Reducing signal-bias from MAD estimated noise level for DCT speech enhancement

Authors:
Md. Kamrul Hasan;Sayeef Salahuddin;M. Rezwan Khan
Affiliations:
Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh;Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh;Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh
Venue:
Signal Processing
Year:
2004

Citing 8
Cited 1

Digital signal processing (3rd ed.): principles, algorithms, and applications

Digital signal processing (3rd ed.): principles, algorithms, and applications
Comparison of a discrete wavelet transformation and a nonuniform polyphase filterbank applied to spectral-subtraction speech enhancement

Signal Processing - Special issue on acoustic echo and noise control
Noisy speech enhancement using discrete cosine transform

Speech Communication
An energy-constrained signal subspace method for speech enhancement and recognition in white and colored noises

Speech Communication
Speech Enhancement with Reduction of Noise Components in the Wavelet Domain

ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97)-Volume 2 - Volume 2
Speech enhancement based on a priori signal to noise estimation

ICASSP '96 Proceedings of the Acoustics, Speech, and Signal Processing, 1996. on Conference Proceedings., 1996 IEEE International Conference - Volume 02
Fundamentals of Wavelets: Theory, Algorithms, and Applications

Fundamentals of Wavelets: Theory, Algorithms, and Applications
De-noising by soft-thresholding

IEEE Transactions on Information Theory

Tiny surface defect inspection of electronic passive components using discrete cosine transform decomposition and cumulative sum techniques

Image and Vision Computing

Quantified Score

Hi-index	0.08

Visualization

Abstract

This paper presents an improved thresholding technique for speech enhancement in the discrete cosine transform (DCT) domain. The signal-bias-compensated noise level is used as the threshold parameter. The probability of speech presence in the noisy DCT coefficients at the finest level is exploited for compensating the signal-bias. The performance of the proposed algorithm is evaluated on speech corrupted by background Gaussian white noise and real noise recorded inside a moving car. A relative comparison in the enhancement of the degraded speech is performed using the uncompensated and compensated noise levels for the same denoising filter. Both objective and subjective evaluations are presented to show improved performance for the case when the compensated noise level, estimated by the proposed algorithm, is used rather than the uncompensated noise level obtained using the median absolute deviation method. Simulation results are also given to show the superiority of the proposed method over a recently reported enhancement algorithm.