Ph.D. Student in Computer Science
Ph.D. Student in Computer Science
Language Engineer
Linux Enthusiastic
Python Charmer
After I have graduated with a master's in Natural Language Processing, I am currently a Ph.D student in Computer Science working on the VoxCrim research project. My thesis is about comparing voices in forensics. My interest in Speech Processing has also allowed me to work on the improvement of Google speech recognition and synthesis.
September 2018 - Present
Laboratoire d'Informatique d'Avignon (LIA)
The topic of the thesis is about voice comparison in forensics: in two audio files, how to determine whether the voice comes from the same source?
November 2017 - February 2018
Snips
In order to improve the functionality of its personal assistant, Snips hires linguists to work on different languages. It implies recruiting the linguists for a given language (native or not), teaching them Snips's tools and needs, delivering linguistic data on time and developing tools to help them.
August 2017 - September 2017
Amazon
Creating Amazon Alexa's French voice. This includes normalizing textual resources through regular expressions, creating natural language test cases for natural language processing modules, labeling linguistic data with semantic, syntactic and morphological features, contribute to the formalization of language models through in depth knowledge of phonological and morphosyntactic aspects of the target language, QA of TTS voices and linguistic data.
September 2016 - June 2017
Google (via Adecco Outsourcing)
Developing Google Assistant’s speech synthesis.
Main tasks: chosing the voice based on phonological and prosodic criteria, writing message patterns, implementing the messages, setting up a large acoustic database from scripts, developing programs to ease the team’s work.
March 2015 - September 2015
Sorbonne-Nouvelle - Paris 3 University
Correction of automatic phoneme recognition in order to study the occurrences of the phoneme /R/ in French oral discourses.
Main tasks: defining the frontiers of the /R/ phonemes from spectrograms.
April 2013 - March 2014
Google (via Berkley Recruitment Ltd.)
Processing linguistic and speech data in order to improve automatic speech recognition (ASR) and speech synthesis services on Google products.
Main tasks: associating acoustic forms to graphemes, identifying the ASR output errors, adding entries into the pronunciation dictionary, modelling language using grammars.
Located in Dublin, Ireland.
March 2016 - July 2016
Laboratoire National de métrologie et d'Essais (LNE)
Setting up a perceptive test evaluating how serious automatic speech recognition errors can be and their impact upon a specific task as part of the ANR project VERA.
Main tasks: sampling data, recruiting subjects, supervising the experience, analyzing results.
July 2015 - September 2015
CNRS - Modyco
Converting automatically lextrees from Decoda format to Orfeo format for the ANR project Orfeo.
Main tasks: Developing a program to convert Decoda lextrees to Orfeo lextrees.
2014 - 2016
Sorbonne Nouvelle - Paris 3 University
Natural Language Processing
2012 - 2013
François Rabelais University of Tours
Linguistics
2008 - 2012
François Rabelais University of Tours
English Studies
2007
Lycée Maurice Genevoix of Ingré
Science specialty Physics and Chemistry
French
English
Spanish
German
Swedish
