Masao Watari

Speaker verification system

Control reference pattern candidates corresponding to a verification reference patterns of a registered speaker are synthesized by connecting unit speech patterns of a plurality of speakers. A plurality of control reference patterns similar to the verification reference pattern are determined from among the control reference pattern candidates. First dissimilarity between an input pattern of a speaker to be verified and the verification reference pattern specified by the registered speaker and second dissimilarity between the input pattern and the control reference patterns specified by the registered speaker are calculated. The speaker to be verified is judged as the registered speaker on the basis of the first and second dissimilarities.

A high-speed dp-matching algorithm based on frame synchronization, beam search and vector quantization

This paper discusses the high-speed DP-matching as the speech recognition algorithm including connected word sequence recognition. The first improvement is the frame synchronization. By this elaboration, an improvement of the speed by approximately one order of magnitude is achieved, compared with the consecutive word recognition of two-level DP-matching type, where DP-matching is iterated by assuming that any time in the input speech can be the word boundary. The second improvement is the introduction of the beam search. n n n nThis paper discusses the practical aspects of combining the beam search and DP-matching. The discussion includes the construction of the work area, control of DP recursive expression and other problems, aiming at an effective reduction of the computational complexity for the recursive expression. The third improvement is the built-in vector quantization. It is shown that an effective reduction of the computational complexity for the local distance can be produced through a skillful integration of the beam search and the vector quantization. n n n nThrough an evaluation experiment for the discrete word, it is seen that there is a possibility of achieving the speed improvement by a factor of 30. This corresponds to the speed improvement of two or more orders of magnitude, compared with the two-level DP-matching for the consecutive word sequence recognition algorithm.

New DP matching algorithms for connected word recognition

In this paper, two new algorithms are proposed to overcome demerits in previous algorithms, such as the CWDP algorithm or the One-Pass algorithm. In the first one, called Blockwise DP matching (BWDP) algorithm, the calculation is carried out in step with the block having BL input pattern frames, instead of one frame used in the CWDP algorithm. This reduces the number of memory access times to 1/BL. However, it cannot handle the finite state automaton control with loop transition rules. In the other algorithm, called Slant-blockwise DP matching (SBDP) algorithm, the calculation block is inclined to the reference pattern time axis. Calculation is carried out in each slant block with BL frame width. This makes it possible to handle the finite state automaton control with loop transition rules. However, the program for this algorithm is rather complex. Further, improvement concerning variable block width is proposed. It can extend the effective block width. Namely, it can further reduce the number of memory access times.

Connected digit speech recognition by using demi-word pair reference pattern (DWPR)

This paper proposes two methods that cope with difficulties in connected digit recognition. One is the Demi-Word Pair Reference Pattern(DWPR) Method, which is useful for dealing with the coarticulation effect appearing at the word boundaries. The other is the Typically Distorted Pattern (TDP) Method. This helps prepare robust reference templates including several types of distortion which often occur in natural speech. An experiment has been performed for 20 untrained speakers in speaker-dependent mode. The TDP was gathered from multiple speakers pattern. The results showed that the TDP+DWPR integrated method achieved considerably higher recognition accuracy (99.0%) than a conventional word-based DP-matching method (97.5%).

A speaker-independent word-recognition system using multiple classification functions☆

Abstract A speaker-independent word-recognition system has been developed using multiple classification functions for separating 100 spoken words. The speech signal is first analysed and then non-uniformly time-sampled by referring to word-structure tables to construct a word pattern vector of 120 dimensions. Equivalently piece-wise quadratic classification functions are calculated based on a linear-programming algorithm using a large number of spoken-word design samples. A hardware system for real-time recognition has been built as a high-speed microprocessor complex. Using the classification functions calculated from design samples of 100 speakers, a recognition rate of 99% has been obtained for 50 unknown speakers.