SIMULATION AND PERFORMANCE ANALYSIS OF ADAPTIVE ALGORITHMS FOR SPEECH IMPROVEMENT
Abstract
The interest in enhancing speech arises from the use of applications for automated speech processing such as Speech Enhancement, Speech Recognition, Mobile Telephony, Human-Machine Interaction and Wireless hearing aid in our everyday lives and Messaging devices. Adaptive filters have become an active research area in the field of Speech Communications. In the Previous, many algorithms have been proposed that can be used for adaptive filtering; Least Mean Square (LMS), Normalized LMS (NLMS), Affine Projection (AP) and Fast Euclidean Direction Search (FEDS) algorithms have been used in a variety of Signal Processing applications because of their simplicity in computation and implementation. By using these algorithms to get better Signal to Noise Ratio (SNR), Mean Square Error (MSE), Root MSE (RMSE) and Distortion values are difficult. In this paper we propose two Hybrid algorithms Normalized Hybrid Projection (NHP) and Fast Hybrid Euclidean Direction Search (FHEDS) Algorithms in noise cancellation for Speech Enhancement. The simulation results show that the SNR, MSE, RMSE and Distortion values of these algorithms are comparable with the proposed design algorithms. Each noise such as Babble, Factory, Destroy Engine, Car, Fire Engine and Train Noises has been modeled and its values have been mention in the Tabular forms. In these calculations, we get better results from proposed algorithms.