DocumentCode :
2484649
Title :
Real-Time Endpoint Detection Algorithm Combining Time-Frequency Domain
Author :
Zhou, Ming-Zhong ; Ji, Li-Xin
Author_Institution :
Nat. Digital Switching Syst. Eng. & Technol. Res. Center, Zhengzhou, China
fYear :
2010
fDate :
22-23 May 2010
Firstpage :
1
Lastpage :
4
Abstract :
Time-domain endpoint detection algorithm is simple and can be implemented easily, but has a poor performance under the background of noise environment. Frequency-domain endpoint detection algorithm is robust in noise proof, but has a higher computational complexity and longer running period. According to the features of above algorithms, This paper proposed an endpoint detection algorithm which combines time domain with frequency domain, which takes a real-time multi-stage detection on speech from telephone channel using the frequency spectrum entropy and the Short-term Energy Zero Value as the decision-making parameters. The simulation results show that this method could detect the starting and ending points of speech signal exactly, and meets the need of real-time processing.
Keywords :
computational complexity; decision making; frequency-domain analysis; real-time systems; speech processing; time-frequency analysis; Frequency-domain endpoint detection algorithm; computational complexity; decision making parameters; frequency spectrum entropy; real time-domain endpoint detection algorithm; real-time multistage detection; short term energy zero value; telephone channel; time frequency domain; Background noise; Computational complexity; Detection algorithms; Frequency domain analysis; Noise robustness; Speech; Telephony; Time domain analysis; Time frequency analysis; Working environment noise;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Intelligent Systems and Applications (ISA), 2010 2nd International Workshop on
Conference_Location :
Wuhan
Print_ISBN :
978-1-4244-5872-1
Electronic_ISBN :
978-1-4244-5874-5
Type :
conf
DOI :
10.1109/IWISA.2010.5473576
Filename :
5473576
Link To Document :
بازگشت