New Inverse-Filtering Technique for Deriving the Glottal Air Flow Waveform during Voicing
Martin Rothenberg · The Journal of the Acoustical Society of America · 1970
It is known that for many conditions of voicing and vocal-tract configurations, the volume velocity waveform at the glottis can be derived by a procedure known as inverse filtering, in which the radiated sound pressure waveform, as picked up by a microphone, is passed through a filter having a transfer characteristic approximately the inverse of the transfer characteristic of the vocal tract at that instant. It is shown that an inverse filter, which uses the volume velocity waveform at the mouth, instead of the radiated pressure waveform, will provide a better replica of the glottal waveform in at least three respects: it provides a reliable zero reference level, much less low-frequency noise, and an easy and accurate amplitude calibration. The required volume velocity waveform at the mouth has been obtained by means of a wire-sceen pneumotachograph based on a specially constructed “circumferentially exhausted” pneumotachograph mask. The mask was designed for a minimal distortion of voiced speech, and when in place will shift formant frequencies less than 10%, as averaged over the different English vowel phonemes. The time resolution of the complete inverse-filter system is presently limited by the pneumotachograph to about 12 msec.