/music-lyrics — Lyrics Transcription

Transcribe lyrics from audio with timestamps using faster-whisper (when available).

Usage

/music-lyrics <audio_file_path>

Steps

• Validate the audio file path

• Run lyrics extraction:

python3 -m music_analyzer lyrics "<audio_file_path>"

Optionally specify model size: --model-size base (tiny/base/small/medium/large-v2)

• Present results:

• Has Vocals: Whether vocals were detected

• Language: Detected language

• Lyrics: Timestamped segments with text

• Full Text: Complete concatenated lyrics

Requirements

• Full tier: faster-whisper must be installed for actual transcription

• Lite tier: Only vocal detection heuristic (no transcription)

Install full dependencies:

pip install -e "~/.claude/plugins/music-analyzer/src/[full]"

Output Fields

Field Description

segments

Array of {start, end, text, confidence}

full_text

Complete lyrics text

language

Detected language code

has_vocals

Boolean — vocals detected

method

"whisper" or "none"