adefossez/demucs
View on GitHub ↗ ★ 3,526 Python
Code for the paper Hybrid Spectrogram and Waveform Source Separation
4 mentions on Hacker News by 4 people: 0 stories, 4 comments · first 2024-07-31 · latest 2025-12-23
20082026
-
Use Demucs bruh https://github.com/adefossez/demucs
comment by hamza_q_ · 2025-12-23 · in Our new SAM audio model transforms audio editing -
Of course there are music models that let you edit and remix music, just add a bit of Audacity: https://github.com/adefossez/demucs
comment by Shorel · 2025-04-25 · in DeepMind releases Lyria 2 music generation model -
I'm using https://github.com/adefossez/demucs to split drums, bass, voice and "everything else". Works pretty well for my personal/hobbyist use (quality also depends on genre and instruments used - synth stuff tends to bleed into voice a…
comment by senko · 2025-01-21 · in Guitar chord karaoke with Vamp, Chordino, and FFmpeg (2022) -
For audio analysis: https://librosa.org/ Audio track seperation: https://github.com/adefossez/demucs demucs works pretty well.
comment by jmkr · 2024-07-31 · in Ask HN: What are some unusual but useful Python libraries you've discovered?