Binary quantization of feature vectors for robust text-independent speakeridentification

In this paper, we present a novel approach to vector quantization in which a feature vector is represented by a binary vector. It is called binary qua ntization (BQ), The performance criterion of vector quantization, distortio n (distance) measure, was employed for investigating the effectiveness of B Q. At 12 b/analysis frame, the average distortion caused by BQ is even lowe r than the intraspeaker average distance between two repetitions of the sam e word (after DTW alignment). Since the output of BQ is a binary sequence, it is possible to combine it with forward Hamming net classifier. In terms of the idea of hierarchical model for describing a speaker individual chara cteristics, a text-independent speaker identification system was set up. Ex perimental results show that the performance of this system is very good. N ot only are the small memory space and little computation required, in the speaker identification system, but, more importantly, it shows strong robus tness in additive Gaussian white noise.