Improving Speech Transcription for Mandarin-English Translation

Marcus Tomalin, Mark J. F. Gales, X. Andrew Liu, Khe Chai Sim, Rohit Sinha, Lan Wang, Philip C. Woodland, Kai Yu

Abstract

This paper describes the development of the CU-HTK Mandarin Speech-To-Text (STT) system and assesses its performance as part of a transcription-translation pipeline which converts broadcastMandarin audio into English text. Recent improvements to the STT system are described and these give Character Error Rate (CER) gains of 14.3% absolute for a Broadcast Conversation (BC) task and 5.1% absolute for a Broadcast News (BN) task. The output of these STT systems is then post-processed, so that it consists of sentence-like segments, and translated into English text using a Statistical Machine Translation (SMT) system. The performance of the transcription-translation pipeline is evaluated using the Translation Edit Rate (TER) and BLEU metrics. It is shown that improving both the STT system and the post-STT segmentations can lower the TER scores by up to 5.3% absolute and increase the BLEU scores by up to 2.7% absolute.
BibTeX
@inproceedings{tomalin2007improving,
  author = {Marcus Tomalin and Mark J. F. Gales and X. Andrew Liu and Khe Chai Sim and Rohit Sinha and Lan Wang and Philip C. Woodland and Kai Yu},
  title = {{Improving Speech Transcription for Mandarin-English Translation}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2007},
  doi = {10.1109/icassp.2007.367172},
}

Loading related content...

Figure