Tomonari Kakino

Template Compression and Distance Normalization for Reliable Text-dependent Speaker Verification

In the field of text-dependent speaker verification, DTW is an effective technique for small foot print embedded systems, due to its low requirements on training data, storage and computation. Research on further reducing the storage has always been active. However, most of the proposed methods achieved compactness at a cost of performance degradation. The purpose of this research is to develop a template compression method for DTW-based systems without performance degradation. At the same time, a novel normalization method for matching distances is proposed. We expect both these two techniques will be effective for not only speaker verification but also other DTW-based tasks

Codebook-Based Pseudo-Impostor Data Generation and Template Compression for Text-Dependent Speaker Verification

DTW-based text-dependent speaker verification technology is an effective scheme for protecting personal information in personal electronic products from others. To enhance the performance of a DTW-based system, an impostor database covering all possible passwords is generally required for the matching scores normalization. However, it becomes impossible in our practical application scenario since users are not restricted in their choice of password. We propose a method to generate pseudo-impostor data by employing an acoustic codebook. Based on the pseudo-impostor data, two normalization algorithms are developed. Besides, a template compression approach based on the codebook is introduced. Some modifications to the conventional DTW global constraints are also made for the compressed template. Combining the normalization and template compression methods, we obtain more than 66% and 35% relative reduction in storage and EER, respectively. We expect that other DTW-based tasks may also benefit from our methods.