Exercise 3.1. Complete table 3.2.

Exercise 3.2. If we are interested in the perception of intonation, we might low-pass filter a speech signal at, say, 200 Hz, in order to create a somewhat muffled version of a recording, which preserves the pitch of the original. How many samples would a moving average filter need to include in order to attain this cut-off frequency?

Exercise 3.3. Try it. Adapt the program meansof4.c into a new program meansof80.c. Hints: Instead of the equation:

    y[i] = b[0]*x[i]+b[1]*x[i-1]+b[2]*x[i-2]+b[3]*x[i-3];

don't bother with the array b, and use a loop to add up the x[i-n]'s:

    for (n=0;n<80;n++) y[i] += x[i-n]/80;

Don't forget to initialize y[i] to 0 before the loop. (If you get completely stuck, sneak a peek at meansof40.c.)

Exercise 3.4. Suppose we have digitized a cassette tape recording at 16000 samples/s. Unfortunately, there is an annoying high-pitched "tape hiss" in the recording, as in the file hissy.dat. However, we decide that we can live with a recording that only goes up to 5500 Hz, because the most important features of speech are below that frequency. What size of moving-average window is needed?

Exercise 3.5. Compile the simple version of the Klatt synthesizer, simple_klatt.c, as follows:

gcc -c simple_parwave.c
gcc -c simple_klatt.c
gcc -o sklatt.exe simple_klatt.o simple_parwave.o

Test it out by synthesizing riy.par using the command sklatt -i riy.par -o riy.dat -r 2. You can then play and listen to riy.dat by importing it into e.g. Cool Edit 96.

Exercise 3.7. The data for figure 3.9 was generated using the command sklatt -i riy.par -o figure3.9.txt -n 0 -q. The figure was plotted by importing figure3.9.txt into a graph plotting program. You can generate a glottal signal file using the command sklatt -i riy.par -o riy_voice.dat -n 0 -r 2, which you can then play and listen to! It sounds just like a buzz, with the right duration and pitch for [ri:], but no articulation.

3.11. Further directions

Klatt, D. H. and L.C. Klatt (1990) Analysis, synthesis, and perception of voice quality variations among female and male talkers. Journal of the Acoustical Society of America 87 (2), 820-857.

IPOX: a rule compiler for the Klatt synthesizer.

3.12. Reading in preparation for the next lecture.

Johnson, K. (1997) Acoustic and Auditory Phonetics. Blackwell. Pages 33-47.

Either: Wakita, H. (1976) Instrumentation for the study of speech acoustics. In N. J. Lass, ed. Contemporary Issues in Experimental Phonetics. New York: Academic Press. 3-40; especially pages 3-15 and 29-33.

Or: Wakita, H. (1996) Instrumentation for the study of speech acoustics. In N. J. Lass, ed. Principles of Experimental Phonetics. St. Louis, Missouri: Mosby. 469-494; especially pages 469-478.

(These two chapters are substantially the same, so it is not necessary to read both.)