A Wavelet-Based Approach for Pitch Classifiers

Abstract

A pitch is a human classification method which allows ordering sounds in discrete harmonics frequencies. Although many of the practical pitch detection methods uses the autocorrelation and cepstrum techniques to detect audio pitch, a wavelet-based classifier was developed using wavelets to create a unique signal coding for each pure pitch signals. A comparative study of how the wavelets coefficients vary on each of the seven proposed pure pitch signals were given in this article. Also mathematical and illustrated examples of this study were presented as well. Key Terms - Autocorrelation, Classifier, KNN, Wavelets, Wavelets Transforms.