🎵｜stable-audio

[cobalt surge]

this doesn't tell me anything about the format you were feeding it a couple days ago and what you're using it now.

[somber cradle]

I download the video and put it on the website

Then it is cut off in 30 seconds, before it worked, now I don't know why it is giving this error.

I download the video, go to the audio option, select it and that's it, the audio goes to normal. But when I use the prombt on the audio of the video I downloaded, it gives this error

[cobalt surge]

mp3, wav, mp4, aiff ? ogg, flac, mod, midi, bink ?

[somber cradle]

Mp4

Sometimes this error occurs too

Since the audio is only 30 seconds long

I use SnapTube to download videos and to cut I use Kinemaster

[gaunt orbit]

maybeeee

try 31 secs

[somber cradle]

When I use more than 30 seconds the site cuts to 30 seconds

[gaunt orbit]

ah oop

[somber cradle]

I don't know if you saw the other error I showed @gaunt orbit

The videos I download are in mp4 format

[cobalt surge]

ok I have the same issue, might be broken at the moment, probably worth opening a ticket for it https://kb.stability.ai/knowledge-base/kb-tickets/new

[somber cradle]

What is this for??

[cobalt surge]

that's the contact link for stableaudio

[somber cradle]

Could it be a bug on the site?????

What do I put on the ticket?

[cobalt surge]

you describe your problem, what you're trying to do, what type of files you're using as input, the error you get, etc

[somber cradle]

The name of the ticket has to include the error that is occurring, right???

[cobalt surge]

a summary of your problem should be fine.

"[stableaudio.com] can't get any output."

something like that

[somber cradle]

I sent it and now what do I do?

I put the name of the ticket as the name of the error, I don't know if it works

The appeared here noreply

What do I do now? Just wait???

[torn hare]

filmmaker

[cobalt surge]

yup, now you wait for them to read your ticket, investigate, reply

[somber cradle]

Do you have any theory for this error???

My theory is that it's a website bug

Why wasn't this happening before?

It only started happening now so I think it must be a bug on the site.

How long does it usually take for them to respond?

[somber cradle]

@cobalt surge A question when they answer the notification appears here on my cell phone

[cobalt surge]

probably a day or so.

[lyric elbow]

what model are you using exactly?

anybody mind telling me about the new open source audio models?

[brisk wind]

Suno

[brisk wind]

is not open source

[cobalt surge]

he s not talking about suno but Stable Audio Open Small

[brisk wind]

the closest thing that I can think of is yue

[lyric elbow]

I have no idea what model to even ask about
I only know about ace step, but it barely supports any language

[lyric elbow]

Too bad
Damn that sounds good

[brisk wind]

most open source sound models are not great at all

https://www.riffusion.com/song/5918c395-ebe4-4bac-90d3-52032afa2b49 my riffusion attempt at other song.

[cobalt surge]

oh ok I thought you were specifically about the new model. Didn t see the "s" at the end.

[brisk wind]

I think I have tried one other than yue but can't remember off the back of my head

[lyric elbow]

Ace step prob
Couldnt find any better than it
How good is yue?

[brisk wind]

yue is pretty bad.

I use yue to produce it and you be the judge

the lyrics is nsfw

[lyric elbow]

yeah it seems like yue doesn't cut it

[brisk wind]

but it is so bad it is kinda good

[somber cradle]

@cobalt surge Where they respond

[somber cradle]

Suno is garbage It doesn't create the style you want, sometimes it's really bad, I prefer Stable audio

[brisk wind]

there is actually something called stable audio

suno 4 or above is good

[somber cradle]

I'm using it but it's giving me this error

[brisk wind]

https://www.riffusion.com/song/14afcc6f-061b-466c-9c62-a0d0a1493dc2 this one I created with riffusion is decent

[somber cradle]

This only happens when I use audio from a game or something non-original.

[brisk wind]

this one https://stableaudio.com/ ?

[somber cradle]

I want a website or an app that lets me guess audios and remix them.

[brisk wind]

they let you remix it

so is suno but the free version limit the length of it

[somber cradle]

Like the Stable audio example takes the Devil May Cry song and turns it into a Resident Evil style

[somber cradle]

But Suno creates something that has nothing to do with what I send. I want a remix with similar audio, but in the style I mentioned.

