No abstract

No abstract

No abstract

Abstract-In our present work, we propose a nature inspired population based speech enhancement technique to find the dynamic threshold value using Teaching-Learning Based Optimization (TLBO) algorithm by using shift invariant property of dual tree complex wavelet transform (DT-CWT). The performance of these proposed methods are evaluated in terms of Perceptual Evaluation of Speech Quality (PESQ) and Peak Signal to Noise Ratio (PSNR). Speech quality of different speech waves are compared for two level wavelet packet decomposition and dual tree wavelet transform using soft threshold. The speech qualities of the waves are better than the other available articles in the literature. Keyword-Speech enhancement, Dual tree complex wavelet transform, Teaching-Learning-based optimization, Perceptual Evaluation of Speech Quality I. INTRODUCTION In many speech related systems, the original speech signal is contaminated with some interference sources. The interference source, i.e the noise signal degrades the quality of the original speech. The speech signal is affected by wide-band noise in the form of white or colored noise and a period noise such as hum noise. The most common type of noise in communication channels is the additive wide band Gaussian noise. Speech enhancement aims at improving the performance of speech communication systems in noisy environments. Speech enhancement may be applied, for example, to a mobile radio communication system, a speech recognition system, a set of low quality recordings, or to improve the performance of aids for the hearing impaired. Several methods [1]- [6] have been proposed in the literature for the enhancement of degraded speech. A majority of these methods can be grouped into spectral processing and temporal processing methods. In the spectral processing methods, degraded speech is processed in the frequency domain mostly using Fourier transform for achieving speech enhancement. In the temporal processing methods, the processing is done in the time domain. Most speech enhancement methods improve the quality of the signal but degrade its intelligibility of the speech. Performance measures like PSNR and PESQ are widely used as the performance of the evolution criterion. For elimination of the Gaussian back ground noise in the communication channels, we have been implementing adaptive thresholding technique using TLBO optimization. The TLBO algorithm [7]-[9] is a global optimization, population based iterative learning mechanism that exhibits some common characteristics with other evolutionary computation (EC) techniques like (GA) [10], Particle Swarm Optimization (PSO) [11], Differential Evolution (DE) [12], and Artificial Bee Colony (ABC) [13]. The TLBO algorithm does not require any algorithm-specific control parameters like mutation and crossover as in genetic algorithm. The TLBO methods provide the learning mechanism in adaptive models. The organization of this paper is as follows. Section II presents speech de-nosing using wavelet thresholding. Section III discusses introduct...