Title :
Reproduction and Recognition of Vowels Using Competitive Associative Nets
Author :
Nedachi, Naoko ; Kurogi, Shuichi
Author_Institution :
Dept. of Control Eng., Kyushu Inst. of Technol., Fukuoka
Abstract :
It is well known that the LPC (linear predictive coding) is a powerful tool for processing speech signals, and in this article we show that piecewise linear predictive coefficients obtained by the competitive associative nets called CAN2 can reproduce vowel signals better than the LPC. Furthermore, we present three distance measures for the CAN2 to recognize vowels, and we examine and analyze the recognition performance via the present measures and the relationship to the LPC
Keywords :
content-addressable storage; linear predictive coding; neural nets; speech recognition; LPC; competitive associative net; linear predictive coding; speech recognition; speech signal process; vowel reproduction; Chaos; Linear predictive coding; Piecewise linear approximation; Piecewise linear techniques; Signal generators; Signal processing; Speech analysis; Speech coding; Speech processing; Speech recognition; Competitive associative nets; Piecewise linear predictive coefficients; Vowel reproduction and recognition;
Conference_Titel :
SICE-ICASE, 2006. International Joint Conference
Conference_Location :
Busan
Print_ISBN :
89-950038-4-7
Electronic_ISBN :
89-950038-5-5
DOI :
10.1109/SICE.2006.315292