Besides, it creates extra time for the music.

[brisk wind]

it takes time to figure out how to use suno

[somber cradle]

I asked chatgpt if suno could do this and he said no

[brisk wind]

https://www.youtube.com/@minwagetrader/videos my whole channel uses suno

[somber cradle]

I want something that does a remake of the audio video that I sent. Stable audio does that.

Do you understand???

Stable audio is giving that error in the image I showed This happens when I use audio even though it is mp3 mp4 wav

I don't know how to fix it, but a guy here told me to send ticket I didn't understand anymore, but he told me to wait for them to respond.

[brisk wind]

Ok

[cobalt surge]

your email

[somber cradle]

Could someone test if Stable audio is still bugged with the audios?

[lyric elbow]

@south dawn possible scam/fishing attempt

[somber cradle]

@cobalt surge They answered me and now what do I do?

[cobalt surge]

I assume you mean stability.ai staff answered your ticket. Then you simply follow their instructions ?

[somber cradle]

They said they are working on this fix and that it is a website bug.

[loud garden]

I love using the cover feature on it for songs and seeing what AI does when giving it 15-25% variantion

[brisk wind]

Didn't know u can do that

[loud garden]

Yh u can

[somber cradle]

Stable audio will have an update on June 31st

Is it to fix that bug I sent to the guys?

[somber cradle]

Has anyone tested whether the audios in stable audio are working???

[supple eagle]

I was trying to get a horse whinney or neigh from stable-audio-open or stable-audio-open-small, and i can't seem to get anything besides weird human-sounding grunts. I'm wondering if there is a more specific prompt i ought to be using, or if this is just out of scope of what either of these models can do

any tips on how to use these for SFX/foley audio

[somber cradle]

I wanted a website where you send your audio and make a remix using prombt like Stable audio

Stable audio can do this but it has this bug and cannot use audios

[somber cradle]

Stop bots

[somber cradle]

Has anyone tested if that audio bug has been fixed???

[merry hound]

ACE ... frequency prompted, latent power edited before decode ... cleaned up in post

https://tenor.com/view/kimiko-five-tails-fortnite-dance-irish-jig-step-it-up-emote-gif-9366678942463289609

[merry hound]

lyrics by Comfyanon (just spiced em up) ... no Latent power editing...

played with noise and double sampling on the older ones

https://tenor.com/view/buffalo-bill-silence-of-the-lambs-basket-lotion-gif-16034350541563927252

[merry hound]

@low fjord #🎵｜stable-audio message

[low fjord]

passed this on

[merry hound]

conjured up a legend O_O

https://tenor.com/view/static-x-music-metal-gif-12843867219337554988

[winged raptor]

Seems like is working now

[somber cradle]

This error???

Did you mean it was fixed???

[somber cradle]

Does anyone have the paid Stable Audio? With 3 minute audios, if so could you make some songs for me????

[winged raptor]

Yes, I think so

[somber cradle]

Does anyone have paid Stable audio?

Why would I want someone to make a song from an audio of my full song since the paid mode is up to 3 minutes

[somber cradle]

Could someone help me with this that I asked

[somber cradle]

I wish someone would make some music for me

But since I don't have the paid version, I only have a 30 second limit.

I would like someone to help me make a remix of the audios with more than 30 seconds if anyone can help me thank you

[worldly copper]

Hy anyone's have hey gen ai membership contact ya who know how to do ai voice dubbing contact me

[hallow scaffold]

Hey! 👋
Working on OBSIDIAN-Neural, a VST plugin that does real-time AI audio generation for live performance.
Basically: type "dark techno kick" → AI generates directly in your DAW while you're jamming.
Current stack:

Stable Audio Open (local + server)
VST3 interface
8-track sampler with tempo sync
MIDI triggering

GitHub: https://github.com/innermost47/ai-dj
Demo video: https://www.youtube.com/watch?v=l4KMC5adxVA
Anyone else working on similar stuff? Curious to see what you're building here!

[somber cradle]

Will no one help me???

[patent hemlock]

https://tenor.com/view/empty-roll-tumbleweed-gif-5514767

[quiet wolf]

👍

[somber cradle]

I don't think anyone will help me

[somber cradle]

Is there anyone please who has paid Stable audio???

