Title :
On the use of higher-order statistics for robust endpoint detection of speech
Author :
Rangoussi, Maria ; Delopoulos, Anastasios ; Tsatsanis, Michail
Author_Institution :
Dept. of Comput. Sci., Nat. Tech. Univ. of Athens, Greece
Abstract :
Third order statistics of speech signals are not identically zero, as it would be expected based on the linear model for voice. This is due to quadratic harmonic coupling produced in the vocal tract. Based on this observation, third order cumulants are employed to address the endpoint detection problem in low SNR level recordings due to their immunity to (colored) additive non-skewed noise. The proposed method uses the maximum singular value of an appropriately formed cumulant matrix to distinguish between voiced parts of the speech signal, and silence (noise). Adaptive implementations are also proposed, making this method computationally attractive. Results of batch and adaptive forms are presented for real and simulated data.
Keywords :
eigenvalues and eigenfunctions; matrix algebra; signal detection; speech analysis and processing; speech recognition; statistical analysis; SVD algorithm; adaptive implementations; cumulant matrix; higher-order statistics; low SNR level recordings; maximum singular value; quadratic harmonic coupling; robust endpoint detection; speech recognition; speech signals; third order cumulants; Additive noise; Colored noise; Eigenvalues and eigenfunctions; Electronic mail; Higher order statistics; Robustness; Signal to noise ratio; Speech processing; Speech synthesis; White noise;
Conference_Titel :
Higher-Order Statistics, 1993., IEEE Signal Processing Workshop on
Conference_Location :
South Lake Tahoe, CA, USA
Print_ISBN :
0-7803-1238-4
DOI :
10.1109/HOST.1993.264597
