Paper Details: Downloads: 581
Serial Number: P1180847507
Title: Modified Time Adaptive Wavelet Based Approach for Enhancing Speech from Adverse Noisy Environments
Authors: M. G. Sumithra and K. Thanuskodi and M .R.Anitha
Abstract: This proposed work considers the musical noise encountered in most frequency domain speech enhancement algorithms is partially due to the large variance estimates of the spectra. To address this issue, a new wavelet based approach is proposed. Enhancement of speech from adverse noisy environment is accomplished through the use of soft thresholding on time adaptive discrete wavelet transform co-efficients followed by post filtering. This approach is incorporated with time domain and frequency domain analysis. Results are measured objectively by Signal to Noise Ratio (SNR), Itakura-Saito (IS) distance, Minimum mean square error (MMSE) and subjectively by Mean Opinion Score (MOS). Objective measures and listening test demonstrated significant improvements over other methods when tested with TIMIT sentences corrupted by various types of colored noise.
Keywords: Discrete wavelet transform, Time adaptation factor, Soft threshold, Band pass filter, Speech enhancemen, Daubechies wavelets.
Journal/Conference: International Journal of Digital Signal Processing
Volume: 9
Issue: 1
Submission Date: 11/21/2008 12:00:00 AM
Review Date: 2/11/2009 12:00:00 AM
Publishing Date: 3/18/2009 12:00:00 AM
Article Downloads: 581
Download:

Facebook