Title :
Talking Condition Identification using Circular Hidden Markov Models
Author_Institution :
Sharjah Univ.
Abstract :
In this work, circular hidden Markov models (CHMMs) have been used to enhance the recognition performance of isolated-word speaker-dependent and text-dependent talking condition identification systems. Our talking conditions in this work are: neutral, shouted, loud, slow, and fast. Our results show that CHMMs significantly enhance the talking condition identification performance compared to the left-to-right hidden Markov models (LTRHMMs)
Keywords :
hidden Markov models; speaker recognition; circular hidden Markov models; isolated-word speaker-dependent recognition; left-to-right hidden Markov models; text-dependent talking condition identification systems; Authentication; Automatic speech recognition; Hidden Markov models; Loudspeakers; Signal processing; Speaker recognition; Speech processing; Speech recognition; Testing; Text recognition;
Conference_Titel :
Information and Communication Technologies, 2006. ICTTA '06. 2nd
Conference_Location :
Damascus
Print_ISBN :
0-7803-9521-2
DOI :
10.1109/ICTTA.2006.1684562
