Characterizing temporal organization in presentation-like Japanese speech using the normalized pairwise variability index

Abstract

Objective acoustic measures for presentation-like speech remain limited. We tested whether the normalized pairwise variability index (nPVI), calculated from vowel-onset intervals, distinguishes Japanese speaking contexts. Recordings of presentation-like, dialogue, and rereading speech in the Corpus of Spontaneous Japanese were analyzed. Speech rate was higher in presentation-like speech than in rereading speech, whereas nPVI was higher in dialogue than in the other contexts. Speech rate and nPVI were not clearly correlated. Thus, vowel-onset interval-based nPVI captures local temporal variation beyond overall speech rate and may provide an objective index for describing local temporal organization across speaking contexts, including presentation-like speech.