodies and vice versa. So even if steps are followed as I mentioned, it may give something off a bit.

The more you can have the melody baked into the instrumental the better.

So make an instrumental out of the a cappella bit first, to see if you can get an instrumental with a baked in melody. Then use my steps. Since you are using a cappella, sometimes Suno hears a capella as an instrument, so cross fingers, it doesnt.

[polar spade]

Even better.... maybe. Uninstall and reinstall the APP then.

[quick sonnet]

Also I am a free user, so I think this feature is not avaible for me.

[polar spade]

Keep in mind they are working on an upgrade to 1.31.1 as we speak, so just check your update. Mae sure its current.

[polar spade]

Oops.... sorry, that does limit things to just writing lyrics exactly the same meter and syllables as the original, then cover the old/original but exchange with parody lyrics. It has worked before back when we only had 3.5. -- but far from perfect

[stuck junco]

Hi there! Thanks for reaching out about this!

Would like to better understand what's occurring here. Do you mind reaching out to me via DM to discuss this?

[crude onyx]

where can you check your version to make sure its the 1.31?

[polar spade]

Yea... built in.... on all paid levels as well. No secret handshake or mantra to remove it.

[polar spade]

Under your account in in the app , scroll down to bottom.

....or just go to your app store to see if there is an upgrade available.

If so, upgrade. If not, uninstall and reinstall.

[quick sonnet]

That's what I did, but like I said the results had 0 similarity to the original. Posibly my prompts are the problem. I told to mach exactly the singing.

[polar spade]

Can 3.5 use phrasal prompting? I havent looked in awhile.

Either way, it may be difficult with 3.0 or 3.5, for sure.

We had fun last with just 3.5 to rewrite hallucination songs. You write phonetically what you hear, then substitute real lyrics as close possible and cover and paste back in. So youre forced to use between 2 and 4 minute songs. But some get close.

[quick sonnet]

Sorry, I am really new, so what is ment by phrasal prompting?

[polar spade]

Phrasal: using sentences and phrases instead of just single prompts and commas

[summer garnet]

Yeah, I am just an amateur at best with this music production business. Trying to learn though. I saw in a video that you could drag in stems from other songs using Suno Studio where if you like the drums from one song over the other and it can be adapted to the beat of the song if I recall correctly. I was hoping you could do that with vocals too.

[polar spade]

New? Dude. You've been doing this couple a months. You're getting old school already. 😅 Youre nearly one of us already. I seen you passing out help. We'll done. 🤘

[quick sonnet]

Sometimes it also works. Posibly I am just using wrong phrases?

[summer garnet]

@polar spade Do know of Suno Studio allows you to adapt vocals from other songs like it can instruments via stems? The vocals would be from other variations of the same persona singer for the same song.

[polar spade]

I forget some things. I just dont remember is 3.5 only responds to single prompts anyway. If phrases are working for everyone, then I guess thats been improved within the model. Thats good.

As for "wrong phrases": it was hard enough already before. So it may be just burning through credits to see if you get lucky. So thats an issue.

[polar spade]

You can drop in stems as well. Yes. (Now i have to run to computer to see if i remember correctly)

Also, in the stem separator tools it has a button, "Edit in Studio" bottom right, as a shortcut to Studio to do this.

Yes... after checking... instrumental and vocals stems can be used in Studio.

[quick sonnet]

Sometimes it works and sometimes not. 3.5 is defenetly not as god as the Pro ones.

[polar spade]

This is a 15 second sample.... an a cappella vocal from one output stem added to the acoustic guitar of another. Done in Studio. Song starts in at 5 seconds. https://suno.com/s/byGwjRvRtf1Ciavl

[summer garnet]

Thanks @polar spade . Yeah, that would make sense. I wonder if you can adapt multiple vocals than just one to an instrument. Say I liked the chorus in one and verse 2 of another same yet different variation of the song. Not sure if it already exists, but that would be an amazing feature if you could prompt or provide it like you can with ChatGPT and other modular AI the parts you like of one song, the vocals and such and how it was sung and have it recreate it with the way it was sung in the sample provided.

[polar spade]

We cant ask Suno to "use the chorus from song 1 and bridge from song 2" that's probably ver 7.5... 😆

But we can manually splice different and all segments we like, from any output, all day long.and with Suno we have remaster and cover to do some nice blending when done.

[polar spade]

Everyone who has difficulties with getting a duet just right. Now with Studio, it just a matter of picking the best of all outputs and splice where needed.

Im an old man which makes me old school and have no issue with the concept of splicing everything manually until something better comes along, rather than playing roulette with my credits.

[hollow vector]

@stuck junco sure can still trying to figure out how to use discord. I will message you

[north plinth]

general chat on discord to see if others had noticed

[sage needle]

is there a link to filters, or words that get flagged? One of my projects keeps getting flagged but it doesnt say why?

[north plinth]

I used to be able to that I'm aware of, must be a recent thing, I don''t download many so was surprised to see it, especially when they're my songs

[near flume]

I wouldn't worry about what can done in anything below v4. All the older versions are being removed in the next few weeks. They announced it shortly after v5 went live.

[north plinth]

what about the free users, does that mean they will have to sub?

[polar spade]

@near flume @north plinth

Fact check:

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

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

[polar spade]

BASIC members still have 3.5 and the 2 added 1 minute 5.0 sample outputs now.

[sage needle]

3.5 does some wild and fun stuff when you feed it Esperanto

[still cargo]

hi guys. Is it possible to recreate the absolutely same verse (with vocals) but with changed lyrica? v5 of course. Right now I'm trynna do it in Editor and always get completely different melody if change lyrics. It keeps the key and style yeah but the melody of verse 2 doesn't match verse 1 in that case

[naive ibex]

I'm confused — are versions 4.0 and 4.5 frozen, or has someone updated them with new features (like in Studio)? I'm using the same style presets I used a lot before, but now the output looks completely different in several styles. Even when I pick the old "not in beta" models, the results don’t match what they used to look like at all.

[gleaming solstice]

Anyone had any luck figuring out the most persistent way to get the engine to render out exact input audio? Or atleast staying really close to the input? Simplying putting audio influence to 100 on the cover feature isnt super consistent

I have had some decent luck with splitting my tracks up until horizontal slices and feeding them to SUNO 1 by 1 and then mixing them all back together

wondering if there is any better way lol

[quiet solar]

If you have Suno Studio, that may be possible. I've yet to have the opportunity to try it out, but splicing is the kind of functionality it's intended to support. Splicing between songs is a bit more tricky if you have to correct for tempo and pitch (which might be why they used drums, to avoid the pitch issue). Without knowing how it does the adaptation process, which presumably involves some sort of regeneration of the sample, it's difficult to say how simple or tricky it might be.

