应用概率LR文法分析器于语音识别系统

SPEECH RECOGNITION BASED ON PROBABILISTIC LR PARSER

将概率ＬＲ文法分析器应用于语音识别系统中．模拟实验表明，系统使用该模型后识别性能优于使用二元文法模型的系统．其中。

Probabilistic LR parser is applied to speech recognition in this paper. Simulation experiments show that the system employing the above language model has better recognition performance than the system employing bi_gram model. Two different rejection mechanisms are important in improving system performance within the system.