Phoneme-based word recognition by neural network-a step toward large vocabulary recognition

  title={Phoneme-based word recognition by neural network-a step toward large vocabulary recognition},
  author={A. Hirai and A. Waibel},
In this paper we present a neural network based word recognition system extendable to large vocabulary isolated word recognition. The system consists of (1) time-delay neural networks (TDNNs) for phoneme spotting, and (2) a higher level network and a DP-time alignment procedure for word recognition. TDNN-based phoneme spotting networks are used whose role is to fire when a particular phoneme is input. A higher level network then improves these phoneme firing patterns in view of an idealized… CONTINUE READING


Publications referenced by this paper.
Showing 1-6 of 6 references

Improvement on Phoneme Spotting Experiment by a Large Phonemic Time Delay-Neural Network

  • H. Sawai, K . Shikano
  • Proc. of ZCASS,
  • 1990

Consonant and Phoneme Recognition by Modular Construction of Large Phonemic Time - Delay Neural Networks

  • H. Sawai Waibel, K. Shikano
  • 1989

Speaker - Independent Word Recognition Using Dynamic Programming Neural Networks

  • R. Isotani Sakoe
  • Neural - Net Classifiers Useful for Speech…
  • 1989

Reconnaissance de la Parole par Reseaux mu 1 ti - couches

  • R. J. Williams
  • Dynamic Programming Algorithm Optimization for…
  • 1988

Similar Papers

Loading similar papers…