嵌入式英语语音识别系统误差自动检测方法研究

Research on automatic error detection method of embedded english speech recognition system

为了提高嵌入式英语语音识别系统发音误差自动检测能力,提出基于时频分析和关联信息熵特征提取的嵌入式英语语音识别系统发音误差自动检测方法。采用时频特征分解方法进行嵌入式英语语音识别系统英语发音信号的降噪处理,对降噪输出的英语发音信号进行特征分解和关联维特征配准,结合小波多层重构方法进行语音信号的重组,提取英语发音信号的关联信息熵特征,根据提取的语音信号的关联信息熵特征进行自动匹配,实现对嵌入式英语语音识别系统误差的自动识别。仿真结果表明,采用该方法进行嵌入式英语语音识别系统发音误差自动检测的准确性较好,对语音信号的分辨能力较好,提高了嵌入式英语语音识别系统的发音误差的检测性能。

In order to improve the automatic detection ability of pronunciation error in embedded English speech recognition system,an automatic detection method for pronunciation error of embedded English speech recognition system based on time-frequency analysis and feature extraction of associated information entropy is proposed.The method of time-frequency feature decomposition is used to reduce the noise of the English pronunciation signal in the embedded English speech recognition system.The output English pronunciation signal is decomposed by the feature decomposition and the correlation dimension feature registration is carried out.Combined with wavelet multi-layer reconstruction method,the speech signal is reorganized to extract the associated information entropy feature of the English pronunciation signal,and the correlation information entropy feature of the extracted speech signal is automatically matched according to the extracted speech signal correlation information entropy feature.The tone recognition of embedded English speech recognition system is realized.The simulation results show that the proposed method can detect the pronunciation error of the embedded English speech recognition system with good accuracy and resolution ability.The performance of pronunciation error detection in embedded English speech recognition system is improved.