Computer Recognition of Interphonemic Transitions in Continuous Speech.

摘要

The objective of this research was to determine the feasibility of recognizing five distinct speech units (interphonemic transitions) in the speech of one speaker. A computer program performed the recognition task based on formant features in the transition regions of the speech data. A comparison of two sequential recognition models revealed that transitions in continuous speech can be more accurately modelled in terms of independent formant sequences than a simple sequence of sounds.