Adapting grapheme-to-phoneme conversion for name recognition

  • Xiao Li ,
  • Alex Acero ,
  • Asela Gunawardana

IEEE Workshop on Automatic Speech Recognition and Understanding |

Published by Institute of Electrical and Electronics Engineers, Inc.

This work investigates the use of acoustic data to improve grapheme-to-phoneme conversion for name recognition. We introduce a joint model of acoustics and graphonemes, and present two approaches, maximum likelihood training and discriminative training, in adapting graphoneme model parameters. Experiments on a large-scale voice-dialing system show that the maximum likelihood approach yields a relative 7% reduction in SER compared to the best baseline result we obtained without leveraging acoustic data, while discriminative training enlarges the SER reduction to 12%.