Tree-trellis forward-backward algorithm has been widely used for N-best search in continuous speech magnition. In conventional approaches, the heuristic score used for the A' backward search is derived from the padial-path scores recorded during the f-ard pass. The inherently delayed use of language model in the lexical tree structure leads to inefficient pruning and the partial-path scorn recorded is an underestimated heuristic score. This paper presents a novel method of computing the heuristic mre that is more accurate than the partial-path =re. The goal is to rewver high-score sentence hyptheses that may have been pruned halfway during the fanvard search due lo thc delayed use of LM. For the application of Hong Kong stoek information inquiry, the proposed technique shows a noticeable performance improvement. In particular, a relative error-rate reduction of 12% has been achieved for top-I sentences.
[1]
Tan Lee,et al.
Lexical tree decoding with a class-based language model for Chinese speech recognition
,
2000,
Interspeech.
[2]
Frank K. Soong,et al.
The use of tree-trellis search for large-vocabulary Mandarin polysyllabic word speech recognition
,
1994,
Comput. Speech Lang..
[3]
Frank K. Soong,et al.
A Tree.Trellis Based Fast Search for Finding the N Best Sentence Hypotheses in Continuous Speech Recognition
,
1990,
HLT.
[4]
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.