• DocumentCode
    3622345
  • Title

    Hierarchical Structures of Neural Networks for Phoneme Recognition

  • Author

    P. Schwarz;P. Matejka;J. Cernocky

  • Author_Institution
    Speech@FIT group, Brno University of Technology, Czech Republic, schwarzp@fit.vutbr.cz
  • Volume
    1
  • fYear
    2006
  • fDate
    6/28/1905 12:00:00 AM
  • Abstract
    This paper deals with phoneme recognition based on neural networks (NN). First, several approaches to improve the phoneme error rate are suggested and discussed. In the experimental part, we concentrate on temporal patterns (TRAPs) and novel split temporal context (STC) phoneme recognizers. We also investigate into tandem NN architectures. The results of the final system reported on standard TIMIT database compare favorably to the best published results
  • Keywords
    "Neural networks","Hidden Markov models","Error analysis","Multilayer perceptrons","Speech recognition","Keyword search","Artificial neural networks","Recurrent neural networks","Training data","Dynamic range"
  • Publisher
    ieee
  • Conference_Titel
    Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on
  • ISSN
    1520-6149
  • Print_ISBN
    1-4244-0469-X
  • Electronic_ISBN
    2379-190X
  • Type

    conf

  • DOI
    10.1109/ICASSP.2006.1660023
  • Filename
    1660023