A custom MCP server that separates a YouTube track into stems (vocals, drums, bass) and extracts a sonic signature: BPM, musical key, stereo width, transient punch, and a 512-dim CLAP semantic embedding. Runs locally on CPU via Demucs and librosa.
machine-learning deep-learning mcp transformers music-information-retrieval librosa clap audio-processing source-seperation demucs audio-ml llm-tools stem-separation
-
Updated
Apr 14, 2026 - Python