Sébastien Le Maguer - Professional webpage

Table of Contents

1 Global information   html

After achieving a Master of Science in Complex System and Algorithms at Université de Lille 1 (Lille, France) in 2008 and a PhD in Text-To-Speech synthesis (TTS), I’m currently a Post-Doc researcher at the Saarland University in the Ingmar Steiner’s MMCI Group. My work focuses on TTS (annotation, corpus building, descriptive features, statistical speech synthesis and corpus analysis) and information retrieval.

Keywords : TTS, HTS, Descriptive features, Corpus annotation, Evaluation

1.1 Maintainer

  • Text-to-Speech synthesis system: Marytts
  • Linguistic annotation data: Roots

1.2 Membership

2 Research

2.1 Research experience   latex

Year Description team/group Place Supervisors
10/14 Post-Doc MSP group - Saarland University Saarbrücken - Germany Ingmar Steiner
02/14 - 09/14 Post-Doc LINKMEDIA - INRIA Rennes - France Vincent Claveau
09/13 - 12/13 Post-Doc CORDIAL - IRISA Lannion - France Damien Lolive / Nelly Barbot
09/11 - 08/13 Research Assistant CORDIAL - IRISA Lannion - France Olivier Boëffard / Nelly Barbot
10/08 - 07/13 PhD CORDIAL - IRISA Lannion - France Olivier Boëffard / Nelly Barbot

2.2 PhD

  • Title : Experimental evaluation of statistical speech synthesis system, HTS, for French
  • Supervisors : Olivier Boëffard, Nelly Barbot
  • Defended the 2nd of July 2013
  • Prix de l’innovation du Trégor

The work presented in this thesis is about TTS speech synthesis and, more particularly, about statistical speech synthesis for French. We present an analysis on the impact of the linguistic contextual factors on the synthesis achieved by the HTS statistical speech synthesis system. To conduct the experiments, two objective evaluation protocols are proposed. The first one uses Gaussian mixture models (GMM) to represent the acoustical space produced by HTS according to a contextual feature set. By using a constant reference set of natural speech stimuli, GMM can be compared between themselves and consequently acoustic spaces generated by HTS. The second objective evaluation that we propose is based on pairwise distances between natural speech and synthetic speech generated by HTS. Results obtained by both protocols, and confirmed by subjective evaluations, show that using a large set of contextual factors does not necessarily improve the modeling and could be counter-productive on the speech quality.

Keywords : Computer science, Speech processing, Text-to-Speech synthesis, HTS

PhD document (in french)

3 Teaching

3.1 Current (2016-2017)

Title Description Attachments
FLST Presentation in front of coli students about the research group slides
Statistical speech synthesis   slides
Reproducible research with org-mode   slides

3.2 Previous

3.2.1 2015-2016

Title Description Attachments
Statistical speech synthesis   slides

3.2.2 2014-2015

Title Description Attachments
Statistical speech synthesis   slides

3.2.3 2012-2013

Title Description Attachments
Unix Programmation   (to appear)
Unix Utilisation   (to appear)
Algorithmique distribuée   (to appear)

3.2.4 2011-2012

Title Description Attachments
Unix Programmation   CM1 CM2 TP1 TP2 TP3 TP3.2 TP4 TP5 TP6 TP7 TP8
Unix Utilisation   CM TP1 TP2
Algorithmique distribuée   CM TP

4 Education

Année Description
2008 - 2013 Doctorate Degree in Computer Science, ENSSAT/Université de Rennes 1 (22, France)
2006 - 2008 Master of science - Complex systems and algorihms, Université de Lille 1 (59, France)
2005 - 2006 Bachelor’s Degree in computer science - A.I and robotic, U.B.O. (29, France)
2003 - 2005 DUT in computer science, IUT de Lannion (22, France)

5 Technical skills

  • Programming :
    • skilled : Perl, C, C++, Java, Python, Php, Shells, Matlab
    • fundamentals : Scheme/(E)Lisp
  • Systems : Linux, Windows
  • Foreign Language : English (Written and Spoken), German (basics), French (Native)

6 Publications

Author: Sébastien Le Maguer

Created: 2018-01-09 Tue 15:42

Emacs 25.3.1 (Org mode 9.1.6)