Page 120 - The Master Handbook Of Acoustics

95
SPEECH, MUSIC, AND NOISE

produced by the vibration of the vocal cords, pulses of sound having a
fine spectrum that falls off at about 10 dB per octave as frequency is
increased as shown in Fig. 5-4A. The sounds of the vocal cords pass
through the vocal tract, which acts as a filter varying with time. The
humps of Fig. 5-4B are due to the acoustical resonances, called for-
mants of the vocal pipe, which is open at the mouth end and essen-
tially closed at the vocal cord end. Such an acoustical pipe 6.7 inches
long has resonances at odd quarter wavelengths, and these peaks occur
at approximately 500, 1,500, and 2,500 Hz. The output sound, shaped
by the resonances of the vocal tract, is shown in Fig. 5-4C. This applies
to the voiced sounds of speech.

Formation of Unvoiced Sounds

Unvoiced sounds are shaped in a similar manner as indicated in Fig.
5-5. Unvoiced sounds start with the distributed, almost random-noise-
like spectrum of the turbulent air as fricative sounds are produced.
The distributed spectrum of Fig. 5-5A is generated near the mouth end
of the vocal tract, rather than the vocal cord end; hence, the resonances
of Fig. 5-5B are of a somewhat different shape. Figure 5-5C shows the
sound output shaped by the time-varying filter action of Fig. 5-5B.

Putting It All Together
The voiced sounds, originating in vocal cord vibrations, the unvoiced
sounds, originating in turbulences, and plosives, which originate near
the lips, go together to form all of our speech sounds. As we speak, the

Sound source Vocal tract Sound
vocal cords time-varying filter output
Slope
10 dB/octave
Amplitude Amplitude Amplitude

Frequency Frequency Frequency
A B C
FIGURE 5-4

Sound spectrograms of human sounds other than speech. AT&T Bell Laboratories.