This project is aimed
towards, improvement of speech signal in
hearing aids. This is fully software based implementation and is tested
on MATLAB. Signal improvement is divided into two sections, the first one is
denoising and the second one is to enhance or suppress particular frequencies
so as to match that with the hearing response in terms of audiogram for that
Audio files which are
without noise are stored in the database. Some random traffic noise is added
into it. Then the signal passes through various wavelet decomposition levels so
as to partition them into the low frequency and high frequency components.
Using the universal thresholding method, high frequency components is
suppressed so as to suppress the noise. Then resultant components are
reconstructed back to get denoised speech signal. Performance of denoising
algorithm is tested by various performance parameters like SNR, MSE,
In the second section,
we are filtering speech signal with standard octave frequencies divided ranging from 250 Hz to
16Khz. As per audiogram obtained, gains or attenuation factor for various
frequency bands is set so as to amplify or suppress particular frequency band.
Then these components are added up to get audiogram adjusted speech.
Sign-in to continue reading Full Article
Pay only for this article and continue reading the full article