Questions

1. Record speech sampled at . Generate 8 bit, 4 bit, 2 bit, and 1 bit quantized speech. Compare the original and quantized speech both graphically and audibly.

2. Record a speech sample with . Design filters using fir1 that cut off at , , and to simulate the signals transmitted by broadcast FM radio, broadcast AM radio, and the telephone network. Play back the filtered signals. Repeat the above for a music signal. Use your judgment in deciding what the filter orders should be. Remember that the higher the order, the more time you have to spend waiting for the computer to determine the filter response; however, the higher the order, the closer the filter frequency response approaches that of the ideal filter.

3. Record speech sampled at . Keep every fourth sample, using Matlab constructs similar to the ones you used in half from lab01. [This process is called downsampling; downsampling followed by filtering is called decimation.] Play back the result. Then design a lowpass filter with a cut off frequency at 5500Hz, and use it to filter the downsampled signal. Play back the result.

4. You are given two signals and . You are also told that, with as input, is the output of a lowpass filter with cutoff frequency . How would you obtain , the output of a high pass filter with cutoff frequency and input ?

5. How would you create a bandpass filter from lowpass and highpass filters? Draw a block diagram to illustrate your idea.

6. Record a speech sample with . Using filters, divide the speech sample into a low frequency band from 0Hz to 1kHz, a mid frequency band from 1kHz to 4kHz, and a high frequency band from 4kHz to 11kHz. Listen to the filter outputs. Can you identify the original speech?

7. Generate a musical note with the electronic keyboard, and sample it at . Cut off the beginning and ending transients. Then analyze its frequency content by using fft. Duplicate that note using the sum of appropriately scaled and shifted sinusoids.