Audio Classifier & M3U Playlist Generator

---

Audio Classifier

Overview

A Python command-line tool for classifying .wav and .flac audio files using rule-based audio feature analysis. It generates a CSV report with classification results.

---

Features

• 🎧 Audio Feature Extraction: Uses librosa to extract tempo, RMS energy, spectral centroid, etc.
• 📊 Rule-Based Classification: Matches extracted features against predefined rules (e.g., "Happy" for 120-140 BPM).
• ⚡ Parallel Processing: Utilizes ProcessPoolExecutor for fast batch analysis.
• 📄 CSV Output: Generates a structured report with feature scores and matched labels.

---

Requirements

• Python 3.6+
• Libraries:

  pip install numpy librosa

---

Usage

python3 eval_to_csv.py --directory <path> --output <file> --min-score <value>

Example:

python3 eval_to_csv.py --directory ./music --output audio_index.csv --min-score 0.7

Arguments:

• --directory: Audio file directory (default: current directory).
• --output: Output CSV file (default: audio_index.csv).
• --min-score: Minimum score threshold for label matches (default: 0.66).

---

Output CSV Structure

Column	Description
file_path	Full path to the audio file
best_label	Highest-scoring label
best_score	Score for best_label
error	Error message (empty if no error)
<label>_score	Match score for each label
<label>_passed	Number of rules passed for the label
<label>_total	Total rules for the label
<label>_matched	1 if score ≥ threshold, else 0

---

M3U Playlist Generator

Overview

Generates .m3u playlists from an audio CSV index, supporting device-specific path transformations (e.g., HiBy, Shanling, generic drives).

---

Features

• 📁 Detects labels from CSV headers (<label>_matched, best_label, etc.).
• 🎧 Creates playlists for matched labels, best labels, or highest scores.
• 🔄 Supports path transformations for different devices.

---

Usage

1. Prepare CSV: Ensure it contains columns like file_path, <label>_matched, <label>_score.

2. Run Script:

   python3 csv_to_m3u.py <csv_file> --output-dir <dir> --base-path <prefix>

3. Variants:

   • varianta: HiBy devices (a:\Music\)
   • variantA: Shanling devices (A:\Music\)
   • variantc: Generic drives (c:\Music\)

---

Example

Input CSV (audio_index.csv):

file_path,genre_matched,artist_matched,best_label,genre_score,artist_score
/music/song1.flac,1,0,genre,0.9,0.8
/music/song2.wav,0,1,artist,0.7,0.9

Command:

python3 csv_to_m3u.py audio_index.csv --output-dir playlists --base-path /music/

Output Structure:

playlists/
├── varianta/
│   ├── genre.m3u
│   └── artist.m3u
├── variantA/
│   ├── genre.m3u
│   └── artist.m3u
└── variantc/
    ├── genre.m3u
    └── artist.m3u

---

Notes

• Skips rows with errors.
• Falls back to highest score if no labels match.
• Normalizes file paths and applies variant-specific prefixes.

---

m3u setup

On Hiby non-Android devices:

• Create a directory playlist_data and place m3u files inside
• On the device in Music > Folder Playlist icon > Playlists, select option "Load playlist" (the operation takes time depending on the size of the files)

On Shanling non-Android devices:

• Create a directory _explaylist_data and place m3u files inside
• On the device in Playlist click on "Import Playlist" button (the operation takes time depending on the size of the files)

TODO on hibyMusic and Poweramp on Android devices.

---

License

MIT License. See LICENSE.