"Show HN: Polymath: Convert any music-library into a sample-library with ML" https://news.ycombinator.com/item?id=34756949#34782253

Looks like there's a Hooktheory Classroom product, in addition to Hookpad and their music theory book which features the music of Avicii.

/r/musictheory/wiki suggests that "Open Music Theory" is the best text: https://www.reddit.com/r/musictheory/wiki/index/

The search keyword "Melodic"

Do Windows VST plugins work with Glicol on Linux with Yabridge, LinVst, Carla, or WineASIO?

Glicol is written in Rust and has its own non-Rust syntax?

Is there a syntax/grammar API doc?

> Show HN: Polymath: Convert any music-library into a sample-library with ML

I tried this the other day. There's several issues regardless of using the pinned versions in requirements.txt and trying to use debian bullseye and python 3.10.10 as implied by the provided dockerfile. At least these I noted:

- fix deprecated np.bool usage

- module 'scipy.signal' has no attribute 'hann' --> change to scipy.signal.windows.hann

- librosa.feature.melspectrogram(y, change to y=y

- needs CUDA 11

I got it to work without gpu acceleration cause of the last point. If I come back to this I'll probably just look at using demucs, librosa etc. directly (polymath.py is only ~600 lines gluing things together) and hopefully get it working with latest nvidia drivers.

Also if I have days to let this compile maybe I could get CUDA 11 from this https://aur.archlinux.org/packages/cuda-11.7

Installing conda-forge/tensorflow-gpu installs conda-forge/cudatoolkit-feedstock ; https://conda-forge.org/blog/2021/11/03/tensorflow-gpu/ :

  conda install -c conda-forge install tensorflow-gpu # cudatoolkit
  mamba install tensorflow-gpy
  pixi install tensorflow-gpu

FWICS the polymath README mentions the reference docker container, which should install all of the dependencies:

  docker build -t polymath ./
  docker run \
    -v "$(pwd)"/processed:/polymath/processed \
    -v "$(pwd)"/separated:/polymath/separated \
    -v "$(pwd)"/library:/polymath/library \
    -v "$(pwd)"/input:/polymath/input \
    polymath python /polymath/polymath.py -a ./input/song1.wav

There's a music utility called "time travel" IIRC, that records all recent audio and you can then save a recent window of that. (Edit)

jellyjampreserve: https://github.com/betodealmeida/jellyjampreserve :

> JellyJamPreserve is a Raspberry Pi project that uses jack_capture to record audio on the toggle of a switch. It continuously records audio into a 5 minute circular buffer, and when the switch is turned on the buffer is dumped to a file and it starts recording audio to the file until the switch is turned off. This way you can preserve any cool improvisations by recording sounds from the past!

swh/timemachine; "JACK timemachine" https://github.com/swh/timemachine manpage: https://manpages.org/timemachine

Is there a better way to do this with Pipewire instead of Pipewire-jack?

Want to let you know that I just spent an hour playing around with glicol. It's really something, and very cool that you even included a collaberative editor. Have you shown it on HN yet? I think lots of people here would be interested

Yes it's showed a while ago. But now it's in the next round of refactoring. The syntax does not change much, but the core speed is optimized. The website needs to be rewritten. I am considering using solidjs, dioxus and egui. In addition to the browser editor, there is also this cli that can be used: https://github.com/glicol/glicol-cli