改代码为音频降噪功能，是基于RNN的音频降噪算法。其中采用的是 GRU/LSTM 模型，支持wav格式。(The modified code is an audio denoising function, which is based on RNN's audio denoising algorithm. The GRU/LSTM model is adopted to support the wav format.)