📚┃suno-school

[charred merlin]

You could hum the tune?

[lucid anvil]

3 old unfinishg projects that are as WIP on soundcloud, they don't work

[charred merlin]

#🪲┃bugs-web #💡┃suggestions-web

[lucid anvil]

all of it is on creative commons license

Yeah let's wait ....

[unreal bolt]

it wont be that easy

they would have to partner with bmi or the gema

and even those cant hand out licenses that would be needed for an ai like this

but they could set something up that would let artists chose to sell licenses to their copyrights for use as trainingsdata

[errant kindle]

Hi all, started using v3.5 entering custom lyrics for a whole song (2180 characters), but it repeats some parts randomly. any advice please?

[daring lava]

Does it happen consistently? If so, if you’d like to share the prompt/lyrics you are working with maybe we can better help identify the problem.

[hollow hare]

Hm, try padding some parts with some instrumental tags (see RayanAIR's Novo Lands which shows how flexible these tags are if you need an example of how to use the tags)

The padding should hopefully reduce the cause of repeats. I also found having more lyrics helped but not always (for this, I would usually copy my current lyrics, use the Generate Lyrics on it, then paste what I had at the front and delete the extra overflow characters... then trim the song by using a v3 extensions with [end] at the point I want the song to stop after)

[wraith raft]

i think beatport et al will be easier since that's people uploading their stuff specifically to be licensed. BMI and all the other music rights management orgs around the world is going to be one hell of a battle. who knows what they'll want.

like assurances that suno has a license to all the training data. that'd probably be a nonstarter

[charred merlin]

Instead of trying to tackle the entire song at once, and giving Suno a lot to handle, try breaking the song up into multiple parts; create the first part, extend from it, and if you want to add a third part, i suggest "get whole song" to extend from so the third part matches the "Feel" of the entire song, not just the second part.

[errant kindle]

thanks, this is an example https://suno.com/song/4a520c56-3b47-4252-9ef7-b48ebf19e397. also this one repeated the otro although it's actually quite catchy https://suno.com/song/e3d811a1-2aa1-4ffd-9f56-5056f7b0492b

[errant kindle]

I'm also curious, is the upload audio feature designed for a full instrumental (like 2-3 mins)? say I play something on the piano can i use suno to add lyrics?

[charred merlin]

6-60 seconds

[viral widget]

Maybe also a cool feature would also be to cut/edit the resulting audio, because sometimes after the end of a song the outro starts again until the time limit is reached.
So if you could just cut the end such that the hallucination in the end is gone that would be nice?

[daring lava]

@unborn quiver what’s going on?

[unborn quiver]

@daring lava So I am not sure why in the world this thing will not let me make it public https://suno.com/song/c4a30968-167a-4008-be44-7ca9a4b842fe

It ticks the thing fine, but when I refresh it is link only

I tried on the phone even

Same result

[daring lava]

Have tried clearing your cache?

[gentle wasp]

Hey guys, I have an idea, but not sure how to do it... I want to combine video game and a tv that are related but how do I do it? in this case the X-men 89 game ny Konami, and tv theme along with the games intro I know it's copyrighted material, but I just want to try...

[daring lava]

Awesome track btw

[unborn quiver]

Thanks, but what does it matter if it won't let me make it public xD I'll try to clear cache but since I already tried on phone and it didn't work I doubt it will do anything

@daring lava nope same thing, I click make public it gives me yellow popup saying it was made public and on refresh it is still link only

I don't see anythign wrong in the instrumental meta tags

I don't think the name can be it can it?

[daring lava]

Not sure, this is from an upload?

[unborn quiver]

I made it myself from scratch

No

It is not using any new features

[daring lava]

That’s odd, have you tried logging out and back in?

[unborn quiver]

clearing cache naturally forced me to

Funnily enough if I make a new track right now, it actually lets me make it public. I just generated a random instrumental with no context and made it public, refreshed and it stayed public, but that song won't work

[charred merlin]

Since you've already tried another device, does the same thing happen with new generations after the track with the issue? Beat me to it 😛

[unborn quiver]

What is it about it?

[charred merlin]

If the part of the song that you like is under 60 seconds, you can clip that part out & upload it as a new track & extend from there 😛

[unborn quiver]

Okay so its not the name as I just tried renaming a test track to the same thing and it stayed public fine

[daring lava]

Not sure what’s going on with this, very odd. Maybe go ahead and start a support ticket for now. Hopefully it will fix itself but if not then at least we’ll have the team involved

[unborn quiver]

You kidding me? I spent 1k credits on this xD

[daring lava]

The team will be able to help best with this, try emailing [e-mail]. Sorry this is happening. I’m sure it’s frustrating. We’ll get this taken care of

[timber river]

@prime gale using vxtwitter domain lets us play the video straight from discord btw eg https://vxtwitter.com/suno_ai_/status/1800932487633207599

[unique stone]

Someone else has the same issue in the bugs channel. I've pointed Coda in that direction in a DM. Seems z is looking into it for the other user already.

[daring lava]

Awesome 👏 thank you

[charred palm]

Uploaded music: is the max length now 3:30???

[unique stone]

I think it's the max length of the selected model, minus the length of the uploaded audio

E.g. with v3.5 if you upload 30 seconds your extension will be up to 3:30

[charred palm]

Makes sense! As all my clips ends at the 29-30 mark. Thank you!

[gaunt ice]

Hey this new feature added. When they mean audio, do they mean voice or actual sounds

[charred palm]

Anything. Experiment! (Which I am doing right now)

[limber linden]

So folks i noticed something...

[cosmic mirage]

Hello, how do you make an album with same voice?

[feral ibex]

Is anyone aware on how to get Jarvis to play Suno songs in discord? I was able to previously before the update, but can’t now.

[neat field]

testing upload feature now, HYPUUUUU

[sly grail]

do uploads always have to be the intro?

[unique stone]

Yes

[daring lava]

What’s that?

[daring lava]

Something quite a few are testing I’m sure with the new upload feature

[limber linden]

i deleted the post, i thought i had figured something out. i was mistaken.

[daring lava]

Oh ok 👍

[limber linden]

so far the 14 second mark is the sweet spot for the stuff i am doing

for the extension on uploaded material.

[cosmic mirage]

I mean the suno team have not done some demo about this? I think it's a pretty popular demand

[daring lava]

Oh I bet they are working in all kinds of stuff, not sure atm. But again, could be possible with the new upload function. Maybe.

[fluid robin]

Best way to make a ambient soundtrack that don't playing music ? been struggling with this one due to that it works the first few minutes and then music directly after

[zenith gale]

Is it possible to extend uploaded audio using v3? The Part 2 output has the v3.5 tag on it even when I select v3 in the Create window.

[fluid robin]

Yes you can combine the diffrent version when extending the first part or second one

a few dose that in order to have the v3 sound effect that 3.5 dosent have

[neat field]

it seems like if your upload is a whispering voice, v3.5 does worse than it might with an anti lullaby v3.5 extension that is extended into a new genre

[modern belfry]

I think Suno is drunk tonight or something. I keep telling it to do a fiddle solo, but it does a banjo solo. I tell it to do a guitar solo. It'll do a fiddle solo. Wasting so many coins on this short song.

[rustic wyvern]

i tried it on mobile yet it didn't work, it put another voice instead of mine

[loud heron]

Do any of you guys have an workflow for enhancing audio quality post generation?

[cursive vortex]

Depends on how bad the quality is. That will limit how much you can do to make it better. Most songs a simple mastering pass from Bandlabs mastering or on your own DAW is enough to clear up the muddy mids.

[low coral]

So, am I understanding it correctly that we can only really extend a sound we upload, rather than using it like of like Midjourney image prompts to generate a song from scratch using the melody as a guide?

Or do I just need to put the prompt info in first before uploading so it has something to attach to?

[twilit merlin]

i have issues with distorted vocals from the middle of tracks

[ebon raven]

I would like that if you upload your own album cover for a song.. that picture will be that background of the video you can download with lyrics. (right now it reverts back to the original random generated picture)

[abstract spade]

If I accidentally delete a song and then delete it from the trash, can I get it back? 🥲

[crimson scaffold]

2 accidents one after the other?

😭😭🫶Please improve the vocal quality particularly the clarity @hybrid arch

[hollow hare]

You could still grab the mp3/4 from the CDN if you still have the identifier [the ID after /song/ ], otherwise no AFAIK (maybe it is something that [e-mail] might be able to retrieve? I think it is unlikely but better to reach out and ask - just be sure to provide them with the identifier)

[hollow hare]

Try using the style prompt "clear [male/female] vocals", it should improve the quality (but still primarily dependent on the genre and RNG)

[abstract spade]

thank you

Another question is how to export in 32-bit WAV? in the future?

[hollow hare]

There is a current option to export as WAV if you have a subscription plan (though I do not remember what its settings are).

Based on the past, each released major version appears to have its quality improved during the training process so it stems that the future versions may very likely improve the quality

[cursive vortex]

Exported wav info:

[forest cliff]

I'm having troubles getting my song to end using the Extend option

[hollow hare]

Extend where you want to end with the extension set to v3.0 when using ending tags (e.g. [end] in lyric box and "end, short" as the style tag) as 3.0 is much more reliable for ending than 3.5

[forest cliff]

thanks, I'll try that

oh wow, the first result is only 10 seconds instead of 2 minutes

[hollow hare]

Because of the Short prompt. I was hoping it would have been < 7 sec so you could have had some credits refunded

[forest cliff]

https://suno.com/song/fd076f95-e00f-4cbd-be3d-8a02b4dd6b71
Worked great 🙂

[abstract spade]

I intend to purchase a monthly account this month, will the music generated be mine? What about my rights to these songs? Can I modify them freely?

[slim summit]

https://suno-ai.notion.site/FAQs-b72601b96de44e5cacd2cd6baa985448#47fc7b6ad1114a1597903b11c73ade77

tldr: all yours

[hollow hare]

Covered in the FAQ (Suno's Help page) and the Terms of Service

Some highlights:
https://media.discordapp.net/attachments/1102676217687195679/1242805116084420739/image0.jpg?ex=666b8435&is=666a32b5&hm=f20c3932c1ceff1c7ab0b0e56452a14672c358db85cce13b9769b0032989b5ea&
https://media.discordapp.net/attachments/1102676217687195679/1249443846492917880/image0.jpg?ex=666b4742&is=6669f5c2&hm=ebd8cac790328862f040d885993c3d3a37dd03e580be6da52e784ebb5ea1de22&
https://media.discordapp.net/attachments/1102676217687195679/1248122915774533693/image0.jpg?ex=666b164c&is=6669c4cc&hm=c26c29a8488777dbed888bdac966ecedca02053059e2bf80e42e808e76e0615b&
https://media.discordapp.net/attachments/1102676217687195679/1249441403956760586/Screenshot_20240609_125633.jpg?ex=666b44fc&is=6669f37c&hm=f4769ebb85220ac34261e9da57e9a2e7903955aa6d4592c018fdf72df60cf929&

[slim summit]

different question: does the "repeat" button in the site app scripting actually have any function?

[hollow hare]

It repeats when I use it. If you mean if there is any execution when clicking it... well, I did not inspect the element but I suspect that it toggles the standard loop attribute in the media control

[slim summit]

ok then i guess it starts playing anew. as in, it stops the playback and starts a new playback from the start of the song. it's the simplest and most naive and therefore a smart way to implement this feature. but if it works for other people, i guess the "reduced motion" accessibility addons in my browser that heavily limit autoplay may be interfering with it.

[heavy solar]

i hope been one 4.1 and 5.1 and 6.0 one day

[abstract spade]

What is the difference between the pro plan and the premier plan other than the amount of music?

[hollow hare]

Just the credit amount that renews each month AFAIK.

[cursive vortex]

Well, and also having 4 more letters in your plan name.

[hollow hare]

Suno is pretty straightforward like that at the moment.
Like the difference between free and subscription is just the amount of credits and ownership (oh, and the cosmetics like swapping song images and your avatar).

There is no downgrade of quality, no special member only versions (well, outside of alpha testing phases), etc.

[cursive vortex]

Um, also some queue prefrences I think. I do know that if there's too high of a demand sometimes generation is disabled for the free tier, but haven't seen that in a while.

[abstract spade]

Thank you Paimon for your attention and kindness in responding.

[charred merlin]

Afaik, the generation queue for pro/premier is the same, but I'm curious about what "custom generation queue" means for the custom plans you'd contact billing for-- whether there's an even faster queue to pay for priority access to, or if it would just be the option to use the shared priority queue to try to shave off the price for a larger plan (eg, more credits, but lower priority)

[viscid radish]

have been playing with the new upload feature in the hopes to extend some personal instrumentals into longer tracks. Set, Custom mode / Instrumentals. set music type to Heavy metal goth cyber punk fusion. Suno keeps imposing gibberish lyrics over every version. Over otherwise amazing extensions. Please let Instrumentals be Instrumentals. Please and Thank You.

[open crown]

any way to get over a minute of audio upload?
like to put it together

[teal raft]

For enterprise level service

As in, corporate level shenanigans

[charred merlin]

I figured that was the most reasonable use case, but I was wondering about the ambiguity in "custom"

[teal raft]

pat has mentioned before that they may offer another tier about what is currently available but it isn't a priority at the moment

[charred merlin]

Day zero signup right here

[teal raft]

This was also a couple months ago so who knows what has changed since then

[pure bison]

how do you guys use the new feature where u upload the audio file? i mean what kind of audio? us singing the melody?

[charred merlin]

Singing, playing an instrument, drumming a beat on your thigh, humming a tune, a cool beat from Garageband, nature sounds, any random noise (breaking a plate, your child's laughter, ...I think SixStringToker has a collection of fart noises they were going to use)

The feature is available to paid memberships, and the upload button shows up right beside the 'Create' button.

[zenith panther]

I can not use this new feature, always failed, please help

[charred merlin]

What audio format is that? .mp3, .wav, etc?

[lone harbor]

I have a song that I did in v3 that I just want to regenerate in v3.5 because the audio output sounds better to me. Is there a way to get Suno just to redo the song as is?

[bronze helm]

What if you try to manually add a file extension to the file? It seems the file you are uploading is missing one, it only has a file name. Try adding ".mp3" so that it becomes "[filename].mp3" while having extensions enabled if you are using Windows.

It could be that the audio data is retained but the file is simply missing its extension

[earnest ledge]

Will it work with voice too? Is it already accessible on the website?

[narrow cypress]

now that we have audio upload, what am I doing wrong? I upload audio, and it just adds it to my library. How do you turn it into a song? Just Extend?

[oak rampart]

Can anyone help regarding voice cloning of bark using custom audios ?

I am facing inference time issues regarding voice cloning

[glacial scarab]

make sure you did not accidentally switch model versions. Otherwise the results should be similar except for the normal AI randomness

[graceful condor]

all my stuff is made with 3.5, but suddenly the styles of my generations completely changed

[glacial scarab]

do you have an example. At least a prompt. If its either too complex or too simple that could explain the randomness

[untold briar]

If you're extending an upload, heads up: Extending from Uploaded Audio is for 3.5 only, but the bug is that the UI lets you pick other models. However it is always using 3.5

[narrow cypress]

try hifi instead of extremely loud maybe, it could be causing distortion

[graceful condor]

it's very strange

now it's giving me heavy metal

and this other generation has no intro

very, very inconsistent

[glacial scarab]

i would also suggest switching extremely loud. it could be detrimental 180bpm is also a tag that won't really work better would be fast or uptempo or italian musical terms. Tags like this can be really inconsistent in their results and give the impression of changes in the model when in reality it is just randomness causing the problems. if you generate enough you will probably get something like the original sooner or later

[graceful condor]

removed that from the prompt

[untold briar]

You can even upload your old clips you like (choose a small section that has the sound, but not the main melody) and Extend

[graceful condor]

but then it'd not have an intro

[alpine widget]

I get a Network Error when I generate Audio from an uploaded clip. Anyone got the same problem?