[quiet solar]

There's currently no way to regenerate an exact copy of the input audio. Audio Influence 100 can be a bit excessive - I find the sweet spot is more typically in the 70% - 90% range, coupled with mid-low style input (e.g. 30%-40%) and low weirdness at 5%-20%, but your experience might be different.

This also works a LOT better on Extend than on Cover (which is specifically intended to change and adapt the underlying track), but there are times you may have to bite the bullet and use Cover anyway depending on what it is you're trying to do. The more complicated things are going on in the song, the less likely Suno is to perfectly replicate it - which might be why you've had more luck by splitting into stems first.

[gleaming solstice]

Im a classical / media composer, so I am mostly playing with ways to basically AI master my work.

Varrying degrees of success so far

Sometimes it works really well and can interpret my work exactly how it is written and it spits our a super realistic sounding AI render of what I wrote. Other times it just basically writes it itself lol

Thats why breaking up my tracks into slices and rendering them then mixing them back together works decent, as long as I tell it to be specific to the BPM and I dont have any crazy tempo changes it stays pretty well vertically aligned. A bit of post editing where needed and I get a pretty amazing sounding render

Its just that the time it takes to do that is probably roughly equivalent to the time it would have taken me to just mix and master my track in my DAW like normal lol although sometimes the SUNO instruments actually sound better than my literal sampled library which is neat but also a little crazy lol

[quiet solar]

Recreating the same underlying melody with new lyrics is definitely possible across any of the current versions. Recreating the exact same melody arrangement is much more tricky.

It can be done in the Editor with a fair degree of trial and error, or by splitting out the vocal/instrument stems and using Extend to create a new verse in which the edited lyrics are sung, then switching the edited lyrics over in an external audio editor like Audacity.

[gleaming solstice]

An AI powered plugin that I could load into my DAW that would just render all my MIDI data with AI would be amazing and I would pay hundreds if not thousands of dollars for it

[quiet solar]

I honestly feel your pain there. xD

If you're trying to replicate, it might be worth paying very close attention to your style prompting - Suno's default prompts that it attaches to uploads to describe them often lack useful info about the song when it's being generated. Especially if you're aiming for it to stick very faithfully to the original upload, combining high audio influence with a good prompt may help reduce the number of times Suno wanders off to do it's own thing.

[quiet solar]

One day, maybe we'll get there. Who knows. xD

[gleaming solstice]

Yeah funny enough with V5 I noticed the prompt it makes automatically for input is getting better and giving me a pretty good output if I just feed it back through

Like this is basically exactly what I wrote:

I think the only thing it missed is around 2:55 the little trill in the string is a major 2nd instead of a minor 2nd but otherwise this is basically exactly as my raw input was (in pieces of course) but it is using the suno instruments

It sounds as good as anything I spent weeks working on mastering and it took me like a day to get done thats a huge time saver even for how much of a pain it was to slice it all up and bounce it one by one

[amber karma]

I want to know am I able to use this app with my own music I crested and still use the ai voices you have?
My music style is nowhere in your app it is basically rap, rock, crap like that. I have a entire different genre of music.
So I would want to use my own music and create a song with ai vocally to add with my much.
If so how do I do such.

[half dragon]

Use upload feature, then Cover or Extend.

[amber karma]

Thank u I'll try that

[hollow elm]

Welcome.. Does anyone know how to change the key of the song (cover) because when I type in the description I want e minor key, it produces a song with the key of the cover

[ashen carbon]

my friend turned me on to this VST called Synplant last week and it's a game changer for production with suno now that we have midi export. i have no financial incentive in sharing it here at all btw, just found it super useful: https://www.youtube.com/watch?v=rN7B8LUY3WU&t=1243s

in a nutshell it lets you recreate any synth sound that Suno generated as an actual playable synth. check out the video, it explains it much better than I could

[halcyon patio]

Anyone know the v3.5 issue?

[lofty merlin]

Does anyone know how to get Suno v5 to sing backing vocals when they could ambiguously be instructions? I have a song where the backing vocal is (Whispers in the dark), but v5 keeps skipping it entirely, presumably because it thinks I'd like it to whisper instead of saying those words. If I remove the parentheses, it sings them as though they're main vocals instead of backing vocals.

[glad perch]

[Style: Genre, Vibe, Tempo, Key Artists/Sounds Like]
[Instrumentation: Core instruments and sounds]

[Verse]
(Main vocalist's lyrics here)
Vocal Style: (e.g., soft, breathy, confident, rapping, powerful belt)
Background: (Describe the music and any background vocals here, e.g., "driving drum beat, subtle synth pads, occasional background 'oohs'")

[Chorus]
(Main catchy chorus lyrics here)
Vocal Style: (e.g., anthemic, layered, euphoric, gang vocals)
Background: (Describe the music and the specific choir/chorus here, e.g., "soaring electric guitars, heavy drums, a full choir singing the harmonies, layered vocal stacks")
Chorus Type: (e.g., "Big, reverb-drenched gang vocals," "Tight, harmonizing trio," "Call and response with a female backing vocalist")

[Verse 2]
(Lyrics for the second verse)
Vocal Style: (e.g., more intense, melodic flow)
Background: (e.g., "bassline becomes more prominent, same background 'oohs' as verse 1")

[Chorus]
(Repeat chorus lyrics)
Background: (You can add a note like "even bigger drums and choir than the first chorus")

[Bridge]
(Lyrics for the bridge)
Vocal Style: (e.g., stripped-down, emotional, spoken word)
Background: (e.g., "music drops out to just piano and vocals, build with strings")

[Guitar Solo / Instrumental Break]
Description: (e.g., "face-melting electric guitar solo, wah-wah pedal")

[Final Chorus]
(Chorus lyrics, maybe with a variation)
Background: (e.g., "full band, maximum energy, choir is the loudest here, ad-libs over the top")

[green walrus]

Tysm for the details, really helps construct the song from start to finish

Also it won’t let me remaster a song because it says there’s copyrighted lyrics even though I WROTE them and it let me make the song in the first place during v3 or whatever

💀💀

[sage needle]

wait thats a format that works?

Like you can get that specific with it?

[glad perch]

thx for the info, looks like some kind of redis issue....

[glad perch]

heave you tried to change a bit the lyrics?

[near flume]

[Whisper: "whispers in the dark"]

[lofty merlin]

Nope, that didn't work; it still skips the line 🙁 I even tried using [Backing vocal sings: "whispers in the dark"] to no avail

[lost ledge]

I've personally had the most success with using [these brackets] to describe rhythm, tone, direction etc. and using (these paranethesis for backing vocals)

[lofty merlin]

Yeah, that's what I had originally that I couldn't make work 🙁 I guess I'm just cutting that line

[lost ledge]

Are you doing a cover or is this the first creation of the song?

