The use of tree-trellis search for large-vocabulary Mandarin polysyllabic word speech recognition

In this paper, we propose the use of a tree-trellis search scheme for the task of large vocabulary Mandarin polysyllabic word recognition. Usually, the task of large vocabulary word recognition is computationally intractable by whole-word based approach. We convert this task into a tree network search problem by considering basic syllables as matching units. A vocabulary of 22 088 polysyllabic words is represented by a tree network linked with basic syllables. Some implementations of tree-trellis search scheme for this special task are investigated. In comparison with other search algorithms, the experimental results show that the tree-trellis search algorithm is a very promising one for this special task.

[1]  R. Schwartz,et al.  A comparison of several approximate algorithms for finding multiple (N-best) sentence hypotheses , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[2]  P. Mermelstein,et al.  Fast search strategy in a large vocabulary word recognizer , 1988 .

[3]  Lalit R. Bahl,et al.  A Maximum Likelihood Approach to Continuous Speech Recognition , 1983, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  D. O'Shaughnessy,et al.  A*-admissible heuristics for rapid lexical access , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[5]  E. Schulze.  Hypothesizing of words for isolated and connected word recognition systems based on phonem preclassification , 1982, ICASSP.

[6]  Louis A. Liporace,et al.  Maximum likelihood estimation for multivariate observations of Markov sources , 1982, IEEE Trans. Inf. Theory.

[7]  Kenji Kita,et al.  Continuous mixture HMM-LR using the a* algorithm for continuous speech recognition , 1992, ICSLP.

[8]  Steve Austin,et al.  The forward-backward search algorithm , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[9]  Lawrence R. Rabiner,et al.  A segmental k-means training procedure for connected word recognition , 1986, AT&T Technical Journal.

[10]  Nils J. Nilsson,et al.  Principles of Artificial Intelligence , 1980, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Roberto Pieraccini,et al.  Strategies for lexical access to very large vocabularies , 1988, Speech Commun..

[12]  Yuqing Gao,et al.  A real-time Chinese speech recognition system with unlimited vocabulary , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[13]  Victor Zue,et al.  Properties of large lexicons: Implications for advanced isolated word recognition systems , 1982, ICASSP.

[14]  Frank K. Soong,et al.  A Tree.Trellis Based Fast Search for Finding the N Best Sentence Hypotheses in Continuous Speech Recognition , 1990, HLT.

[15]  Amir Averbuch,et al.  An IBM PC based large-vocabulary isolated-utterance speech recognizer , 1986, ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing.