Aghilas SINI

Avenue Olivier Messiaen, F-72085,Le Mans - France aghilas.sini@univ-lemans.fr

Since my childhood, language and speech have always been at the center of my concerns. That is why I am a speech researcher and engineer now. My research interests mainly focus on expressive speech production and recognition. I am also interested in pathological voices and second language acquisition. Specifically, I spent three years and a half working on topics in expressive speech generation during my Ph.D. training. Recently, I am working on pronunciation assessment in the case of migrants and children.

If you would like to reach out to me, my email address is aghilas.sini at univ-lemans dot fr.


Experience

R&D Engineer

Learn&Go

  • Children Speech Recognition
  • Pronunciation and prosodic assessment for developing language acquisition systems

February 2022 - present

Research PostDoc

Université de Rennes 1 - IRISA Lab.

Under NADINE project which is an european innovation project

  • Multilingual speech analysis, concerned languages: English, French, Spanish, and Greek
  • Pronunciation and prosodic assessment for developing language acquisition systems

March 2020 - January 2022

Ph.D Candidate

Université de Rennes 1 - IRISA Lab.

SynPaFlex (Flexible speech synthesis for expressivity)

  • Thesis title: Characterisation and generation of expressivity in function of speaking styles for audiobook
  • Thesis Supervisors: Elisabeth Delais-Roussarie and Damien Lolive.

December 2016 - Febrary 2020

Software Engineer

LORIA-INRIA Lab.

IFCASL (Individualized Feedback for Computer-Assisted Spoken Language Learning)

  • Automatic pronunciation diagnosis and error detection for Germans learning french
  • Modification and re-synthesis of learner audio samples using teacher audio samples based on Pitch
  • Synchronous Overlap and Add algorithm
  • Feedback to correct devoicing of final consonants in French spoken by German learners

January 2016 - November 2016

Software Engineer

LORIA-INRIA Lab.

ORTOLANG (Open Resources and TOols for LANGuage)

  • Development of syntactic-semantic analyser tool called J-Safran for spoken documents in French language
  • Contributing to development of JTRANS tool for semi-automatic alignment of speech and textual corpus
  • Contributing to development of JSnoori an Interactive tool for speech signal processing and phonetics

November 2014 - January 2016

Education

Université de Rennes 1

Ph.D
Computer Science - Artificial Intelligence - Language & Speech Technology
December 2016 - October 2020

Université Paul Sabatier - Université de Toulouse III

Master
Computer Science - Artificial Intelligence, Pattern Recognition and Robotic
September 2011 - August 2014

Université Mouloud Mammeri - Université de Tizi Ouzou

Bachelor
Automation and Control Systems
September 2008 - August 2011

Supervise

Teaching Assistant

Temporary Teacher
ENSSAT Lannion, Computer Science Departement
  • Software Engineering, Web Services and Artificial Intelligence
Febrary 2019 - August 2019
Temporary Teacher
ENSSAT Lannion, Computer Science Departement
  • Operating system
  • Software Engineering, Web Services and Artificial Intelligence
  • Automatic language and speech processing
September 2017 - August 2018
Client-side Web programming course
IUT Charlemagne, Université de Lorraine, Nancy, France

Languages taught JavaScript, JQuery, Ajax.

January 2016 - March 2016

Co-supervisor

Master Student - Thibault Gaudier
ENSSAT Lannion, Université de Rennes 1

Deep-learning based voice conversion: application to characters switching for story-telling

Febrary 2021 - Present

Training

Workshop and Summer School attended

Summer School on Big Data & Speech
Station Biologique de Roscoff, CNRS, Sorbonne University
July 2018
Speech Synthesis: Advancement in Modern Speech Synthesis Engines
University of Crete, Heraklion, Crete, Greece
July 2016
Workshop on Feedback in Pronunciation Training
Hofgut Imsbach, Northern Saarland, German
November 2015

Internships

PhD Internship at CSTR
funded by Doctoral School MathStic, under Supervision: Prof. Simon King
September 2018 - December 2018
Master Internship at LORIA-INRIA
Mapping of a sound environment for a mobile robot (Master thesis in French)

Control of a mobile robot movements to localize a sound source as quickly as possible. The belief about the source position is represented by a discrete grid and a dynamic programming algorithm was introduced to find the optimal robot motion minimizing the entropy of the grid .

