Thanks Richard!

“New form of synthesis” is exactly how I see it too.

Just like FM brought new timbres in the 80s, or granular
opened new textures in the 2000s, AI-driven sample
generation is another tool in the creative palette.

Not a replacement, an addition.

Appreciate you getting the vision.

I am the developer of this plugin. Generated samples are usable commercially under the Stable Audio license, provided your annual revenue does not exceed $1 million USD.
https://huggingface.co/stabilityai/stable-audio-open-1.0/blob/main/LICENSE.md

What about the other stuff he said?

this one is just a sad little footnote

congratulations on foisting yet another faeces picasso upon the arts

you’re using the term ‘producers’ pretty similarly to what i experiance after morning coffee

– All 8 tracks now output in stereo
– Stereo waveform display
– Stereo VU meters
– Complete stereo pipeline

Download: https://github.com/innermost47/ai-dj/releases/tag/v231

AI generated sounds cannot be copywritten but music made with it can. it means that you can use the samples without fear of legal action but the music made is human construction and can be copywritten

AI samples are legally not allowed to be copyrighted so a plugin that can generate AI samples can make sounds on the fly without worries about legal action

“AI”

Honestly, I get it. Even as the developer of OBSIDIAN Neural, I have conflicting feelings too.

When I see tools like Suno or Udio generating entire tracks from A to Z, yeah—I worry. Because the fun IS in creating music, working through ideas, composing. I’ve been making music for 23 years. I understand that joy deeply.

What I tried to do with OBSIDIAN is different: it’s a plugin that lets me perform live, mixing AI-generated elements with my hardware synths. I trigger drum loops, bass, pads, and ambient sounds via MIDI while I improvise on my gear. It’s just another “instrument” in my setup—not a replacement for composition.

Generative AI is here. Rather than reject it entirely, I asked myself: How can I use it creatively and non-destructively? How can I integrate it with existing tools without replacing them?

That’s my philosophy. If some people disagree, that’s fine—they can follow their own path. I won’t be offended if, as a guitarist, someone doesn’t want to hear about guitars. Everyone has their thing.

I also like the unpredictability—asking the AI for a sample and getting something slightly different than I imagined. That adds an element of surprise I find creatively interesting.

Yes, there’s a SaaS with a subscription. But the entire codebase—VST and inference server—is open source and free: https://github.com/innermost47/ai-dj

Anyone can download it, self-host, and have total control over the infrastructure. It’s completely free that way. You just need a decent GPU (my RTX 3070 generates samples in ~30 seconds).

I built the SaaS for users without GPUs. And yes, this project takes significant time—as a developer who wants to work on it full-time, I charge a subscription to pay myself and cover third-party services (I use fal.ai for inference models on the SaaS).

That’s the full picture. Use it, don’t use it—either way, I respect your choice.

“This plugin is worthless.”

Have you tried it?

that was my first thought also.