Effect of temporal modulation rate on the intelligibility of phase-based speech

Abstract

This study investigated the effect of temporal modulation rate on the intelligibility of speech synthesized with primarily phase information using two methods: Phase-based vocoded speech (preserving phase cues and discarding envelope cues) and Hilbert fine-structure stimuli (summing up the multi-channel Hilbert fine-structure waveforms). Listening experiments with normal-hearing participants showed that the intelligibility of the two types of phase-based speech was significantly improved when synthesized using a high temporal modulation rate (or short frame) compared to that synthesized using the whole speech segment. This intelligibility advantage appears to be attributed to better preservation of the temporal envelope cues in phase-based speech.