Title :
Speech enhancement based on hearing masking properties and subspace
Author :
Qi, Ding ; Wang, Xu ; Jinfu, Xu ; Bingxi, Wang
Author_Institution :
Inf. & Eng. Univ., Henan, China
fDate :
31 Aug.-4 Sept. 2004
Abstract :
The DFT is used to asymptotically approach the KLT in signal subspace based speech enhancement so that the computation complexity is reduced from O(N3) to O(NlogN). The eigenvalues are perceptually filtered based on PEAQ masking model to suppress the music noise. The results tested in different background noise demonstrate this method can reduce the distortion to the speech signal effectively.
Keywords :
computational complexity; discrete Fourier transforms; eigenvalues and eigenfunctions; filtering theory; music; noise; speech enhancement; PEAQ masking model; computational complexity; eigenvalues; hearing masking property; music noise; signal subspace; speech enhancement; speech signal; Acoustic noise; Auditory system; Discrete Fourier transforms; Distortion; Filters; Masking threshold; Noise reduction; Psychoacoustic models; Speech enhancement; Vectors;
Conference_Titel :
Signal Processing, 2004. Proceedings. ICSP '04. 2004 7th International Conference on
Print_ISBN :
0-7803-8406-7
DOI :
10.1109/ICOSP.2004.1452643