[dapper fulcrum]

I opened it today and what happened to suno? I spent all the credits and not once did the generation make a song with the addition of the genre "dark trap" and even with the generation of only the genre "dark trap" it turned out to be dance music, what kind of nonsense? (((
https://suno.com/song/726d625c-d210-4fae-8393-12e1b6dd838f
https://suno.com/song/f96c2253-7bad-4ccf-8b2c-e3d7a7867b09

[bleak bobcat]

How i can get a melancholic ballad, only with piano, violin and cello? Always turns to pop. I'm using v3.5. I think with 3.0 is less changer.

[foggy jasper]

am I allowed to make songs from UI sound effects from video games?

[jade pike]

Love Suno but , Really disapointed of the upload function I only have tryied upload 60 secounds acapellas and 3.5 is good when u put lyrics in but less good for intrumentals. but I wounder if we going to get the ability to save multi-tracks in the future? 🤔

[foggy jasper]

I feel you. I've just been tricking it. Even though it only lets me extend with 3.5, I extend a third time on the 3.5 extension with v3.

sometimes it works 🙂

usually 2-5 seconds after is good

[real pewter]

Hi guys, Suno rulz but I have one problem. On my generated last songs is not good quality audio sound. With interference, noise, double voice. Can you help me to improve my quality song before generate? There are some tips for me? Regards!

[foggy jasper]

never mind im stupid

"vocal tracks are private"

makes sense why

[small parrot]

I bought a $30 monthly version, hoping that my music can be sold or commercialized. Where can I download the copyright certificate to prove that my music can be commercialized? After I uploaded the music to the music website, the music website failed to pass the review. I need to upload the authorization file of the ai ​ ​ creation platform or the voice authorization of the ai ​ ​ performer. What should I do? Is there a special authorization document to solve this problem? Please give me an official reply from suno official. Thank you.

[storm grail]

Does anyone else have this problem when automatically generating songs? I enter my prompt, I get a perfect song, but the instrumental section is then sung, so I can then delete the song, here is my prompt:

Create a lively and joyful song that celebrates the freedom of dancing in the rain. with intro and instrumental sections, at least 3 minutes, at least 5 verse

Here what i get ( a part):

[Chorus]
Oh we dance dance dance in the rain
Lose ourselves in joy and pain
No more worries no more chains
Only freedom in the rain

[Instrumental Break]
(Energetic electronic beats with catchy melodies, perfect for dancing)

The singer then strangely sings the part in brackets?
Is there a problem with my promt?

thx Mike

[untold briar]

Use [energetic ...] Instead of (energetic...) things (in parenthesis are supposed to be sung in background voices)

It sometimes still sings very long complicated instructions though. Versus simple ones like [guitar solo]

[dense merlin]

I used a program to extract this channel to chat and now trying to get chatgpt to turn that into a pdf with every possible [tag] I might also include some prompts for having chatgpt craft lyrics better than it's primary school rhyms it does. that I use.

[storm grail]

thank you, but the style promt is automatic generatet from suno, i only make the song description promt (no custom creation)...

[hollow field]

👍

[untold briar]

You can use custom mode and write the song description in lyrics, the click "make random lyrics". Then you can edit them before clicking Create.

[storm grail]

Yes, that's a good idea, but I think I'll have to enter the style of music myself - but no problem. I just like to be surprised by what comes out automatically when I just enter a song description 😄 usually not that bad (if it works with the instrumental sections)...

[untold briar]

Well, Suno should know better than to do (Energetic electronic beats with catchy melodies, perfect for dancing) instead of [Energetic electronic beats with catchy melodies, perfect for dancing]. If it happens very often, you could report that in the bugs channel actually.

[storm grail]

I've already written to support, but unfortunately there's no reasonable answer other than a reference to this forum. Whatever, I hope suno will solve this at some point... thank you👍

[cunning edge]

when i want it to end, it never ends, when i want it to extend, it ends asap, i'm depress.

[bronze helm]

Try using the v3 model to end your tracks. It seems to listen to the [End] and [Outro] tags unlike v3.5

[cunning edge]

Oh..

[humble shadow]

Hello, I have a question about credits. I'm a suscriber of Pro Plan. But when are the credits renewed? They said every month, but from the date of payment?

[bronze helm]

Let's say you subscribe on June 13 (today). Your credits should be renewed on July 13.

[humble shadow]

Okay. I suscribed on May 13, we're on June 13 but I haven't be renewed

[bronze helm]

🤔 Here's my plan, I believe I subscribed on April 13 and have been paying on the 13th of each month ever since

What does your Renewal Date say?

[humble shadow]

Okay weird...

But thanks I'll wait tomorrow ^^

[bronze helm]

If it says May 13, 2025, it means you have made an annual subscription I believe, so your credits should be renewed today either way

[humble shadow]

Thanks 😉

Btw, does Suno Chirp Bot is working ?

[bronze helm]

I think the Chirp Bot is obsolete if I'm not mistaken, as everyone can access the website

[heavy solar]

i cant add no Michael Jackson songs in here Suno.AI what the rip off

Uploaded audio is too long (currently 277.7 seconds). Maximum duration is 60 seconds. WTF

[unique stone]

Cut the audio up in an audio editor such as Audacity.
Don't upload songs that you do not have rights to.

[unique stone]

If you are uploading content you do not have rights to, this is against the Suno Terms of Service and you are reminded not to do this when you first upload audio.

[heavy solar]

Banned me i not care i was stop im am a michael Jackson fan not fair

[onyx swift]

Hi,

Recently several of my tracks made on sumo.ai website are speeding up the tempo about half way through a track.

Is there a way to prompt to avoid this? Or is this a bug suno are aware of?

Help

https://suno.com/song/fa632899-8b18-4fa2-830a-74e5e3f70c2a

Listen from 00:40
This one goes off the rails completely

[wanton dagger]

v3.5 seems to have erased the heavier genres completely. Exact same prompts that worked in v3 no longer do at all. It has erased the entire Modern metal genres and sub genres. It always pumps out some generic radio rock/pop or old man metal when trying to get anything modern and heavy with melodic parts and singing. They sacrificed creativity for quality.

[leaden wave]

If I was to use pre-cleared samples of existing music for the new upload feature; for example If I was to sign up to this website https://www.tracklib.com/ could I then use any sample I get from there?

[daring lava]

As long as it matches this criteria, you should be good

[daring lava]

Yep, I hear it. Looking into this. Thanks for sharing

[chrome stream]

hi guys just joined here, does anyone know if suno annoce they have pla to release API ?

[charred merlin]

No comment regarding that so far

[heavy solar]

i hates is take my money back

[charred merlin]

[e-mail]

[onyx swift]

Thanks a lot David

[neat field]

i haven't tested upload in the middle yet, can one extend a song with an upload ?

[dreamy sparrow]

Will the recent update also fix distorted/robotic voices?

[charred merlin]

No, it's only for the beginning of songs at this point

[whole kiln]

hello all!
Quick technical question... Does anyone have an idea on how to add an isolated tap dancing phase to a song? 🙂

[charred merlin]

If you get any songs that notably have degraded audio quality, make sure to select Bug > Poor Audio Quality, then give it a thumbs down. That's the best way to actively contribute to overall better sound quality.

[dreamy sparrow]

Due to the AI having more capabilities in harmonic melodies, it starts to imitate auto-tune and handle the vocals more as an instrument than a vocal. This is where the distortion or loss of quality usually happens.

[ivory owl]

Bonjour ! I wasn't able to find info about how the MUSIC part of the songs is generated ? Are the instrumental parts trully AI-generated, or, as Pi AI suggests, they are actually tracks composed by real musicians with traditional means ? Is this information available on Sona website ? 😮 Thanks !

[naive quail]

Didn't even know that was a thing, despite it being right in front of me this whole time.

[dreamy sparrow]

"Chirp", developed by Suno to create the instrumentals, and "Bark", also developed by Suno to generate the vocals. More info on Bark at https://github.com/suno-ai/bark.

[charred merlin]

Pi is hallucinating.

[ivory owl]

Thanks ! Any info about the "Chirp" process then ? Is it generative-AI music, or put together in a more "traditional" way ?

[ivory owl]

Yes Pi hallucinates most of the time, I agree !

I did find this but I'm not sure it means that the music parts are really generated by AI, not just bits and samples (or even entire segments of tracks), pieced together...? "Suno was the first to offer AI generated instrumental music with compatible vocal melodies or rapping on top. Tracks are generated end-to-end, with lyrics and instrumentation created simultaneously, making for lightning-fast delivery. "

[charred merlin]

Where did you read that? AI-cited blurbs shouldn't be included in research notes.

[dreamy sparrow]

The wording looks AI-generated.

[naive quail]

Looks like it's from something posted on Audio Cipher back in May

[hybrid marsh]

how do i make 10 songs at once?

[charred merlin]

Click "create" 5 times in a row 😉

[tiny tangle]

Hi, fairly new to Suno.

If I were to extend a song, does it override the original version, or duplicate it and extend the copy?

Thank you 😊

[charred merlin]

Welcome! ❤️

When you make an extension, it leaves the original track intact and creates a new file that starts from the extension point you selected. If you click "Get Whole Song" on the newly generated clip (shown as Part 2 in the library), you can generate a new track from the beginning to hear all of your clips put together.

Getting the Whole Song can be helpful when making longer songs, or using multiple extensions at once, because it's the difference between making the extension "sound like the whole song", or just making it sound like a continuation of the clip you're extending from.

Excited to hear what you come up with!

[dapper fulcrum]

I opened it today and what happened to suno? I spent all the credits and not once did the generation make a song with the addition of the genre "dark trap" and even with the generation of only the genre "dark trap" it turned out to be dance music, what kind of nonsense? (((
https://suno.com/song/726d625c-d210-4fae-8393-12e1b6dd838f
https://suno.com/song/f96c2253-7bad-4ccf-8b2c-e3d7a7867b09

[tiny tangle]

Legendary, thank you so much!

[ivory owl]

Yes agreed on that

[limpid badge]

You need to add more specific info, type of vocal / rapping, a lil bit instruments description (sinister piano for example), also add mood (cold, chill etc) It depends what exactly you want

[ivory owl]

On Audiocipher ( https://www.audiocipher.com/post/suno-ai-chirp. ) — but what I'm waiting for, is to find an official document explaining that the music IS indeed AI-generated, not an assembly of samples or pre-existing "blocks" made by musicians. Do you know where to find this information on how it works ?

[dapper fulcrum]

you are wrong because I indicated the basis, even if there was no basis, the elements would be in the compositions, but no, these tracks have nothing to do with what I indicated. If you are related to this Suno company, then why are you deceiving? it looks like the code is deliberately distorted so that people have to spend more credits

[hollow hare]

Use dashes instead of spaces for a compounded terms will help get the correct genre (e.g. dark-trap). See Jonathan Fly's post about 3.5: #📚┃suno-school message

As is, the AI is likely thinking it as "dark" (emotion) & "trap" (genre), but if you had the dash/hyphen, it would see it as "dark trap" (genre).

If the dash did not work: there are some genres that 3.5 "forgot" or having a hard time getting, like phonk, so what you can do as an option is make it in 3.0 first (imo, make it an intro instrumental piece) then Extend it as 3.5.

I assume that you are using custom mode? If so, take advantage and describe other settings too like moods, tempos, etc. See #💬┃general-chat message

[pearl forge]

can any one help my .wav download is broken ...

[limpid badge]

I spent thousand of credits creating such kinda trap, i know what im talking about, if your samples sound so differently you 100% do sometjing wrong

[limpid badge]

If you dont wanna help and just wanna complain and cry dont ask anything bruh

[slim summit]

there is nothing geneating and hasn't been for 12 minutes

[dapper fulcrum]

your words sound like you need to do this when requesting “rock”, this is as absurd as if the network shows me a person instead of a robot when requesting

[dapper fulcrum]

I condemn the rudeness, I made a complaint to the developers, let them correct the code or the database so that there is no idiotism in substituting genres. and there is no question of any specifics, because I indicated the basis, if I write a violin, then it should be

[limpid badge]

you literally started blaming me out of nowhere and being rude when I was just trying to help

Again, if you dont wanna listen tips, just dont talk

[final hazel]

Hi Noeliza, I understand your frustration with the genre generation. To help the AI produce more accurate results, try using hyphenated metatags like "dark-trap" instead of "dark trap". Also, specifying additional details about instruments (e.g., "sinister piano") and mood (e.g., "chill") can improve the accuracy. If these adjustments don't help, consider using Custom Mode to fine-tune your parameters further. If you believe there's an issue with the AI's performance, please report it using the "Report Bug" tool in Suno. Thank you for your patience and understanding.

[dapper fulcrum]

I wasn't being rude to you because nothing was directed at you personally, but I was asking if you were cheating, which is a better question if you're a Suno developer. sorry for my english. I use a translator.
your advice was not needed, because I don’t need to explain to the robot what violin, piano, electro, dark trap music is.
therefore, let them check the code or database by genre, otherwise if there is a dog instead of a cat, this equals a non-working service. it's not even in my interests, but in theirs

[dapper fulcrum]

It worked fine for me for several months and yesterday, and here you are telling me about clarifications. what an excuse for the developers to fail....

[limpid badge]

Yes you were rude. And im not a suno developer, im user just like you. Despite how quickly develops AI it still makes mistakes, i dont know what do you ecxpected

[hollow hare]

Did you try refreshing the page?
(I would recommend doing Shift + F5 so it bypasses any cached materials in case that somehow played a factor but I really doubt that is the case here)

I know that, while generating, the loading icon may disappear but the final time will show the load icon on the generating tracks if it is in progress (and display the song length when it is finalized)

But 12 minutes is a very long time and should have been done by then...

[slim summit]

did you switch to model 3.5? it has severe problems with some of the more ...esotheric genres. my recomendation is to generate a beginning using model v 3.0 and doing part 2 using 3.5

[slim summit]

it seems to be related to streaming the rest of the song. when the generation finishes, the whole song is not downloadable immediately.

also the the key command you are referring to is ctrl+F5. not shift

[charred merlin]

~~If nothing else has worked, copy the "share song" link for your song, grab the long string of characters separated by hyphens and paste it into https://cdn1.suno.ai/YOUR_SONG_ID.wav~~

Only seems to work for .mp3 and .mp4

[dapper fulcrum]

show messages containing rudeness or not lying. and secondly, you yourself are rude by inattentively reading the comments, I have already said twice that the request worked fine for several months and last day too, but now I have never been given a track with the addition of elements of this genre, not a single percent

[final hazel]

First of all, AI is unpredictable. It was and still wil be. If you had good results in the past, it more a question of luck (any of the devs will tell you so). And if you are using V3.5, this model is different. This is unavoidable. Because this is new tech and devs know what they "feed" it but don't know how it fully will "respond". V3.5 is know for better quality, but listens less to prompts. So sometimes you need to add a "hyphen" and sometimes you need te repeat the words in your prompt to give it more weight. This is just the reality of the current state of AI music. There nothing malicious going one, this is not to make you pay more credits. This is not cookie cut science.