[austere fossil]

Does Comfy UI have UI for Stable audio?

Any recommended open source UI for Stable audio?

[quiet wolf]

amazing

[somber cradle]

Nobody wants to help me 😭😭😭

I'm lost for anyone who has paid Stable Audio to make some songs for me but they're ignoring me.

[winged raptor]

LOL that is probably because you actually pay for each generation

[somber cradle]

Basically you told me to get by

[winged raptor]

sorry,
If I remember correctly the free version let you create 10 songs about 3 min or so
But yeah, at some point is better to have a paid account

[somber cradle]

If I buy the cheaper paid version, will I be able to send 3-minute audios???

The audios I send are cut at 30 seconds

[winged raptor]

oh wait, I was thinking about https://stableaudio.com/, and you?

[somber cradle]

On this site the audios are cut at 30 seconds. I wanted to know if I get the paid option, at least the cheapest one, can I send audios longer than 30 seconds?

[winged raptor]

yep, you can send audios up to 190 seconds

I see now that this Is kind of new thing, before was limited to 30 sec.

[somber cradle]

I wish someone had the paid mode to make some music for me

I wanted to do some remixes of some audios that I have, but as they are more than 30 seconds long it's not possible. If I keep cutting it into parts it will be horrible.

190 Seconds is 3 minutes???

[winged raptor]

190 sec is 3 minutes + 10 seconds

[wicked prairie]

👍

[quiet wolf]

I love

[wicked prairie]

ohoh

[blazing vault]

fr

[dusk oyster]

stable audio open 1.0 is really cool. any source or documentation where it is described what this stuff does ?

[shut saddle]

You're more likely to get detailed assistance on the Harmonai discord.

[daring surge]

A cow is grazing on the grassland, and from time to time it lowers its head to look at the crowd in the distance

[patent hemlock]

https://tenor.com/view/spongebob-wait-gif-5140632

[blissful mauve]

why

[brisk wind]

Making my own music coz npc music is lame https://youtu.be/GrzXLq9VfTY https://rumble.com/v6w5ca8-crimson-sky-v2-music-video.html

[patent hemlock]

https://tenor.com/view/middle-of-nowhere-tumbleweed-desert-no-water-gif-12085928

[brisk wind]

that's sad

[main osprey]

hello

[somber cradle]

Are you guys getting a new audio error in Stable Audio???

[somber cradle]

I'm getting this error and I can't get into Stable Audio.

[mental drift]

Greetings! I'm wondering if any of the devs out there know if it's possible to be able to either multi-select and download or given access to your generated files through some other means. I do alot of multi-prompting and prompt refining .. generating 4 variations multiple times with very similar prompts. This causes a stacking effect and as a user it's easy to miss some of redownload the same file (also compounded by overwrite protection / auto indexing the filenames). Any information you have would be greatly appreciated. I love trying to get AudioSparkz 2 to generate the least through-composed things 🙂

[cedar crystal]

Hi everyone! 👋 A group of musicians and industry experts are currently collaborating to make an online course on how you can earn a 6-figure income through AI music (navigating the watermarking issue, promotion strategies, and every other caveat).

The course is on track to go live a few months from now. In the meantime, feel free to sign up for our email list at https://aimusiccreatorsacademy.com/ and interact with our pages:

• https://www.youtube.com/@aimusiccreatorsacademy
• https://www.instagram.com/aimusiccreatorsacademy/
• https://www.facebook.com/aimusiccreatorsacademy/
• https://www.tiktok.com/@aimusiccreatorsacademy
• https://x.com/aimcacademy
• https://bsky.app/profile/aimcacademy.bsky.social
• https://www.linkedin.com/company/ai-music-creators-academy

[vast prairie]

https://suno.com/song/77874a64-ab63-454a-8059-685f8ef9ff97

[bright dragon]

Hi friends, do you know of any open-source AI that creates sound effects? I create my videos with WAN 2.2, but I need to add sound. Open-source is preferable because it shouldn't be censored.

[pearl flint]

Any tools I should know about aside from the stock code in the official stable-audio-tools repo before I dive into training stable-audio-open and/or the small model?

[tidal lava]

@low fjord

[harsh coral]

this is actually pretty good

[dry furnace]

Wher tf mod is remove this shit

[crystal stream]

Has anyone else used stable audio output with soundsynth.com?