[lofty merlin]

First creation

[lost ledge]

You may want to mess with the generation sliders a bit. Try lowering the "Weirdness" slider so it follows your instructions more closely

[green walrus]

I haven’t because I am quite satisfied with the current lyrics but suno won’t even tell me which lyric is copyrighted so I don’t know which one to change

[lofty merlin]

No success 🙁 I tried using (Whispers in the dark) and bringing the Weirdness setting all the way down to 30%. Still skips the line every time.

[lost ledge]

Hmm.. Have you tried using the editor to just replace the section until it gets it right?

[lofty merlin]

A bit, but after a few tries and failures, I lost patience for it 😅

[knotty lynx]

play!

[gleaming ferry]

ho

[cosmic lintel]

Hi, does anyone know how I can recreate this music? Or if someone could do it for me?

[twilit comet]

Is there a reason I cannot make a persona based on a song I made with my persona or is it just because it was made from a persona?

[solid crater]

🤣 was scratching my head on this for a bit.... Looksl like when you CREATE a PERSONA, it binds it to the original song 'as if it was created from it'. Thus, the song technically was created from it. There's some recursive logic in there, but it made sense for at least 5min.

[frigid pivot]

Is there anyway to get my instrumental songs longer in v5??? All of my songs are songs are around the 2 mins mark or less. I don’t get any longer songs like in v4.5? I would to get some tracks between 3 to 5 mins, if possible.

[livid crescent]

I've had decent luck using **** Lyrics for backing vocals and then another set of asterisks after

[pearl torrent]

https://www.tiktok.com/@csv.cria/video/7556172038063344908?is_from_webapp=1&sender_device=pc

[frigid willow]

Tried this and it only extending my song by like 15 seconds not worth the credits

[trail fulcrum]

It's a shame that Suno5 doesn't transfer the original uploaded audio 1:1. I've now uploaded several audio files with vocals, but Suno 5 hasn't transferred them 1:1 from the original. Suno needs to work on this so that the original vocals/instrumentals are used.

[devout minnow]

Yea just a 15 seconds isnt long at all, i had a quick go generating an instrumental yesterday putting 8 minutes in the style box and didnt get it, the extensions i got were about 1 min 49 secs whereas the instrumental track itself was 1 min 44 secs

[brisk flint]

I was pretty successful yesterday with V4.5+ creating tracks between 3:50 and 7:59 using this approach, but note, for V5 this approach did fail: https://suno.com/s/69j3sdoo9Tj22kG9 - Feel free to check my prompts.

[round bobcat]

I hope v5 will get a better audio quality. It really sounds really dull on several of my loudspeakers. Annoying. And I dream of a filter which removes all shimmering of the old songs as doing them again with da new model changes their character too much.

Song lenght depends on the server load - my observation - which means the time of the day.

[rose zinc]

Hi there IceMetalPunk. try using ..(.."..Whispers in the dark.."..), I think by adding the dots before and after the spoken word, you might get it in.

[near flume]

build out a structure in the lyrics box using tags as you would for a regular song ie: intro, verse, drop, breakdown, etc. Also put some descriptive tags within each section. Here are some that I use [Atmospheric_Wash][Break_Patter_Complex][Groove_Burst_135BPM]

[halcyon sundial]

extensions arent working after 10 min?

[summer garnet]

The sound quality isn't bad for me and I have it playing on the Bose. It depends on what is generated. I find that extracting the stems makes it lounder and a bit strange to me at times. I try to avoid stems extraction if I can help it at the moment. Not sure if it is only me on that.

[summer garnet]

I agree that some of the generations does have some dull sound at times.

[round bobcat]

I mean v5 The older versions don't have this problem, but the shimmer - more or less.

Interestingly the songs start in a bad quality and get better after a few seconds. Like here https://suno.com/s/fRUowb6Fz759XPIQ

[main crescent]

I heard there was a community driven bark voices library or an audio prompts channel for Bark voices but im not seeing it anywhere, is there a specific Bark related Discord Channel?

[hard beacon]

bark or spark ? 😄

[raven turtle]

Anyone else finding that stem extraction in Studio is better than Suno or splitting outside the app?

[main crescent]

Bark, I need voices for an audiobook progect im working on

project

[hard beacon]

godverdomme.. i am so sick of this "oh oh oh" intro vocals... is there a negative prompt about this? 😄

[round bobcat]

You can rob the voices of your family and friends with F5-TTS.

[lost ledge]

I've found that my songs were also being created at the 2 minute mark on V5.

My solution has been either:

1. Just making the song longer by including more prompts for lyrics / insturmentals

2. In the style I try to make mention of how long, or possibly even altering the BPM so that it needs more time to get through the whole of my song

3. Probably the most effective, is using the Editor to extend the song or right clicking a song in your workspace and choosing "Remix/Edit > Extend" and then adding segments to the end or in the middle to manually lengthen the song

[neat yew]

Has anyone had luck with creating backing tracks for guitar? No matter how specific my prompts get I always end up with some solos being played throughout that cant be separated in stems.

[polar spade]

You can test and use a prompt such as "ambient" to see that Suno heavily gauges a songs length by the genre it thinks youre trying to emulate. This was true in 4.5 as well, but perhaps 5.0 has a different "alggorhythm" for us to work out.

As others have recented posted, the use of spacing and some well place metatags can give you length as well from the lyric box. The crux might be this needs to be done with NO lyrics between the metatags and spaces, then hope you can Cowver with .lyrics without getting truncated again from attempting a Cover.

First things first, try tweaking the genre by using "ambient", New Age", "orchestral" or othe genre which are typically long winded, placing them near front if possible but almost anywhere thing the stytle prompts.

Counterintuitively, as well, removing prompts can, in some instances, lengthen the outputs.

@lost ledge suggestions are spot on as well. #📚┃suno-school message

[tranquil breach]

https://suno.com/s/Ww1kQNNe2AdApV70

[polar spade]

For me, I believe setting any slider to 100% is like giving Suno free range to do what it wants. So you got disco😀

[tranquil breach]

but it appeared to be sweet ..........might get surprise if you listen

what woody

you called me a scammer?????

how dare you

[radiant breach]

lets play a game... what do you think Suno did with this in 6 generations?
A. Multiple guitar riffs to open the song
B. A quick guitar solo, followed by the first verse
C. Straight to the verse, forget those riffs.

[round bobcat]

It is not a good idea to set weirdness to 100% - I got this: https://suno.com/song/fd8158e7-e9d1-410f-a9e3-9dee5aae071f

[tranquil breach]

you done with 100% weirdness ? cool it is !

uuh but

it got messed in the end, may i try your prompt?

[round bobcat]

100% Weirdness, 81% audio influence and 21 Style influence - sounds like aliens on the radio

Ah, you ment the other guy

[tranquil breach]

shall i remix and give it a try?

[tranquil breach]

A?

