Most communication environments suffer from noise. The noise decreases the
recognition of the speech and annoys the listeners. It is too important to remove
the noise to enhance the quality of the communication system. The noise may
result from the background of the recoding environment or from the devices used
in recoding like the microphones and the speakers. There are a lot of research
works in the area of speech enhancement to remove or at least reduce the noise.
The thesis discusses the previous work done in this area including the basic
spectral subtraction, Wiener filter and wavelet denoising techniques and
proposes implementation of these techniques in real time. Ten Simulink models
are built using MATLAB Simulink blocks. The models are for normal spectral
subtraction, multi band spectral subtraction, spectral subtraction with noise
update, multi and spectral subtraction with noise update, normal Wiener filter,
multi band Wiener filter, improved Wiener filter, multi band improved Wiener
filter, wavelet Denoising, combined Wiener with wavelet. The Real Time
simulation is done using Texas Instrument TMS320C6711 through its code
composer. The implemented real time algorithms are tested using two noisy
speech corpuses with different signal to noise ratios.