modules/nsf_hifigan/utils.py

import matplotlib

matplotlib.use("Agg")


def init_weights(m, mean=0.0, std=0.01):
    classname = m.__class__.__name__
    if classname.find("Conv") != -1:
        m.weight.data.normal_(mean, std)


def get_padding(kernel_size, dilation=1):
    return int((kernel_size*dilation - dilation)/2)