TR2003-110

Design of the CMU Sphinx-4 Decoder


    •  Lamere, P., Kwok, P., Walker, W., Gouvea, E., Singh, R., Raj, B., Wolf, P.P., "Design of the CMU Sphinx-4 Decoder", Eurospeech, September 2003.
      BibTeX TR2003-110 PDF
      • @inproceedings{Lamere2003sep,
      • author = {Lamere, P. and Kwok, P. and Walker, W. and Gouvea, E. and Singh, R. and Raj, B. and Wolf, P.P.},
      • title = {Design of the CMU Sphinx-4 Decoder},
      • booktitle = {Eurospeech},
      • year = 2003,
      • month = sep,
      • url = {https://www.merl.com/publications/TR2003-110}
      • }
  • Research Areas:

    Artificial Intelligence, Speech & Audio

Abstract:

The decoder of the sphinx-4 speech recognition system incorporates several new design strategies which have not been used earlier in conventional decoders of HMM-based large vocabulary speech recognition systems. Some new design aspects include graph construction for multilevel parallel decoding with independent simultaneous feature streams without the use of compound HMMs, the incorporation of a generalized search algorithm that subsumes Viterbi and full-forward decoding as special cases, design of generalized language HMM graphs from grammars and language models of multiple standard formats, that toggles trivially from flat search structure to tree search structure etc. This paper describes some salient design aspects of the Sphinx-4 decoder and includes preliminary performance measures relating to speed and accuracy.

 

  • Related News & Events

    •  NEWS    Eurospeech 2003: 2 publications by MERL researchers and others
      Date: September 1, 2003
      Where: Eurospeech
      Brief
      • The papers "Classification with Free Energy at Raised Temperatures" by Singh, R., Warmuth, M., Raj, B. and Lamere, P. and "Design of the CMU Sphinx-4 Decoder" by Lamere, P., Kwok, P., Walker, W., Gouvea, E., Singh, R., Raj, B. and Wolf, P.P. were presented at Eurospeech.
    •