Study on formants extraction and recognition of Chinese voiced sound based on fuzzy rules

Author

Kai, Zhang ; Ye, Sheng ; Pu-qiang, Yan

Author_Institution

Dept. of Precision Instrum., Tsinghua Univ., Beijing, China

Volume

1

fYear

1996

fDate

14-18 Oct 1996

Firstpage

793

Abstract

The questions of feature selection and extraction in the recognition of continuous Chinese speech are discussed. First, the characteristics of Chinese and a knowledge of enunciation, audition and Chinese phonetics are exploited completely, and then Chinese voiced sound recognition based on the spectrum structure is presented. Finally, the extraction of the formants, one of the most important voiced sound spectrum features, is discussed in detail. A series of rules for the extraction of Chinese voiced sound based on a fuzzy knowledge process are put forward. In addition, we discuss some problems that are easily found in the formant extraction processes of the voiced sound. We also give solutions by using a typical example and the result is optimistic

Keywords

feature extraction; fuzzy systems; inference mechanisms; knowledge based systems; natural languages; spectral analysis; speech processing; speech recognition; Chinese phonetics; Chinese voiced sound; Chinese voiced sound recognition; audition; continuous Chinese speech recognition; enunciation; feature extraction; feature selection; formants extraction; formants recognition; fuzzy knowledge process; fuzzy rules; rules; spectrum structure; voiced sound spectrum features; Character recognition; Data mining; Feature extraction; Instruments; Loudspeakers; Pattern recognition; Robustness; Shape; Spectrogram; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing, 1996., 3rd International Conference on

Conference_Location

Beijing

Print_ISBN

0-7803-2912-0

Type

conf

DOI

10.1109/ICSIGP.1996.567382

Filename

567382