CU-HTK APRIL 2002 SWITCHBOARD SYSTEM
Phil Woodland, Gunnar Evermann, Mark Gales, Thomas Hain, Andrew Liu, Gareth Moore, Dan Povey, Lan Wang
May 2002
This paper describes the Cambridge University HTK (CU-HTK) system developed for the NIST April 2002 evaluation of English conversational telephone speech transcription (Switchboard). A range of new features were integrated and are described in detail. Major new features include the use of Minimum Phone Error training (MPE), third differential features in conjunction with an HLDA transform, speaker adaptive training (SAT), the use of a special single pronunciation dictionary (SPRON) and a new LVR decoder (HDecode). A faster version of the full system that ran in less than 10 times realtime was also developed. Results are presented for the 2001 development test set and the 2002 evaluation set. In the April 2002 Switchboard evaluation the CU-HTK system achieved an overall word error rate of 23.9%, which was the lowest word error rate by a statistically significant margin.
If you have difficulty viewing files that end '.gz',
which are gzip compressed, then you may be able to find
tools to uncompress them at the gzip
web site.
If you have difficulty viewing files that are in PostScript, (ending
'.ps' or '.ps.gz'), then you may be able to
find tools to view them at
the gsview
web site.
We have attempted to provide automatically generated PDF copies of documents for which only PostScript versions have previously been available. These are clearly marked in the database - due to the nature of the automatic conversion process, they are likely to be badly aliased when viewed at default resolution on screen by acroread.