Title :
Voice activity detection over multiresolution subspaces
Author :
Erdol, Nurgun ; Schultz, Robert
Author_Institution :
Dept. of Electr. Eng., Florida Atlantic Univ., Boca Raton, FL, USA
Abstract :
In this paper, voice activity detection (VAD) is posed as a binary detection problem of an unknown speech signal in both stationary (vehicular) and nonstationary (babble) noise environments. Optimal detection methods are applied on the wavelet transform coefficients of a signal segment to determine the presence of speech. Theoretical analysis is done to justify the effectiveness of multiresolution decomposition on the computation of the noise eigenvalues and vectors and sufficient statistics. VAD results are compared to optimal detection without wavelet transformation and to an energy based method which is used as control. The results show the superiority of the proposed method as increased accuracy in detection
Keywords :
acoustic noise; eigenvalues and eigenfunctions; optimisation; signal detection; signal resolution; speech processing; wavelet transforms; babble noise; binary detection problem; detection accuracy; energy based method; multiresolution decomposition; multiresolution subspaces; noise eigenvalues; noise vectors; nonstationary noise environment; optimal detection methods; signal segment; speech signal; stationary noise environment; sufficient statistics; vehicular noise; voice activity detection; wavelet transform coefficients; Covariance matrix; Discrete wavelet transforms; Eigenvalues and eigenfunctions; Energy resolution; Karhunen-Loeve transforms; Noise level; Phase noise; Signal resolution; Speech enhancement; Working environment noise;
Conference_Titel :
Sensor Array and Multichannel Signal Processing Workshop. 2000. Proceedings of the 2000 IEEE
Conference_Location :
Cambridge, MA
Print_ISBN :
0-7803-6339-6
DOI :
10.1109/SAM.2000.878001
