Implementation of Adaptive Filtering Algorithm for Speech Signal on FPGA

Abstract: Th is project gives the study of the principles of Adaptive Noise Cancellation (ANC) and its Applications. Adaptive noise Cancellat ion is an alternative technique of estimat ing signals corrupted by additive noise or interference. In signal processing methods of removing noise, levels of noise rejection are not attainable without prior knowledge about speech signal and noise. But in this method of noise cancellation with no a priori knowledge of signal or noise, noise rejection can be achieved satisfactorily. FPGA implementation of adaptive filtering algorith m is studied in this project wo rk. Two adaptive Filtering Algorithms are imp lemented LMS and wiener. LMS filter is designed in VHDL. Here wiener filter is implemented in adaptive manner to accommodate the varying nature of speech signal. The adaptive wiener filter is implemented in time domain rather than in frequency domain. This adaptive wiener filer is uses two method for speech enhancement TSNR and HRNR. The basic principle of wiener filter is to obtain the estimate of speech signal corrupted by noise. The noise reduction process applies spectral gain to short time spectrum value of noisy speech signal. This gain is expressed as function of priori SNR which is estimated using decision - directed approach. TSNR is used to eliminate the drawback of decision directed approach and retains its advantage. But in noise reduction process some harmonics which a part of o rig inal speech signals are suppressed. For that HRNR method is used to recover these harmonics. The resulting artificial signal is produced in order to refine the a priori SNR used to compute a spectral gain able to preserve the speech harmonics.

Keywords: Adaptive noise cancellation, LMS Filter Weiner filter, adaptive wiener filter, TSNR, HRNR