The CUHTK-entropic 10xRT broadcast news transcription system
JJ Odell, PC Woodland, Thomas Hain · 1999
This paper describes the development of the CUHTK-Entropic 10xRT Broadcast News Transcription System. Previous HTK broadcast news transcription systems have focused on maximising accuracy with few constraints on compute power available. In order to develop a system running in under 10 times real time on a single CPU, detailed investigation and optimisation of the system architecture and mode of operation was required. This paper outlines those developments and discusses the way in which operation under 10xRT was ensured despite variability of the data to be recognised. On the 1998 test the system produced an average word error rate of 16.1% running in 9.5xRT. 1. Introduction As the state-of-the-art in broadcast news transcription continues to improve, interest is growing in the range of applications for the transcription of such audio sources. However, most research on such systems is heavily focused on increasing accuracy and there has been relatively few investigations into the tra...