[dapper fulcrum]

I already said that yesterday everything was fine with 3.5 and before, when it appeared. but today all the credits are just wasted, not a single track that I deleted was not a single percent from the dark trap genre

[limpid badge]

Your first replying to me, you started projecting at me some stuff. And in your first message you didn’t say all the information, it seemed to me that you might be a newbie and just started to use suno

[charred merlin]

When you keep getting the incorrect results, do you back up & rethink your approach? Or are you trying the same thing over & over, expecting different results?

The community-driven suno wiki has a lot of helpful tips for trying to coax specific sounds out of the tool.

[pearl forge]

thanks I'll give that a try I'm a bit worried it's permanently broke for me not sure though as I had a lot of these errors recently and I'm keen to still master from wav format.

[charred merlin]

Not sure if you saw the edit, the method I shared only seems to work for mp3 and mp4. Maybe you could extend from the "corrupted .wav" song, get the whole track, and then try to download that .wav?

You'd need to tidy up the end, maybe even cut off the undesired ending with another extension point-- not elegant, but just trying to think of ways around the problem immediately without needing to reach out to support email ❤️

[pearl forge]

ok yeah interesting point I'll see if I can find the first one where it messed up..

[dapper fulcrum]

The point is that never before this day has generation made a mistake with the concept of dark trap. sorry for my bad english

[wanton dagger]

v3.5 seems to have "forgotten" many genres. Continue to use v3 or a mix of both to get the results you want, that is what I am having to do

[dapper fulcrum]

with version 3.5 until today everything was fine for me((

[abstract spade]

If we made a parallel with music publishing, would SUNO be a music label and recorder ? I want to mention them like this in the songs generated here even after editing them... it would be correct, wouldn't it?

[toxic leaf]

I can’t return to the basic subscription, what should I do? I already canceled my subscription, but nothing changes

[hollow hare]

Ah, right, that is dependent on the browser. I looked into the keyboard shortcuts (for future reference) and it appears that the following are used for the cache-bypass reload/refresh:

• Chrome uses Shift + F5 or Ctrl + Shift + R
• Edge uses Shift + F5
• Firefox uses Ctrl + F5 or Ctrl + Shift + F5
• Opera uses Ctrl + F5
• Safari depends on the version: v11.1+ uses Command + Option + R while v9+ also has Command + Shift + R

[hollow hare]

Cancelling the plan will go into effect in the next cycle renewal (so the current month would be the subscription then goes back to free plan) and all that entails (e.g. as always, the credits will renew/refresh, not carry over)

If this is your first day or you had cancelled but Suno still thought you were subscribed, contact [e-mail]

Fixed the email, you want billing

[hollow hare]

I would say that Suno would be more along the lines of a service or person-for-hire than those. You can list yourself as the music label, pretty sure Top Music Attorney (website, YouTube) covers topics related on how to be one if you are interested in that.

(EDIT: Found the video for being your own producer which was what I was thinking of, sorry)

Remember that most distributors hates AIs (especially poor quality or very obvious AI sounding media) so make sure it sounds quite clean if you are going that route. Hence why I would recommend not mentioning Suno (if the song was generated while on the free plan, remember you cannot monetize but must credit Suno such as like in the description of YT videos)

Before you sign up for a distributor, investigate and read the terms of service and check the chat here for the history if they will accept your music.

The more obvious your song sounds like AI generated, the more likely it would be getting rejected.

Generally, what I had seen was:

✅ DistroKid and Ditto are known to accept and distribute AI songs and
❌ TuneCore and CDbaby are known to reject
#💡┃suggestions-web message

I recommend taking a look at Top Music Attorney coverage regarding different distributors ( #🗣️┃chill-chat message )
It will give you a good understanding of what to look for in their Terms of Service.

You especially want to find out how they distribute royalty and for any clauses that will make them retain it instead of giving it to you if they terminate your account (see TMA video about TuneCore and Symphonic as they had those very shady clauses)

[abstract spade]

Thank you again for your attention, Mister Paimon

[mossy herald]

Hello everyone, do you know if the Tracks upload service that is only available for paying users will at some point in the near future also be available for freemium users?

[hollow hare]

You're welcome. An edit I made was that I found the video for being your own producer which was what I was thinking of, sorry

[fierce bone]

This has probably been brought up multiple times already (would be good with some forum channels to more easy find topics) but i am slightly dissapointed with Suno's new "upload your own sound" feature, after also trying out Udio's version of it.
Like i feel like Suno probably stressed it out and released it because Udio beat them to it, and now its not what was meant to be! Ofc that can be fixed with time and patches, but i still feel like they released a not so good feature bc they were stressed and its hurting the rep of Suno in the ongoing "Suno vs Udio battle" if you will?

I am currently a yearly sub of Suno and also sub to Udio on a monthly basis, but have lately been feeling more and more that i get slightly dissapointed with Suno's updates and neww feature and been using Udio more. Even tho i still think Suno generally produces better things "musically" whilst "udio" has better voices. But right now i can generally create a good instrumental track with suno and then use Udio's upload feature and ad song to it and i get a great result. But it just makes me slightly sad that it feels like they are not adressing this or even talking about it.

Whats the point of my message you ask?
Well honestly i dont have a specific point, i just want to ventilate and discuss and see if others have the same feelings and thoughts as me.

[hollow hare]

I really doubt it will become available for free users for two reasons:

1. Need some carrots to have subscribers, more than just uploading album covers or changing profile pictures

2. There is already some users trying to upload copyrighted materials (and trying to bypass the content ID check). Having it available for free users will multiply that problem and affect the resources used.

With having it set to a subscription only service, it helps enforce people to have (relatively) good behavior (because paying users would not want their accounts terminated) while having it on free plan would encourage some bad actors to keep creating new accounts in order abuse such a feature if they can

[daring lava]

@hollow hare just want to say thank you for your contribution to this channel and to members. Definitely appreciated!

[unique stone]

Suno announced audio upload was coming, and early-access testers could already use it before Udio released theirs.

There was definitely no rush from Suno. Upload already worked at that point, just needed some finishing touches.

I am using both, and get different results with the same audio clips so it's hard to do a 1:1 comparison. Personally I find it interesting to take parts from one service, upload it into the other, and see what it can do with it.

[mossy herald]

oh ok .. thanks for the answer 😁👍🏻 Totally agree

[unique stone]

Basically I don't see it as a battle, but a toolshed where I try to be productive with what's available rather than having one tool and deciding to use only that for everything.

[fierce bone]

Fair enough, but to be honest i think the same is true for Udio, no? Like they probably had it done and hade certain users testing it before they released it? They just didnt announce it and just dropped the bomb as it released.
And i mean i have take actual songs (that i own, or friends own that make their music) and uploaded both to Udio and Suno, and Udio can like 30% of the time copy more or less exactly what i uploaded and continue in the same voice even, with new lyrics and same music/melody. Suno dont even come close to that, like what ever suno makes is for sure "inspired" by what you uploaded but its still no where close to matching and continuning from where the sound left off.

And i agree, i use both tools as i said and pay for both of them! My ventilation tho is mostly bc i am starting to feel that most things Suno does Udio just does better, and i wanna have them kinda be good at different things or in this case Suno to kinda "catchup" a little with some things. Bc it feel like they are being left behind right now in terms of how good the services are at actually generating stuff.

Once again im saying this as person that is a yearly sub to Suno and just pays moth by month with Udio.

[unique stone]

It really varies. I get good and bad generations from uploaded audio in both, ranging from "wow that turned out great" to "what the heck happened here". And sometimes I'll get continuation of a song, other times the AI will take bits of it and do something creative with them... In both cases I've managed to upload part of a song generated by the other and have it continue.

The only thing I wish were better is the audio quality here.

[unreal trench]

hi peeps, i can't publish my song... how come?

[charred merlin]

Can you elaborate?

Just on a whim, I'm going to guess that by publish, you mean set your song to "public" on the suno website, and I'm going to infer from the timing of last night's release that the song starts off with an uploaded audio track that includes vocals and/or copyrighted material?

[unique stone]

Think this just got answered in general channel

[unreal trench]

hi there.. yes it's an audio upload and i'm already getting lots of infos in the general... thanks a lot....

[grizzled slate]

I'm working in Custom Mode. I've been adding song ideas into the, Lyrics section, then clicking, Generate Lyrics.

I have tried different procedures:

1. Add idea, Generate Lyrics, leave the Style of Music blank, then Click Create.

2. Add idea, Generate Lyics, add Style of Music, then Click Create.

3. Add idea, Add Style of Music and then Click Create.

The Output is different, as you'd expect, but my question is about the Generated Lyircs.

Question: Are the Lyrics Generated independent of the Style of Music, or influenced by them?

Hope you understand and have an answer too.

---

---

Upload Audio question:

I have an Upload Audio file of 50 seconds. I've extend this and set the extend time at different points, 12, 20, 30, 43 and 50 Seconds.... But there doesn't seem to be a major difference in the OutPut.

Can you please give me a brief description of what is happening with the Extended Uploaded Audio?

Many thanks

[austere minnow]

What does Suno's new audio upload tool do? I don't understand it.

[grizzled slate]

It's great! It will take what you give it, ie and Uploaded Audio file, or your own rhythm or song, which you then Extend. Then Your song is based on Your ideas. Hope tha helps!

[open mortar]

My experience is that you should always see the lyrics window as a seperate prompt. in fact you can write two verses and then write "finish tis song with chorus and outro" or "write two more verses" with different outcomes, if you write "write one pop-verse and one country-verse"

[elfin shard]

is it just me or is suno not loading in the browser

[charred merlin]

Just you

[elfin shard]

damn

nvm it's fixed

[alpine venture]

https://suno.com/song/8cc3b70a-8abc-4d85-9feb-9d1aaf881616

I think this establishes that when you toggle the instrumental toggle, [instrument] is in the prompt

as it cleanly reads it

so that's interesting

Also an interesting method to get promot information

that's hidden

[timber river]

all you have to do is get the link you want to share and change x to vxtwitter in the link and it should work

[modern belfry]

I have a question about uploading audio to Suno. I have a youtube channel where I'm basically bringing back to life old songs from the 1900s to 1950s. So I found one today that would be cool to do as a remix (recorded in 1927). Now any song that was recorded from 1929 down is now considered public domain. Would uploading this song on Suno cause issues? I know I don't own the song, the copyrights are expired. Soo?

[grizzled slate]

Hi and thanks for the reply. So you don't think the Style of Music and Lyrics generated are connected? I didn't realise you could determine the style of each verse. If I didn't add a Style of Music, I could determine the Style in the lryic prompt? Do you have any more tips that work in the Lyrics section, please. There doesn't seem to be enough information, so we all end up doing the same thing and that's just a was of talent. Many Thanks. Any advice and links that would help, please let me know! 👍

[grizzled slate]

No... It's now Public Domain... so there is no copywrite issue! Hope it goes well. Would like to hear the result!

[silk basin]

How about music recorded from back in the day that’s technically not owned by anybody like the MP3.com archive?

[open mortar]

No i dont think they are connected. I would add "make a dance lyric" and the more "prompt" the better, and pressing the generate button can always make things disepear so save what you like in an textfile during generations. Example: i wrote "make four verses" it wrote four new verses and deleted the first verse i had. but writing "write two more verses and chorus" the first verse stayd.

[modern belfry]

Ill take a chance on it lol

[modern belfry]

heres my channel if you wanna check out songs I already covered. https://www.youtube.com/channel/UCnS44Js78Ef-6DXMea5YzaA Mix of many styles on there.

[sterile finch]

I don't know what happened since yesterday but all the generations that I create the Ai is incorrectly using prompts as lyrics rather than following the request. I used the same prompts I used yesterday to test to see if it's just what I had written but it's doing the same. There's also been a huge dip in quality. I'm not sure if it has anything to do with uploaded sound coming in play and possibly affecting the servers or whatever but even when I'm not using uploaded audio it's been sounding like it's being played through a tin can and not using the prompts correctly as they previously did.

[charred merlin]

https://cdn.discordapp.com/attachments/1102676217687195679/1250765434035372133/Screenshot_2024-06-11_013622.png?ex=666c2195&is=666ad015&hm=ece4d0154a6c198d3a3bc11a1ccf6b0f16576356aba2227279c76decfc939bee&

Do you "own" or exclusively control all rights to that content?

If the answer is anything other than "yes", you risk getting your Suno account terminated. It's simple as.

[silk basin]

Not really? It's in a murky area though, abandoned music of a website that doesn't exist anymore that was lost media for nearly two decades after MP3.com got purchased by CNET. Legally nobody owns it or has rights to it anymore.

Then again, I don't know how we treat stuff uploaded to the Internet Archive

[charred merlin]

nobody owns it
"I understand that I am not allowed to upload content I do not own or exclusively control the rights to it, and that if I do so in spite of this prohibition, I will (among other problems) be breaching a contract with Suno."

If you want to go for it anyway, that's your prerogative, but you did agree not to do so, at least through the tools provided by Suno.

[unique stone]

Even if you can't obtain the music any more, it would still have been created by someone who would hold the copyright of it.

(Edited because what's up with my ability to type legible sentences today)

[charred merlin]

There's other open-sourced music gen software out there, you can plop "The Beatles - Hey Jude" into, and turn it into an 18 minute rock opera without anyone stopping you.

[weak kiln]

is there a way to download all my audio from suno?

[charred merlin]

Not in bulk, yet

[weak kiln]

i was hoping in bulk. I've done a few pages and i would like to download and post process them

[prime vortex]

I wonder if it would be all right to use a sound effect of a rooster crowing from the website called Pixabay in Suno's audio upload feature. I did check on Pixabay's website, it seems like it would be allowed. I would like to see if I could make an opera song with lyrics based on the crow of the rooster. Could someone please give me some advice, I don't want to accidentally do something that's not allowed.

[open mortar]

Yes if the downloaded clip is under free use you are allowed to "Modify or adapt Content into new works"
But they also have things that could involve copyrighted things so be careful with what sounds you download.
You could also use elevenlabs to create your own soundclips. Then you can get sounds like this:

[prime vortex]

@open mortar okay thanks very much, I'll look into that and keep it in mind. Thanks for your help.

[tiny tangle]

If I want an epic, final note - a crescendo for the vocalist, what would be a suitable prompt?
I'm thinking "crescendo final-note", does this line up with how Suno takes on board the style prompts?

[floral sonnet]

hello everyone... quick question : is there anyway to force an exact tempo from Suno >?... like if your 'upload audio' is 140 bpm can you tell Suno to output at 140 bpm?

[remote flume]

Wow, we woke up to a big present

This is like Christmas

[naive quail]

I'd try putting 5 to 6 seconds of a beat at the beginning of your upload and then extend from just after that. It's done a good job of matching the tempo of beats that I've tried, tho they've been the only thing in the upload. Worth a shot, at least.

er... or maybe at the end and extend from just before it so that it doesn't throw that back in if you merge it all together.

[floral sonnet]

ok thanks... I'm in a bit of a brain fog about extending and editing parts together...I think i need to go back to' extending / time reference / make a complete song '-- school! ...

