Federico Flego
I am a research associate at the
Cambridge University Engineering Department and I
belong to the
Speech Research Group
in the Information Engineering Division.
Research Interests
Statistical Machine Translation
- Hierarchical phrase-based translation
Large vocabulary continuous speech recognition (past)
- Adaptive and discriminative training
- Model compensation for noise robustness
- Rapid environment adaptation
- Environment change tracking/detection
- Model compensation for speaker adaptation
- Key word spotting in extremely noisy speech (S/N < 10dB)
Research Projects
- 2012-         :Broad Operational Language Translation
(BOLT)
                   
(DARPA funded project)
- 2011-2012: Robust Automatic Transcription of Speech
(RATS)
                   
(DARPA funded project)
- 2008-2010: Rapid and Robust Environment Aware Processing
(RREAP)
                   
(Toshiba Research Europe Limited funded project)
Activities
Publications
Journal articles
Conference proceedings
- Model-Based Approaches for Degraded Channel Modelling in Robust ASR
M. J. F. Gales and F. Flego, Interspeech 2012
- Factor Analysis Based VTS Discriminative Adaptive Training
F. Flego and M. J. F. Gales, ICASSP 2012
- Factor Analysis Based VTS and JUD Noise Estimation and Compensation
F. Flego and M. J. F. Gales, ICASSP 2011
- Discriminative Adaptive Training with VTS and JUD
F. Flego and M. J. F. Gales, ASRU 2009
- Incremental Adaptation with VTS and Joint Adaptively Trained Systems
F. Flego and M. J. F. Gales, Interspeech 2009
- Transforming Features to Compensate Speech Recogniser Models for Noise
R. C. van Dalen, F. Flego, and M. J. F. Gales, Interspeech 2009
- Combining VTS Model Compensation and Support Vector Machines
M. J. F. Gales and F. Flego, ICASSP 2009
- Incremental Predictive and Adaptive Noise Compensation
F. Flego and M. J. F. Gales, ICASSP 2009
Technical Reports
Teaching
E-mail:

,
Office: +44 (0)1223 339 762