Cambridge University Offers a PhD Studentship in Speech Recognition

The Speech Group of the Engineering Department at Cambridge University in England is offering a PhD research studentship associated with a new grant funded by the EPSRC. The project, “Generative Kernels and Score Spaces for Speech Classification of Speech,” aims to significantly improve the performance of speech recognition systems in highly challenging environments, such as high levels of background noise.

The project will combine and develop acoustic models, based on hidden Markov models, with discriminative classifiers, such as support vector machines and log-linear models. Thus, the work will involve research in both general machine-learning classification approaches and advanced speech recognition schemes. State-of-the-art speech recognition systems developed at Cambridge based on the publicly available HTK toolkit (http://htk.eng.cam.ac.uk/) will be extended to incorporate these combined classification schemes.

The studentship offers an opportunity for study towards a PhD while working in a research group. There are opportunities for publication and travel to international conferences. 

Eligbility:
Candidates must have a very good first degree in a relevant discipline and a masters degree (or equivalent) is an advantage. Since the project will use and extend the HTK large vocabulary speech recognition system, familiarity with the HTK toolkit and a good knowledge of C is a distinct advantage.

Funding:
The research studentship will start in January. The studentship will cover all United Kingdom and European Union university fees and a maintenance allowance at the current EPSRC award. For non-EU students, additional funding for university fees may be required.

Closing date: 12 October 2010.

Contact:
Principal Investigator
Dr Mark Gales
Engineering Dept, Trumpington Street, Cambridge
Email: mjfg@eng.cam.ac.uk
Tel: +44 1223 332733