Table 1 Hyper-parameters used in the first stage of the BF-net
From: A multichannel learning-based approach for sound source separation in reverberant environments
Symbol | Parameter | Description |
|---|---|---|
NFFT | 1024 | Number of FFT |
Wlen | 512 | Window size of FFT |
Hop | 128 | Hopsize of the window |
B | 256 | Channels in bottleneck |
H | 512 | Channels in convolutional blocks |
P | 3 | Kernel size in convolutional blocks |
D | 6 | Convolutional blocks in each repeat |
R | 4 | Number of repeats |