This is a Python library for working with audio. It comes in two flavors - a regular Python version called pyaus, contained in the pyaus directory, and a Cython version called caus, contained in the caus directory.
- 
If you are running Windows, you can simply run the command pip install aus-pythonto install a prepackaged version from PyPi.
- 
If you are not running Windows, or if you want to build and install the package manually, follow these instructions: - 
Before building this package, first decide whether you wish to use the Python version or the Cython version. Copy all of the .py files from pyausorcausto theausdirectory.
- 
You will need to have a Python virtual environment with buildinstalled in it. Alternatively, you can simply installbuildin your local Python installation. You may also need to install the packagesCython,numpy, andsetuptoolsto build the package.
- 
Run the command python -m build. The package will be located in thedistdirectory, in both.whland.tar.gzformat.
- 
To install the package in your local Python installation, simply navigate to the distdirectory and run the commandpip install filename.whl(replacefilenamewith the actual name of the file in thedistdirectory).
- 
You are now ready to import ausinto your Python code. Note that the package name isaus-python, because PyPi would not let me use the nameaus. However, when importing into your Python code, you use the lineimport aus.audiofileorimport aus.operationsetc.
 
- 
Documentation is available at https://aus.readthedocs.io/en/latest/.
You will need the following Python libraries: matplotlib, numpy, pedalboard, regex, scipy. You will also need Cython if you want to build the Cython version.
The package is divided into 8 modules:
Tools for spectral analysis and analysis of audio waveforms. Many of these tools are based on formulas from Florian Eyben's "Real-Time Speech and Music Classification," published by Springer in 2016. Among other things, this module computes spectral centroid, entropy, slope, and flatness.
This module is for reading and writing audio files, using either the pedalboard library or using (slower) code provided here.
Funtionality for grain extraction
This module has various operations that can be performed on audio, such as spectral frame swapping, equal energy forcing, dc bias removal, and beat envelope generation.
Plotting functionality for audio and spectrum
Tools for extracting samples from audio
Tools for spectral analysis
Tools for generating simple waveforms