DocumentCode
3451915
Title
Phoneme segmentation of speech signal
Author
Goh, Y.H. ; Raveendran, P.
Author_Institution
Dept. of Electr. Eng., Univ. of Malaya, Kuala Lumpur, Malaysia
fYear
2009
fDate
14-15 Dec. 2009
Firstpage
1
Lastpage
3
Abstract
This paper proposes a phoneme segmentation method based on pattern similarity measurement between two consecutive time-series voiced speech signal. System first performs pitch detection by using dyadic wavelet transform. Then, pattern similarity measurement has been carried out to act as phoneme boundaries detector. Output result has been compare to manual phonemic segmentation. Proposed method proves effective in finding phoneme boundaries.
Keywords
signal detection; speech processing; time series; wavelet transforms; dyadic wavelet transform; pattern similarity measurement; phoneme boundaries detector; phoneme segmentation; pitch detection; time-series voiced speech signal; Autocorrelation; Band pass filters; Detectors; Frequency; Signal analysis; Speech analysis; Speech processing; Speech recognition; Transient analysis; Wavelet transforms;
fLanguage
English
Publisher
ieee
Conference_Titel
Technical Postgraduates (TECHPOS), 2009 International Conference for
Conference_Location
Kuala Lumpur
Print_ISBN
978-1-4244-5223-1
Electronic_ISBN
978-1-4244-5224-8
Type
conf
DOI
10.1109/TECHPOS.2009.5412045
Filename
5412045
Link To Document