| A pitch tracker for identifying voiced consonants (1999) | |||||||||||||||
Abstract | |||||||||||||||
| We report on one aspect of our research into text-free, speech recognition-free lip synchronization (lip-sync). By lip-sync we mean inputting freely spoken speech and outputting parameters for animating a mouth, which can then be synchronized with the input speech. We first describe a “pitch tracker ” that we find helps to expedite the identification of voiced consonants. Our approach to lip-sync is to subdivide the speech signal into specific classes of sounds, and then apply specialized lip-synch algorithms to determine the mouth shapes (visemes) corresponding to the individual members. One such class is the voiced fricatives. They are the medial consonants of over, bozo, seizure, and other, designated as /v/, /z/, /zh / and /dh/. This paper describes a method for identifying voiced fricatives in the speech signal when they occur between vowel sounds. The technique is currently being extended to handle fricatives that occur in other contexts. The method is based in part on a sudden shift in the measurement of the fundamental frequency F0 of the speech signal at the start and end of the voiced fricative. F0 is determined by the rate at which the vocal cords vibrate or indirectly by the glottal pulse (GP) — the period of a single vibration. Its actual measurement is affected by rapid phase changes induced by the motion of the articulators — acting, in effect, as a time-changing filter — so that the F0 we measure does not always correspond to the actual rate of vibration of the vocal cords. We have developed an algorithm for measuring F0 based on the fact that the amplitudes of the odd harmonics of a periodic function with period P are zero when the function is expanded in a Fourier series whose coefficients are determined by integrating over 2P instead of the usual P. That is, if f is integrable and periodic with period P, and t0 is an arbitrary point in time, then the following hold. t0+ | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||