September 2018 - December 2018

Skills

Programming Languages & Tools
Machine Learning Libraries

Contributions

Corpus
Software

Interests

Apart from being an teacher and researcher, I enjoy my full-time job as a father and husband.

I am a Judo practice. I enjoy watching Japanese animation, and I love to cook. I am good enough at drawing and as a classical guitar player.


Publication

  1. Aghilas Sini (2020), “Characterisation and generation of expressivity in function of speaking styles for audiobook synthesis” PhD dissertation. Universit́e de Rennes 1.

  2. Aghilas Sini, Sébastien Le Maguer, Damien Lolive, Elisabeth Delais-Roussarie. Introducing Prosodic Speaker Identity for a Better Expressive Speech Synthesis Control. 10th International Conference on Speech Prosody 2020, May 2020, Tokyo, Japan. pp.935-939

  3. Aghilas Sini, Damien Lolive, and Élisabeth Delais-Roussarie. Annotation automatique des types de discours dans des livres audio en vue d’une oralisation par un système de synthèse. In Traitement Automatique des Langues Naturelles(TALN’18), Rennes, France, 2018

  4. Aghilas Sini, Damien Lolive, Gaëlle Vidal, Marie Tahon, and Élisabeth Delais-Roussarie. ``Synpaflex-corpus : An expressive french audiobooks corpus dedicated to expressive speech synthesis`` . In Language Resources and Evaluation Conference (LREC'18),Miyazaki, Japan,2018

  5. Deng, Boyuan, Denis Jouvet, Yves Laprie, Ingmar Steiner, and Aghilas Sini ``Towards confidence measures on fundamental frequency estimations.`` In Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on, pp. 5605-5609. IEEE, 2017.

  6. Ghosh, Sucheta, Camille Fauth, Yves Laprie, and Aghilas Sini. ``End-to-End Acoustic Feedback in Language Learning for Correcting Devoiced French Final-Fricatives.`` Proc. Interspeech 2017 (2017): 349-353.

  7. Ghosh, Sucheta, Camille Fauth, Aghilas Sini, and Yves Laprie. ``L1-L2 interference: The case of final devoicing of French voiced fricatives in final position by German learners.`` In Interspeech 2016, vol. 2016, pp. 3156-3160. 2016.

  8. Vincent, Emmanuel, Aghilas Sini, and François Charpillet. ``Audio source localization by optimal control of a mobile robot.`` In Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on, pp. 5630-5634. IEEE, 2015.

RESEARCH WORK IN PROGRESS

  1. Aghilas Sini, Damien Lolive, and Nelly Barbot ``Investigating inter and intra-speaker voice conversion in audiobook``. submited for Speech Prosody 2022.

  2. Aghilas Sini, Marie Tahon, Damien Lolive and Elisabeth Delais-Roussarie ``Sentiment Analysis of French Audiobooks: Categorical vs. Continue Approach Speech And Text Prosodic Correlations``.

  3. Aghilas Sini, Lily Wadoux, Damien Lolive, Gaëlle Vidal, Antoine Perquin and Elisabeth Delais-Roussarie ``MUFASA: MUltispeaker French Audiobooks corpus dedicated to expressive read Speech Analysis``. Language Resources and Evaluation Journal.

  4. Aghilas Sini, Arnaud Delay, Damien Lolive, and Gwénolé Lecorvé ``Multi-language pronunciation assessment system for migrant's linguistic skills evaluation`` en préparation.