基于改进语音特征提取方法的语音识别

Speech Recognition Based on Improved Speech Feature Extraction

在分析语音特征提取方法基础上提出一种改进组合算法,并采用HMM声学模型和Viterbi算法进行模式训练和识别.实验结果表明,该算法在噪声环境中具有较好的鲁棒性,能有效提高噪声环境下中文连续语音识别的正确率,增强语音识别整体性能,因此在噪声环境下的语音识别系统中具有一定的实用价值.

This paper analyses speech feature extraction and makes improvement of it, also adopts the HMM acoustic model and the Viterbi algorithm to do model training and recognition. The experimental results show that the method has good robustness in the noisy environment. It can effectively improve the accuracy of the continuous chinese speech recognition, enhance the overall performance of speech recognition, and has some practical value to the speech recognition system in the noisy environment.