[round bobcat]

not yet - as it a project in progress.

[radiant breach]

https://tenor.com/view/false-wrong-fake-incorrect-untrue-gif-23252753

[tranquil breach]

don tell me poor SUNO skipped them all lollll

[radiant breach]

yeah

[tranquil breach]

lollll

[radiant breach]

[intro guitar riff]

[guitar solo]

[interlude]

[break]

[verse]

this worked better in my simple 3.5 prompting brain

[tranquil breach]

v5 got too personal

lolll

v5 got its own temper i guess

[radiant breach]

yeah, but I thought being distinct in there being 2 different riffs that v5 would be smart enough to pick it up

[tranquil breach]

try 100% style and 100% lyrics?

0% weirdness

[round bobcat]

My current game - cover v5 songs in older models

[dusty drum]

Hello everyone, why does this error message keep appearing in Suno when I try to upload my own audio sample and then create my own beats with it? Can someone help me? Thank you very much.

[unique stone]

Your "Extend from" time is 0:00.00 which is likely why you get this error.

If you want to continue from what you uploaded you need to change this by typing a time in the box (maybe 0:06?) or dragging from the left of the waveform.

[dusty drum]

I left the timeline as it was and started from 0, and then, unfortunately, the same thing happened. And if I deactivate the pink area, the marker, the sample isn't processed in the beat.

I used to be able to extend the samples without any problems, but since the latest update it is no longer possible.

[unique stone]

What happens if you type 0:06 in the "Extend from" box?

(it looks like a 6 second sample, this is trying to extend from the start i.e. nothing)

[dusty drum]

then only a small piece of the end is used and not the whole loop

[tight zephyr]

Hi, I am curious to hear if users really hear a difference between MP3 and WAV files? Because if tehre is a lo quality audio generated by the Suno model, there is not going to be much difference between MP3 and WAV.

[near flume]

You can hear the difference if you're doing clear, high range music (acoustic guitars, classical music). The mp3 tends to sound a little dull.

[solid swift]

GM so in V5 im trying to make an instrumental Hardcore Gabba type track, with chopped vocals etc, i get the vocals at the start how i want it and in the middle but the track always ends up so short like 31 seconds or 50 seconds how do i make a longer track like 3mins?

[rose quiver]

Is the yearly special really over? It took several days to transfer money to my card and I go to purchase and it shows the normal price.

[polar spade]

Theres the present promotion for everyone of 20% off annual, which they had last year and this.

A couple others specific ones ( mostlyt for new users and invited) which ended Oct 5th.

[gilded mortar]

Hi guys how are you. When I'm trying to upload a song from my phone. The uploads stops and a red alert pops up say this song already exists. What does that mean? Even though I never uploaded it before

[rose quiver]

Thanks for answering. I was looking forward to a year at that very low price. I have to stick to the 2500 credits per month.

[polar spade]

It shouldnt be about an audio already existing on Suno in your library. We should be able to upload the same audio as often as we wish.

It may be recognizing a melody or pattern which is similar to something in its censor data as an already publicly available audio/pattern. And THAT can include if ytour own audio is on YouTube, for instance and is listed with Content ID.

[gilded mortar]

Thank you for the reply. So is there a work around or a fix ?

[polar spade]

Is it an audio of an already publicly available audio (?) So it might get censored as something already existing , or just something similar to you, say, singing on a mic or playing piano/guitar at home?

[gilded mortar]

Yes its an ost from TV series to be honest

[hollow hare]

#📚┃suno-school message

[polar spade]

Its simply recognizing that then, it seem. Suno's main purpose is to avoid any audio which could be controversial or risk copyright infringement, but also to keep repetitive "public" sounds and text from being used.

Thats a good thing over all, to keep us creative and come up with our own designs and melodies rather than just throw other's art into the crankcase.

You might even come up with a far-from-similar melody if you have a simple keyboard and try that to see what you get.

Sorry for it being inconvenient to our creative flow, though, its cool to think "what if" and remix other things, but in fairness, we should get permission , right?

[gilded mortar]

Thank you very much for the help Mr dasvon

[polar spade]

If you can come up with a melody in your head that doesnt sound familiar to anything you recall, then ask for help here to come up with some similar prompts to get that "vibe" then you can be amazed at how you can create something not exact, but has that same vibe as the TV Series. Thats always cool too.

[summer garnet]

One thing I have noticed with the song editor is that sometimes when I regenerate an alternative with the replaced lyrics, it bugs out and removes the newer alternatives. The only way I can use or go back to it seems to be when I go back to the sidebar where they are and click on them. Its kind of frustrating that happens. I'm not sure its because of something I did (the song is chopped up by its components: verse1, pre-chorus, chorus, etc.)so sometimes some of what I want to change with lyrics ends up bleeding over into another section and I have to highlight it.

Anyone else have that issue?

[twilit comet]

I am saying I cannot create a new persona from the already existing persona

Song A: Created, turned into Persona
Song B (New): Persona used, song made
Wanted to make a persona from song B but could not

[half dragon]

What if you opened song B in the editor - then hit Save as New Song which will create song C --> see if you can create a persona from C

[twilit comet]

I'll give it a go.

[quick sonnet]

So just as an example would like to make something like the chorus in this track, based on some other one. But defenetlly not going to use this one or any of its lyrics, or the ones of the original. https://www.youtube.com/watch?v=V6bIFWT8L_o

[hallow grove]

Anyone having issues with extracting stems. Been sitting here for 30 mins and it just keeps spinning.

[half dragon]

Try to leave the page, then go back in - they might show up - It could be a refresh issue.
Stems extraction happened on the server so you don't have to sit there and wait.
But be careful not to waste more credit to extract them again, use the VERY small link that says "See extracted stems"

[quick sonnet]

If you are a Pro user should have no problems just downloading your track and then you have many posibilities for extracting stems.

[hallow grove]

Thanks yaall im a pro user, pulled track into studio and I’ve tried logging out and refreshed a couple times but no luck. Do I have to download it to my local first then upload?

[half dragon]

You don't have to download first - they will be extracted on the server.
You may have experienced some hiccup or load on the server - check back in later times.

[quick sonnet]

There are also other tools than suno you can use to extract stems. As a free user I had to use that tools and the rsults were not as bad, but now after suno has blocked download option for free users, can't do it any more.

[restive turtle]

good morning folks I need help

what

I hate you woodhouse

ugh anyways

https://suno.com/s/W4i5XKMR3OK35Ijj

so tldr I just want the "I s*kk at math" part from the intro to sound the same just like how the AI delivered that same line throughout

is it possible to fix it without remixing and editing it manually?

coz if you listen to it, the intro messed up the line but it was okay on the other parts

[viral pond]

My suno library is clogged with sound effects now after i made this thing https://suno.com/s/yCoV1EIGdo3qhHXy 🤣🤣🤣🤣

studio really made things easier, so that i would not needed to download a song then reupload it if i needed some atmosphere ot it, but it kills the remove/cover origins, persona and style prompts

[sweet flume]

Anyone got any tricks for Remastering a track to have a wide and more defined sound signature? I'm dealing with Bass Music and I can't quite seem to wrangle it into something that sounds good.

