Skip to main content

Table 1 Trainable parameters of Sinc filters, where N is the size of filter bank

From: Trainable windows for SincNet architecture

Window functions

No. of trainable parameters

Gaussian, Exponential, Kaiser, Taylor, Chebwin, Tukey, Slepian,

\(2N + 1\)

Cosine-Sum

\(2N + (K+1) \;\;\ ; K > 0\)

Fixed windows

2N