[naive quail]

it's definitely capable of pulling the beat from the uploads, at least. I tapped out a quicky with a ballpoint pen to test that this morning: https://suno.com/song/6ae32d49-e552-4593-a797-6f7211677999

[open mortar]

Sometimes just writing the text as LOOOOOONNNGGG! accually works. but also maby try adding [high notes] or [vocal] even [outro] can do wonders sometimes

[tiny tangle]

Thank you, I'm going to add some tags into the custom lyrics to see how they influence things ❤️

[open mortar]

here are an example too https://suno.com/song/948bc3c9-c47b-47cc-b4d3-e7c8c410fec5

[tiny tangle]

Ah, okay - so multiple tags in a row can add layers of instruction. Big help, thank you 🙂

[floral sonnet]

😦 no the tempos dont match

so annoying... it would be so simple to program it so the output could be fixed at original tempo...

[neat field]

https://suno.com/song/0fa458ab-41c9-4716-b29d-8c21ef217cce/ my first attempt with v3.5 to create an audiobook short story equivalent with a nice voice

Experimental theatre, indie radio drama, female spoken word seems ok, maybe TTS would be more emotive, maybe not

[hollow hare]

Maybe adding a tempo keyword would help? Try one of the Italian keywords #📚┃suno-school message

[wet fox]

Is there a way to make a song with the lyrics and everything and then somehow take the voice out of the track? I'd like to actually vocally myself sing on some of the tracks

[tiny tangle]

Sorry if I'm mixing things up, when you refer to tags, do you mean words that go into the style prompts, or [words like this in brackets] you put into the script?

[tiny tangle]

Omg karaoke versions would be incredible 😍

The function could be called karAIoke 😁

[hollow hare]

In this case, they would be part of the style prompt in custom mode. I do not know if using it as a metatag (the prompts in the square brackets) would work, something to try out.

If you were not aware about the wiki, be sure to check out https://suno.wiki

[tiny tangle]

Ah, thank you for the link.
I've tried to pick up tips from reading all the convos here, but there's a lot - browsing the wiki will be so helpful 😁
Cheers!

[hollow hare]

The following regarding the weighting among the style prompt may help you as well #💬┃general-chat message

[slim ether]

hi, can we download other users' music if they allow it?

[neat field]

this is not a Suno feature exactly, and yes if they upload somewhere else and grant permission

[final hazel]

In this song I use a babies heartbeat which will stay steady into the whole song. So it can be done with the right sample.

https://youtu.be/dLUHgVUXaQA?si=PkKcwvHw-tuAXK9i

[charred merlin]

@tiny tangle search in this thread for “stem separation”, “UVR” (ultimate voice remover), or “OpenVino”

Happy karAIoke’ing!

Not a Suno feature (yet?) but possible with external software ❤️

[wet fox]

i tried voiceremover, but it kinda just distorted the portion of the song the vocals would be at

[tiny tangle]

Thank you 😁
Very helpful, time to plan an AI themed karaoke night 😜

[wet fox]

trying openveno now

openvino*

[bright isle]

Hello, if i have a question about the usage of Suno in a video who do i speak to, and yes i have read the faq. thankfull for awnsers.

[vestal rapids]

what do you need to know?

[bright isle]

if a song is made with a free Suno account, and is used in a Youtube video but you demonetize the video and give Suno Attribution is that ok?

[vestal rapids]

do you have monetisation on the YouTube channel?

[bright isle]

it is not for my youtube but for a recruitment video for a Discord rp server for a game but i do not think they have monetization on the YouTube channel

[hollow hare]

Covered in the FAQ (Suno's Help page).

The free plan songs can be posted the songs on free social media (e.g. YouTube) but you must credit Suno (e.g. "Made using Suno.com v3.5" or variants as simple as "Made with Suno" in the description). You cannot monetize it either if it was made during the free plan.

[alpine venture]

so, on technical gen stuff

I wanted to really say, people are a little bit sleeping on (or maybe I just have weird ears) using style "noise" to influence style

I basically now have an entire album using this concept while the "getting is good" bc frankly, I've never heard anything that sounds like the most recent stuff, before, and I really love the idea of making new genres or musical progressions using AI tools, especially ones nearly impossible with regular instruments

the most obvious example being from today, where the "aaaaaaaaaaaaaaaaaaaaaaaaaaaaa" at the end alters the entire progression

https://suno.com/song/5f6d1c00-401d-4caf-a101-060331725d86

in general, throughout the recent stuff, the "hallmark" has been using syllabic noise to produce both instrumentation AND vocalization, and use that other further amplify 3.5s tendency

towards bright, fast pace, etc

using sounds within the "music style" section at the end. It concatenates it into your lyrics usually, and will sometimes mix it in, creating sometimes very odd rhythms

https://suno.com/song/694738f8-948c-42e3-add5-639ada5627aa

another clean examples

click on both of the above

and look at the prompts

in particular, note the first one is not dunstep

dubstep

not at all, but it is

it's like, glitch hardstyle dubstep

but without any of that prompted

so it retains the speed of hyper-X, the deeeeep bass of suno trap, and the complexity of math

listen to nigun of closing doors

it's still more trad

but hang on

[glacial scarab]

i think especially V3.5 benefits a lot from complex (and weird) prompts

[alpine venture]

it certainly breaks much farther from 3.5

https://suno.com/song/4cda59a3-435e-4114-a21c-9296e92a2b35

that's more extreme of an example

In any case, we can talk pedantic

about what odd means, but I don't mean outsider art, if I want that I can go to 2

I want strong progression and interesting coherence, but I want to force 3.5 into a very awkward space

and this method does it imo

I may be in the minority on suno, but broadly in regular music, I personally think people want things that are somewhat close to that "edge" bc it surprises them

so that's basically the rub: pleasant progressions where you're left asking "why did I like that/did I like that"

nice; you can even just stack special characters. I find that doing all of one type often will get a strong opening instrumental, but it likely depends on your genre

[night burrow]

Hello when i type /chirp it shows this message please help

[naive quail]

Chirp has been decommisioned. You'll need to use the website: https://suno.com/create

[night burrow]

@naive quail is this not working on discord app, i have app and i am trying from discord app. i am first time user

[prisma wraith]

Chirp is still gone, so you've gotta go through https://suno.com/ if you wanna generate music

[night burrow]

how can i generate music from my discord app. is it generating from discord app or not

[torpid quarry]

no

website only

[night burrow]

is suno have his app or not

[torpid quarry]

suno have his website

[night burrow]

how can i purchase plan

[torpid quarry]

on website

[night burrow]

i can't show any option of subribe plan in website

[night burrow]

is there any relax mode like midjourney?

[torpid quarry]

no

[night burrow]

where i can download my creations. for example i have generate 100 songs so where i can download all my songs directly. is there any archive,

@naive quail if i am a paid member, so i can use generated songs for youtube, for monitize?

[naive quail]

That is my understanding, but I'm just another users, so don't take my word for it!

[night burrow]

@naive quail sorry brother can u give me any name of support

[naive quail]

moderators are the ones dressed in blue 🙂

Ping them directly and you may catch some heat, tho. Someone should chime in soon.

[night burrow]

@naive quail thank you

[unique stone]

@night burrow You can monetize your songs if they were created during a paid subscription (even after that subscription ends.) This is covered in the Help on the website -> https://suno.com/docs/faqs

There isn't a way to download multiple songs in one go I'm afraid (I am downloading a bunch of mine one-by-one at the moment and it's taking me a while...)

[charred merlin]

https://youtu.be/lj8utBd_IUM

Watch this video that 3Daizy made, it will help you get started ❤️

[hollow hare]

Here is the link for the latest version btw

https://youtu.be/hsecBccWETM
(Check the video description for the links to other languages if needed)

[charred merlin]

You're a doll ❤️

[sullen harness]

Suno is down, how my life is going to be without it

[hollow hare]

Was it the CloudFlare screen? Try again in a minute

[sullen harness]

yes

[past cliff]

so how is it determined that something can be public or not when using pre recorded samples?? i recorded my voice making "mongolian throat singing" sounds and nothing i extend from that is able to be public,

the craziest part also is that some songs extended from the same voice clip are able to be public while some arent

[naive quail]

pretty sure this is a bug right now. I've run into the same thing

[past cliff]

interesting. appreciate it. i see other audio clip upload extensioins that are able to be public also so weird

[primal socket]

Hi, everyone! How do I stop Suno from generating 2 songs every single time?

it's basically wasting my credits with that functionality

[charred merlin]

This further makes me believe that Suno model uses some form of audio diffusion, because I've heard instances where a tone can morph between instruments or vocals for a split second. If some noise (literally, opposed to image noise :P) is added to the uploaded clip, it depends how close it is to discernable vocals for the extension.

Throat singing is ambiguous enough that it could diffuse into someone singing "hey oh oh oh", or an unintelligible rhythmic noise, so that would influence whether the track can be made public or not.

[charred merlin]

Not wasting credits, that's the basic functionality

[naive quail]

I've had a verse sung with a didgeridoo voice 😄

[past cliff]

that actually makes sense

[primal socket]

2,500 credits renew monthly (500 songs), but in reality it's 250 songs and their closest variants

[runic dock]

I am trying to use the song upload feature using a 60s acappella song. How extend this song with full instrumentation ? It seems ignoring all styles i give . I get another accapella and sometimes one instrument at the end. Any advice?

[charred merlin]

#💡┃suggestions-web

[lone oak]

Saw the new update, does anyone know what the Audio Input feature is suitable for?

[strong falcon]

Is it possible to create an audio with lyrics, and then, if we like it, get the same instrumental but without the lyrics parts ?

[charred merlin]

search in this thread for “stem separation”, “UVR” (ultimate voice remover), or “OpenVino”

[lone oak]

UVR 5?

[primal socket]

thank you I wrote in #💡┃suggestions-web . I can't believe the option to select outputs for a prompt is not available 😄

[lone oak]

Can the Audio Input feature train human voice?

Are new features just a prelude?

[past cliff]

for now. you dont have to keep the prelude custom part after you extend

[junior sun]

If I upgrade to the premium plan from pro. And I already used up all the 2500 pro credits. The potation thing wants to bill me 30$ even tho i payed 10$ for pro so... proration logically would be the diffrence between pro and premium... so 20$ and i get the extra 7.500.... So am i getting 10k or 7.5k for this 30$ Seems like we getting screwed if we upgrade... unless its really 10k then that would be fine I guess. Its unclear

[lone oak]

Seems to have saved some troubled arrangers

[strong falcon]

thx i will check at will

[lone oak]

There is no inclusion relationship between the two membership levels

[junior sun]

i have pro active right now. they have the option to upgrade imedialty to premium but they charging me 30$ instead of 20$

[lone oak]

This means that you will get 12,500 points by opening two levels of membership

[junior sun]

am i getting 7.5 or 10k for the 30

ok

site should say that its unclear

the word proration makes that confusing

cause what your saying its not prorated

[lone oak]

I hope you can use up all the points. The first time I did this I couldn't use up all 12,500 points.🤣

[junior sun]

loll

yeah im debating i got 12 days to burn that 10k if i do this haha

[remote flume]

I just signed up for the yearly due to this update

Good job suno

[lone oak]

Is it the update on the 12th? Can you share what the recent update is for?

[placid sapphire]

yes you can monetize your music, you own it once its created.

[charred merlin]

• obligatory "Provided you have a paid account, otherwise Suno owns the song and you're not allowed to monetize it, and are required to credit them when posting."

[nimble citrus]

👏 🧹

[lilac garden]

trying to fuse 2 genres on an instrumental song, 3.0 and 3.5 are picking only one of the genres... any tricks to resolve this ?

[charred merlin]

Try separating them with hyphens? country-metal opposed to "country metal"

[velvet bolt]

Do you want it to do multiple genres at once? Or switch between the two? Try specifying which instrument is which style

[lilac garden]

i'd like both genres at once

[vernal harness]

is the upload audio working for anyone?

[hollow hare]

You need to have a subscription to use it

[vernal harness]

I do 😦

I keep getting:

[hollow hare]

Ah, I see. Unfortunately I do not know how to solve that one.

Maybe try another browser or incognito in case something is interfering? Maybe a VPN if you have one?

Hopefully someone with more experience will respond soon

[vernal harness]

Thank you!!!

[soft thistle]

Anyone else get this message yet? I used Suno yesterday to make an idea based off a beat I made. It came up with a good melody so I used it today and reuploaded to try some new ideas. Then I got this.
The original uploaded beat:
https://suno.com/song/b4a42652-ee63-4ee5-a397-e0eedae23aba
The idea Suno made:
https://suno.com/song/213f5761-3f7f-4969-ae3d-8e82b8c44025

The new beat using the melody from Suno's generated idea that I'm currently trying to upload to try out more ideas.

[silent viper]

I can't use the feature: "Extend".
Neither in version 3 nor in version 3.5.

Anyone else having this problem?

[grizzled slate]

Nice idea....

[vestal pelican]

Hi guys, I'm facing a issue with a song where ever I mentioned male or female for vocals it's gives the opposite also sometimes randomly, any suggestions

[worn comet]

one of your prompts is strongly associated with a particular gender

remove that one

pop is female, alternative rock is male, that kind of thing

[alpine venture]

https://suno.com/song/e259d41b-cf50-4d58-b393-0f420b5f32de

well that one got very stuck 😄

[floral sonnet]

hi everyone... still asking aobut the tempo... can we lock the output to the same tempo as the imported audio >?... all my generations are 1, 2 or 3 bpm different... which is super annoying if one wants to use them together later on ....

also is I upoad audio for the intro lets say can I upload a new chorus and then somehow edit the two ideas together ?... i don't know how to keep one idea going without starting a new idea... i now I'm not really explaining ti well.. sorry... but if anyone can help thsat would be cool..

[floral sonnet]

Also I am looking for one on one guidance from anyone experienced... pm me plz

Also ok.. lets say I've uploaded audio that Ive asked Suno to extend.. is there a way I could have the output starting at the start of the new stuff that Suno generate ... like a checked box a bit like the 'Include Source Image In First Frame' that you have in Kaiber (AI visual)... ??

ok so message to the developers "there are key areas where this program could be so much better"... (ok, ok, so its pretty amazing already..as it is.. BUT it could be AMAZING! ..) 😄 and in this biz the first company that implements these things will win!..usually..

[versed plover]

hi,everyone , can someone talk with me? How to collaborate with Suno Company? We are an independent development team from the Chinese community, working on accessibility projects. We hope to gain access to Suno's API.

[unique stone]

Press the "..." on the song, select "Get Whole Song". This will join the uploaded audio with the extension and create a new clip in your library.

Heh... Usually people want the opposite (exclude the uploaded audio when using "Get Whole Song")

You can then generate more extensions off the "whole song" it creates if you wanted to.

Being able to upload audio and insert it somewhere else in the song would be good (e.g. uploading a chorus). If you have ideas on what could be improved, the best place to put them is #💡┃suggestions-web @floral sonnet 🙂

[gray crater]

If I dm chirp the command ./chirp it says the application did not respond. How to fix?

[rotund mica]

I have paid subscription and today my Library does not open at all. What is the matter?

[trail fulcrum]

it would be nice if they worked on the quality of the songs. Sometimes it sounds really bad

[unique stone]

The Chirp bot has been retired.
The website at https://suno.com is where to make music now

[gray crater]

Can you pay with PayPal somehow? To get credits

[gritty olive]

How to get rid of drums? I want a sad song with lyrics and just a guitar and it keeps pushing drums in there, so I change it and I want a sad classical song with lyrics, drums again, so a sad orchestral one with lyrics, it put drums in there again, so I write without drums, it always stuffs the same or rhythmically similar drums in there every time as soon as it makes the transition to the chorus and during the chorus

[floral sonnet]

is there any reason why the upload is not working at the moment?

