Sixth European Conference on Speech Communication and Technology
This paper describes a number of recent improvements to the HTK Broadcast News Transcription System. Changes to the system include the use of more acoustic training data; use of cluster-based variance normalisation and vocal tract length normalisation; the use of interpolated language models and enhanced adaptation using a full variance transform. These changes produce an reduction in word error rate of 13%. A simplified version of the system has also been constructed that runs in less than 10 times real-time and gives a 2.3% absolute higher error rate than the 300xRT full system.
Full Paper (PDF)
Bibliographic reference. Woodland, P. C. / Odell, J. J. / Hain, T. / Moore, G. L. / Niesler, T. R. / Tuerk, Andreas / Whittaker, E. W. D. (1999): "Improvements in accuracy and speed in the HTK broadcast news transcription system", In EUROSPEECH'99, 1043-1046.