Post-doctoral position in Speech Recognition

Application Deadline: 30th November 2004

Start date (latest): 1st March 2005

Applications are invited for a Post-Doctoral position to be held 9 months in Technical University of Crete, Chania, Greece followed by 9 months at LORIA, Nancy, France.

The successful applicant should have a PhD in the area of Computer Science, Statistics, Engineering, Mathematics or Physics. The candidate sould have a strong speech/signal processing background with emphasis in speech recognition. Good knowledge of statistical modeling and front-end techniques for robust speech recognition is a plus. Strong software skills are important (C/C++, script languages).

The project involves the development of novel feature extraction algorithms and statistical models for automatic speech recognition. The collaboration is part of the MUSCLE Network of Excellence EU Project (www.muscle-noe.org) and extends over 18 months. The parties involved are the (Speech and Dialogue Group) at the Technical University of Crete (LORIA) and the (Computer Vision, Speech and Signal Processing Group ) at the Technical University of Athens.

The stipend is 29660 Euros/year tax free with social security paid by MUSCLE. More information about MUSCLE Fellowships can be found at:
Please send a CV and the names of 3 referees to Alex Potamianos potam@telecom.tuc.gr and Khalid Daoudi daoudi@irit.fr by Oct 30th 2004.

For further information, interested candidates can contact any of the following:
  • Fellow: Eduardo Sanchez SOTO
  • Nationality: Mexican
  • Host 1: Prof. Alexandros Potamianos (TSI-TUC)
  • Host 2: Dr. Khalid Daoudi (INRIA-Parole)
  • Starting date: April 1, 2005 (duration: 9 mths + 9 mths)
 In the first 9 months, the post-doc will investigate novel features for ASR, including features motivated from nonlinear speech models (AM-FM features), time-domain features and suprasegmental features. This work is in collaboration with the ICCS-NTUA. During the last 9 months, the postdoc will investigate how to best integrate new features in the speech recognition framework. Early vs. late intergration will be investegated and various statistical models (Bayesian network) will be evaluated for feature combination. The end goal is to significantly improve on the state-of-the art of speech recognition systems.

Note that for the first 9 months, the candidate will be based at TUC Chania (or NTUA Athens), Greece and for the the last 9 months the candidate will work at LORIA, Nancy, France.