The creator of the Linux operating system, Linus Torvalds, has released a new open source hobby project, AudioNoise, a digital guitar effects pedal.
Torvalds started making effects pedals from kits recently as a holiday project, and as a way to learn analog electronics. He describes building effects pedals as, “LEGO for grown-ups with a soldering iron.”
Here’s what he has to say about AudioNoise:
“These are — like the analog circuits that started my journey — toy effects that you shouldn’t take seriously. The main design goal has been to learn about digital audio processing basics. Exactly like the guitar pedal was about learning about the hardware side.”
What has surprised many is that Torvalds says that he used vibe coding in developing the pedal.
‘Vibe coding’ is a new approach to coding that essentially delegates programming details to AI. It’s done using large language model (LLM) systems, which generate a result to match the user’s prompt. The idea is that the developer can focus on the ‘vibe’ of a project – the high-level vision and requirements – and AI acts as a dedicated junior developer.
Torvalds does not tout his project as a great effect, but more as a fun way to learn about coding and digital signal processing. The project may be less significant on its own than as a sign of where things are heading with AI-augmented development.
You can get a sense of the project from the video embedded above, by SavvyNik, and check out Torvalds’ repo at Github.
Fuck ai
We should be open to AI tools that give us new creative tools and options – like AI stem separation, sample creation, etc. There’s a lot of cool stuff happening in this area.
But yeah, fuck AI that’s designed to replace musicians.
I’m not dead set against the use of AI or “vibe coding”, but there’s a high potential for disastrous results if your entire codebase is vibe coded. There will inevitably be bugs, and if you don’t know what’s going on with your code, you might have a tough time finding the bugs, let alone fixing them. Proceed with caution. AI is a tool, not a replacement for software engineers.
“AI is a tool, not a replacement for software engineers.”
For musicians, too. I don’t want to listen to AI-made music, but do see places where I’d use it in music production.
I’m interested in tools that help with things I don’t know much about – like solving mix problems or mastering tracks for different platforms. Ideally, the tools should be transparent about what they do, so we can learn more as we use them, and ultimately tweak the controls to get exactly what we want.
You know who the article is about, right? Pretty sure he knows how to code!
As a synth developer, I see AI as a great tool for creating stuff that’s in my mind more easy and faster, so that’s perfecly fine for creating audio hardware or software devices. But creating music directly by AI is a total no-go, imo. Big difference!
Only the visualizer tool was vibe-coded. Directly from Linus’s GitHub:
“Also note that the python visualizer tool has been basically written by vibe-coding. I know more about analog filters — and that’s not saying much — than I do about python. It started out as my typical “google and do the monkey-see-monkey-do” kind of programming, but then I cut out the middle-man — me — and just used Google Antigravity to do the audio sample visualizer.”
Speaking as a non-tech person, the problem with AI is I never know who it is. Sometimes it’s smart, insightful, useful. Other times it’s a dangerous, lying, moron. The problem is you never know which AI (or what mix of the two) will show up in any given instance. AI is just some random guy.
Interesting interview with Linus where he says you want a pet project where you’re free to fail, so you can learn things
I also love the Linnux platform and GitHub.