References
Rossing, The Science of Sound, Chapter 15
"The Throat Singers of Tuva," by Theodore C. Levin and Michael E. Edgerton, Scientific American magazine.

Apparatus
Microphone w/amplifier circuit
Fast Fourier Transform spectrum analyzer - Stanford Research SR 760
Digital Oscilloscope - Tektronix TDS 210

Introduction
The basic elements of speech production are: (1) the vocal cords, which normally produce triangular puffs of air with a well-defined pitch, and (2) the vocal tract, which can be modified to produce variable-frequency resonances called formants. The formants filter the sound produced by the vocal cords to produce sounds we have been trained to recognize and understand. Important speech sounds include voiced sounds (vowels and certain consonants) and unvoiced sounds (other consonants), depending on whether the vocal cords are vibrating or not, respectively.

Vocal cords are contained in the larynx, or voicebox. The pharynx connects the larynx to the oral cavity, and contains valves at either end, the epiglottis and the soft palate. Both valves can move to adjust the pharynx length slightly and change the location of the vocal formants. The oral cavity is adjustable through action of the soft palate, tongue and lips. The soft palate serves as a valve between the oral cavity and the nasal cavity; the latter cannot be adjusted. Manipulation of the elements of the vocal tract causes formant frequencies to change, thereby altering the way in which sound waves from the vocal cords are filtered.

Procedure

Turn on the FFT analyzer, oscilloscope and microphone and notice how the waveform and spectrum of your voice appears when you make a steady sound, such as "oo." Adjust the frequency span fo the FFT analyzer to 3.125 kHz. Try raising and lowering the pitch of your voice (this often happens when you talk), and notice the effect on the spectrum and waveform. Do this by using the pause button to freeze the spectrum after a steady sound is recorded. Press the marker key and turn the knob to measure the frequency of several sharp peaks, which are vocal cord resonances. Record the frequency of the lowest four or five in your notebook.

Question 1. Are the vocal cord resonant frequencies harmonic?

Use the FFT analyzer in signal average mode to record the spectra of each lab partner speaking the ten vowel sounds in Table 15.3 of Rossing (a photocopy is attached). Make hard copies of the vowels /I/ (as in "bit") and /u/ (as in "food") for each partner, while recording the other formant frequencies in your lab notebook without plotting the spectra.

You'll need to produce steady sounds for a few seconds if you want to average 100 or more spectra to obtain clean signals. Practice a few times to be sure that you can observe the formants. You will also be able to see, especially in the region of the first formant, individual peaks that are harmonically related. Determine the fundamental (pitch) for each vowel.

Clues regarding the spectra:
1. If a large peak appears at about 60 Hz (line noise), set the start frequency to 90 Hz or so.

2. Try repeating your first run a few times to check the reproducibility of the first three formant frequencies.

3. Determine what your frequency "span" should be based upon the expected formant frequencies in Table 15.3.

C. Whistle a comfortable note and take a spectrum of it. Sing the same pitch and compare the frequency of that pitch to the whistle spectrum. Lastly, determine the highest and lowest frequencies you can produce while whistling.

Analysis
Create your own table with the formant frequencies and amplitudes, similar to Rossing's. Comment on how your results compare for each of the first three formants.

Question 2. What is the fundamental frequency for your vowel sounds? What does it relate to physically?

Question 3. Explain the whistle spectrum and its perceived pitch.