A comparison of two acoustic measurement approaches to the rhythm continuum of natural Chinese and English speech

The recent interest in investigating the prosodic features of speech rhythm (speech timing patterns) using acoustic measurements instead of just perception has brought about several different metrics. Two of the main approaches are: (1) to compare vocalic segment (Vs) percentage against the standard deviation of consonantal segments (Cs) or (2) to compare the pairwise variability index (PVI) of normalized Vs against the raw PVI of Cs. This study compares and contrasts these two metrics using a large corpus of multispeaker data of more naturally occurring speech from American English and Mandarin Chinese broadcast news.