| Presentation # | 7 |
| Session: | ASR I |
| Location: | Kallirhoe Hall |
| Session Time: | Wednesday, December 19, 10:00 - 12:00 |
| Presentation Time: | Wednesday, December 19, 10:00 - 12:00 |
| Presentation: |
Poster
|
| Topic: |
Speech recognition and synthesis: |
| Paper Title: |
IMPROVING LF-MMI USING UNCONSTRAINED SUPERVISIONS FOR ASR |
| Authors: |
Hossein Hadian, Sharif University of Technology, Iran; Daniel Povey, Johns Hopkins University, United States; Hossein Sameti, Sharif University of Technology, Iran; Jan Trmal, Sanjeev Khudanpur, Johns Hopkins University, United States |
| Abstract: |
We present our work on improving the numerator graph for discriminative training using the lattice-free maximum mutual information (MMI) criterion. Specifically, we propose a scheme for creating unconstrained numerator graphs by removing time constraints from the baseline numerator graphs. This leads to much smaller graphs and therefore faster preparation of training supervisions. By testing the proposed unconstrained supervisions using factorized time-delay neural network (TDNN) models, we observe 0.5\% to 2.6\% relative improvement over the state-of-the-art word error rates on various large-vocabulary speech recognition databases. |