Gender-based Speaker Classification using ANFIS: An Adaptive Neuro-fuzzy Approach with MATLAB

The core problem of this research work is automated gender based speaker classification, that is, to imbibe into a machine the capability of identifying the gender of a human speaker, that is, to identify whether a human speaker is male or female. A gender-based speaker classification tool has been developed to automate the classification process. The tool is based on speech signal processing. With the help of speech signal processing, a speech signal is analyzed first, followed by extraction of features from the signal. During speech analysis, the speech signal is processed and classified for enabling the extraction of pitch. The feature extraction is performed over the classified speech by using the ‘Cepstral analysis’ to determine the pitch of a speech signal. This allows the gender-based speaker classification to be accomplished on the basis of the pitch value.