ultimate-music-manager（终极音乐经理）

Sonic Phoenix

A multi-phase pipeline that transforms a disorganised local music collection into a pristine, fully-tagged library sorted by <Language>/<Artist>/<Album>/<Artist> - <Title>.<ext>. Works with any language — Hindi, English, Japanese, Spanish, or anything else your collection contains.

The pipeline never deletes audio files. Suspected duplicates are moved to a staging folder for manual review.

Quick Reference

Environment Setup

Prerequisites

• Python 3.12 (required — shazamio-core only ships wheels for 3.10-3.12)
• Git (to clone the repo)
• FFmpeg (optional — only needed if your collection contains non-MP3 formats like FLAC, OGG, or WMA that Shazam needs to decode)

Installation

# Clone into your music folder (recommended) or anywhere else
cd /path/to/your/music
git clone https://github.com/drajb/sonic-phoenix.git
cd sonic-phoenix

# Create virtual environment with Python 3.12
python3.12 -m venv .venv

# Activate
source .venv/bin/activate          # macOS / Linux
# .venv\Scripts\activate           # Windows PowerShell
# .venv\Scripts\activate.bat       # Windows cmd

# Install dependencies
pip install -r requirements.txt

Configuration

Create a .env file in the repo root (copy from .env.example):

# REQUIRED — absolute path to the root of your music collection
MUSIC_ROOT=/path/to/your/music

# OPTIONAL — override any of these defaults:
# SORTED_ROOT=/path/to/your/music/Sorted
# DATA_DIR=/path/to/your/music/.data
# SHAZAM_CONCURRENCY=20
# ITUNES_COUNTRIES=US,GB

Verify configuration:

python config.py

This prints all resolved paths and confirms MUSIC_ROOT exists.

Directory Layout After Setup

<MUSIC_ROOT>/
├── sonic-phoenix/           # This repo
│   ├── config.py            # Central configuration (single source of truth)
│   ├── .env                 # Your local overrides (gitignored)
│   ├── 01A_extract_metadata.py
│   ├── 01D_shazam_all_files.py
│   ├── ...
│   └── config/
│       └── language_hints/  # Optional language classification overrides
│           └── examples/    # Templates to copy and customise
├── Sorted/                  # Pipeline output (auto-created)
│   ├── English/
│   │   ├── Adele/
│   │   │   └── 25/
│   │   │       └── Adele - Hello.mp3
│   │   └── ...
│   ├── Hindi/
│   └── <any language langdetect finds>/
├── .data/                   # Machine-generated JSON artifacts (auto-created)
│   ├── metadata_catalog.json
│   ├── shazam_final_results.json
│   ├── catalog.json
│   ├── final_catalog.json
│   └── ...
└── <your unsorted audio files>

The Pipeline

Happy Path (Minimum Viable Run)

For a quick, effective sort of your library, run these scripts in order:

python 01A_extract_metadata.py     # Extract ID3/FLAC tags
python 01D_shazam_all_files.py     # Acoustic fingerprint every file via Shazam
python 02A_catalog_music.py        # SHA-256 hash catalog + duplicate detection
python 02D_organize_music.py         # Sort into Language/Artist/Album hierarchy
python 03A_consolidate_by_artist.py --dry-run   # Preview artist folder merges
python 03A_consolidate_by_artist.py              # Execute merges
python 03D_titanium_resort.py      # Final structural enforcement pass
python 04I_polish_and_enrich_v6.py # Enrich tags via iTunes + embed artwork
python 05I_finalize_catalog.py     # Lock the master catalog

Each script is resumable — if interrupted, re-run it and it picks up where it left off.

Phase 1: Extraction and Identification

When to use 01B vs 01D: If your collection has clean tags on most files, use 01A then 01B (faster — only fingerprints untagged files). If tags are unreliable (generic names like "Track 01", meaningless album fields, garbled metadata), use 01D to fingerprint everything and cross-validate against the acoustic results.

Shazam rate limiting: If you see HTTP 429 errors, lower SHAZAM_CONCURRENCY in your .env (try 5-10).

