def _loss()

in src/losses.py [0:0]

• 17 lines of code
• 2 McCabe index (conditional complexity)

    def _loss(self, data, target):
        '''
        :param data: predicted wave signals in a B x channels x T tensor
        :param target: target wave signals in a B x channels x T tensor
        :return: a scalar loss value
        '''
        data, target = self._transform(data).view(-1, 2), self._transform(target).view(-1, 2)
        # ignore low energy components for numerical stability
        target_energy = th.sum(th.abs(target), dim=-1)
        pred_energy = th.sum(th.abs(data.detach()), dim=-1)
        target_mask = target_energy > self.ignore_below * th.mean(target_energy)
        pred_mask = pred_energy > self.ignore_below * th.mean(target_energy)
        indices = th.nonzero(target_mask * pred_mask).view(-1)
        data, target = th.index_select(data, 0, indices), th.index_select(target, 0, indices)
        # compute actual phase loss in angular space
        data_angles, target_angles = th.atan2(data[:, 0], data[:, 1]), th.atan2(target[:, 0], target[:, 1])
        loss = th.abs(data_angles - target_angles)
        # positive + negative values in left part of coordinate system cause angles > pi
        # => 2pi -> 0, 3/4pi -> 1/2pi, ... (triangle function over [0, 2pi] with peak at pi)
        loss = np.pi - th.abs(loss - np.pi)
        return th.mean(loss)