基于支持向量机的语音识别研究

Research of Speech Recognition based on Support Vector Machine

为了得到较好的语音识别效果,构建了基于线性核函数支持向量机的非特定人孤立词语音识别系统,取得了较高的识别率,并将该实验结果同基于HMM的识别结果进行了比较,显示出了支持向量机在基于有限样本情况下进行语音识别的优势。

In order to get better results of speech recognition, we construct a non-specific, isolated word speech recognition system based on linear kernel function support vector machine. Through doing a lot of experiments, we get higher accuracy. By comparing the results with that of speech recognition based on HMM, we find that the support vector machine has advantage on speech recognition when the training data is small.