Abstract

Gender audio classification is considered one of the most significant methods in audio processing. In this paper, an algorithm involving audio features (mean, standard deviation, zero crossing, Amplitude) and Support Vector Machine (SVM) is presented to perform speaker gender recognition. For each audio, the highlights vector is utilized as an info vector in the SVM algorithm. An example of 2270 audio, include 1132 female audio with 1138 male audio is analyzed based on this algorithm. With only the four features, the average prediction error is 5%.

Keywords

The article was added to IASJ on 2018-05-27

454 Total full text downloads since the date of addition