Would be cool if you could also prompt styles and instructions for remasters so you can explain how you want Suno to change it.

Anything I'm not thinking of in Studio, or whatever tricks? I've tried distilling into 4.5, and then remastering for further refinement back into v5 does some pretty good results, but just caaaaaan't quite get it.

[ebon crest]

can someone help me understand how to force some lyrics to be male and some female? none of the tags i use work

[half dragon]

There will be no reliable way to do it. Either:

• keep trying
• use editor to edit each section
• use studio (if accessible)
  None of these ways are reliable either but you could get lucky. Who knows!
  Point is, be patient and keep at it.
  Making duet is challenge and not cheap, but like anything else, the result is rewarding.

[ebon crest]

this is infuriatingly frustrating. Thank you! I got a version of a song which is fantastic until like the end and then it just veers off and ruins the whole thing.

[placid mesa]

I bet the easiest way would be to cover the track with a persona that is a different gender and sounds the way you like, then separate the stems on both tracks, take it into the studio, and just alternate the vocal tracks by turning sections off and on.

[lost ledge]

^

Was literally just about to say that lol

[ebon crest]

unfortunately don't have access to studio 🙁

[lost ledge]

You could use a free program like Audacity if you can get both versions of the song and just download their WAV or their MP3.

It'll require a bit more tuning and tweaking though, since the instrumentals may differ

[placid mesa]

Ya. You could totally do it in any audio editor if you pop the stems out. Then you can pick the background instrumentals and beats you like before studio is available. Hopefully it won't be long before everyone has access. 😄

Does anyone know if it's better to generate the stems on a track before popping it into studio or just generating the stems in the studio? I'm having an issue where the stems are taking forever to generate, and I wonder if it's because I'm actively doing it in studio.

[half dragon]

I did it both ways and saw no differences.

[placid mesa]

Alright. I just need to be patient, then. Thanks!

[quiet solar]

If you take Prils' suggested route and use Audacity, I'd recommend splitting out the instrumental and vocal stems - then pick the instrumental you prefer and use that throughout, and only switch the vocals as needed. It's a lot easier to adjust the timing of the vocals to make sure they match up than it is to blend one instrumental track (or combined vocal/instrumental track) in and out with another.

However, stitching together two (or more) versions in Audacity is much easier than trying to get a perfectly clean run of male and female vocals switching over consistently, in all the correct places, in a single generation in Suno.

[fringe swan]

/chirp

[flint haven]

The /chirp command is no longer working

Chirp is retired, and no longer on discord. You can make music now on our website: https://suno.com

[woeful sand]

Try giving a sample dual channel. Same sample twice, 1 max right the other max left 😉 make sure you align them of ofcourse to avoid getting echo instead of wide stereo.

[sweet flume]

Manual panning to create a stereo effect, yeah, though I don't see a means of automating that, is there?

It'd be great if I had an LFO to work with, then I could draw out the panning if I've gotta do that manually, but there are other things than just a stereo effect in itself. Just simple clarity in some cases and sections as well as how the sounds are manipulated. Sometimes Suno will generate a pingpong effect which is really sweet in some sections with a remaster and sometimes it won't with others, or the tone and sound style is very different, but still the same source sound, so I'm trying to figure out how to wrangle it. Even if there'd be a means to "Copy Master" from stuff already generated by Suno, that'd be great, but currently I'm evidently stuck.

Though mind you, the music I'm generating is Bass Music which is very effects and distortion heavy and aggressive.

An example of what I'm looking to refine, just getting Suno to recognise the background bass sounds took 2k credits worth of prompts to get it to this state which was sweet as hell, but I'm open to any suggestions at all.

[woeful sand]

Automating it would be a real treat. In Studio you need to Tinker with the layers. Like left fade in, right fade out and a third center fading in and out at the same points to get a swoosh effect. A bass drop would be vice versa for the first two layers i guess... All in all a lot of work. Studio isn't nearly as versatile as it should be compared to other similar software. No effects like reverb etc. It still depends too much on what samples are generated alas. Although it gives more flexibility in creating a track it is more a mastering tool then a creative tool.

[sweet flume]

I think I've figured something out for myself in the meantime. I noticed with the Stems you extract, you can remaster the specific stems if you want which is potentially interesting. I'm gonna tinker with that a little bit and see if it gives me results.

Might be able to just sew in bits and parts for my use.

[woeful sand]

Let me know how that works out for ya 👍

[sweet flume]

Will do. So far Studio's fighting with me by sample length. Figures the remastered samples are longer and slower BPM than the project which is tweaking me out so once I figure that out, might be cookin'.

[woeful sand]

There is a follow tempo button. Click on the sample and it appears on the upper right. I usually use the main track, they should align then

Just need to tinker with it

[sweet flume]

Yeah, it seems to set automatically, but doesn't snap to it. Oh well, I'll figure something out.

[woeful sand]

Good luck, you'll get the hang of it👍 😉

[sweet flume]

Many thanks.

[sacred flax]

Hey, I'm looking for someone who could make music/backing for a text I already have. Am I in the right place? I don't have money to pay, so is anyone willing to do it for free? The entire text is in Polish, and I will give more information in private.

[bleak furnace]

https://suno.com/s/H20K2b0N4PxQ7bSy hello everyone, does anyone know what genre of music this is, if possible give me some suggestions about prompt, thank you very much experts

[prime blaze]

Hi. I use Suno to create instrumentals. However, the output is only around 1 minute 30 seconds, or 2 minutes 30 seconds at most. How can I make the beat or the song longer than 3 minutes?

[jagged eagle]

hello can i have identify cods to log in beacasus i alresy used me 3 free cridts

[near flume]

You need to bulk out a structure in the lyrics box. Suno doesn't completely understand time or durations so that doesn't work

[restive pebble]

Forgive me if this has been addressed somewhere on here but what gives with the stems combining things such as Guitar and FX. On many of my stems the AI combines these or other instruments with either the Vocals or FX rendering both track useless. Is there a known workaround or resolution for this?

[lyric kestrel]

Can you make the vocals lower if they are too high somehow?

[tranquil breach]

Tune them in DAWs

[polar spade]

It could certainly fit within " New Age" music.

[polar spade]

If there is a fainter but noticeable instrument on the vocal track, that is called a bleedover or artifact. As suno creates generative stems, this shouldn't occur. However, garbage in-garbage out, as they say -- there may be times when something may be perceived as a vocal -- it had high timbre, vibrato, distorted, etc-- and so gets tossed in with generating a vocal stem. This could be the case (?)

