| | |
| | | mask = torch.reshape(mask, (b, -1, 1)) |
| | | if mask_shfit_chunk is not None: |
| | | mask = mask * mask_shfit_chunk |
| | | inputs = inputs * mask |
| | | |
| | | inputs = inputs * mask |
| | | x = inputs.transpose(1, 2) |
| | | x = self.pad_fn(x) |
| | | x = self.fsmn_block(x) |
| | | x = x.transpose(1, 2) |
| | | x += inputs |
| | | x = self.dropout(x) |
| | | return x * mask |
| | | if mask is not None: |
| | | x = x * mask |
| | | return x |
| | | |
| | | def forward_qkv(self, x): |
| | | """Transform query, key and value. |
| | |
| | | # print("in fsmn, cache is None, x", x.size()) |
| | | |
| | | x = self.pad_fn(x) |
| | | if not self.training and t <= 1: |
| | | if not self.training: |
| | | cache = x |
| | | else: |
| | | # print("in fsmn, cache is not None, x", x.size()) |
| | |
| | | # if t < self.kernel_size: |
| | | # x = self.pad_fn(x) |
| | | x = torch.cat((cache[:, :, 1:], x), dim=2) |
| | | x = x[:, :, -self.kernel_size:] |
| | | x = x[:, :, -(self.kernel_size+t-1):] |
| | | # print("in fsmn, cache is not None, x_cat", x.size()) |
| | | cache = x |
| | | x = self.fsmn_block(x) |