This model encodes audio files into vectors of 100 dimensions. It was trained on 240,000 Spotify playlists and on 30 second samples of over 4 million songs. The details can be found (here)[https://github.com/teticio/Deej-AI].
To encode an audio first install the package with
pip install audiodiffusion
and then run
from audiodiffusion.audio_encoder import AudioEncoder
audio_encoder = AudioEncoder.from_pretrained("teticio/audio-encoder")
audio_encoder.encode(<list of audio files>)