| Saturday Afternoon (15 May, 2004) |
| 3 |
Acoustic Modelling / Improved Training Algroithms, 13:45-15:30 |
| 13:45 |
Update on Discriminative Training, Recent results with MPE, MMI
(PDF slides)
Thomas Colthurst, Spyros Matsoukas, Owen Kimball, BBN |
| 13:55 |
Precision Matrix Modelling for LVCSR
(PDF slides)
Khe Chai Sim, Mark Gales, Cambridge |
| 14:20 |
Decision-tree versus genonic clustering of HMM states
(PDF slides)
Venkata Gadde, Dimitri Vergyri, Jing Zheng, Horacio Franco, SRI |
| 14:30 |
Novel Approaches update
(PDF slides)
Nelson Morgan, ICSI and Andreas Stolcke, SRI |
| 15:00 |
Discussion on Acoustic Modelling / Training Algorithms, 15:00-15:30 |
| 15:30-16:00 Break |
| 4 |
Language Modelling and Fast Algorithms, 16:00-17:20 |
| 16:00 |
Language Modelling Using Fisher Data
(PDF slides)
Holger Schwenk, Gilles Adda, Jean-Luc Gauvain, LIMSI |
| 16:15 |
Adaptive Language Models
(PDF slides)
Langzhou Chen, Jean-Luc Gauvain, Lori Lamel, Gilles Adda, LIMSI |
| 16:35 |
Reducing storage and computation in training
(PDF slides)
Jeff Ma, Spyros Matsoukas, Rich Schwartz, BBN |
| 16:55 |
Quick update on faster CTS decoding
(PDF slides)
Andreas Stocke, Jing Zheng, SRI |
| 17:05 |
Towards efficient cross-site system combination
(PDF slides)
Rohit Prasad, Spyros Matsoukas, Rich Schwartz, BBN
Jean-Luc Gauvain, Holger Schwenk, Lori Lamel, LIMSI |
| 17:25 |
Discussion, 17:05-17:20 |
|