Audio Glitch Detector

Detect audio glitches and discontinuities in sinusoidal audio signals by analyzing the signal derivative.

Important: This tool is will only work with sinusoidal audio signals.

Limitations

The algorithm detects discontinuities, so glitches that does not directly produce discontinuities might not be picked up. It can detect single sample errors in a sine wave. The test files in test_files/ show cases where the algorithm works well.

Installation

From PyPI

pip install audio-glitch-detector

From source with uv (development)

git clone https://github.com/kaspernyhus/audio-glitch-detector
cd audio-glitch-detector
uv sync --dev

Quick Start

Command Line Usage

Analyze a WAV file with automatic threshold:

audio-glitch-detector -f path/to/audio.wav

Monitor live audio stream:

audio-glitch-detector

Save glitch blocks for analysis:

audio-glitch-detector -f audio.wav --save-blocks

Library Usage

see examples/ for usage examples.

Visual Analysis

When using --save-blocks, glitch blocks (max 50) are saved as .wav files and with waveform visualizations showing the derivative analysis:

Test Files

Validate the detector with test files with a known number of discontinuities:

audio-glitch-detector -f test_files/sine_discont_2_mono.wav

Expected output:

Number of discontinuities detected: 2
0:00:01.892857
0:00:03.288367

How It Works

1. Derivative Analysis: Calculates the first derivative of audio samples
2. Threshold Detection: Uses 99.5th percentile of derivative distribution
3. Peak Identification: Finds samples exceeding the threshold

Development

Setup

uv sync --dev

Run the application

uv run audio-glitch-detector --help

Run tests

uv run pytest