mirror of
https://github.com/index-tts/index-tts.git
synced 2025-11-28 18:30:25 +08:00
72c09ec0b7* indextts2 * update lfs for audio files --------- Co-authored-by: wangyining02 <wangyining02@bilibili.com>
46 lines
1.2 KiB
Python
46 lines
1.2 KiB
Python
# Copyright (c) 2023 Amphion.
|
|
#
|
|
# This source code is licensed under the MIT license found in the
|
|
# LICENSE file in the root directory of this source tree.
|
|
# This source file is copied from https://github.com/facebookresearch/encodec
|
|
|
|
# Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
# All rights reserved.
|
|
#
|
|
# This source code is licensed under the license found in the
|
|
# LICENSE file in the root directory of this source tree.
|
|
|
|
"""LSTM layers module."""
|
|
|
|
from torch import nn
|
|
|
|
|
|
class SLSTM(nn.Module):
|
|
"""
|
|
LSTM without worrying about the hidden state, nor the layout of the data.
|
|
Expects input as convolutional layout.
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
dimension: int,
|
|
num_layers: int = 2,
|
|
skip: bool = True,
|
|
bidirectional: bool = False,
|
|
):
|
|
super().__init__()
|
|
self.bidirectional = bidirectional
|
|
self.skip = skip
|
|
self.lstm = nn.LSTM(
|
|
dimension, dimension, num_layers, bidirectional=bidirectional
|
|
)
|
|
|
|
def forward(self, x):
|
|
x = x.permute(2, 0, 1)
|
|
y, _ = self.lstm(x)
|
|
if self.bidirectional:
|
|
x = x.repeat(1, 1, 2)
|
|
if self.skip:
|
|
y = y + x
|
|
y = y.permute(1, 2, 0)
|
|
return y
|