[modest oasis]

Is there any way to keep a song you like but change 1 or 2 words in it without creating a whole new song that sounds different

[hollow hare]

Not yet unfortunately. The in-painting (or w/e it was called) is very likely going to be in v4

You can try extending but sometimes that may change how it sounds because of how it works (but you can try adjusting for a few seconds until you get a similar/same soundand manually overlap in a DAW if the rest does not sound the same)

[modest oasis]

Thanks for the information

[modest oasis]

Discovered the app yesterday and just hit 100 songs

[hollow hare]

Awesome 😮 Since you just joined, you may not be aware of these resources:
https://suno.wiki
https://youtu.be/hsecBccWETM
(Check the video description for the links to other languages if needed)

[hollow hare]

(See above in case you missed that)
Information about Extend can be found here: https://www.suno.wiki/faq/making-music/how-do-i-make-the-song-longer/

My blurb about it:
The issue is that the AI does not remember the original style used. What it does, for each extension, is study the style ~1 min (unsure if this got extended to 2 min for v3.5) prior to the continuation point to learn/mimic the styling you had (which is why extensions in v3.0 are 1 min long as the other min was reserved for studying).

So if you had a loudness or degradation issue within that minute, it would start becoming emphasized as that is what the AI picked up on and all future extends builds up the issue as long as it exists within the 1 min of continuing.

It also explains why the style can change when you vary the starting point by a few seconds

[modest oasis]

Thanks for the information. Love this program

[charred merlin]

Hrm, 3Daizy's v3 video had more timestamps than the v3.5 version -- just as helpful, but looking at the "chapters" for the v3 video you could see sections for downloading, etc. (not a critique Daizy, just an observation!)

But then again, if someone can't be bothered to watch a 3 minute video to help them learn everything they need to know to use Suno, I doubt anything beyond explicit instructions for their exact questions would help.

[surreal mango]

Are there any plans to add a division of music into components?

[charred merlin]

Search this conversation for "stem separation", it's not built into Suno (yet?) but it's available through external software

[foggy jasper]

@final hazel learned some pretty informative stuff from your dubstep track. i had no idea to arrange meta tags that way

[undone sleet]

i want to extend one of my old songs but i cant see it anymore in the creat tabe since too many had already been made and theres no way to change the page

[charred merlin]

Extend it from the library page

[undone sleet]

i tried that but it just send me to the creat page

with the bar that would usually have the song still not filled

[charred merlin]

Once you click "extend", it will show the song you're extending at the bottom of the page

[undone sleet]

yes and its missing when i do it

like usually the song youre trying to extend is shown at the bottom along with where you want to extend it from right?

but when i do it the song aint there and its acting like im still making a new one from scratch

nvm

logged out, refreshed the page, logged back in and it started working

was worried for a bit there since this would be the first song im going to officially publish as an ai artist on spotify

[naive quail]

This happens to me regularly. A refresh is all it takes to get it working again. You can probably skip the log out/in step next time.

[old saffron]

Please wait.
Please wait for your other generations to finish.

this now always comes when i want to create a new song

i have credits left

last time i was able to create one he did 2 new lines but then they disappeared after some time

now it doesn't work anymore

[unique stone]

The server is down at the moment - #💬┃general-chat message

[old saffron]

ah okay.. so need a bit of patience

[quick magnet]

Is it possible to change our display name and handle? I've been trying but just getting a spinning wheel of doom and then nothing..

[charred merlin]

#📚┃suno-school message

[primal socket]

well, I have had 10 songs generated out of thin air, because of that problem, how do I get my credits back for those?

they keep appearing in my freaking library - now there are 14

[charred merlin]

I'm curious if the songs generated "out of thin air", or if you hit the create button a bunch of times in a row wondering why it wasn't doing anything, and now all of those generations are processing.

[old saffron]

mh.. now the things i wanted got created but the 2 songs i did just before server down are "unfinished".. they break in the mid of the song

[primal socket]

Yes, that's exactly what I did. I assume there was some latency created between the commands I sent and the problem with the server and now they are coming back from the queue.

16 songs now

And why is it generating TWO SONGS every single time? Arrrrgh, I hate it... it's basically taking half of my credits for nothing.

[charred merlin]

Nope, not taking half of your credits for nothing, this is the default functionality of the website.

I suppose you could email [e-mail] and explain that you clicked the button a bunch of times without realizing it would create songs for each press, and that you would like your credits back?

[primal socket]

ok, will do that, thanks

and no, that's not "functionality", functionality is the the quality of being suited to serve a purpose well, and taking the money for 2 songs every time I want one song created does not equal serving a purpose well

well it doesn't serve my purpose anyway

[charred merlin]

In the future, if doing something doesn't give you the results you're hoping for, it can be helpful to investigate before doing the same thing 8 times in a row expecting different results.

[robust granite]

I'm having a problem. Do you know how to solve it?

"Generations have been temporarily disabled while we investigate a system issue."

I searched on the internet and found that it will take some time, but I am asking because the problem continues to occur even after 2 hours.

[unique stone]

Thanks for answering this @charred merlin

[charred merlin]

https://tenor.com/view/watching-umm-hmmm-gif-18025084

[unique stone]

(I can't give detailed responses atm as I'm working and just glancing at what's going on occasionally)

[charred merlin]

Best of luck, devs. ❤️

Btw, you can still listen to & download your existing tracks while we wait!

[robust granite]

There is no other way for me to solve it and all I can do is wait, right?

I need to support the developers...

Thanks for the reply!

[solar palm]

You could fix the system issue, I guess 👀

[charred merlin]

https://jobs.ashbyhq.com/suno/9e6da9b6-8562-4d9e-ae8e-c3319f76bdba
😛

[robust granite]

I hope it gets resolved quickly. I was using it well.

[charred merlin]

It's funny how quickly Suno becomes an integral part of our lives.

It's like when the power goes out at night & you think, "Okay, well I'll just put on an old movie. Oh wait, no power. Okay, well I'll just flip on a light and read. Oh yeah, no power."

[wanton siren]

hi. will there be an easy way to get the duplicate generations refunded? sry, I don't think it's my fault when the webpage isn't giving me proper feedback of what's actually being generated and what's not.

[primal socket]

[e-mail]

[charred merlin]

The site is undergoing maintainence at the moment-- no statement has been put out at the moment regarding duplicate generations. From what I gather, it seems to be from people clicking "create" and not seeing an immediate result, so they click the button multiple times, leading to delayed generations.

Gogu linked to the support email if you want to reach out to the team directly.

[wanton siren]

oki, thx for the infos

[mystic hound]

bless you for this. i didn't realize they were hiring. ❤️

[charred merlin]

Lots of jobs on the Careers section, as well as a general application if nothing stands out, but you still want to put your name in there.

https://jobs.ashbyhq.com/suno

[tranquil estuary]

This audio input feature.... is it good sound quality wise or does it still have that fuzzy sound that the top end always had back in ver 3? It sounded like a constant de-esser or something. I'm not too keen on paying for a subscription if the output doesn't sound good. Can anyone enlighten me? 🙂

[ashen bane]

hey am i the only one with this

[charred merlin]

#📚┃suno-school message

There's a lot of backlogged generations, we're getting back to normal after a speedbump

[unique stone]

The overall quality hasn't changed. The v3.5 model was released to address some of the more annoying issues (quality degrading as the song progresses, vocals turning into choirs etc.) but even with audio input it'll go through the v3.5 model and not be perfect quality. There are often examples posted in #💬┃general-chat of things made with audio uploads, so it's worth checking there to get a feel for how it sounds.

v4 is being worked on. I'm hoping it'll have better quality but no info on that yet.

[ashen bane]

ok thanks

for ur answer

[jagged latch]

Hello Dev ..
Is there any Payment Method in the Next Update ?
Coz a lot of people in my Country (Indonesia) Too Hard to pay with Credit Card.. Most of people in here Pay with "DANA" Or "GOPAY" Like Discord Payment Method .. Thanks

[tranquil estuary]

I see, luckily I haven't had the choir issue so far. But the de-essed highend is unfortunately a huge problem if I wanted to extract and use the vocals for example.... Perhaps it's better if I wait until v4?

[charred merlin]

#🪲┃bugs-web message
They're working on it, it will just take time ❤️

[unique stone]

There's possible workarounds like using a stem splitter and then using another service to replace the vocals, but that's obviously a lot of time/effort and my results doing that haven't been great.

You can always try out some free gens to judge the quality yourself (no audio uploads on free accounts though) but it might be worth waiting to see what v4 brings.

[ancient glacier]

Other languages videos don't show the lyrics

Anyone help please

[hollow hare]

I assume you mean it shows block figures instead of lyrics in the video generated?

Unfortunately, you may want to report that to #🪲┃bugs-web and ask them to have unicode support in whatever program they are using to generate the video.

IMO, if you are planning to make it available to the public in YouTube, just make an AI video for the song instead and use the native caption/subtitle tool in YT (or a 3rd party that supports your language to make a caption file)

[opal sand]

em how do I do vocal chops with suno lol

[opaque canopy]

when i make a metal song and put [drum solo] or [bass solo] nevers works but [guitar solo] works 50% of the time.... how do i get my prompt to make a bass or drum solo i like guitars but i mean ..... versatibility

and.... next update please make sure the [end] prompt work ?

[ancient glacier]

Yes

[weary wharf]

I usually create rock songs, but this works for me %90 of the time.

[Rock Guitar Solo]

[Rock Drum Fill]

[Wild Saxophone Solo]

Not just the solo, but with the style or genre works well.

For ending, drop to v3, the quality will be the same, but the time window drops, so make sure you do not have lyrics from the middle or start of the song, they will sound different.

[opaque canopy]

nice i will try those prompt!

[weary wharf]

With 3.5 this works around 50% also

[Interlude]
[Instrumental: Drum solo]

[wanton siren]

is the songradio feature considered to be bugged atm or am I missing something? coming from english metalcore for example, it just starts a list of seemingly random genres, styles and languages. On the other hand, some songs (even with 'easy' tags like Dark Synthwave Rock) won't bring up any list at all.

[daring lava]

Thank you @weary wharf for helping members here! Truly appreciated ❤️

[daring lava]

Thank you for bringing this up, looking into it now

[sturdy flax]

Do cover images matter? Like, can a great one be the difference between a new public song trending or not? thoughts?

I am implementing the strategy now with my latest, but not sure how much difference it will make - bit if it gets people to click then...

[charred merlin]

More of a social engineering question than technical, but I wouldn't be surprised.

[brittle apex]

I think it makes a difference. If it is a deformed bird or some weird tie-dye it will get less clicks than a polished image

[charred merlin]

I'm sure having a photo that shows some sort of thought will garner more attention than the guitar-playing bird ^

[sturdy flax]

That's fair

Really I like that there's not people spamming their songs here LOL - thanks for tue feedback friends

[placid sapphire]

Absolutely matters ,just like YouTube Thumbnail's are most important to unsubscribed viewers, Not sure the Image builder on Suno is something i would sue though to promote music, They're not the best pipeline ,try Leonardo AI to design covers for your video downloads art.

[sturdy flax]

Oh for sure, I made my latest with Dalle and Canva

[charred merlin]

If it helps anyone get started making cool art, I love the Copilot app because you get 4 variations on your image prompt, and the mobile app lets you swap between a square & landscape version (similar layout, but generated differently), so that makes 8 variations per prompt, but you can apply different filters to regenerate as well, so there's a crazy amount of variations you can get through basic experimentation.

Then taking the photos into a free editing app like SnapSeed lets you make a bunch of in-depth edits without having to train yourself on all of the crazy settings Photoshop has

[sturdy flax]

Good point, I applied for the Suno Community Lead role and thos perspective is useful 💯

[charred merlin]

Oooh, good luck! I'm not in Boston/NY, so until they accept remote work, Canadians like me are out of the running.

[sturdy flax]

Well, I'm in Seattle but the Community Lead role was listed as remote

[charred merlin]

I think it's more of a hybrid thing from what I've been informed.

[sturdy flax]

Dang 😅

[charred merlin]

APPLY ANYWAY

[sturdy flax]

Well, here's hoping they make an exception or carve out a special role for me LOL

[sturdy flax]

Oh I did, and reached out to a team member I sorta know

They are here in Seattle

I've been pushing

Posting tons of memes too LOL

[charred merlin]

I'm even crossing my toes for you.

Back onto image gen for Suno; I don't think I've shared any of my songs yet, so I'll throw a youtube track onto the suno showcase to show an example of the simple yet effective image gen/editing workflow I was talking about

[wintry saddle]

I assume we still have some backend challenges going on, I had some in flight gens reeapear done, and vannish again since.

[brittle apex]

That is my go to as well

[covert shard]

can you use chords?

[brittle apex]

You can upload a sample of a chord progression and extend. I've seen chord progressions used in prompts but it might not always work

[sturdy flax]

Not really, but eventually

[sharp kraken]

it definitly doesnt Always work - but i do it all the time. you can get alot out of time signature and scales

check it https://suno.com/song/99d92972-b8ec-4d0d-b070-6e060b749cf3

[still otter]

Hello, I am hesitant to take the premium version for the audio inputs but I would like to have more details on this one. How it works?

[sharp kraken]

it plays SOME of those chords, and freestyles a few in there

but i mean, whats the chances of prompting "D#dim_𝅘𝅥𝅰" and getting that same chord? kinda oddball chord you dont hear often

[red hollow]

Hello, I have a question, when you access the monthly payment, can you remove the horrible sound that is heard in the background of the songs?
And the excess reververance and lack of brightness in the treble sounds or is it still the same as the free version?

[sharp vector]

The generation is exactly the same, you just get more credits to try stuff

[wintry saddle]

thats what mastering process is for.

[red hollow]

It is not worth paying if the sound is still poor. The generation of songs and rhythms is excellent but the poor sound quality in some songs leaves something to be desired.

[charred merlin]

As a side note, maybe I'm absolutely clueless, but I'm not sure what this "widespread phenomena" is. Some songs have different intonations, but there's only been a few times that I've forcibly had to turn something off because of poor audio quality, even then I just reported it as a bug,

Admittedly, when I'm listening to extensions of a song they seem quieter than the original segment, but when you "get whole song", the audio is consistent throughout.

[daring lava]

Hey @still otter thanks for reaching out! Here’s the run down on how to use it and what to expect…

To use the audio upload feature you will:

1. Go to your library and upload an audio file (6s-60s long)

2. Extend from that upload

3. Include tags to influence the generation extended from your upload as you would a regular generation

This tool is great for extending ideas, building more depth, exploring possible song directions, creating fun music from pretty much any material (given you own the rights to it), and influencing music you’d like to generate.

Hope this helps.

[glacial lion]

What a crazy prompt, but I like the song

[charred merlin]

There's quite a lot of us who don't seem to have the same issues. Perhaps if the state of Suno's text-to-music generation isn't quite to your liking, you can try some other services, or have patience as this literally bleeding-edge technology advances from week to week.

[red hollow]

but that process only appears in the paid version?

[charred merlin]

Yeah, David wasn't replying to your comment

[sharp vector]

The upload and extend is interesting. It does show where the weaknesses in Suno are. If I make something with a very wide dynamic range it'll really struggle

[daring lava]

Is this regarding the new feature? Or just generating in general?

