Automatic speaker verification based on fractional Brownian motion process

A novel text-independent verification system based on the fractional Brownian motion (M_dim_fBm) for automatic speaker recognition is presented. The performance results of the M_dim_fBm were compared to those achieved with the Gaussian mixture models (GMM) classifier using the mel-cepstral coefficients. A speech database, obtained from fixed and cellular phones, uttered by 75 different speakers was used. The results have shown the superior performance of the M_dim_fBm classifier in terms of recognition accuracy. In addition, the proposed verification scheme employs a much simpler modelling structure as compared to the GMM.