Title :
DTW speech recognition algorithm of optimization template matching
Author :
Zhang, Jing ; Qin, Benzhuo
Author_Institution :
Department of Computer Science and Technology, Guangdong University of Foreign Studies, China
Abstract :
For the Traditional DTW algorithm in speech recognition, long time is need for the matching of test speech and reference speech template, against which an improved algorithm of searching path was proposed, the algorithm improved operation speed, especially when the two time sequence involved comparison long, the effect of improvement was more visible. In order to improve the recognition rate of DTW, template matching method was improved, that optimization template matching method. Experimental results show the speech recognition system with improved algorithm effectively reduces the amount of data to be processed and recognition time, and the recognition rate of the system was increased as well. The algorithm had more obvious advantages with the number of speech signal to be identified increased.
Keywords :
DTW; Real-time; Recognition rate; Searching path; Speech recognition; Template matching;
Conference_Titel :
World Automation Congress (WAC), 2012
Conference_Location :
Puerto Vallarta, Mexico
Print_ISBN :
978-1-4673-4497-5