Estimation of vocal tract area functions in children based on measurement of lip termination area and inverse acoustic mapping

Although vocal tract area functions for adult talkers can be acquired with medical imaging techniques such as Magnetic Resonance Imaging (MRI), similar information concerning children's vocal tracts during speech production is difficult to obtain. This is largely because the demanding nature of the data collection tasks is not suitable for children. The purpose of this study was to determine the feasibility of mapping formant frequencies measured from the [i, ae, a, u] vowels produced by three children (age range 4 to 6 years), to estimated vocal tract area functions. Formants were measured with a pitch-synchronous LPC approach, and the inverse mapping was based on calculations of acoustic sensitivity functions [Story, J. Acoust. Soc., Am., 119, 715-718]. In addition, the mapping was constrained by measuring the lip termination area from digital video frames collected simultaneously with the audio sample. Experimental results were augmented with speech simulations to provide some validation of the technique.