LINGUIST List 20.2617
Mon Jul 27 2009
Diss: Phonetics/Phonology: Oliveira: 'From Grapheme to Gesture:...'
Editor for this issue: Di Wdzenczny
<dilinguistlist.org>
1. Catarina
Oliveira,
From Grapheme to Gesture: Linguistic contributions for an articulatory based text-to-speech system
Message 1: From Grapheme to Gesture: Linguistic contributions for an articulatory based text-to-speech system
Date: 24-Jul-2009
From: Catarina Oliveira <coliveiraua.pt>
Subject: From Grapheme to Gesture: Linguistic contributions for an articulatory based text-to-speech system
E-mail this message to a friend
Institution: Aveiro University
Program: Linguistics
Dissertation Status: Completed
Degree Date: 2009
Author: Catarina Monteiro Oliveira
Dissertation Title: From Grapheme to Gesture: Linguistic contributions for an articulatory based text-to-speech system
Linguistic Field(s):
Phonetics
Phonology
Subject Language(s): Portuguese (por)
Dissertation Director:
João Torrão
António Teixeira
Dissertation Abstract:
Motivated by the central purpose of contributing for the construction, inthe long term, of a complete text-to-speech system based in articulatorysynthesis, we develop a linguistic model for European Portuguese (EP),based on TADA system (TAsk Dynamic Application), that aimed at theautomatic attainment of the articulator's trajectory from the input text.The specification of this purpose determined the development of a set oftasks, namely the 1) implementation and evaluation of two automaticsyllabification systems and two grapheme-to-phoneme (G2P) conversionsystems, in view of the transformation of the input in an appropriateformat to the TADA; 2) the creation of a gestural database for the EPsounds, in so that each phone obtained at the output of the g2p systemcould have correspondence with a set of articulatory gestures adapted forEP; 3) the dynamic analysis of nasality, on the basis of an articulatoryand perceptive study.
The two automatic syllabification algorithms implemented and tested makeappeal to phonological knowledge on the structure of the syllable, beingthe first one based in finite state transducers and the second one afaithful implementation of Mateus & d'Andrade (2000) proposals. Theperformance of these algorithms - especially the second - was similarto the one of other systems with the same potentialities.
Regarding grapheme-to-phone conversion, we follow a methodology based onmanual rules combined with an automatic learning technique. The evaluationresults of this system motivated the exploitation of others automaticapproaches, finding also to evaluate the impact of the syllabicinformation integration in the systems.
The gestural description of the European Portuguese sounds, anchored on thetheoretical and methodological tenets of the Articulatory Phonology, wasbased essentially on the analysis of magnetic resonance data (MRI), fromwhich all the measurements were carried out, aiming to obtain thequantitative articulatory parameters.The several gestural configurationsproposed have been validated, through a small perceptual test, whichallowed identifying the main underlying problems of the gestural proposal. This work provided, for the first time to PE, the development of a firstarticulatory based text-to-speech system.
The dynamic description of nasal vowels relied either on the magneticresonance data, for characterization of the oral gestures, either on thedata obtained through electromagnetic articulography (EMA), for the study of the velum dynamic and of its relation with the remaining articulators.Besides that,a perceptive test was performed, using TADA and SAPWindows, toevaluate the sensibility of the Portuguese listeners to the variations inthe height of velum and alterations in the intergestural coordination.Thisstudy supported an abstract interpretation (in gestural terms) of the EPnasal vowels and allowed also to clarify crucial aspects related with itsproduction and perception.
|