I have used suno or a third party app and re-stem that track out to see if the instrumental could be recognized and separated out. Not successful in one try often, but personally I dont factor in credits much when looking for results, so, often successful eventually.

Another try is to Cover the "mixed" vocal as another vocal or prompt for vocals with the least instrumentals, such as Cover the vocal as an acoustic piece, or a cappella. Then run instrumental/vocal stemming to separate out the vocal again. This may give a vocal thats far from your original, or something useful.

Another is to use multi-stem as before , but go back and just use vocals/instrumental stem separation, to see if it yanks the guitar or other back into the instrumental half.

Fourth, is to use a third party stem separator. Its good to have a backup in your back pocket. See if any different. If it works, then you have work ahead of ytou, to upload and blend back in as best possible. Or at least patch all the tracks offline first then upload to Suno for further.

[polar spade]

Alternatively, you could remaster the original output once more before multi-stemming, to let Remaster make all instruments more clear, so they separate out better as the stemming tracks.

[polar spade]

Lower in pitch or lower in volume?

Volume: in Edit UI, the editor, , you can select a section or all, and volume control is in the sections menu choice. In studio, volume of a track is also a slider in the track panel controls on the left of each track.

Pitch: You could adjust speed down, but dont toggle "keep pitch" . That will lower both vocal pitch and music. Then click "keep pitch" and re-adjust speed up an amount equal to match the original speed.

[tepid moth]

I did notice a new welcome login to studio today, so I guess there have been some improvements.?. No idea if it is because I reuploaded a song after I worked on it in my DAW and I get much cleaner stems because of that, or if it have been a major upgrade too it. First time I tried to upload a song where I play much myself (using my own sounds) and then uploading that version to Studio. The guitar and bass stems been very good and clean, and actually usable. Especially the guitar stems (becides one artifacts that I can remove) been top notch. If it is because I uploaded a clean version of the song, then I know what my workflow will be in the future.

[bleak furnace]

thank you 👍

[polar spade]

Studio is in beta. So everyday is likely different (as in improvement or sometimes just a change) 😁 But I do see the "welcome to studio" or " Try V5" banner frequently, especially on a log out/in scenario.

Yea, the cleaner , crispier, widened, more separated upload, the better. But its still amazing I can give it a crappy-singing-with-a-room-fan-on rendition and it come back with a cool version.

[restive pebble]

Appreciate the direction.

It's not faint it's in your face audible FX. I don't mind slight bleed over even though I'm new to mixing and mastering, have to learn by doing. My lyrics are tight, properly tagged and the styles do not force the timbre etc.. a occasionally add a small vibrato but not much. I do notice the stems have serious solo sound issues for volume, instrument sound etc...An example is Bass, it comes across as muted and low volume but when mixed into a song it sounds as expected. Makes it a challenge to find the sound. (whole different topic)

Credits aren't an issue for me either. I don't mind using them but I wanted some direction before I blew more needlessly. I didn't think of using the Cover, but have tried alternate stems with no success, however I haven't tried just the vocals\instrumental separation. I'll also try the cover and see what I get.

External stem separators aren't a bad idea if suno fails with the suggestions. Either a desktop DAW and Studio will be for reassembly. Studio seems more automated but limited and the DAW makes me earn the mix\master which I like.

thanks again for the guidance.

[restive pebble]

That too is a good idea. 👍

[polar spade]

I finally am working with Fadr at the moment for the stems that Suno still bleeds/mixes some sounds into. But even Fadr, for instance, could not separate a cuica i could hear in an FX stem, I guess because cuica are mostly a squeaking noise.

Back when stems first were introduced, as many complained of artifacts, many of us would just Cover the vocal stem again, which changed the vocal tract to either vocals wifh extra ooos and ahhhs or vocals with cleaner discernible instruments. Then re-stem to have just vocals. This worked quite well when it stemmed separated by frequencies. Anyway, sometimes "old school" still works.

[restive pebble]

Old school in age but not so with DAW's or music in general. I'll play around and see what comes from it. Statistically it'll have to separate them eventually.
In truth, none of it matters as I'm using Suno to bring my words to life, probably like millions of others are and I do not expect any of my songs to 'hit' or give me anything more than a few likes. I'm more interested in learning the art\skill of mixing and mastering just because I have the opportunity. Suno does most of it inside studio so I get quick results but I enjoy the manipulation of the tracks in the DAW. Tedious and a challenge, hell yeah, but I learn and hear subtle differences.

What about when suno add in obvious profanities to the background adlibs? Does suno permit publishing of those? If I recall profanity was forbidden in lyrics.

[polar spade]

The jury always seems out of that. I have no idea. Sometimes , i think some things are allowed "in the name of both 'art' and 'inclusiveness' ", whereas other songs may get censored for seemingly having no "substantial worth" in regards to its use of profanity or slurs.

TOS notices their policy is to censor outputs which are "offensive" to Suno and/or users. I dont know yet how that is gauged.

[digital kraken]

Since V5, I've noticed that most of my tracks have intros with (often disgusting) vocals from the singer.
Do you have this too? And do you know how to avoid them? (Even when you carefully specify the type of intro and the instruments, it almost always happens.)

[polar spade]

That seems to be a thing more v5. I think we're all looking for an option or to prompt around it when its not wanted.

Ive tried things like putting a double-intro, or elongating or forcing some intro vocals, to be cut out later. Nothing has been consistent enough to suggest in here yet though.

One of us will figure outsomething, im sure.

[polar spade]

Its kind of stereotyping, for example, that all hip hop genres should have "uh yea" in the intro, but its recently too often to be by chance. I agree. It comes close to "appropriating" a cultural style to a genre when most genres are much broader than that.

I feel it would be better to ratchet that back a bit and let the user put in interjections if they wish.

[polar spade]

V5, relatively, is much stronger to pushing the stereotype of a genre. Thus the grunts and interjections before a hip-hop song, as I mentioned.

But as an example, songs i once put "gospel" in the prompts for a mild vamping and minor key vibe, now all have a chorus of a church choir backing if I cover and keep "gospel" in the prompts. Recently im having to remove prompts to avoid these stereotypical in-your-face outputs instead of using them as nuance and vibe technique prompts as i would have before.

A whole new learning curve. Im used to it and ultimately dont mind, but its a 1-step forward, 2-steps back process for awhile -- them full speed ahead once things become rote.

[torn spoke]

I've had success with going "-like" to lower the effectiveness of the keyword

as in, gospel-like

[polar spade]

Oooo... good idea. Ill try that as well. Thanks.

[torn spoke]

Also, if you want less stereotypes of genres, specify the area and decade of the sub-genre/style you want

[digital kraken]

In the lyrics or in the prompt ?

