Learn More
This paper presents a superpositional model of prosody where linguistic structures are directly encoded into the prosodic parameters via global melodic and rhythmic contours. This model is applied here to the generation of prosody specific to six common attitudes in French. Perceptual tests using high-quality TD-PSOLA re-synthesis show that predicted(More)
The majority of research in the analysis and generation of prosody for use in speech synthesis systems has focused on prosodic features that ease the syntactic parsing of an utterance or highlight certain parts of it. It is well-known that prosody-and especially final lengthening in French – may reflect very fine and complex attributes of the syntactic(More)
We present here a trainable generative model of French prosody. We focus on the sentence level and design SNNs able to generate both rhythmic and intonation contours for diverse attitudes. First results of a perceptual test show that listeners are able to retrieve the right definition of attitudes by listening to synthetic PSOLA stimuli. 1. THEORETICAL(More)
This paper presents a methodology for evaluating the performance of synthetic prosody. The methodology focusses on the functional equivalence of natural and synthetic prosodic contours in terms of both distinctiveness and coherence. A perceptual experiment was designed to test the performance in terms of distinctiveness for a simple phrasing task. Results(More)
A data-driven method based on a new paradigm is introduced in this paper. We assume that cognitive representations of the discourse are prosodically encoded by means of global multiparametric prototypes. The generation of adequate prosodic contours is then obtained by retrieving and combining these elementary prototypic contours accessed by linguistic or(More)
A new paradigm for modelling prosody is introduced. We assume that global melodic prototypes are built and stored in a "prosodic lexicon". The actual generation of adequate prosodic contours is achieved by retrieving and combining these elementary global contours accessed by linguistic keys. Two automatic F0 generation procedures have been used: The first(More)
The topic of this paper concerns automatic training of prosody for a text-to-speech synthesis system. We propose a method that permits the system to be specialised for a concrete application by training it on a representative corpus of data for that application. Design, training and evaluation of the system is carried out for an application concerning the(More)
  • 1