Convolutional Recurrent Neural Network Based Progressive Learning for Monaural Speech Enhancement
share
Recently, progressive learning has shown its capacity of improving speech quality and speech intelligibility when it is combined with deep neural network (DNN) and long short-term memory (LSTM) based monaural speech enhancement algorithms, especially in low signal-to-noise ratio (SNR) conditions. Nevertheless, due to a large number of parameters and highly computational complexity, it is hard to implement in current resource-limited micro-controllers and thus, it is important to significantly reduce both the amount of parameters and the computational load for practical applications. For this purpose, we propose a novel progressive learning framework with convolutional recurrent neural networks called PL-CRNN, which takes advantages of both convolutional neural networks and recurrent neural networks to drastically reduce the amount of parameters and simultaneously improve speech quality and speech intelligibility. Numerous experiments verify the effectiveness of proposed PL-CRNN model and indicate that it yields consistent better performance than the PL-DNN and PL-LSTM algorithms and also it gets results close even better than the CRNN in terms of various evaluation metrics. Compared with PL-DNN, PL-LSTM and state-of-the-art CRNN models, the proposed PL-CRNN algorithm can reduce the amount of parameters up to 77%, 93% and 93%, respectively.
READ FULL TEXT