I like messing with audio things. I was testing some AI audio and noticed that even if they clean incredibly some really dirty files, sometimes they mess up really badly.

Also every audio cleanup tool bills by the minute (why should a 61-second file cost the same as a 120-second one?).

So I spend some times and build a phoenix API with a rust service doing spectral denoising, compression, eq,de-essing and LUFS normalization. Two-pass: it analyzes the file first, then adapts the processing chain. Added some optional AI but it's not a generative one.

I think it's pretty cool. You can try it, free tier is 3 hr/month, billed by the second.

Let me know if you have questions, feedback, or you like the mascot!