论文部分内容阅读
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 binary tree based on Statistic Language Model (SLM) in speech pause prediction. Based on the large-scale corpus, the ternary model Trigram is used to establish the statistical language model. The SLM is used to generate the corresponding binary tree for the sentence to be processed. The information contained in the binary tree is applied from different perspectives to the prediction of voice pause. The experimental results show that the binary tree generated based on SLM can better contribute to the prediction of speech pause.