[runic orchid]

Hi, I've created a vibde data and AI engineering tool. And I would love for others to test it. Anyone who is in?

[runic orchid]

Greetings! I'm wondering if any of the devs out there know if it's possible to be able to either multi-select and download or given access to your generated files through some other means. I do alot of multi-prompting and prompt refining .. generating 4 variations multiple times with very similar prompts. This causes a stacking effect and as a user it's easy to miss some of redownload the same file (also compounded by overwrite protection / auto indexing the filenames). Any information you have would be greatly appreciated. I love trying to get AudioSparkz 2 to generate the least through-composed things 🙂

[runic orchid]

Are you guys getting a new audio error in Stable Audio???

[plush robin]

Hey everyone!

Wanted to share a project I've been building that might interest folks here who are into generative AI.
Obsidian Neural is an open-source VST3 plugin that brings Stable Audio Open directly into music DAWs for real-time generation.

What it does:
Instead of generating static audio files, it acts as a live instrument:

Type a prompt → AI generates a loop
Trigger it via MIDI like a sampler
8-track system with instant switching between variations
Everything syncs to your DAW's tempo

Think of it as Stable Audio Open meets live performance.

Why it's relevant here:

Built on Stability AI's Stable Audio Open model
Explores real-time inference for creative workflows
Open source (AGPL v3.0) — all code is public
LLM-powered contextual prompts for better generation

Similar vibe to what Stable Diffusion did for images, but for musicians who want to play with AI rather than just generate finished outputs.

Links:

GitHub: https://github.com/innermost47/ai-dj
Website: https://obsidian-neural.com
Demo videos: https://www.youtube.com/playlist?list=PL9PCUNVx6wp8gMdbo59a1k3M7sdSPmo5A

[red locust]

Stop the channel

[red locust]

What problems have you encountered

[wide breach]

https://youtu.be/Y0sFunHFpoQ?si=QswN7h0hGvHHN5Vl

[faint perch]

talented

[plush parrot]

Interested in this too.

[narrow galleon]

elevenlabs has a thing for this

[plush parrot]

Anything locally?

[plush robin]

macOS Update!
OBSIDIAN Neural - AI-powered VST3 for real-time music generation with text prompts & MIDI control.
Now signed & notarized by Apple - no more security warnings!
Perfect for live performance, jamming & improv. Generate unique audio loops on-the-fly and layer them with your gear.
https://obsidian-neural.com
https://github.com/innermost47/ai-dj

[lyric radish]

hey@all thank you for your great work!! I want to donate a large amount of ethically sourced audio data files (40+ million files + the tools to algorithmically increase the amount available), all types of hq audio assets (songs, loops, instruments, compositions etc) for ai training purposes to the stable diffusion project. Just send dm me

[plush robin]

I added a "Draw-to-Audio" feature to my AI music generation VST - sketch your sound instead of typing prompts

So I've been working on OBSIDIAN Neural, an open-source VST3 for AI music generation focused on live performance, and just added something weird: a canvas where you can draw what you want to hear.
How it works:

Draw on the canvas (lines, shapes, whatever)
Vision LLM interprets your drawing
Translates it into audio generation prompts
~10-20 seconds later, you've got a sample

Examples:

Chaotic scribbles → distorted aggressive rhythms
Smooth flowing curves → ambient pads
Sharp geometric shapes → structured sequences

It's not meant to replace traditional prompting, but gives you another creative input during composition/live sessions - especially useful when you're in the flow and don't want to stop to type.
The whole project is open source, presented at AES AIMLA 2025 in London. Built for musicians who want AI as an instrument, not a songwriting robot.
Links:

GitHub: https://github.com/innermost47/ai-dj
Website: https://obsidian-neural.com

Would love feedback from other producers/performers experimenting with AI tools!

[somber cradle]

I don't know if anyone will answer me, but could someone send a message to the stable audio developers to fix a bug in version 2.5?In this version, the audio tracks you add are not modified.They don't undergo any changes or anything.I've already tried setting up several prompts, but the audio doesn't change.Only the older version, 2.0, works well, but that version is terrible.I already tried messaging them, but I didn't know how to explain the problem properly, but I tried and it didn't work.

