摘要： 在分析了传统语音识别方法的特点后，提出一种规则结合模型的语音识别思想，并实现了”特征空间轨迹匹配＋种经网络”方法作为例证。实验数据反映该方法能够分别规整语音信号的时域确定因素和特征域不确定因素。实践表明,该方法是一种行之有效的实现特定领域汉语语音识别的实用方法。

关键词: 语音识别, 特征空间轨迹匹配, 神经网络

Abstract: After analizing traditional speech recognition plan, a rule-and-model recognition idea is introduced. A method named "Feature Space Trace Matching+Artificial Neural Network" is carried out and proved to be effective to normalize the determinate factors in time domain and the undeterminate factors in feature domain respectively. Experiments show that the method is a valid way to implement Chinese speech recognition in specific field.

中图分类号: 

• TN912.34