Abstract not found

This paper presents the main approaches used to synthesize talking faces, and provides greater detail on a handful of these approaches. No system is described exhaustively, however, and, for purposes of conciseness, not all existing systems are reviewed. An attempt is made to distinguish between facial synthesis itself (i.e the manner in which facial movements are rendered on a computer screen), and the way these movements may be controlled and predicted using phonetic input.

Abstract not found

In the research field initiated by Lindblom & Liljencrants in 1972, we illustrate the possibility of giving substance to phonology, predicting the structure of phonological systems with nonphonological principles, be they listener-oriented (perceptual contrast and stability) or speaker-oriented (articulatory contrast and economy). We proposed for vowel systems the Dispersion-Focalisation Theory (Schwartz et al., 1997b). With the DFT, we can predict vowel systems using two competing perceptual constraints weighted with two parameters, respectively λ and α. The first one aims at increasing auditory distances between vowel spectra (dispersion), the second one aims at increasing the perceptual salience of each spectrum through formant proximities (focalisation). We also introduced new variants based on research in physics- namely, phase space (λ,α) and polymorphism of a given phase, or superstructures in phonological organisations (Vallée et al., 1999) which allow us to generate 85.6 % of 342 UPSID systems from 3- to 7-vowel qualities. No similar theory for consonants seems to exist yet. Therefore we present in detail a typology of consonants, and then suggest ways to explain plosive vs. fricative and voiceless vs. voiced consonants predominances by i)

The Speech Mapping concept posits that speech sequences can be represented by trajectories in a multiparametric space, whose elements are related to each other by relationships representing speech production mechanisms. This concept is viewed as a useful framework for pronunciation training, in a scheme where the teacher uses a Virtual Talking Head, to manipulate audio-visual speech stimuli in order to fulfil a double task: (1) evaluating and improving the learner's perception of the target language sounds, and (2) helping the learner produce the corresponding articulations by acquiring the internalisation of the relations between articulatory gestures and resulting sounds. We describe a set of data and models, including a virtual talking head, that can be useful for pronunciation training, present a few experiments supporting this approach, and suggest some directions for the future. 1. Introduction A second language learner can be considered phonologically deaf, i.e. he/she may not...

This paper investigates the organization of the vowel space in French speakers. Speakers from 4 years of age to adulthood were recorded in order to generate significant betweenspeaker variability. Each speaker produced repetitions of the ten French oral vowels /i y u e ø o � œ � a/. Acoustic analyses show that despite considerable between-speaker variability in the relative positions of the vowels within the vowel space, speakers tend to

Recovering the motions of speech articulators from the acoustic speech signal has a long history, starting from the observation that a simple concatenated tube model is a reasonable model for the origin of formant resonances. In this work, we take a different approach making minimal assumptions about the interdependence of acoustics and articulators by estimating the full joint distribution of the two spaces based on a corpus of paired data, derived from an articulatory synthesizer. This approach allows us to estimate posterior distributions of articulator state as well as finding the maximum-likelihood trajectories. We present examples comparing this approach to a related, earlier approach that did not incorporate prior distributions over articulator space, and demonstrate the advantages of learning the models from realistic utterances. We also indicate benefits available from jointly estimating particular pairs of articulators that have high mutual dependence. Index Terms: articulatory inversion, speech acoustics 1.

Positional targets for lingual consonants defined using electromagnetic articulography