Analysis of Tones in Cantonese Speech Based on the Command-Response Model

Abstract

As one of the major Chinese dialects, Cantonese has a tone system consistingof nine lexical tones and three additional changed tones, which is considerablymore complex than that of Mandarin. The most important acoustic feature characterizingthese tones is the contour of the voice fundamental frequency (the F₀ contour).In this article we present an approach to modeling F₀ contours of Cantoneseutterances, based on an extension of the command-response model. Analysis-bysynthesisof F₀ contours of the utterances with a fixed carrier frame, in which a targetsyllable with each tone type is embedded, shows that each tone type can berepresented by a specific pattern (polarity, timing, and amplitude) of tone commands.These patterns are found to be essentially maintained in F₀ contours of theutterances with unconstrained text. With the definition of these tone commandpatterns, the command-response model not only provides a novel phonologicaldescription of tones, but also gives high accuracy of approximations to F₀ contoursof Cantonese utterances and allows one to analyze various tonal phenomenain quantitative terms. Quantitative distinctions between various tones are thenrevealed by statistical analysis of the timing and amplitude of tone commands.Especially, systematic alignment in timing is found between the onsets/offsets oftone commands and the rhyme of a syllable, and hence a set of constraints can beintroduced, which together with those on tone command amplitudes and phrasecommand parameters, is then applied for generating F₀ contours of Cantoneseutterances. The validity of the approach is verified by perceptual evaluation of thesynthetic speech stimuli with model-generated F₀ contours, both on the intelligibilityof tones and on the naturalness of prosody.