Matlab implementation of a pitch/unpitched (voiced/unvoiced) detector (Arturo Camacho )

Subject: Matlab implementation of a pitch/unpitched (voiced/unvoiced) detector
From:    Arturo Camacho  <acamacho@xxxxxxxx>
Date:    Tue, 4 Nov 2008 00:04:27 -0800

Dear members of the list: I want to share with you my Matlab implementation of the pitched/unpitched (voiced/unvoiced) detection algorithm I presented in ISMIR 2008 [1]. It works by dynamically determining clusters of pitch and unpitched sound using as criteria the maximization of the distance between the clusters' centroids. Although the method was created for doing pitch/unpitched detection, it is so general (i.e., except for computational efficiency related parameters, it has no intrinsic free parameters,) that I foresee its use in other areas in which the identification of two relatively stable and separated classes in a time series is required (e.g., sound/silence detection). The algorithm is available in the URL under the name 'ts2means.m' (time-series 2-means). It takes as input the pitch strength trace of the signal (which can be approximated as the height of the autocorrelation function of the signal or using more sophisticated methods like in [2-4]) and produces as output a binary trace where 1s represent pitched sound and 0s represent unpitched sound. (BTW, that URL also contains a Matlab implementation of the pitch and pitch strength estimator in [3-4] under the name 'swipep.m'.) I hope many people find this algorithm useful. Sincerely, Arturo [1] Camacho, A. (2008). "Detection of Pitched/Unpitched Sound Using Pitch Strength Clustering," Proceedings of the Ninth International Conference on Music Information Retrieval, pp. 533-537. [2] Shofner, W. P., and Selas, G. (2002). "Pitch strength and Stevens' power law," Percept. Psychophys. 64, 437450. [3] Camacho, A. (2007). "SWIPE: A Sawtooth Waveform Inspired Pitch Estimator for Speech and Music," doctoral dissertation, University of Florida, Florida. [4] Camacho, A. (2008). "A sawtooth waveform inspired pitch estimator for speech and music," J. Acoust. Soc. Am. 124, 1638-1652. -- __________________________________________________ Arturo Camacho, PhD Alumni Computer and Information Science and Engineering University of Florida E-mail: acamacho@xxxxxxxx Web page: __________________________________________________

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University