Gaussian Components Optimization for a Robot Controlled by Speech Commands in Mexican Spanish

In this paper we describe two experiments using speech commands in Mexican Spanish language, for these experiments two techniques were used, dynamics time warping (DTW) and Gaussian mixture modeling (GMM), these are successful methods in speech recognition context, but rarely used in robotics domain. The position of a planar robot could be controlled by a microcontroller receiving signal commands from a computer interface. The computer receive speech commands which are processed using either, templates paradigm in the dynamic time warping case, or robust Mel frequency cepstral coefficients (MFCC) using the GMM method. Over the past several years GMM and MFCC have become two of the dominant approaches for modeling speaker and speech recognition applications, nevertheless we obtained similar results in experiments with DTW. In this paper we focus in the optimization of Gaussian components number.