Synthesizing dialogue speech of Japanese based on the quantitative analysis of prosodic features

Abstract

Through the analyses of fundamental frequency contours and speech rates of dialogue speech and also of read speech, prosodic rules were derived for the synthesis of spoken dialogue. As for the fundamental frequency contours, they were rst decomposed into phrase and accent components based on the superpositional model, and then their command magnitudes/amplitudes were analyzed by the method of multiple regression analysis. As for the speech rate, the reduction rate of mora duration from reading-style to dialogue-style was calculated. After normalizing the sentence length, the mean reduction rate was calculated as an average over utterances without complicated syntactic structure.Results of the above analyses were incorporated in the prosodic rules for dialog speech synthesis. Using a formerly developed formant speech synthesizer, synthesis was conducted using both the former rules of read speech and the newly developed rules. A hearing test showed that the new rules can produce better prosody as dialogue speech.

Development and evaluation of a spoken dialogue system for academic document retrieval with a focus on reply generation

“…The prosodic rules used to control prosodic features of the reply speech in both the original and the new systems basically consist of those that determine the positions and magnitudes/amplitudes of the phrase, and accent commands of the F 0 contour generation process model [10]. They are a dialogue speech version [12] of rules originally developed for read speech [13], in which the phrase and accent commands are represented by the phrase and accent symbols, respectively.…”

Corpus-based generation of F<inf>0</inf> contours of Japanese based on the generation process model and its control for prosodic focus

“…From these considerations, we have developed a corpus-based method of synthesizing F 0 contours in the framework of the generation process model (F 0 model) and realized speech synthesis in reading and dialogue styles with various emotions [3,4]. The model represents a sentence F 0 contour as a superposition of accent components on phrase ones; each type of components assumed to be responses to step-wise accent commands and impulse-like phrase commands, respectively [5].…”

Corpus-based generation of prosodic features from text based on generation process model

“…From these considerations, we have developed a corpusbased method of synthesizing F0 contours in the framework of the generation process model (F0 model) and realized speech synthesis in reading and dialogue styles with various emotions [3][4][5]. The model represents a sentence F0 contour as a superposition of accent components on phrase ones; each type of components assumed to be responses to step-wise accent commands and impulse-like phrase commands, respectively [6].…”