Music Duplicate Finder

A Python script for finding duplicate and similar music files using spectrogram analysis with GPU acceleration support.

Features

• GPU Acceleration: Uses CUDA (RTX 4090) for faster processing when available
• Multi-format Support: MP3, WAV, FLAC, M4A, AAC, OGG, APE, WMA, ALAC, AIFF
• Advanced Analysis: Spectrogram and MFCC feature extraction
• Dual Detection: Exact hash duplicates and similar audio content
• Batch Processing: Efficient processing of large music libraries

Usage

python music_duplicate_finder.py /path/to/music/library

Command Line Arguments

Argument	Description	Default
library_path	Path to music library directory	(required)
--threshold	Similarity threshold (0.0-1.0)	0.95
--output	Output JSON filename	music_duplicates.json
--cpu	Force CPU mode (disable GPU)	False
--batch-size	Processing batch size	1000

Examples

Example 1: Basic usage

python music_duplicate_finder.py "D:\Music"

Example 2: Custom threshold and output:

python music_duplicate_finder.py "/home/user/music" --threshold 0.9 --output "results.json"

Example 3: Force CPU mode:

python music_duplicate_finder.py "/music/library" --cpu

Requirements

librosa>=0.10.0
numpy>=1.24.0
scikit-learn>=1.3.0
cupy-cuda12x>=12.0.0
cuml-cuda12>=23.8.0
cudf-cuda12>=23.8.0

GPU Requirements (Optional)

NVIDIA GPU with CUDA support
cupy, cuml, and cudf packages for CUDA acceleration