daw-master-audio-analyzer

name	daw-master:audio-analyzer
version	1.0.0
description	Advanced audio analysis and feature extraction using librosa with optional Vamp plugin support via sonic-annotator
author	Hermes Agent
license	MIT
metadata	{"hermes":{"tags":["audio","analysis","feature-extraction","librosa","vamp","statistics"],"provides":["analyze","probe","extract_batch"],"requires":["python","librosa","soundfile","numpy"],"optional":["sonic-annotator"]}}

daw-master:audio-analyzer

Comprehensive audio analysis skill for extracting musical features, spectral characteristics, and metadata from audio files. Integrates with the daw-master meta-skill pipeline.

Capabilities

• Feature extraction: Tempo, key estimation, loudness, spectral features, MFCCs, harmonic/percussive separation
• Batch processing: Recursively analyze directories with progress tracking and CSV/JSON output
• Vamp plugin support: Optional sonic-annotator integration for advanced analysis (qM Tempotracker, NNLS Chroma, etc.)
• Graceful degradation: Works with librosa alone; uses sonic-annotator when available

Operations

Operation	Description	Parameters
analyze	Extract comprehensive feature set from a single audio file	file, features (list, optional subset)
probe	Basic file metadata (duration, sample rate, channels, format)	file
extract_batch	Recursively analyze directory, write results to CSV/JSON	directory, pattern, output_format, output_file

Dependencies

Required (Python):

pip install librosa soundfile numpy

Optional (Vamp plugins):

sudo apt install sonic-annotator
# Install Vamp plugin packs:
#   qm-vamp-plugins: tempo tracking, chroma, onset detection
#   vamp-example-plugins: basic features

Examples

from skills.daw_master.audio_analyzer import analyze, probe, extract_batch

# Single file analysis
features = analyze("input.wav")
print(f"Tempo: {features['tempo']} BPM")
print(f"Key: {features['key']}")

# Batch processing
extract_batch(
    directory="./samples",
    pattern="*.wav",
    output_format="csv",
    output_file="features.csv"
)

Output Format

analyze() returns a dictionary containing:

{
  "duration": 180.5,
  "sample_rate": 44100,
  "channels": 2,
  "tempo": 128.0,
  "tempo_confidence": 0.85,
  "key": "C major",
  "key_confidence": 0.72,
  "loudness": -12.3,
  "spectral_centroid": 2450.0,
  "spectral_rolloff": 5800.0,
  "spectral_flatness": 0.15,
  "mfcc": [mean_vector, std_vector],
  "harmonic_ratio": 0.68,
  "percussive_ratio": 0.32,
  "onset_count": 240,
  "zero_crossing_rate": 0.08
}

Integration with daw-master

All audio-analyzer operations are pipeable:

# Chain with sox-engine for preprocessing
from skills.daw_master.sox_engine import trim, normalize
from skills.daw_master.audio_analyzer import analyze

# Normalize then analyze
normalize("raw.wav", "normalized.wav")
features = analyze("normalized.wav")

Tips

• For accurate tempo/key: use high-quality, full-bandwidth audio (not MP3 128kbps)
• Batch mode: set pattern="**/*.wav" for recursive directory scanning
• Sonic-annotator features appear under vamp_plugins key in output
• Memory usage scales with file duration; process long files in chunks if needed