Applied Signal Processing

Projects,  to be handed in before February 15.

2. Speech Recognition

Many of the techniques that we have learnt in these lectures can be applied to the processing of speech. For this project you will make use of the VOICEBOX software, a package for MATLAB

(http://www.ee.ic.ac.uk/hp/staff/dmb/voicebox/voicebox.html).

1. Build a system that records an input speech signal.

2. Analyse this signal, filtering out noise. Try to go as far as you can:

- Recognise different letters and words, whether the speaker is a man or a woman, etc.

- Try to go as far as possible with your analysis.

3. Sample the signal with as few bits as possible, and show how the signal changes as a function of the number of bits used.

4. Write out all the files that you have produced in WAV format.

For a lot of information about audio processing see Chapter 22 of Smith (http://www.dspguide.com/ch22.htm).

Information about data compression can be found at

http://www.data-compression.com/speech.shtml