Is anyone else here having this problem too?

[wide breach]

https://youtube.com/playlist?list=OLAK5uy_nbQDb8X7lPuTUgf6V1FqHxybaRe7f8zQw&si=fOOYJzu9voCiIiuX

🧢 New album! Hip hop / synthwave, inspired by Stranger Things–style themes.

[wide breach]

Grove Lake - Textured Echo Lights [hip-hop, synthwave, intrumental] (2025)
https://www.youtube.com/watch?v=Cn8wqwDOOoc

[terse dagger]

A slight modification to the default workflow in the Comfy templates section. Supplying a random number to SuperPrompt produces a random prompt that the audio generator tries to match.

 a group of astronauts in spacesuits stand in front of a glowing orb, their faces etched with determination. The orb is a vibrant red, with intricate patterns etched into its surface. The astronauts wear helmets and helmets, and their faces are filled with excitement and anticipation. The scene is set in a futuristic space station, with glowing orbs and glowing screens.

[olive cedar]

hey is there API for this?

[cobalt surge]

depends what you mean by "this".
https://platform.stability.ai/docs/api-reference#tag/Stable-Audio-2

[frozen jay]

What is currently the best way to do AI audio

Im a bit new to SD - really great tool - looking into audio as well

[topaz mortar]

​🚀 LOOKING FOR A CREATIVE PARTNER (AI ANIME SERIES) 🚀
​The Vision: I am a 15-year-old creator building a high-quality AI Anime Web Series. The goal is to build a solid portfolio and eventually pitch to major platforms like Netflix/Crunchyroll.
​Who I am looking for:
​🎨 AI Artist: Someone who can generate consistent Anime characters (Stable Diffusion/Niji).
​🎬 Video Editor: Someone who can animate images into cinematic scenes.
​🌍 Age: Preferably 15-18 years old (Beginners are welcome!).
​The Deal: * This is a Long-term Partnership (50/50 Profit Share).
​No upfront payment, we grow and earn together as a team.
​Let’s build a masterpiece from scratch!
​Interested? Send me a DM

[lost marten]

Anyone know of any tools like Suno’s audio cover where you can sing and Suno will give back its own polished rendition, but same cadence?

I don’t mean cloning or auto tuning

But it gives you back an actual polished rendition (with a different voice )

Wondering what the tech is behind this

[sharp crow]

Anyone here that know if it would be possible to generate continues stream of music such as soundscapes etc?

@low fjord sorry for the ping but here is a scammer, couldn't find a moderator role to ping.

[worldly nebula]

@sharp crow hi

[sharp crow]

Allo?

[weak depot]

👋 Hi there!

I train flux /sdxl lora for your socials and also Onlyfans and patreon. If u need ur AI Influecer I'll be happy to help u with it.

🔗 Portfolio & custom LoRA with stable face:

https://www.behance.net/gallery/243708697/Stable-AI-Influencer-Private-Flux-Face-LoRA

[devout raptor]

sent a dm

[wet quarry]

I am trying to understand how I can input a particular sample, and then make its output sound more like sound fx, than music - if you know what I mean. Sort of musique concrete-ish. Are there prompts that can make my input sound like "tumbling pieces of cardboard" or just "metal", as in the object/material "metal"?

[coral meteor]

https://soundcloud.com/adren6kr6m6/sets/deadmedia

[sharp crow]

?

[sharp crow]

why do you keep pinging me?

[plush robin]

https://admin.obsidian-neural.com/

[orchid bobcat]

Hello

So what are people using these days for AI music/remixes? I never used one yet, only image generation

anyone can point me out toward something not too hard to use?

[frigid dune]

AI ADS

[wide breach]

Hip-hop, trap

https://open.spotify.com/intl-pt/track/1UwKdbhuX8wEK7kcEwhPlv?si=593a1bdb164349e7

[carmine obsidian]

Meet The Ultimate Saturday Morning Idiot Menace: LASER PIRATE!

[wide breach]

Hip-hop, Trap
https://open.spotify.com/intl-pt/track/4LTXX5RMaMsuqtH2Sr84Cv?si=8a55252005b24a96

[orchid bobcat]

asking the same thing as last time

What are people using these days for AI music/remixes? I never used one yet, only image generation
anyone can point me out toward something not too hard to use?

