EE342 Problem Set 12

DUE F 05/07/1999

1. Download the data file cowboy.dat from my web page. It is the sampled speech (sampling frequency of f_s = 8192Hz) of a man saying "cowboy" with some frequency specific noise included.
2. Plot the speech data versus discrete-time sample number.
3. Compute the DFT of the speech data and plot its magnitude spectrum versus frequency in Hz. Where are the largest frequency components?
4. Assuming the man's voice is the smallest of the substantial frequency components, design a digital filter to filter out the higher frequency noise.
5. Filter the speech data via a recursive difference equation representing your filter.
6. Plot your filtered signal versus discrete-time as well as its magnitude spectrum versus frequency in Hz. Also listen to the filtered speech data if possible. Comment on the effects of your filtering. Did you remove the noise?