Ingeniería de características para clasificación de señales sonoras

Abstract

The classification of sound signals is an area in which there is a great research activity with very diverse applications in numerous areas of science and engineering. These applications include, among its first steps, the need to represent sound through a set of features. The set of techniques that allow the transfer of raw data to the most appropriate characteristics for its subsequent application is commonly known as features engineering.On the other hand, the processing and classification of anura songs have attracted the attention of the scientific community, both from a biological point of view, and as indicators of climate change. This work focuses on the parameters used to characterize the anura sounds. Jobs can be grouped into three categories: 1. Comparison of alternatives in the characterization of auro sounds. Determination of the optimal values ​​of such characterization 2. Analysis of the role played by the symmetry of integral transformations in the characterization of sounds based on cepstral coefficients. 3. Analysis of the computational effort required in the different stages of the characterization and classification process of anura sounds.