Application of recurrent U-net architecture to speech enhancement
T Grzywalski, S Drgas - 2018 Signal Processing: Algorithms …, 2018 - ieeexplore.ieee.org
T Grzywalski, S Drgas
2018 Signal Processing: Algorithms, Architectures, Arrangements …, 2018•ieeexplore.ieee.orgIn this paper a recurrent U-net neural architecture is proposed to speech enhancement. The
mentioned neural network architecture is trained to provide a mapping between a
spectrogram of a noisy speech and both spectrograms of isolated speech and noise. Some
key design choices are being evaluated in experiments and discussed, including: number of
levels of the U-net, presence/absence of recurrent layers, presence/absence of max pooling
layers as well and upsampling algorithm used in decoder part of the network.
mentioned neural network architecture is trained to provide a mapping between a
spectrogram of a noisy speech and both spectrograms of isolated speech and noise. Some
key design choices are being evaluated in experiments and discussed, including: number of
levels of the U-net, presence/absence of recurrent layers, presence/absence of max pooling
layers as well and upsampling algorithm used in decoder part of the network.
In this paper a recurrent U-net neural architecture is proposed to speech enhancement. The mentioned neural network architecture is trained to provide a mapping between a spectrogram of a noisy speech and both spectrograms of isolated speech and noise. Some key design choices are being evaluated in experiments and discussed, including: number of levels of the U-net, presence/absence of recurrent layers, presence/absence of max pooling layers as well and upsampling algorithm used in decoder part of the network.
ieeexplore.ieee.org
以上显示的是最相近的搜索结果。 查看全部搜索结果