[restive pebble]

I've been having some moderate success using descriptors such as R&B with a whisper of gospel OR R&B sprinkled with funk and find the output is not so over the top or in your face.

[torn spoke]

prompt

[slender rapids]

is there a duo tag ?

[torn spoke]

you mean duet?

Just describe what ye want

[digital kraken]

I do this, but it doesn't work.

[Intro]
[Four measure instrumental intro with fast electro-swing beat and walking double bass]
[-like vocalise, voice, scat, choirs]

[remote mountain]

guys how can i start with music/

?

i want to make my own music

[slender rapids]

Basically, I'm wondering if there is a tag for duets. The ones I know [choir] allow several people to sing, and I'm wondering if there is something similar for a [duet], where just two people sing in the music.

And I also have another question about persona: if we choose our persona, can we make three of them sing? And how do we use them?

[graceful condor]

Man if I want SUNO anything to do finally right it is not misspelling words, mumbling words or ignoring words and or just singing something that is not there at all. I hate this so much and it ruins perfect songs just because of one damn word misspelled. Is there any work around it? Because now with V5 even Remaster is not fixing it anymore. Before you could remaster and in most cases if fixed the misspelled word in the new version but now it does it again at the same spot. I freakin hate this so much...

[jagged oar]

if suno mispronounces a word a lot, type it phonetically until it makes the sound correctly

each generation you do will maybe ignore that and say it even wrongerer so you may have to isolate that section and regen just that section until it says it correctly

[graceful condor]

the problem is that if a song is perfect and just such a word is spoken wrongly it really ruins the song. And even if I edit it in the Editor by and try to fix that small section the generated part never or hardly ever sounds suitable. Either the sound is totally off to the rest which mean not loud enough or muffled. you can clearly hear where the selection was regenerated but it does not sound clean or totally off. I wish this function would be more advanced within the editor. First of all it is way to expensive in credit costs just to generate such a small section and like I said it sounds literally like trash in 99% of the cases. Sadly.

[jagged oar]

thre is a heal feature you can apply to it that iirc fixes that issue

[graceful condor]

Someone mentioned before this "heal" function. Never seen it. What should this be?

[polar spade]

This is a basic way to start a song, by just describing what you wish https://help.suno.com/en/articles/2462273

You can then use the custom tab to be more creative.

[torn spoke]

It's usually a sign of sub optimal lyric structure if it has issues following them

[torn spoke]

Unless you mean with the new vocal track generation. Cause that's completely broken.

[graceful condor]

You mean V5 or what do you mean with Vocal Track generation. Honestly I do not see an issue with the lyrics.

[torn spoke]

I mean the "regenerate" function in suno studio.

Cause it doesn't work for vocals.

I can't see your lyrics so can't really diagnose anything, but jumbled pronunciations usually happen with mismatched syllables

[graceful condor]

For example at the end of a line is the word "cracks" she sings something like "hacks"

And german is a mess in V5

[fading glacier]

Is there a way to use prompts to edit songs after creation, (ie faster bpm, or changing tone etc) or is that what studio is

[torn spoke]

Interesting. I've never actually tried to generate anything in my primary language, I should give it a go

[graceful condor]

I do all kinds of languages even ancient ones like ancient Egyptian, ancient greek, different native dialects, but mostly Japanese, Chinese, Korean besides Englisch and yeah some German because it is my mother tongue.

In japanese the Ai also sings some stuff in many cases that is not even there. The good think at least with japanese is that the "other" stuff at least makes sense and sometimes is even better. But in English the Ai just kills words, butchers them or does not even sing them. Or repeat stuff that is not even part of the lyrics anymore especially in extends. But which really bothers me the most is the "misspelling"

[torn spoke]

I wonder how good the nsfw filter is in Norwegian

Damn. Ran out of tabs in my google doc

[torn spoke]

wow, it's really good at the oslo dialect at least

[tepid narwhal]

I dont know whats more wild in this picture the fact you have a google sheet with that many tabs or the fact your keyboard is hexagons and not in a standard layout

[sour dune]

Hey is there a way to have some layer to the vocal? i mean some chorus that sing with the main singer. and not replace him.
if i put ( ... ) the chorus sing. but i would like the main singer to sing, but some background singer to layer the vocal.

[lost ledge]

I would try adding the layering instruction to the lyrics prompt and then using paranthesis to denote the background on the same line as the main singer. Would look something like this as an example.

[Chrous]
[Background vocals overlap]
Yeah, I'm singing a song (singing a song)
These words are words (Words are words)
Look at all these words (So many words)

You may also find more success with different versions. If you use the Version tab on the creation page you can try generating songs with different versions and find which one most accurately reflects your aesthetic choice or is most compatible with your way of typing prompts.

For example, I personally enjoy v4.5+ and I know a lot of people are big fans of the v3/v3.5 models

[torn spoke]

It's so much easier to type on

[tepid narwhal]

That would screw me up big time not being in standard QWERTY layout

Are you someone who basically uses a phone all the time and doesnt use a real keyboard at a computer much?

[olive quest]

Hey all. I'm trying to create a drum track in studio, but most of the times it ends up adding extra instruments into the track, despite choosing drums as the instrument. Anyone have any recommendations on getting it to behave?

[raven turtle]

Same issue. It usually gives me full tracks no just drums like I ask for. Yeah it’s beta, but those credits are like cash every time!

[olive quest]

I've just found that I have been having better luck turning the audio influence down to 10 or below, and ensuring the prompt is simple.

But that's after about 20 tries

[heady schooner]

When you make lyrics, how do you make it use the lyrics only once? Specifically if you only wanna do lyrics in the beginning of a song, and not the rest of it?

[magic shale]

today again V5 generates a bad sound... sometime the sound is good, sometimes bad. Hard to understand why

[vernal elm]

With the Premier Plan can I edit the vocals, not the lyrics, if so how much control do I have? Examples would be appreciated

[heady pike]

Sleazy Playful 80s Synth-Pop, 105 BPM, Juno pads, Linn-style drums, and a bright, repetitive synth hook. Energetic, slightly breathy male vocals with a confident, teasing tone, chorus lyric "Cunning look, electric shock", soundtrack to a neon-lit arcade chase where the prize is a kiss.

any ideas how to update this prompt to get lyrics who actually have meaning not like randomized lines who rimes

[fading glacier]

Just write them and then have use the cleanup

[queen linden]

I am putting those vocals to the test with musical quick patter, and lots of wording lol

[solid crater]

Sometimes wonder if there is a SEED assigned each day... and some seeds are 'just bad'. Only a guess, but have noticed 'a general change of character' day-to-day when coming back to a prompt from a previous day.

[polar spade]

If you like the percussion part of the track, then stem the drum track to get just another drum track?

