Muser

Muser is a “smart” music visualizer, made by Jon Shamir. It is an experiment to test how maching learning technology can be used to enhance music visualization.
The idea for muser and the visualization itself is inspired by Wassily Kandinsky (1866-1944). Generally credited as the pioneer of abstract art, his work is well-known for its musical inspirations. Kandinsky associated specific tones and instruments to shapes and colors, thus “visualizing” a musical composition. [jonshamir/muser]

Muser是由Jon Shamir制作的“智能”音乐可视化工具。它是一个实验，以测试机器学习技术如何用于增强音乐可视化。
Muser的构思和可视化本身受到瓦西里·康定斯基（1866-1944）的启发。康定斯基通常被认为是抽象艺术的先驱，他的作品因其音乐灵感而闻名。康定斯基将特定的音调和乐器与形状和颜色关联，从而“可视化”音乐作品。 [jonshamir/muser]

git clone --depth=1 https://github.com/jonshamir/muser
cd muser

Create requirements.txt:

audioread==3.0.1
librosa==0.8.1
musicnn==0.1.0
numpy==1.16.6
pandas==1.1.5
scikit-learn==0.24.2
scipy==1.5.4
soundfile==0.12.1
tensorflow==2.3.4
resampy==0.2.2
ipython==7.16.3

conda create --name muser python=3.6.13
conda activate muser
pip install -r requirements.txt
pip install matplotlib

Edit tools/tagger.py, playlist-creator.py:

# %matplotlib inline

python tools/tagger.py
pip install eyed3
python tools/playlist-creator.py

npm install
npm install --save-dev cross-env
npm run start
# set NODE_ENV=development && node tools/bundler.js

reference

• #deepwiki jonshamir/muser
• batch_muser_tagger.py

muser_batch-script.mp4, (bilibili