Aulyzer FREE — Scientific Audio Measurement Instrument

Aulyzer is a cross-platform command-line tool for detailed audio analysis. It is designed for engineers, researchers, and audio professionals who need precise, repeatable measurements on audio files, either individually or in comparison with a reference. Aulyzer is built for unattended operation and straightforward automation through scripting.

Purpose

The primary purpose of Aulyzer is to provide a comprehensive set of audio-quality and signal-characterization metrics. It is particularly effective in two scenarios:
- Single-File Analysis: Extracting a broad range of metrics from one audio file in order to assess its properties. This is useful for quality control, file validation, and general signal characterization.
- Pairwise Comparison: Comparing one or more probe audio files against a reference file to quantify differences in level, timing, and frequency-domain behavior. This is especially useful for testing audio-processing chains, evaluating codec quality, and verifying signal integrity after modification.

Key Features

- Comprehensive Measurements: Peak/true-peak (ISP), RMS, crest factor, DC offset, clipping statistics, activity analysis, bandwidth, spectral centroids, and more.
- Advanced Pairwise Comparison: Cross-correlation based lag detection, RMS and peak deltas, band-power differences, log-spectral distance, and frequency-response deviation metrics.
- Batch Processing: Feed one reference and many probes in a single command; each probe gets its own console and JSON report plus optional exports.
- Fragment Analysis: Target specific time ranges from each file without modifying the sources.
- Configurable Limits & Alerts: Attach min/max/delta constraints to nearly every measurement; out-of-range values are flagged in reports for automated QC pipelines.
- Optional Processing Steps: Controlled resampling, dithering, band-pass preparation, lag compensation, and compensated audio renders for deeper investigations.
- Extensive Parameter System: Every measurement, limit, action, and formatting option is addressable through key=value arguments. Run aulyzer --list-params to browse defaults.
- Flexible Reporting:
* Console output: human-readable sections with optional long descriptions or alert-only filtering.
* JSON export: every run produces machine-readable JSON files under the report directory for downstream tooling.
- Visualization Outputs: Generates Gnuplot scripts and PNGs for waveform and spectrogram views with customizable palettes, fonts, and dimensions.
- Cross-Platform: Works on Windows and Linux.

Measurement Categories

Single-file metrics:
     - Duration & detected sample rate / bit depth (nominal and inferred true bit depth).
     - Peak and true-peak level with timestamps, RMS window statistics (peak/mean/min/total).
     - Crest-factor trace, frame power, spectral centroid, DC offset, clipping percentage, event counts, and click detection via abs-delta PCM.
     - Activity timing and bandwidth estimation when enabled.

Pairwise metrics:
     - Lag detection via cross-correlation with optional automatic compensation.
     - RMS difference vectors (abs/peak/mean/min/total) with optional level compensation.
     - Band-power deltas, log-spectral distance, per-band frequency-response deviation, and configurable export of aligned or EQ-adjusted probe WAVs.

Optional actions & exports:
     - Controlled resampling
     - Dither injection for single-file and pairwise flows
     - Waveform and spectrogram PNG creation with full styling control (by means of gnuplot script generation).
     - Intermediate audio renders: aligned probe/reference, frequency-response compensated probe, and filtered variants when requested.

Outputs & Artifacts

* Console report summarizing the reference, each probe, and pairwise comparison sections
* JSON reports for the reference and every probe comparison
* Optional PNG images and WAV intermediates
* Alerts clearly flagged when any measurement violates configured limits

Supported Formats & Requirements

* Input format: mono PCM WAV files using one of the supported sample rates (2 kHz - 192 kHz)
* Fragment notation: wrap file name and optional start/stop seconds in braces {"file.wav",start,stop}
* Multichannel files must be pre-converted to mono before analysis