Transcription of broadcast news with a time constraint: IBM’s 10xRT HUB4 system

Ellen Eide, Benoît Maison, D. Kanevsky, P. Olsen, S. Chen, L. Mangu, M. Gales, Miroslav Novak, Ramesh Gopinath

We describe a system which automatically transcribes broadcast news in less than 10 times real-time. We detail the system architecture of this system, which was used by IBM in the 1999 HUB4 10xRT evaluation, and show that the performance of this system is over 20 percent more accurate at the same speed than the system we used in the 1998 evaluation. Furthermore, we have closed the gap in word recognition accuracy between an unlimited resource system and this which runs in under 10 times real time from 45 percent to 14 percent.

doi: 10.21437/ICSLP.2000-403

Cite as: Eide, E., Maison, B., Kanevsky, D., Olsen, P., Chen, S., Mangu, L., Gales, M., Novak, M., Gopinath, R. (2000) Transcription of broadcast news with a time constraint: IBM’s 10xRT HUB4 system. Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000), vol. 2, 851-854, doi: 10.21437/ICSLP.2000-403

@inproceedings{eide00_icslp,
  author={Ellen Eide and Benoît Maison and D. Kanevsky and P. Olsen and S. Chen and L. Mangu and M. Gales and Miroslav Novak and Ramesh Gopinath},
  title={{Transcription of broadcast news with a time constraint: IBM’s 10xRT HUB4 system}},
  year=2000,
  booktitle={Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000)},
  pages={vol. 2, 851-854},
  doi={10.21437/ICSLP.2000-403}
}