Page 19 - Fall 2011

vibration of the vocal folds serves as the sound source. The supra- glottal cavity, which is the part of the vocal tract that extends above the vocal folds, is the filter which shapes the sound generat- ed at the vocal folds. Differences in the size and shape of the vocal folds contribute to inter-speaker dif- ferences in pitch and voice quality. Larger vocal folds with more mass will vibrate more slowly, producing a voice that is lower in pitch than that of a talker with smaller vocal folds. The oscilla- tion of the vocal folds provides the fundamental fre- quency, which lis- teners perceive as the pitch of the voice, and har- monics, which occur at multiples of the fundamental frequency. “Voice quality” refers to the variations in the sound of a talker’s voice, ranging, for example, from breathy to modal to creaky. It is determined largely by the relative speed and dura- tion of vocal fold closure in the course of vibration, and by whether full vocal-fold closure is achieved. (The vibration of the vocal folds is a complex process and we are glossing over many details here.) The size and morphology of the supra- glottal cavity determines the resonant properties of the filter. The larger this oral cavity, the lower the resonant frequencies produced by the vocal tract. Manipulating the configuration of the vocal tract modifies the resonant frequencies it pro- duces. These resonant frequencies provide valuable informa- tion about a speak- er’s vowels within the acous-tic-pho- netic space of human vocaliza- tions. Power spectra made from the first third of the final vowel in tomato from a male (top) and female (bottom) speaker are shown in Fig. 2. There are several differences between the female and male productions. Let’s first turn our attention to the lowest frequency component of the spectra. This is the fundamental fre- quency. The fun- damental frequen- cy is lower in the male’s voice than in that of the female. The com- ponents going up in frequency are the harmonics. The fundamental and the harmonics are produced by the vibrating vocal folds. Note also that the high amplitude reso- nance peaks in the Fig. 2. Power spectra from the first third of the final vowel in the word tomato from a male (top) and a female (bottom) speaker. female-produced spectrum are generally higher in frequen- cy than the high amplitude resonance peaks in the male- produced spectrum. One intuitively expects to find these predictable differences in speech produced by males and females, due to the relative differences of their vocal tract sizes. Differences in filter shape and source vibrations, how- ever, would arise even when comparing spectra for two female speakers, or two productions from the same speaker. The number of potential acoustic parameters is limitless, and speakers do not have complete control over all of these parameters. Speech, simply, is immensely variable. 18 Acoustics Today, October 2011