Using spectral coherence between temporal envelope and mouth movements to reveal rhythmic regularities in speech

This paper features a method to characterize the rhythmicity in speech from the temporal envelope and mouth opening-closing movements using spectral coherence [He (2022) J. Acoust. Soc. Am. 152, 567–579]. Such a coherence reveals the joint regularities in both domains of signals. For illustrations, recordings from both L1 and L2 English utterances are used for comparisons. The recordings utilized for this Prosody Visualization Challenge (PVC-2023) have been chosen from the EMA-MAE corpus [1],1 which is openly accessible. As such, there are no ethical concerns associated with their current scientific and not-for-profit usage.