pyACA 0.1.4

scripts accompanying the book An Introduction to Audio Content Analysis by Alexander Lerch

pyACA

IN WORK: Currently untested

Python scripts accompanying the book "An Introduction to Audio Content Analysis" (www.AudioContentAnlysis.org)

This package comprises implementations of simple algorithms and features for audio content analysis.

Please note that the provided code examples are only intended to showcase algorithmic principles – they are not suited to be used without parameter optimization and additional algorithmic tuning. More specifically, the python code might violate typical python style conventions in order to be consistent with the Matlab code at https://github.com/alexanderlerch/ACA-Code

The majority of these python sources require the numpy and scipy installed. Several functions (such as MFCCs and Gammatone filters) are based on implementations in Slaney’s Auditory Matlab Toolbox.

Please feel free to visit http://www.audiocontentanalysis.org/code/ to find the latest versions of this code or to submit comments or code that fixes, improves and adds functionality.

The top-level functions are:

• computeFeature: calculates instantaneous features
• computePitch: calculates a fundamental frequency estimate
• computeKey: calculates a simple key estimate
• computeNoveltyFunction: simple onset detection
• computeBeatHisto: calculates a simple beat histogram

The names of the additional functions follow the following conventions:

• Feature*: instantaneous features
• Pitch*: pitch tracking approach
• Novelty*: novelty function computation
• Tool*: additional help functions such as frequency scale conversion, dynamic time warping, gammatone filterbank, ...

Example: Computation and plot of the Spectral Centroid

    #import numpy as np
    import matplotlib.pyplot as plt 
	import pyACA

    # read audio file
    [f_s,afAudioData] = pyACA.ToolReadAudio(cPath)
    #afAudioData = np.sin(2*np.pi * np.arange(f_s*1)*440./f_s)

    # compute feature
	cFeatureName = "SpectralCentroid"
    [v,t] = pyACA.computeFeature(cFeatureName, afAudioData, f_s)

    # plot feature output
    plt.plot(t,v)