该方法首先利用HMM生成最佳语音状态序列，然后用函数逼近技术产生对最佳状态序列进行时间规正，最后通过RBF神经网络进行分类识别。

Here, the HMM is employed to produce a best speech state sequence which is warped to a fixed dimension vector and the RBF neural network is used as classifier.

youdao

该方法首先利用HMM生成最佳语音状态序列，然后用函数逼近技术产生对最佳状态序列进行时间规正，最后通过RBF神经网络进行分类识别。

Here, the HMM is employed to produce a best speech state sequence which is warped to a fixed dimension vector and the RBF neural network is used as classifier.

youdao