[sharp crow]

Suno is one example.

[orchid bobcat]

oh i forgot to say, i want something local and free like stable diffusion webuis and so on, if possible

[sharp crow]

through ComfyUI you could, if you start from a audio workflow there shouldnt be much hassle.

[wide breach]

Hip-hop, Trap
https://open.spotify.com/intl-pt/track/4zjPI0ny7cRkWbFIyXZWi9?si=f2bf28e922ee4820

[wide breach]

HIp-hop, Trap
https://open.spotify.com/intl-pt/track/0X7n39EUWQFsI0Cz8gBGwq?si=78625e1c73be4bae

[pallid whale]

Try ACE-Step v1.5.

[orchid bobcat]

thanks, but man i feel so dumb when reading these gitup install stuff... sometime you get the easy thing, sometime they say "clone shit" copy that, without saying where or how, lol

[orchid bobcat]

managed to install it, it works, but it all sounds pretty bad, i'm trying to remix something, but it never follows the music style i'm telling it to, even if i lower the strength of the original and so on
do you have exeperience using it @pallid whale ?

[pallid whale]

Yeah - I've been using it through ComfyUI, and I wrote an assistive layer using an LLM that supports structured requests.

It takes some practice to learn how to prompt it and get decent results. Also, make sure you're using 1.5 and not 1.3. There's huge difference between the two.

Suno is a more sophisticated commercial model, but I like using my own hardware and software to do it.

I can post one of my samples, but I'm not sure if I'm allowed to.

I guess everyone else is.

[orchid bobcat]

i'm just using the normal WEBUI cause i don't like comfy, but i wonder how you simply make the instruments you tell in your prompt to appear, it all sound like old school midis, and what are all the different options

i can't really talk much right now, but if you're okay with it, you could help me through DMs later

[pallid whale]

Certain genres are also represented better than others. It depends, really.

[pallid whale]

Sure. Feel free to send me a DM later. I'll get back to you if/when I'm free. 🙂

[surreal jolt]

#✨｜sdxl 帮我生成图片：28岁女生，身材曲线火辣，胸围饱满，腰臀比明显。她穿着一件贴身的白色短袖T恤和一条深蓝色的高腰紧身牛仔裤，脚上踩着一双白色运动鞋。站在城市街道的斑马线一端等红绿灯，身后是车流和建筑。她双手自然垂着，单肩背着一个黑色小包，身体微微侧对镜头，头转向马路对面的红灯，表情安静而自然，眼神放空地等待着，嘴唇自然闭合，整个人散发出一种日常出行的松弛感。

摄影规格：8K超高清画质，用富士GFX100 II中画幅相机搭配80mm f/1.8镜头拍摄，f/2.8光圈，快门1/800秒，ISO 200。利用下午四点的自然光，光线柔和带一点暖色。捕捉她等红灯时放空望向前方的自然神态，眼神平静不呆板，白色T恤的纹理和牛仔裤的质感清晰，背景是虚化的车流和城市建筑，有动态模糊感。

[stuck dirge]

Hello everyone! This is my first post on this server. Anybody used the stableaudio wrapper formation-1 (https://huggingface.co/RoyalCities/Foundation-1) and have any good or bad experience with it?

[plush robin]

Anyone with an RTX 3070+ collecting dust?

OBSIDIAN Neural is building a decentralized GPU network for real-time AI music generation — no big cloud, no middleman. Providers get 85% of subscription revenue split equally each month via Stripe.

The infra is live. The network needs builders.

https://github.com/innermost47/obsidian-neural-provider

[ancient echo]

this is absolutly sick! love it

[meager wedge]

https://youtu.be/ZujlxUiERMo?si=p-0dPfI-QiYpTBZ6

[carmine obsidian]

Lyrics fail but somehow decent.

It's meant to be a Disney-like parody about a ridiculous comic adventure between 2 pokemons 🤣

[knotty ravine]

If someone is still interested...
https://github.com/Stability-AI/stable-audio-3

[oblique patrol]

Not able to test it rn, does it support lyrics?

[knotty ravine]

I haven't seen any support for it.

[wide breach]

https://youtu.be/UXJ14BkGZCA?si=9_-k8RnBVTPhNSpr