Page 109 - Master Handbook of Acoustics

distributed spectrum of noise due to air turbulence resulting from constrictions in the vocal tract. (B)
The time-varying filter action of the vocal tract. (C) The output sound resulting from the filter action
of the distributed sound of (A).

Frequency Response of Speech

The voiced sounds originating in vocal cord vibrations, unvoiced sounds originating in turbulences,
and plosives that originate near the lips, together form our speech sounds. As we speak, the formant
resonances shift in frequency as the lips, jaw, tongue, and velum change position to generate the
desired words. The result is the complexity of human speech evident in the spectrograph of Fig. 5-6.
Information communicated via speech is a pattern of frequency and intensity that shifts rapidly with
time. Notice in Fig. 5-6 that there is little speech energy above 4 kHz. Although it is not shown by the
spectrograph, there is also relatively little speech energy below 100 Hz. It is understandable why

presence filters peak in the 2- to 3-kHz region; that is where human speech sounds resonate.

FIGURE 5-6 Sound spectrograph of a sentence spoken by a male voice. (AT&T Bell Laboratories.)

Directionality of Speech

Speech sounds do not have the same volume level in all directions. This is due primarily to the
directionality of the mouth and the sound shadow cast by the head and torso. Two measurements of

speech-sound directionality are shown in Fig. 5-7. Because speech sounds are variable and complex,
averaging is necessary to give an accurate measure of directional effects.