[olive quest]

I've tried that a few times, and made some interesting results. It seems to happen with any instrument, not just drums. I've found better results using a very low Audio Influence setting, as I've had less blended results doing that.

[fluid linden]

suno has became stupid after the update it's not useabule giving me tuneless songs even when my prompt is clear. skipping beats alot more. and in genaral being the worst i've known it ever be

[astral fossil]

hi DreamDuck, make sure you rebuild your relation with SUNO that's the only clarity to perceive on your side.

[ripe bobcat]

I can’t get a song past two minutes for some reason today. Using my same formulas as before. Doesn’t make sense why it’s doing this. I’m just wasting money at this point.

😵‍💫😵‍💫😵‍💫

and I also get the terrible sounds every once in a while too

[astral fossil]

Hi there Kobra F. okay it's not a question of wasting money, key is you understand the changes and conguate

[ripe bobcat]

So what was updated because I’ve been using version five the day it came out. This just started happening today.

[astral fossil]

and cash his virtuosity in delight

[ripe bobcat]

At least for me

[astral fossil]

no need to panic friend

[ripe bobcat]

What?

I’m just asking a question

[astral fossil]

look there is a problem that has to be solved

no offence okay im good guy so sorry if you feel like tht

[ripe bobcat]

What are you talking about? I’m so confused.

[astral fossil]

i like to help you out

This stressful situation isn't permanent; it's about how someone experiences the changes, even if u are using your formulary or secret doses, changes means adaptive care or quit . It's an evolving update, just like birthdays are to you and me. SUNO is growing, and you know that adolescent behavior is strange. Enrichment is a flow not a chocolate bar

so care for some level-up advise perhaps it's right here to find

meantime relax, no need to suffer as like a bad experience, revise to exert solutionwise, meantime have a nice proceeding, bey Kobra hope you find some value in my interceptive action. As that's my solace 🙂

[north plinth]

I used to get the v5 previews but no more, what happened? I still don't know how many free downloads per month I get either

[summer garnet]

Quick question. When you use studio and put a song on the track. Say you want to add another track for vocals. When you generate the lyrics for those vocals, it doesn't always give you something. Sometimes it gives me nothing unless I am doing a certain part of the song. Also, is there a way to direct the style to use your persona to sing the vocal and not some random voice?

[torn spoke]

that mechanic is broken atm, and you can click the little up arrow on the regenerate UI to bring up advanced options

[summer garnet]

Hmm, I did see the advanced options and I put audio to 100 as it states persona or audio in there. weirdness to 0 and style to 100. It looks like it will give me some lyrics, but it generally will throw in parts of the other lyrics from the song as well along with it. Wish it would just put in my words only.

[torn spoke]

that's why I said it's broken lol

[raw oasis]

How do you inspire your vocals to do runs muiscally?

I want to try and push the vocals. Anyone using Studio for this?

"A vocal run is a series of fast, consecutive musical notes sung on a single syllable or within a single word, showcasing vocal agility and breath control. "

Would be amazing if we could plan out the vocal notes like a melody.

[honest heath]

try the keyword melisma or melismatic

[raw oasis]

you are the best

[honest heath]

its a good word 😄

[round bobcat]

Argglll... v5 annoys me - in almost every song I have a femals voice singing 'uh hu hu' in the intro - and now I get oh wow hooo simular to the intro 'uh hu hu' of the intro - argggl! Everything sounds like the same - sigh.

[honest heath]

try putting the word melisma in the exclude styles box 😄

also, are you using lyrical meta tags to guide the structure of the song?

[torpid haven]

V5 electronic music songs sound very similar from some genres

"No melodies. No chord progressions." Is always ignores for me

Everything gives me Ben Böhmer vibe. Good artist but not what I'm trying ot achieve.

[round bobcat]

And the intro is always singing hu hu hu - argg.... Almost no song without it. 😠

Reminds me of old time when I turned the radio off because every song had an instrumental interlude with the same Yamaha DX7 sounds as this synthesizer was new and everybody bought it. Sigh!

[raw oasis]

Any thoughts on how I can influence more Melismatic? lol (more cowbell!)
https://suno.com/s/cY0w4SfhjhwiRz1S

[void obsidian]

i have a question can anyone tell me how they fix their midi downloads mine seem to always be out of the keys when i move them to my daw.

[lament rapids]

Any help is appreciated.

I have a good track and I need to add my chorus to the end after my tag or bridge.

Except…. SUNO wants to change my song to a different melody and etc.

Any ideas or tricks to help me get the same but an extended result of my original? Live or studio just re-record. I could just do whatever.

Here? I’m stuck.

[narrow lynx]

I’ve had the same experience. Not very impressed with studio. It wasn’t ready for beta.

[raw oasis]

good to know before I upgrade.

The develop so fast, should be polished soon I imagine.

[narrow lynx]

When it does what it claims to do it’ll be a game changer but I don’t see the difference between studio and editor, except that editor works.

[raw oasis]

I am trying to get more whammy bar and harmonized vocals on a specific line. what am I doing wrong or what advanced technique (editor?) can I try. https://suno.com/s/gr3SmnVWrafedWNi

tried [harmonized vocals] but should I move it to [Verse 2 | Harmonized vocals]?

also Suno may not know what a whammy bar is

can try tremolo bar

[true star]

Quick question on screaming: I've tried almost everything I can think of to have screaming be consistent, but (especially with choruses) it almost always refuses to scream for a full verse/bridge most of the time. Does anyone have any ideas on how to work with that? For context its more metalcore/swancore so it has a mix of clean vocals and screams

[torn spoke]

Don't use studio for vocals

[torn spoke]

you'll legit simply have to cut the song off right before the screaming part

and then extend with a huge emphasis on screaming

[true star]

How so? Would it be okay if you gave a quick example of how I'd structure it? I'd really appreciate it

[torn spoke]

I'm saying that, you crop the song up until you're not happy with it, then you extend from there

or you can frankenstein parts you like, and then cover it to smooth over the transitions

[true star]

Crop the song up?

[torn spoke]

and sure you can give me the structure

Do you have studio?

if not, you'll have to patchwork the song together in something like audacity, and then upload it to suno again

[true star]

*[Pre-Chorus]
[Cleans]
Every pulse a broken prayer,
Drifting in corrupted air…
[SCREAMING]
🗣 "DISCONNECT! DISCONNECT!"

[Chorus]
[SCREAMING]
🗣 "SWEAR I HEARD YOU IN THE NOISE /
🗣 A FRACTAL VOICE / A VOID DISGUISED!
🗣 SCREAMING IN SIGNALS / CALL MY NAME!
🗣 BUT THE CODE WON’T TRANSLATE THE PAIN!"*

What makes it weird is right after there's a verse that's just screaming and it usually nails that

[torn spoke]

Why have you got forward slashes in there