Classification of Audible Signals by Characteristics of the Human Vocal Apparatus

In this work, the performance of the Knn-algorithm for classifying different kinds of signals will be analyzed. In particular, the classification will be between two groups: voice and music signals. The distinctive features of speech signals will be exploited to separate them from musical ones. The classification will be based on mean and deviation of the amount of peaks from each spectogram line. In order to adapt the concept of line, thresholds have to be set. Finally, some improvements will be proposed, based on the obtention of other features and the setting of new thresholds for enhancement of performance.