PARALLEL GENETIC-BASED HYBRID PATTERN-MATCHING ALGORITHM FOR ISOLATEDWORD RECOGNITION

Dynamic Time Warping (DTW) is a common technique widely used for nonli near time normalization of different utterances in many speech recogni tion systems. Two major problems are usually encountered when the DTW is applied for recognizing speech utterances: (i) the normalization fa ctors used in a warping path; and (ii) finding the K-best warping path s. Although DTW is modified to compute multiple warping paths by using the Tree-Trellis Search (TTS) algorithm, the use of actual normalizat ion factor still remains a major problem for the DTW. In this paper, a Parallel Genetic Time Warping (PGTW) is proposed to solve the above s aid problems. A database extracted from the TIMIT speech database of 9 5 isolated words is set up for evaluating the performance of the PGTW. In the database, each of the first 15 words had 70 different utteranc es, and the remaining 80 words had only one utterance. For each of the 15 words, one utterance is arbitrarily selected as the test template for recognition. Distance measure for each test template to the uttera nces of the same word and to those of the 80 words is calculated with three different time warping algorithms: TTS, PGTW and Sequential Gene tic Time Warping (SGTW). A Normal Distribution Model based on Rabiner( 23) is used to evaluate the performance of the three algorithms analyt ically. The analyzed results showed that the PGTW had performed better than the TTS. It also showed that the PGTW had very similar results a s the SGTW, but about 30% CPU time is saved in the single processor sy stem.