If from uploaded, keep in mind the quality of the upload will direct the quality generation.

If just generating in general, there are many options to better learn how to best utilize Suno such as SUNO’s community-made wiki or this community, to name a couple.

[charred merlin]

Have you tried moving the extension points?

You're right that complex tracks seem to diffuse less predictably than simple tracks, but there's still wiggle room for experimentation that I've found can give drastically different results.

[daring lava]

Also, thank you @charred merlin You are an amazing help! 🫡

[sharp vector]

I have a suspicion that in order to extend tracks, the audio you upload essentially gets encoded in its neural network - because it has some of that same weirdly echoey/bright sound when you get the 'full' song where it's playing your original uploaded segments.

https://suno.com/song/7e327f36-123d-4fe8-ba9d-069330b419fa - Original, uploaded track (my own material)
https://suno.com/song/9f6b72ad-e164-4767-8f5c-c2b6ec0479fa - Extended track, notice how different the start sounds

[charred merlin]

I don't think that Suno incorporates your tracks into it's "memory", but closer to something like Stable Diffusion's img2img (not saying you're wrong, just my thoughts on the matter <3)

[still otter]

if I captured everything you can take any sample and suno uses it to make music ?

@daring lava

[sharp vector]

Check out those two and see if you still feel the same way, because it really has the same tropes of something generated by Suno in the first instance

You'd have to have some super-weird encoding artefacts to cause that

[red hollow]

I only asked the question because I tried the free version and the sound is not very clear but it does not mean that I am not amazed by this technology for me it is incredible and of course I have patience it is not a complaint I just wanted to know to subscribe or wait for a little since I use it to entertain myself, it is not critical and I see that they are improving day by day

[sharp vector]

It's a little all over the place sometimes, sometimes I've had it generate very clean sounding output

[charred merlin]

I didn't listen to the entirety of the songs, and only based this off the first couple seconds of each, but this sounds like something I'd expect from an audio version of something similar to img2img.

It doesn't retrain the model, and the results are determined by the precise settings (graphical denoising, cfg, steps, etc)

The user Zoject Promboid mentioned uploading a clip of them doing mongolian throat singing which was allowed to be set public (despite it being vocals). Some of the extensions were allowed to be made public, while others were restricted.

From that, I'm deducing that the original content is mixed with generative noise, and diffused into the prompted content, as depending on how it's processed, mongolian throat singing could "diffuse" into vocals singing "hey oh way oh", or an instrumental digeridoo, explaining why some tracks are given the uploaded-vocal restriction while others aren't.

~~ ~~

It's so cool this is all evolving as we speak!

[red hollow]

So could it be something on my computer that is causing it?

[charred merlin]

Easier if we have agreement on what noise you're talking about. Can you share a track where it's particularly notable?

[red hollow]

https://suno.com/song/0fc4ffce-0b8a-41f6-b277-30d455598466

[charred merlin]

Is there a timestamp you can share of when this sound becomes particularly noticeable? Please be descriptive to help us know what you're trying to show

There's tracks I've produced while only using "experimental" as the style prompt that legitimately had screeching and feedback that I needed to report as an audio bug, but I'm honestly confused what the "obviously horrible sound in the background" is.

If you're talking about the white noise ~1:23, that's reverberation from a Chinese cymbal. You could 1) fix it in post, as suggested above, 2) generate the song in smaller extensions, cherrypicking the best sounding generations (which influence the rest of the song's extensions as well!), or 3) convince the AI that it should only use Zildjian's.

[unkempt shale]

The "Extend" on 3.5 is not working right. I get a song that's only 1:27, so I extend it. I start at like 1:15, but I get 2 versions that take it right back to the same end time. I just wasted 10 credits for that?

I am not keen on the audio ending at such a short time, but when I extend, 17 seconds is not an extension.

Would like to see a "Suggested extention time". Sometimes I make a song that I want to go for 10-15 minutes. So to have an option that allows me to use the whole amount of time in a render and not end the song (until I say so) would be much appreciated.

Thank you.

[red hollow]

I am going to try what you propose but first I am going to see that it is not a problem with my computer. Thank you very much for answering your questions. If the problem persists, I will contact you again.

[charred merlin]

I've only run into a couple songs that absolutely refuse to extend, but typically it's an issue of trying to extend from a "musical conclusion", or right at the end of a call & response.

I do agree though that it would be nice to specify the length of the extensions to better control the results, like if it's overpowered by wanting to resolve the composition, you could force it to try to go longer instead.

[unkempt shale]

I don't have much time to figure out where it will start to think about fading or ending, since the times it happens, the songs are relatively short. I haven't tried extending a song that is hitting the limit.

[golden plover]

Hi, I have uploaded a sample and the AI is not using the sample. I am a paying customer and I have wasted over 100 credits because of wrong outputs. Please help me on this. Thanks! The gmail account used is [e-mail]

It was working after uploading 3 samples, then after the fourth sample the interface is blatantly ignoring the instruction and is generating a song which has nothing to do with the sample.

[carmine osprey]

hmm my completions in like 20-40s uploaded clips are 3:32 for some reason ; is there something which affects the total generation length (assuming it doesn't END the song before max) ?

typical extensions are 2:00

not complaining - I am getting around 3:00 added just curious what affects it; sometimes the total time differs but among a set of completions tends to be teh same from the same base.

[cerulean jackal]

Agreed. I get so much static that I can't keep anything it generates.

[charred merlin]

Can you share a relevant clip, with the timestamp & what we should be looking for? The more specific you can explain, the better

Again, if you'e getting songs with audio quality you don't like, click ... > Report > Bug > Poor Audio Quality

[patent sparrow]

Greetings fellows, is there any way to specify how many bars I want in a specific part? For example a 16 bar intro

[charred merlin]

Not yet! As for now, you'll need to play around with generations, extensions & metatags. https://www.suno.wiki/

https://www.suno.wiki/faq/metatags/song-structure/

[patent sparrow]

Alright, thanks for the quick reply

[silent viper]

This is happening to me too.
Since yesterday I have not been able to extend an audio.
In both 3.0 and 3.5, the 2 options last no more than 1 minute.
As I try, my credits are being removed.
Frustrating...

[cerulean jackal]

https://suno.com/song/82570675-410c-48b5-a30c-9c60e079560b

Around 4:10 it starts to fall apart.

It's been getting worse lately

[fluid robin]

another instrument that works in suno = muffled trumpet

[cerulean jackal]

It's static.

[charred merlin]

@cerulean jackal Thank you for the timestamp! I can hear a similar distortion around 2:17, and it seems to be caused by the bass clipping.

The clip Nashgull shared earlier sounds like it might be closer to an issue of Chinese Crash cymbal (nicknamed "trash cymbal" for the crispy, tinny sound), this is a clear case of the bass amplification going too far.

Basic housekeeping out of the way; the best way you can directly help prevent stuff like that is to report each clip that has poor audio quality, ... > Report > Bug > Poor Audio Quality, that combined with thumbs down helps build cleaner training data.

But out of curiosity, do you recall where the extension points were in the song?

[naive perch]

With the uploaded audio feature, is that audio then sent into the eco system of training data? Does it have a permanent impact on the training data?

[cerulean jackal]

Thanks for the feedback. I get a lot of clips I'd love to keep but have to throw away due to the distortion, staticy sound.

[charred merlin]

One way to avoid that would be to use extension points to go back before the distortion starts and try to generate a clip that "goes hard" without just simply cranking the volume past 100%, even though it might take a bit more active work & editing, opposed to getting the entire thing done in a generation or two.

But I would be curious if you recall when any extension points would be so I could analyze the song with more context (eg. comparing peaks in the waveform to the extension points to see when the distortion is more prominent in relation)

[cerulean jackal]

On the one I posted, it was a 3.5 generation, so it carved out a 4 min clip to start, and I extened it once.

[cerulean jackal]

Interesting. So, I have control over the volume when generating a clip?

[charred merlin]

Not directly, the metatags can help give guidance; off the top of my head, if you put [soft breakdown] it might help influence a mellow bridge-type solo opposed to a song-ending ear blaster if you put [intense solo]

[cerulean jackal]

I don't usually put anything like that in my metatags. So kinda just default stuff.

But some of the distortion comes in hard even when it's trying to do soft parts.

[charred merlin]

The community wiki has a ton of gold in regards to that; https://www.suno.wiki/faq/metatags/

I rely more on style tags and hyper specificity in extension points to get exactly what I want. Not uncommon for me to roll 10+ extensions and pick one because it was heading in "the correct direction" according to my mind's ear, and I find those "encouragements" or "suggestions" to be way more effective than trying to bully the AI into a specific direction by repeating a tag or style prompt 10 times in a row, or getting upset that the one-shot generations aren't perfect.

Maybe one day someone will convince me to make a tutorial video for my workflow

[cerulean jackal]

I try this approach alot, but sometimes the song just loses the cadence of the verses and the extended parts do not match the original part. I then have to throw the song out because no matter how hard I try to extend it, it ends up sounding like two different songs.

[charred merlin]

Ahh, did you "get whole song" to extend from, or were you extending from an extension?

BIG difference.

[cerulean jackal]

Yeah, I do a clip, then "get whole" and then extend from there.

I never extend from a "clip" unless it's the first clip.

[brittle apex]

extending from the extension can be useful if you want it to shift styles (eg change vocals from male to female)

[charred merlin]

😛 It can be a cool way to specifically take a song in a different direction, but it sounds like you're already cognizant of it being a pitfall.

Not that it's an immediate fix, but in those instances, ... > Report > Bug > Doesn't Match Style Instructions ❤️

[abstract spade]

An important detail: who are the singers/voices of SUNO?

[charred merlin]

🤖 🤖

[abstract spade]

Better yet, who served as the model for AI?

[cerulean jackal]

I will definately be using the Report function more.

[charred merlin]

I don't know if they've shared any of that information, but Suno is presumably built on a similar framework as Bark, their speech synthesis project. That's open sourced, so perhaps you could look into that for better insight? Unfortunately I don't think that anyone here knows, and if they do, they're likely under NDA to speak about the specifics.

[neat field]

this question is actually like 5 questions, but it's definitely not just 1 question, and it's the wrong question

like asking "how does a person make a round square?"

[coral violet]

Does anyone know a prompt in the lyrics itself to make a word whispered instead of sung?

[charred merlin]

I wouldn't be surprised if there WERE a handful of voice clips used in the original training, but that's like asking which art styles, artists & photographers are StableDiffusion @abstract spade ❤️

[charred merlin]

https://www.suno.wiki/faq/metatags/

[neat field]

https://suno.com/playlist/ca526d7f-8abb-482a-b398-68c7bcca7981 check these prompts 🙂

[charred merlin]

https://tenor.com/view/shapes-geometry-circle-square-gif-14837657

[abstract spade]

ok...

[coral violet]

Does anyone know how to elongate a word for instance again to aggggaaaaiiiiiin

[neat field]

LOVE IT, see and my answer was wrong, you CAN make a round square if you think outside 2 dimensionality

[coral violet]

Thankyou. I tried [whispered] and it didn't work.

[charred merlin]

Metatags aren't explicit instructions, they're directions that the community has discovered work better than others; still takes some experimentation-- the style prompt might be overpowering the metatags?

Hoo boy, we're still figuring this out from day to day. I'd love to be a fly on the wall at Suno.AI to hear the kinds of discoveries they're making and trying to control to give us V4

[humble bison]

How do you PROPERLY structure a prompt for music and voice?

[coral violet]

Please, Please, Please, Please, Please.

[neat field]

oftentimes i have had the model honor spelling out longer vowals between consonants

Escape
becomes
Esc Aaaaay p

[neat field]

the models currently do not understand music theory directions very well

you might try "guitar tabs" in the prompt and then alternating lines of lyrics and guitar chords though, as that may exist in the dataset for the model (untested by me)

[elder dune]

So, today I recorded my Martial arts teacher playing a riff he's working on - with my crappy phone sitting in front of his amp - and suno came up with this banger: https://suno.com/song/97866e38-2f79-4b3a-8a52-320e9003e5e0

Crap I meant that for the general channel. Sorry

[abstract spade]

Thank you for your attention, Mr TheHarber.

[elder dune]

Feel free to delete it you want

[coral violet]

"Maybe one day someone will convince me to make a tutorial video for my workflow" Please, Please, Please, Please, Please. That would be much appreciated.

[sinful oyster]

I had a quick question for anyone who may know the answer... is there a way (be it Style of Music prompt or Lyric prompt) to have the AI generate vocalizations similar to that of what you would find in Two Steps From Hell music? I've been playing with a style of music similar to them, and am trying to see if I can add more dynamic to the songs, vocalizations (ooos, aaahs, etc... without manual typing of it as lyrics) in order to reduce some of the repetitious nature I've come across with this style. But yeah, any input would be appreciated. Thanks in advance 🙂

[naive quail]

I've had some luck using both [Whispered] and [Spoken, Whispered], but they're like gentle suggestions, at best. Hit or miss even when the style of music seems well suited to whispers. Freakin difficult when it doesn't.

Here's an example cut down to ~ 1 minute: https://suno.com/song/50e7a19e-3027-43c1-a9a6-088942122ca6

[coral violet]

Thankyou.

[coral violet]

99% of the time it is missing my 1st verse. It contains the expletive MF. Is this the reason? Is it a hard barrier?

[charred merlin]

https://suno.com/song/a18bfdf6-9db2-4c18-a748-cc8ec2b37744 No 😛

[charred merlin]

Perhaps there could be an issue if you're breaking the expletive into 2 separate words? Might seem silly, but that same phrase can be interpretted colloquially or literally depending on whether it's one word or two.

[naive quail]

Rawr needs a harmonica solo

[coral violet]

Thanks. It is one word.

[charred merlin]

I use MFing, are you using MFer?

I haven't tested any of this, just curious 🙂

[naive quail]

Rawr works with mf'er as one word. Harmonica solo wasn't as good, tho. heh

[coral violet]

oops got blocked coz I spelled it out. It goes "sexy MFer" it creates the second half of the pre chorus (following on from the uploaded part) then skips the first verse, the pre chorus and the chorus and goes straight to the second verse. ..........It just occurred to me why. The first verse is the uploaded part. I'm trying to get a better produced version of the uploaded first verse. It probably thinks I don't want the first verse again. Update. That's not it. Uploaded second verse and it still ignores the first verse. It's a pity because it is giving me awesome stuff in every other respect including sounding like me.

[faint wigeon]

What are y’all using to post-process your songs (damage repair, mastering, cleaning up, etc.)? I’ve got some iZotope tools on mine, but wondering if there’s any other magical tools that can improve quality/reduce glitchy noise of some generations

[valid mango]

Why the lyrics affect so much to the style of music. Like if lyrics says something about partying, and my style says metal etc, it makes a party song. And other way around. Almost every time

[charred merlin]

... > Report > Bug > Doesn't Match Style Instructions, each reported song, along with each thumbs up and thumbs down, help train the AI as to what you expect to hear.

[opaque canopy]

so i should thumb down every generation i dont like before delete them ?

i delete something like 95% of what I generate.....

if I put like on every 'good' version you think the algorytm will push that AI to give me more of what I like?

