Adapting grapheme-to-phoneme conversion for name recognition

@article{Li2007AdaptingGC,
  title={Adapting grapheme-to-phoneme conversion for name recognition},
  author={Xiao Li and Asela Gunawardana and Alex Acero},
  journal={2007 IEEE Workshop on Automatic Speech Recognition & Understanding (ASRU)},
  year={2007},
  pages={130-135}
}
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… CONTINUE READING

Figures, Tables, Results, and Topics from this paper.

Key Quantitative Results

  • 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%.

Citations

Publications citing this paper.
SHOWING 1-10 OF 19 CITATIONS

Acoustic data-driven pronunciation lexicon generation for logographic languages

  • 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2016
VIEW 2 EXCERPTS
CITES METHODS
HIGHLY INFLUENCED

Learning Personalized Pronunciations for Contact Name Recognition

VIEW 6 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

Acoustic data-driven pronunciation lexicon for large vocabulary speech recognition

  • 2013 IEEE Workshop on Automatic Speech Recognition and Understanding
  • 2013
VIEW 6 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

Pronunciation Learning for Automatic Speech Recognition

VIEW 3 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

Automatic pronunciation verification for speech recognition

  • 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2015
VIEW 1 EXCERPT
CITES BACKGROUND

On statistical machine translation method for lexicon refinement in speech recognition

  • 2015 IEEE China Summit and International Conference on Signal and Information Processing (ChinaSIP)
  • 2015
VIEW 3 EXCERPTS
CITES BACKGROUND & METHODS

References

Publications referenced by this paper.
SHOWING 1-10 OF 10 REFERENCES

Discriminative training of language models for speech recognition

  • 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing
  • 2002
VIEW 2 EXCERPTS

Dynamic generation of proper name pronunciations for directory assistance

  • 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing
  • 2002
VIEW 1 EXCERPT

Similar Papers

Loading similar papers…