@iska@mk.starnix.network You can do it the old algorithmic way. Beat detection is rather basic thing, removal might be more complex, but still possible — Ableton Live did that long before neural nets became the buzzword. Voice removal plugins also were a thing long ago — not that good, but no one expects perfection.
@aoi@miruku.cafe
Follow
@iska@mk.starnix.network
BTW you can even use NN and train it on tracks you have sources of or tracker music, where you can easily replace drums with meow sounds. I haven't really used anything other than libfann for OCR, but I'm pretty sure modern libraries are even more flexible — if you can feed it the source and the desired result, you can train it 🤓
@aoi@miruku.cafe