[charred merlin]

Thumbs down have been encouraged, though I imagine more attention / weight is given to legitimate bug reports. Not just reporting the style instructions because the style of metal isn't "heavy enough", but making sure you report if you ask for metal & it gives legitimate pop. Not simply reporting audio quality because you don't like the way it sounds, but because there's a describable issue (eg. screeching, runaway amplification)

[charred merlin]

Not immediately, but it helps it understand when it's doing good, and if everyone does that, it will massively speed up the process. Same thing with different bug reports-- using them appropriately will 1) reduce static/clipping when bass or other instruments get too loud, 2) follow lyrics & metatags better, 3) give improved understanding for specific genres (including subgenres and genre-blending), 4) make songs just sound BETTER

Thumbs up and down are helpful, but a legitimate, actionable bug report is worth 50 of them.

[opaque canopy]

the reason i deleted a song is.... i dont like the singer voice....it ignores my prompts... sometime the melodie is good but ignores my [guitar solo] prompt ... sound quality degrading too much tube effect the longer the song goes....cannot [end] a song in 3.5 is painful....

[loud kernel]

when trying to share links people are getting this right now

[charred merlin]

Same, I'm heading to bed right now, but I'd throw that into #🪲┃bugs-web

[loud kernel]

thx, just did.

[sterile anchor]

Not just sharing links. It’s been doing that to me every time I try to see the homepage for the last few days.

Just kept forgetting to say anything. 😂

[marble parrot]

Is there any way to tell Suno to have two vocalists? Nothing I've used for tags has worked. I enjoy bands like Blink 182, Linkin Park, SOAD, etc who regularly have two people doing vocals on any given song.

[midnight scarab]

How to cancel Suno subscription?

[sharp kraken]

I do it by making two vocal in style. Like this: male punk vocal, female soft vocal

Then you can try adding the same descriptors to the verse you want them to sing, like [Punk verse] / [soft verse]

[sharp kraken]

https://suno.com/song/d4ea2ef9-0675-483f-9fdc-a55b50cb7912

^ male/female duet example

[coral violet]

The left hand side of the suno web page just after half way. Click on subscription.

[midnight scarab]

it's a blank page.

[coral violet]

Mine says manage subscription and cancel subscription is half way down. Maybe you haven't got a paid subscription?

[modest widget]

This is really fun and I learned lot. Thanks!

[midnight scarab]

I GOT, and I'm very scared of my card being billed by a not working app

What I can do is I locked my card

[mental adder]

Question about instrumental songs:

I have now created some instrumental songs and very often instead of the intended 4 minutes I only have 1:30 - 2:59 long songs, which is not bad in itself, but also the extension very often leads to a 15 - 45 second long end that is automatically inserted and finally ends the track.

Is there a trick so that it a) uses the full 4 minutes and b) doesn't want to end directly with every extension?

Maybe someone can help me with that! Thx.

[daring lava]

Have you tried extending from earlier in the song? It could be possible that the reference point from which you are extending is during a part of the song that has already decided to end the piece. Sometimes going back a little more allows you to go a little farther. Look for a signal of strong musical flow and go from there

[mental adder]

yes, I have already extended it from the middle

would need a button that says instrumental, extend song or end song.

Suno, I have a wish 😄

5 x extended

[daring lava]

If you’d like to share the link to the piece (prior to the extend) I can have my hand at it to see what my help

Also “epic, epic, epic” could help, but wouldn’t want you to waste your credits testing if it’s a bother when I could

[mental adder]

thanks for the offer and the help, but it's not just about one particular song, it's about every third song. sometimes it's 4 minutes with a beginning and an end.

i generally want to extend in the last third and usually get an end. especially with songs that have become only 1:20 or 1:40 long instead of 4 minutes with v3.5 and have a good start, this is of course annoying.

[compact sky]

suno can you plz allow bad words?

because everytime when i try to use the word Ass or wett suno says no...

or examples like pussy... its normal slang in hiphop

[mental adder]

https://tenor.com/view/cat-love-tell-me-more-salem-gif-13645379

[compact sky]

i think by the k-i its a bad word filter aktive that not let your write the song

[mental adder]

thats right

[compact sky]

Artistic freedom does not need this filter

i hope in the future it will be open

[daring lava]

Definitely appreciate the annoyance of this and hope to help you find a way to find a resolve. Would be easiest if I knew the specifics of the approach you are going with when generating. If you’d like to share a link so that I can have a more in depth look at what’s going on, that would help a great deal.

[compact sky]

you can fix that issue when your make at the end of the son space, enter, and text in the next line : .........

[mental adder]

i'm not annoyed. i'm still happy about suno and what it can do. it's just annoying that suno is acting a bit strangely, maybe i didn't express myself properly 🙂

[compact sky]

[Bridge]
Heartbeat
Rhythm of my soul
(hey-yeah)
Together
Masterpiece of love
Of love
In your arms
Find my refuge
(ooh)
Souls entwined
Chains of love
Of love

..............

looks like that

[mental adder]

i think you can only really judge that in context and i don't know if a ai can do that

[mental adder]

instrumental pieces unfortunately have no lyrics ^^

[compact sky]

ai can handle bad word.... just global companys restricted it because of getting hate but for Artistic freedom ( that you have when you write songs or create songs) it should be allowed

[daring lava]

Definitely room to grow for sure lol. But isn’t that the case for all this crazy future-ai stuff. Still early days I say 🙂 imagine what we’ll be able to do down the road 💪

[mental adder]

send you a dm

[daring lava]

Okay thanks 👍

[real hound]

just logged in to my suno just to see that the payment was not processed on my subscription since January so cleared the due but still cant compose any songs. Any idea why

[compact sky]

the same problem got copilot when i take bad words inside....its text something but its broke up and say "sorry we cannot show you"

[hollow hare]

Can you provide a bit more detail about the cannot generate song part.

Is it because your credit is always zero? Not appearing? Etc. (A screenshot would likely be helpful in this case).

Anyways, for your card processing, contact [e-mail]

For other support, you can contact [e-mail]

[real hound]

i so wanted to do this small verse for a vietnamese girl

seems like ill have to postpone it for now

if i go to update billing info there is nothing much

this is what the problem is

it shows I have 50 credits with an additional 550 which cant be used nor can i buy more

the worst is the renewal date which is at 3rd Feb

[compact sky]

maybe somebody should add a" song cut tool" on the website

[hollow hare]

Yeah, unfortunately the 550 extra credit cannot be used without the subscription is enabled, though you should have been able to use the 50 credit (likely an unintentional side effect, it was probably overlooked for this scenario).

The only solution to this is to contact [e-mail] to fix the account for you

https://media.discordapp.net/attachments/1069381916492562585/1238249445192237106/image0.jpg

[charred merlin]

...you tried this, right?

[real hound]

already tried

[charred merlin]

It says your payment didn't go through. Have you contacted your card issuer?

[real hound]

it just processed my pending payment from january

i even got a text from my bank

[charred merlin]

This seems similar to when I worked with Apple, you're unable to create songs because you have an outstanding balance with Suno to address. Contacting their [e-mail] is likely the best way forward.

[real hound]

seems like that

[charred merlin]

Like when people do a chargeback for a subscription they didn't want (instead of cancelling 48 hours ahead of time), Apple freezes their iTunes/App Store account until they address the issue. Not implying that you did anything like that, just sharing a situation that appears to be handled similarly.

[real hound]

seems so

[rose plank]

i feel like they need to consider negative prompts

[compact sky]

jep

i send support an email

[charred merlin]

#💡┃suggestions-web is probably the better avenue.

[forest cliff]

I want to make a klezmer wedding bottle dance, which requires several tempo changes. Any tips on how to indicate a tempo change in an instrumental piece?

the only lyric is "Hey", so I'm finding it's difficult to time anything

[final hazel]

There is not much control in that. Things like that often happen spontaneously, and even then they are rare.

[forest cliff]

I saw somewhere how to indicate a specific tempo and key, but I'm not finding that now

[final hazel]

Some people tried that, but it is not really working.

[distant seal]

Hi. v3.5 is great, but it really tries to fill all 4 minutes, even if song is short. Is there any way to say "do not duplicate texts aggressively"?

[rose plank]

Yea this seems like a recent development it didn't do this at the start of 3.5

[charred merlin]

... > Report > Bugs > Doesn't Follow Lyrics / Poor Song Structure (whichever you feel apply) @rose plank

[true carbon]

I am not the best at working with upload but i would like to see how good it can be, i made a beat with FL studio so if someone interested in doing an upload with that beat it would be nice. Let me know

here it is https://youtu.be/92cOg32KTqY?feature=shared

[charred merlin]

In my opinion, it would be more appropriate to share this in #🐣┃suno-showcase and ask for collaboration?

[true carbon]

yeah right sorry

[rose plank]

Iv been reporting just about every generation now

[dry spoke]

Is it possible to regenerate song with the absolutely same style, but slightly different lyrics and structure? Generate versions of the same track?

For example, I made a mistake in the accent in a word and I want to fix it, but I don't want to get a completely different track?

Or replace a word in a lyrics

[daring lava]

Your best shot would be to extend from a part prior to the mistake. That would ensure that some of what was produced will carry over to influence future generations

[dry spoke]

Hmmm... will check 🙂 Thanks!

[daring lava]

Very welcome, happy to help!

[wanton cave]

I am getting really frustrated with Suno's inability to END the song when needed, even with an "[end]" tag. It artificially streches the song to ~4 minutes and even then it doesn't end it properly. If I try to extend the song, and only put [outro] and/or [end] -- it doesn't follow instructions.

[hollow hare]

Extend where you want to end with the extension set to v3.0 when using ending tags (e.g. [end] in lyric box and "end, short" as the style tag) as 3.0 is much more reliable for ending than 3.5

[covert python]

ok so this is a feature of v3.5 if you want to end a shorter song. there is a couple of ways you can do it. 1: revert back to 3 extend and then use the extension with the same style tags. no lyrics. just
[end, short]
[end, short]
[end, short]
[end]

OR

use the extension feature of 3.5 to lengthen the song with same tag sett up. it too can gen a 15 - 30 secand extension.

IF your lucky. it works for me more often then not in 3.5. but may take a couple of gens to not get a broken extension.

[wanton cave]

thank you for the tips, I will try them

[covert python]

my oppologies i didnt realise you had answerd already. i was too bizzy typing.

[hollow hare]

No problem 🙂 I did that as well before a few times

[covert python]

also. how did you manage to give the same answer in a lot shorter answer lmao. so jelous. 😄

[hollow hare]

In my mobile Clipboard function, I pin the most common responses

[covert python]

THATS CHEATING!!! 😛

[hollow hare]

I like to call it utilizing all resources made available :x

[runic dock]

Suno depending if it has enougth context or not may go to à totally différent melody
What is asked here is the so called inpainting feature that is a promise not yet released

[hollow hare]

Inpainting has almost always been stated as part of the v4 release AFAIK.

The release date for that ... is soon(TM)

[covert python]

my turn for question time!!! 😄 i would like to generate an intro of rolling thunder transitioning to a rhythmic drum beat. is this achievable with current levels of A.I at sunos end. ? as i have done a fair amount of "for sciencing" and have yet to find a method of achieving this.

[final hazel]

Inpainting will only work with V4 (at least in the beginning, not sure about the fututure)

[daring lava]

Oh yes, definitely lots to come. Very excited for each and every update 👍

Patience is key as great becomes greater 💪

[hollow hare]

Ah, I wanted that as well but no luck yet figuring it out. We'll, kinda... I was just thinking about looking up what specific percussion is used in orchestra to make that sound but I haven't tried it yet

[final hazel]

You might be able to do that with the upload function. Find a good thunder roll and maybe add a beat. So the Ai will pick over from that.

[covert python]

well ive dabbled with taiko drums a fair bit. but cant seem to merge the two thunder and drums. when i do suno just goes NOPE no intro for you. 😄

[daring lava]

To add to what 3Daisy said (when using upload) I’ve learned from others is that by repeating tags within the style of music can help add weight to the influence (eg pop, pop, pop) during an extreme transition

[covert python]

i must admit i have not yet touched the upload feature. hmmm. an interesting thought. thank you.

[final hazel]

I made one starting with the heartbeat of a baby.. turned out pretting good

[covert python]

oh doubling down is always good for adding weight. but the question was ability to "merg" specific intros. thunder into drum beat. or gun shots into perccusion i think there called. the high hat tinny things.

[final hazel]

Here the example..
https://youtu.be/dLUHgVUXaQA?si=YQAMcgp2dc-WwYpt

You can find more here: https://youtube.com/playlist?list=PLi8taL4TbG_9x9FH7KZvTaQAD65mSq8t2&si=z3KoXFWCB0G4nXH5

[daring lava]

Definitely agree a mixture of upload and heavy tags will do the job best

[runic dock]

About the upload function, i've given a try . It seems hard to make something correct with a pure accapella song. I mean with only singing voice (and with no rythmic pattern). Have you already tried that?

[covert python]

thats exactly what im looking for. well not the baby there expencive. but the merge. 😄 😛

also congrats? O_o

[modern belfry]

I need a Suno expert lol.. I want a short horn intro for a song. Kinda like Metallicas Unforgiven 1. I think it'd be suitable for a metal song

how would I go about that ?

[final hazel]

Like this? https://youtu.be/6rzt1WNGUSc?si=_HxyaBl4PDYWREmo

[final hazel]

Suno doesn't have much horn in the current models. Hard to get them. So I recommend the upload function. You can probably find some royalty free samples here: https://pixabay.com/it/sound-effects/search/horn/

[covert python]

is there anyinstruction for uploading? sorry to be a newcence im kinda helping several people via dm and struggling to fix my own shit. lmao

[final hazel]

It's self-explanatory. Just upload, extend, and add the timestamp where you want the sample to stop and Suno to take over. 🤷‍♀️

[floral sonnet]

hello you guys!

[covert python]

hmm. ok i will definitly try this.

thank you so much for the time and support and advice you guys are AMAZING!!!!

all the love

[modern belfry]

ahh ok I see. Thanks !

[French horn]
[fade out]

[instrumental]
(Yeah, Yeah!)

[chorus]

thats what I was trying lol

[final hazel]

The models of Suno contain a bit of everything. But when it come to "niche" sounds or instruments, it will be come hard to trigger that.

[modern belfry]

Ill give your method a go. Thanks you prob just saved me a ton of coins lol

[floral sonnet]

still looking for a definitive and exhaustive 'style, genre, type 'list for 3.5 ??... anyone have one? .... also would be handy to have all the meta-tags...that are currently recognized in the [ ] square brackets...

[final hazel]

There is none. The developers are not witholding us of that (some think they do). They know what they have learned the model, but they don't know what it will understand for the full 100%. That's just very unpredictable still.

All data is created by human, it learns lyrics, music and genre tags. But those are created by humans too. And humans are not consistent. So is the outcome sometimes too.

[floral sonnet]

hmmmm....

[final hazel]

For instance, some genres are having a hard time giving you an intro, while you prompt for it, while in other is works flawlessly.. I blame human inconstancy for that.

[floral sonnet]

so do you have to have a style ... then a genre and then a type ? like it suggest here --.https://www.suno.wiki/faq/style-and-lyrics/styles-and-genres/

[final hazel]

These are just some things we've found to work, but not always. That's the point—nothing is guaranteed. Have you ever created images with AI? If so, you know there are no set rules. Sometimes it nails your request, other times it completely misses. It's hard to get AI to do exactly what you want because words don't always perfectly capture your vision. If we both read the same book, we'd describe the details differently since our imaginations see things differently.

[floral sonnet]

id like it to be able to 'get whole song' but exclude original upload....

