基于DCNN的语音识别降噪方法研究

Research on DCNN-based noise reduction method for speech recognition

语音信号是当前人工智能领域的一个重要研究方向,传统语音信号识别结果易受噪声干扰,使得信号识别效果不理想,为了提高传统语音信号识别正确率,设计基于深度卷积神经网络(DCNN)的多噪声语音识别方法。首先,分析当前语音信号识别的研究进展,找到不同方法的局限性;然后从硬件和软件对语音信号进行降噪操作,提高语音信号质量;最后采用DCNN进行语音信号识别仿真实验,测试结果表明,该设计方法可以提高语音信号的信噪比,同时可以提高语音信号的识别正确率,相对于对比方法,可获得更加理想的语音信号识别结果,解决了当前语音信号识别过程中存在的难题,具有更高的实际应用价值。

Speech signal is an important research direction in the field of artificial intelligence.The results of the traditional speech signal recognition are disturbed by noise easily,which makes the effect of the traditional speech signal recognition unsatisfactory.Therefore,a multi-noise speech recognition method based on deep convolutional neural network(DCNN)is designed to improve the accuracy rate of the traditional speech signal recognition.The research progress of the current speech signal recognition is analyzed to find out the limitations of different methods.The speech signals are denoised in the manners of hardware and software to improve the speech signal quality.The DCNN was used in simulation experiment for speech signal recognition.The test results show that the designed method can improve the signal-to-noise ratio(SNR)and the recognition accuracy rate of the speech signals,and can obtain more ideal results of speech signal recognition in comparison with the contrasted method.Therefore,it can solve the problems existing in the process of the current speech signal recognition and has a higher practical application value.