1篇
您的检索式：作者名="BAI Haojun"
|
|
|
题名
|
作者
|
年代
|
出处
|
被引量
|
| 1 | Monaural speech enhancement combining accurate ratio mask and deep neural network显示文摘Aiming at the problem that the supervised speech enhancement ignores the impact of the similarity of amplitude spectrum between clean speech,noise,and noisy speech on the enhancement effect,a method combining accurate ratio mask(ARM)and deep neural network(DNN)is proposed for monaural speech enhancement.Firstly,an accurate ratio mask based on ideal ratio mask in the time-frequency domain is designed,which utilizes the normalized cross-correlation coefficients of amplitude spectrum between clean speech and noisy speech,and between noise and noisy speech.Then,the target mask is estimated by the output of the baseline DNN which takes the amplitude spectrum of clean speech and noise as training target,and further uses the target mask to optimize the baseline DNN and gets the enhanced speech from noisy speech.Moreover,considering the discriminative information between clean speech and noise,a discriminative training function is used to replace the mean square error(MSE)as the objective function of the DNN,thus making the output of network more accurate.The experimental results show that the discriminative training function improves the enhancement effect of baseline DNN and the overall joint optimization network.Compared with other common DNN methods,the proposed method has achieved higher average PESQ,STOI and better noise suppression effect under matched and unmatched noise,and the enhanced speech also retains more speech components. | BAI Haojun ZHANG Tianqi LIU Jianxing YE Shaopeng | 2022 | Chinese Journal of Acoustics2022，41，4： | 0 |
您所在的机构暂未开通此部分的功能，请联系管理员