Model of speaker gender recognition based on audio data
A.T. Mukhamatkhanova, T.M. Oparina · Matematičeskie struktury i modelirovanie · 2023
In this article considered the task of recognizing the speaker's gender based on audio data using the KNN method. The authors propose an algorithm for extracting features from spectrogram of audio recordings based on calculating the centroid, bandwidth, and decay of high frequencies. The extracted features are then used as input data for the KNN method. The results show that the proposed method allows achieving high accuracy in recognizing the gender of the speaker.