Phase 2: Cataloguing and Deduplication

Phase 3: Consolidation and Structural Cleanup

Phase 4: Metadata Enrichment

iTunes rate limiting: The script rotates queries across country codes (default: US, GB). Add more via ITUNES_COUNTRIES=US,GB,AU,CA in .env.

Phase 5: Finalization

Supporting utilities: 05A (JSON repair), 05B (sanitize results), 05C (confidence audit), 05D (residue purge), 05E/05H (empty-folder vacuum).

Phase 6: Spotify Sync (Optional)

Requires a free Spotify Developer account.

Spotify setup:

1. Create an app at https://developer.spotify.com/dashboard
2. Set Redirect URI to http://127.0.0.1:8888/callback
3. Add to .env:

   SPOTIFY_CLIENT_ID=your_client_id
   SPOTIFY_CLIENT_SECRET=your_client_secret

1. Run python 06B_spotify_setup.py — browser opens, approve scopes, token is cached.

Language Classification

The pipeline uses langdetect (NLP) to classify tracks by language automatically. No hardcoded language list — it creates whatever folders your collection needs.

Language Hints (Optional)

For artists that NLP can't classify from Latin-script titles (e.g. Bollywood artists romanised as "Arijit Singh"), create hint files:

config/language_hints/
├── examples/           # Copy and customise these templates
│   ├── English.json
│   ├── Hindi.json
│   └── README.md       # Full schema reference
├── Hindi.json          # Your custom hints (create from examples)
└── Spanish.json

Hint file schema:

{
  "artists": ["Arijit Singh", "A.R. Rahman", "Shreya Ghoshal"],
  "dna": ["bollywood", "bhangra", "desi"],
  "keywords": ["ishq", "dil", "pyaar"],
  "lang_codes": ["hi", "pa"]
}

• artists — exact artist names to classify into this language
• dna — substrings to match in folder/file paths
• keywords — substrings to match in track titles
• lang_codes — ISO 639-1 codes that langdetect might return for this language

Configuration Reference

All configuration lives in config.py and is driven by environment variables (set in .env or your shell).

Supported Audio Formats

.mp3, .m4a, .flac, .wav, .ogg, .wma, .aac, .opus

Safety Guarantees

• No deletion. Audio files are never deleted. Duplicates are moved to Duplicates_Staging/ for manual review.
• No writes into the music hierarchy. All working state lives in <DATA_DIR> or the repo directory.
• Resumable. Every script can be interrupted and re-run safely.
• Non-destructive tags. Enrichment writes new tags but does not remove existing valid data.

Troubleshooting

Scripts

Helper scripts that automate common workflows. Run from the repo root.

Usage

# Check environment is ready
bash ultimate-music-manager/scripts/preflight.sh

# Preview what the pipeline will do
bash ultimate-music-manager/scripts/run-pipeline.sh --dry-run

# Run the full pipeline
bash ultimate-music-manager/scripts/run-pipeline.sh

# Run including Spotify sync
bash ultimate-music-manager/scripts/run-pipeline.sh --spotify

# Check pipeline status at any time
bash ultimate-music-manager/scripts/status.sh

Hook Integration

A safety guard hook prevents accidental execution of destructive utility scripts.

Setup (Claude Code)

Add to .claude/settings.json:

{
  "hooks": {
    "PreToolUse": [{
      "matcher": "Bash",
      "hooks": [{
        "type": "command",
        "command": "./ultimate-music-manager/hooks/safety-guard.sh"
      }]
    }]
  }
}

The hook intercepts commands referencing 05D_force_delete_residue, 05F_final_scrub, total_scrub, or absolute_zero_sort and injects a warning requiring explicit user confirmation. Silent (zero overhead) for all other commands.

See hooks/HOOK.md for details.

References

Detailed documentation for deeper dives:

Design Principles

1. Config-driven. One config.py module, one .env file. No hardcoded paths anywhere.
2. Language-agnostic. No baked-in language lists. langdetect + optional hint files handle any language.
3. Phased and incremental. Each phase builds on the previous. You can stop after any phase and have a useful result.
4. Auditable. Every move, merge, and enrichment decision is logged to JSON in <DATA_DIR>.
5. Zero-config happy path. Clone into your music folder, create a .env with MUSIC_ROOT, and run.