讨论基于统计语言模型SLM(Statistic Language Model)的二叉树在语音停顿预测中的应用。基于大规模语料,利用三元模型Trigram,建立统计语言模型;基于SLM为待处理句子生成相应的二叉树;将生成的二叉树所包含的信息,从不同角度应用于语音停顿的预测。实验结果表明,基于SLM生成的二叉树能够较好地为语音停顿的预测做出贡献。
This paper discusses the application of bintree based on SLM (Statistic Language Model) in speech pauses' prediction. It constructs Trigram statistic language model based on large-scale corpus, and builds corresponding bintree for the sentence waiting disposal; and then it predicts speech pauses at two different angle using information provided by tree. The results of experiments show that the bintree based on SLM can make contribution to speech pauses' prediction effectively.