Single Channel Speech Separation Based On Modulation Frequency Domain

Author(s):

Farhana Yousaf , Ilahia College Of Engineering And Technology, Ernakulam; Lekshmi M.S, Ilahia College Of Engineering And Technology, Ernakulam

Keywords:

ASA, CASA, Onset-Offset, Short Time Fourier Transform, Wavelet Packet Transform

Abstract:

Computational auditory scene analysis (CASA) is the study of auditory scene analysis (ASA) by computational means .In recent literature CASA focus for speech separation from monaural mixtures. It is based on the cochlear modeling using short time Fourier transforms (STFT).While the segregation efficiency and perceptual quality is not good; hence it is not applicable for an efficient hearing aid. The proposed system decomposes the input signal into segments by using Wavelet Packet Transform (WPT).Then the pitch range in each frame is calculated by Onset and Offset algorithm. Speech separation is performed by a mask extracted from the modulation spectrogram. Systematic evaluations show that the proposed system outperforms the previous system.

Other Details:

Manuscript Id	:	NCTTP004
Publication Date	:	06/05/2017
Page(s)	:	6-10

Download Article