Emmanuel DIDIOT


     Email: didiot@loria.fr



Current Position


 

I'm now Research Engineer CNRS in ABC team at LORIA since september 2013. I work on M-SVMs applied to protein secondary structure prediction.

 


Old Positions


 

·  Expert Engineer INRIA in SYNALP team at LORIA

·  Research Engineer (Postdoc) in the Spoken Language Processing Group at LIMSI-CNRS

·  Project Manager at Hewlett Packard (HP PSF Luxemburg)

·  Research Engineer and Project leader at TNS Media Intelligence (now Kantar Media News Intelligence)

·  ATER at Nancy2 University ,IUT Charlemagne Computer Science Dept.

·  Member of ABC team at LORIA

·  ATER at UHP Nancy1 Computer Science Dept.

·  Member of Speech team at LORIA

 


Research interests


 

·  Speech/music discrimination

·  Wavelets in signal processing

·  Pattern recognition (SVMs, neural networks,...)

·  Kernel methods in genomics and computational biology

·  Bioinformatics (computational biology)

·  Model selection for multi-class SVMs

·  And many other things...

 


Teaching Experience


Nancy2 University ,IUT Charlemagne Nancy2

202 hours -- Lecturer at Computer Science Dept. (2007--2008).
- Object-oriented programming using Java,
- Database administration,
- Computer architecture, 
- Web programming : xHTML, PHP, 
- Relational database management system (normalization, security, ...).

 

Nancy1 University , Computer Science Dept.

90 hours -- Lecturer at Maths & Computer science Dept. (2006--2007).
- Algorithms and programming for general students,
- Object-oriented programming using Java,
- C programming,
- Database concept,
- Artificial intelligence,
- Network concept and programming.

 

Nancy2 University ,IUT Charlemagne Nancy2

167 hours -- Lecturer at Computer Science Dept. (2004--2006).
- Object-oriented programming using Java,
- Algorithmic,
- Computer architecture.

 


List of Publications


A Wavelet-Based Parameterization for Speech/Music Segmentation
Didiot E., Illina I., Fohr D., Mella O.
Journal:  Computer Speech and Language (CSL)
Volume 24, Issue 2, April 2010, Pages 341-357.


A Wavelet-Based Parameterization for Speech/Music Segmentation
Didiot E., Fohr D., Haton J.-P., Illina I., Mella O.
Proceedings of Ninth International Conference on Spoken Language Processing - INTERSPEECH 2006
September 17-21 2006, Pittsburgh, Pennsylvania, USA. pp: 653-656.

Speech/music discrimination based on wavelets for broadcast programs
Didiot E., Fohr D., Haton J.-P., Illina I., Mella O.
Proceedings of International Conference on Signal Processing and Multimedia Applications - SIGMAP 2006
August 7-10 2006, Setubal, Portugal. pp: 151-...

Une nouvelle approche fondée sur les ondelettes pour la discrimination parole/musique
Didiot E., Fohr D., Haton J.-P., Illina I., Mella O.
Proceedings of XXVIes Journées d'Etude sur la Parole - JEP 2006
June 12-16 2006, Dinard, France. pp. 209-212.


Conception et mise en oeuvre de M-SVM dédiées au traitement de séquences biologiques
Didiot E.
Stage de DEA. Rapport de stage.
2003, Nancy, France. 32 pages.

 


Ph.D. Thesis


Host Laboratory

Speech team at LORIA, France

Supervisors

Jean-Paul Haton (Professor) and Irina Illina (Associate Professor)

Thesis title

Speech/music segmentation for automatic continuous speech transcription

Date of defense

13 Novembre 2007 ( LORIA.)

Abstract

In this thesis, we study the segmentation of an audio stream in speech, music and speech on music (S/M). This is a fundamental step for all application based on automatic transcription of radiophonic stream and most commonly multimedia. The target application here is a keyword detection system in broadcast programs. The application performance depends on the quality of the signal segmentation given by the speech/music discrimination system. Indeed, bad signal classification can give miss-detections or false alarms. To improve the speech/music discrimination task, we propose a new signal parameterization method. We use the wavelet decomposition which allows an analysis of non-stationary signal like music for instance. We compute different energies on wavelet coefficients to construct our feature vectors. The signal is then segmented in four classes : speech (S), non-speech (NS), music (M) and non-music (NM), thanks to two apart class/non-class classification systems. These classification systems are based on HMM. We chose a class/non-class architecture because it allows to find independently the best parameters for each S/NS and P/NP tasks. A fusion of the classifier ouputs is then performed to obtain the final decision : speech, music or speech on music. The obtained results on a real broadcast program corpus show that our wavelet-based parameterization gives a significant improvement in performance in both M/NM and S/M discrimination tasks compared to the baseline parameterization using cepstral coefficients.