[naive quail]

Seems like you'd be able to do it using the upload feature, especially if you throw some percussion into your thunder audio sample. This isn't exactly what you were asking for, but it's along those lines. I think Suno cuts in at 4 or 6 seconds. https://suno.com/song/3e572c78-0c5e-4845-97cf-dde077a86465

[covert python]

thats close to what i was looking for. yeah others suggest upload feature im yet to dabble with that. ill have to give it a go and see what i can come up with. thank you for the info. 🙂

would it be at all possable to get a rule / server etiket clarification from an overlord boss type person? 😄

(staff) so was posting in suno showcase channel and asking for feedback requests. when someone suggested i post in general or dm people with song and feedback. my question is. would i post in suno showcase asking for feedback and suggestions on song. or general? as the user fellt that showcase channel is purely for vids only. and i figured it was more just hey post vids here to keep them all in one location but you can still ask for text based feedback. in no ways a huge issue. i was just asking so as not to annoy / clutter incorect channels.

[runic dock]

But in case of music, a list of known music styles and genre exists.
At least, suno devellopers can communicate what they know that have been used to train the model. If the model is consistently not reproducing the style, the user will flag it as a style bug and suno devs will work on the issue.
Wikipedia has a list and spotify too.
https://en.m.wikipedia.org/wiki/List_of_music_genres_and_styles

Today if a style does not work, the user doesnt know if there is an issue in the style name spelling or if it an unsurported music style. This is actually the real issue in our use-model.

[limber bramble]

Could you implement for fine tools to select from? Like for example if i want a deep male voice the system do not put it, you could add a menu from different voices and even an a way to custom made the tone of the voice like pitch, tone etc.
Also the option to merge 2 different tracks

Once i have extended a track and I want to make a change in the style of the music is almost impossible, the system do not read the new words that describe the new style of music

[finite prism]

Yes, but it doesn't work, because the majotiry do not understand whats the difference between the majority of genres - then the model will get a lot of wrong labels and the usage of this labeling for modelling could lead for model quality drop.

[runic dock]

It will not be the case if a clear documentation is given to users. This was done with chirp v1 and no update has been done since v1.
Maybe devs are not able to say all what are the styles
/genres suno is able to reproduce.
Today everybody add what he wants as style/genre and expect simply magic to operate.

[finite prism]

And I'm sure they do not implement it anymore, because it is useless. Dirty data is useless for modelling.

[runic dock]

What do you mean with the last sentence?

[finite prism]

Please, ask ChatGPT what does it mean in context of machine learning in simple words.

[runic dock]

We are in the context of suno. What are dirty data for suno?

[finite prism]

Then ask ChatGPT what are the examples of dirty data in machine learning for text-to-audio modelling and what does it mean in general.

Dirty data in dataset labeling problem

[slim ether]

Hi. Is it possible to create songs with only vocals without instrument? I tried a few times, but couldn't succeed

[charred merlin]

That's a pretty specific request-- tags like Acapella might help, but it could be tricky to get a song with no instrumentals at all. Might be easier to get the vocals you want & separate out any instrumentals using a stem separator.

[covert python]

just tried acapella . does not work. just an fyi comment.

the closest ive found is the ambiance tag. there is still instrumentals but vary soft its a vary heavy vocals style

[final hazel]

It should be written as a cappella

[covert python]

https://tenor.com/view/ooops-noah-is-gone-two-by-two-all-creatures-big-and-small-oops-whoops-gif-19891523

oh.... lmfao. then not only did i type it wrong i also spellt it wrong.

then yes a capella does in fact work. ( just word better then what i does.

[final hazel]

Here an example of " a cappella".. but completely without background music is hard:

https://suno.com/song/ae066b7a-24e7-4506-8edd-0a6dd36ce97e

[covert python]

my oppologies.

[charred merlin]

Even then, might help, metatags are suggestions, not implicit instructions.

Interestingly enough, Acapella is the anglicization of the phrase, but Suno seems to be trained specifically on the classical spelling. That's an interesting note 🙂

The english version of another word ❤️

[covert python]

thank you. i deleted cus i hate asking dumb questions and took me a minute. (cus im dum) to realise it was a dumb question. and should google it.

[charred merlin]

Not a dumb question at all! And taking it upon themselves to Google something because they want to learn isn't a sign of a dumb person-- be kind to yourself ❤️

[covert python]

thank you so much.

i can be a bit harsh on myself.

[charred merlin]

If we never had this conversation, we wouldn't have realized that "acapella" (accepted by spellcheck) gives a totally different style than "a cappella". 🥳

I genuinely would have thought separating it into 2 words would have given a less specific result. I'd be curious what other musical terms would give different results given their classical or English names?

[covert python]

you do make a vary intresting point. also side note. (i made my auto correct dyslexic) so i cant evan trust that. lmao

[acoustic hornet]

Does anyone know if there is a way to force your created song to NOT go into some kind of instrumental riff after it sings out the first verse, labeled as [verse 1] when the next verse is clearly labeled [verse 2] with no extra tags between the end of the first verse and the tage for [verse 2] ?? That's worded goofy, but I hope SOMEONE understands...

[hot moth]

does anyone know a good prompt for house music, to stop suno going into a break every 8 bars. It is really annoying. I waste so many credits getting songs with breaks and breaks. Useless for dance music as you need a good 32 bars to rock to. Thanks to anyone who end my frustration. I´ve tried break, instrumental, dance section etc.

[hot moth]

I´m trying to make a house track and it stops into some break all over the place. IT is really annoying, it doesn´t understand a strong dance break. And it doesn´t understand 34 bars of epic dance break. hmph.

[carmine cloud]

Is there a good way to stop the AI from repeating stuff? Right now it really loves to repeat the chorus after verse 1 even tho its posted 1 time, instead of going on to verse 2 it does the chorus multiple times before moving on, and no matter how many times I try to generate it keeps doing it

[fringe widget]

Hi all, is there a prompt or way to limit Suno's length of track? I like the 4 minute setup but I want to create shorter tracks sometimes with a proper cutoff rather than having to fade out each track i want to be shorter.

[coral violet]

f I send a playlist to trash, do the songs in the playlist also go to trash?

[bleak bobcat]

At the moment, the only way is to extend the song before the repetition . The pity is that it can change the rythm. Or you can download the song and edit by yourself. I have to work sometimes in this way, but then you can't share it in suno platform.

[charred merlin]

Sending a song to a playlist doesn’t remove it from your base library— you can have multiple instances of the same song if you add it to multiple playlists.

Moving a song to the trash removes it from all playlists, though

[random bobcat]

Is there any way to retain the same style and instrumentals of songs between generations? I noticed some people achieved that somehow, is there a trick to do that?

[carmine cloud]

Fair, thanks for the reply, hopefully its fixed in the next model, that being 3,6 or 4,0 🙂

[coral violet]

When I use upload audio option it seems pretty consistent between generations.

[random bobcat]

Wait, is the upload audio option used directly to set instrumentals of the song?

I was under the impression it is used to just add external songs to my playlists 😂

[unique stone]

Noo it's not for that at all 😆
Upload is used to influence what gets generated when you extend from the uploaded audio

For example, the first 12 seconds or so of this are an uploaded clip, the rest is what Suno generated from it https://suno.com/song/343a2d9b-69ea-4ece-b7bd-196aa03cd6c3

[random bobcat]

Can I also upload other tracks into it? Or is it just plain for instrumentals?

[coral violet]

Yes it is. If I upload a strummed guitar with me singing, it retain the acoustic strum throughout the whole song and the vocal even sounds like me.

[random bobcat]

Wow, that's actually cool

[coral violet]

It is magical.

[unique stone]

Anything between 6 seconds and 60 seconds long. Clips of music, sounds, vocals. I did some low effort beatboxing, imitated a trumpet etc. and it works with that.

Important though: you must have rights to the uploaded audio so don't upload copyrighted stuff hoping to remix your favourite musicians work. It has something in place to block this but even if you get around it, that's against the terms of service.

[earnest pumice]

Can someone help me out with making an ai song??

[coral violet]

Thanks. I mean't sending an existing playlist containing tracks to the trash do those songs go to trash as well. I just tested and they don't which is a pity because it would have been a way of "organizing" my trash folder.

[somber fern]

why is it singing the brackets that are in bold??

[Intro]
[Build-up with aggressive electronic beats and distorted guitar riffs]

[Verse 1]
In the heart of the chaos, where darkness looms,
We rise with fury, breaking out of tombs.
Through the shadows, Your light pierces through,
In the battlefield, our strength is renewed.

[Pre-Chorus]
[Drums intensify, guitars thrash]
(Fire blazing, voices ignite)
With unyielding fire, we claim the night.
(Hearts defiant, spirits unite)
In Your name, we shine our light.

[abstract spade]

Is there any way to pay with credits for the songs I've already created on the free account, as they seem to be better than when I paid...

[charred merlin]

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

[earnest pumice]

Thanks

[shrewd lantern]

how to make more instrumental parts without voice in songs?

[modest oasis]

Is there a way to search your own previous songs . I put in the song name and it can't be found and doesn't show in my normal list anymore

[charred merlin]

You can search your library for style prompts, lyrics, or song titles.

[modest oasis]

I have . The title was Sprintcar mayhem and I have made about 25 versions but it can't find any of them

[charred merlin]

You might have spelled the title differently, but try the other things I suggested instead of title.

[modest oasis]

Thanks

[coral violet]

Another idea is to upload one of your suno tracks and extend that for consistency in style of the whole song.

[harsh mirage]

hi, i made the ai sing my lyrics with suno, but i have the free version. I would like to know if I can still publish it on my YouTube account.

[charred merlin]

If you give attribution to Suno, since it's technically owned by them.

[sudden pike]

I just need some advice about posting my song on YouTube. I uploaded the vocals from some video clips and used the lyrics from the same videos. Will I get into any legal trouble for this?

[covert python]

that dippends on what the video whas. you cant re publish others work.. (to my knowledge) wich granted is not the greatest

i would say if you got lyrics from an actuel artist in part or music from them then i would suggest playing it safe and not making that song publick on youtube

[sudden pike]

any lawyers in here?

ai music lawyer?

[covert python]

put simply. what was the sauce of the clips used?

[final hazel]

is this about Suno?

[sudden pike]

of course , song made with SUNO AI

[final hazel]

Do you "own" the material in the video?

[sudden pike]

nope

[final hazel]

If it is not "royalty free", you could get in trouble yes

[sudden pike]

it a movie

[abstract spade]

All material entered by you is yours, but that generated by suno belongs to it in the free account, as described in its terms and FAQ

insert

[covert python]

thats not the issue. the issue is the "clips used"

[final hazel]

Copyright is copyright.. if you take something from a movie, you can get in trouble.

[sudden pike]

and voice

a famous voice

[covert python]

yeah then you cant use publish or make public anything with clips from any already owned movie, music or song lyrics.

owned as in main stream.

[weary wharf]

I played opening of "Carmina Burana" to Suno, but I'm not able to drive the music.
Trying to convert it to western-classical, symphonic, film-score, movie-soundtrack, epic, cinematic, anthemic, energetic, agressive, choir
But it's always just piano...

Any tips on guiding Suno for uploaded audio?

[final hazel]

Not sure if you are familiar with the voices of chatgpt? Sky was removed because it "sounded" too much like "scarlett Johansson".. So what does that say about your case?

[sudden pike]

lol

you read my mind

actually im making a song about her

[covert python]

to my knowledge you cant really tell suno to give you a music style revolving around a specific pre existing song or artist.

due to once again. copyrite issues.

sorry daizy ill stop.

dont mean to step on mods toes.

[weary wharf]

Aah... If it's still in copyright, I'll remove it.
Will try Mozart or Beethoven...

But still, I'm not able to move it to any other instrument, it's always piano.

[final hazel]

@sudden pike anyhow... it's a risk. Normally they will warn you first though (not always though), and only sue when you are really doing damage to a brand, or if you are not removing it from platforms.. so fully up to you

You cannot share it here either.. 😬

Copyrighted stuff are not allowed in discord

Sorry 😦

[weary wharf]

Carmina Burana is a copyright work, with Schott Musik International (in Mainz, Germany) being the copyright owner.
Hmm... Thought it was older. Removed uploaded + part2s.

[sudden pike]

so many rules

[covert python]

instead of trying to duplicate exact existing styles with there already existing names. what you could try is go to chat gpt. and ask it what style or genra of music it is. gpt will then break it down for you. AND then. you can duplicate the "feel" of the pre existing main stream song. without duplicating the lyrics of cause.

[covert python]

not really. if you create / suno generates the lyrics. then your golden. so long as your not copying another famous person.

[weary wharf]

No lyrics, just instrumental. I was trying to learn how to guide Suno with only a piano tune that I wrote/gave to it.

[sudden pike]

i just thought it would be funny since its a hot topic

[hollow hare]

Beware that LLMs can lie to you with ease https://media.discordapp.net/attachments/1159929934270181437/1250454992196861992/Screenshot_20240611_141809.jpg

A copy-paste of one of my response so do not feel targeted please:

Using existing artists or song titles will likely be ignored, if not censored. Very likely such information was stripped out from the training to avoid lawsuits (assuming if they were included in the first place).

You will need to use style keywords in custom mode to get as close to the genres you want.

E.g. you may want to try Ibiza Trance as one of the styles for ATB-like music (https://rateyourmusic.com/artist/atb)

Though I recommend looking at a specific song for more accurate styles. There are other sites as well to identify song genres like https://sonoteller.ai and Chosic's music genre finder (Spotify only, <
https://www.chosic.com/music-genre-finder/>)

[covert python]

oh ok. then yeah gpt is your friend. breakdown a classic. with gpt it should tell you the specific part that the original usis. then basickly adapt to tag language for suno.

[covert python]

...... oh

https://tenor.com/view/ill-just-leave-now-the-simpsons-homer-simpson-good-bye-bye-gif-14711241

ok ima leave helping people now.

[weary wharf]

The problem is, uploaded song is played in piano -- by me.
Even if I give it "alternative-gothic-rock" it's still playing in piano... feels "rocky" but nothing else changes.
No tags or styles are adding additional instruments. Or even lyrics... Only piano.

[covert python]

ignore me

[hollow hare]

It can tell the truth, especially regarding old music, but you will need to criticize the results you get from it because you do not know when it will lie

[covert python]

yeah i get that.

i jsut assumed that it was atleast semi accurate. concerning basic song breakdowns. and the likes.

[covert python]

i dont think suno can "edit" uploaded clips as such. more "tweek"

[alpine venture]

https://suno.com/song/d0f37b16-21e8-4f8e-87ad-6e6500483f23

[covert python]

AH!

[alpine venture]

examples of how you can get 3.5 to read stuff

not pub

bc it isn't up to par

[covert python]

ware did you come from? 😄

ok im leaving yall to yolke. they know way more about.. well everything then i do.

and paimon.

[alpine venture]

or for example halfway through here I get it to say "gimme da spicy meatball"

https://suno.com/song/3e179f51-601e-4900-88ff-8b97c5290c5b

also not pub

[final hazel]

https://tenor.com/view/mama-mia-thats-a-spicy-meatball-gif-27063016

[alpine venture]

I'm so sorry yall

there is no meatball in that version

that one was vegan

https://suno.com/song/8ba237d9-ea57-42d1-83bc-06436812ff8f

this one contains meatball

at 1:10

to 1:15

also "hoooooo yaaaa"