Skip to content

conv1d

Author: Heli Qi Affiliation: NAIST Date: 2022.09

Conv1dPostnet

Bases: Module

The Conv1d postnet for TTS. Usually used after the Transformer TTS decoder. This prenet is made up of only Conv1d blocks each of which contains: 1. a Conv1d layer 2. a BatchNorm1d layer 3. an activation function 4. a Dropout layer.

Reference

Neural Speech Synthesis with Transformer Network https://ojs.aaai.org/index.php/AAAI/article/view/4642/4520

Source code in speechain/module/postnet/conv1d.py 
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
class Conv1dPostnet(Module):
    """
    The Conv1d postnet for TTS. Usually used after the Transformer TTS decoder.
    This prenet is made up of only Conv1d blocks each of which contains:
     1. a Conv1d layer
     2. a BatchNorm1d layer
     3. an activation function
     4. a Dropout layer.

    Reference:
        Neural Speech Synthesis with Transformer Network
        https://ojs.aaai.org/index.php/AAAI/article/view/4642/4520
    """

    def module_init(
        self,
        feat_dim: int = None,
        conv_dims: int or List[int] = [512, 512, 512, 512, 0],
        conv_kernel: int = 5,
        conv_stride: int = 1,
        conv_padding_mode: str = "same",
        conv_batchnorm: bool = True,
        conv_activation: str = "Tanh",
        conv_dropout: float or List[float] = None,
        zero_centered: bool = False,
    ):
        """

        Args:
            feat_dim: int
                The dimension of input acoustic feature tensors.
                Used for calculating the in_features of the first Linear layer.
            conv_dims: List[int] or int
                The values of out_channels of each Conv1d layer.
                If a list of integers is given, multiple Conv1d layers will be initialized.
                If an integer is given, there will be only one Conv1d layer
                -1: same size as the previous convolutional layer's dim
                0: same size as the input feat_dim
            conv_kernel: int
                The value of kernel_size of all Conv1d layers.
            conv_stride: int
                The value of stride of all Conv1d layers.
            conv_padding_mode: str
                The padding mode of convolutional layers. Must be one of ['valid', 'full', 'same', 'causal'].
            conv_batchnorm: bool
                Whether a BatchNorm1d layer is added right after a Conv1d layer
            conv_activation: str
                The type of the activation function after all Conv1d layers.
                None means no activation function is needed.
            conv_dropout: float or List[float]
                The values of p rate of the Dropout layer after each Linear layer.
            zero_centered: bool
                Whether the output of this module is centered at 0.
                If the specified activation function changes the centroid of the output distribution, e.g. ReLU and
                LeakyReLU, the activation function won't be attached to the final Linear layer if zer_centered is set
                to True.

        """
        # Convolution arguments checking
        assert isinstance(
            conv_dims, (List, int)
        ), "The dimensions of convolutional layers must be given as a list of integers or an integer!"
        assert isinstance(
            conv_kernel, int
        ), "The sizes of convolutional kernels must be given as an integer!"
        assert isinstance(
            conv_stride, int
        ), "The lengths of convolutional strides must be given as an integer!"
        if conv_dropout is not None:
            assert isinstance(
                conv_dropout, (List, float)
            ), "The dropout rates of convolutional layers must be given as a list of integers or an integer!"

        # input_size initialization
        if self.input_size is not None:
            feat_dim = self.input_size
        else:
            assert feat_dim is not None

        # register convolution arguments
        self.conv_dims = conv_dims if isinstance(conv_dims, List) else [conv_dims]
        self.conv_kernel = conv_kernel
        self.conv_stride = conv_stride
        self.conv_padding_mode = conv_padding_mode
        self.conv_dropout = conv_dropout

        # Conv1d layers initialization
        _prev_dim = feat_dim
        _tmp_conv = []
        for i in range(len(self.conv_dims)):
            # 0 means go back to the input feat_dim
            if self.conv_dims[i] == 0:
                self.conv_dims[i] = feat_dim
            # -1 means equal to the previous layer
            elif self.conv_dims[i] == -1:
                self.conv_dims[i] = self.conv_dims[i - 1]
            _tmp_conv.append(
                # don't include bias in the convolutional layer if it is followed by a batchnorm layer
                # reference: https://stackoverflow.com/questions/46256747/can-not-use-both-bias-and-batch-normalization-in-convolution-layers
                Conv1dEv(
                    in_channels=_prev_dim,
                    out_channels=self.conv_dims[i],
                    kernel_size=self.conv_kernel,
                    stride=self.conv_stride,
                    padding_mode=self.conv_padding_mode,
                    bias=not conv_batchnorm,
                )
            )
            # BatchNorm is better to be placed before activation
            # reference: https://stackoverflow.com/questions/39691902/ordering-of-batch-normalization-and-dropout
            if conv_batchnorm:
                _tmp_conv.append(torch.nn.BatchNorm1d(self.conv_dims[i]))
            if conv_activation is not None:
                # no 'ReLU'-series activation is added for the last layer if zero_centered is specified
                if i != len(self.conv_dims) - 1 or not (
                    zero_centered and "ReLU" in conv_activation
                ):
                    _tmp_conv.append(getattr(torch.nn, conv_activation)())
            if conv_dropout is not None:
                _tmp_conv.append(
                    torch.nn.Dropout(
                        p=(
                            self.conv_dropout
                            if not isinstance(self.conv_dropout, List)
                            else self.conv_dropout[i]
                        )
                    )
                )
            _prev_dim = conv_dims[i]
        self.conv = torch.nn.Sequential(*_tmp_conv)
        self.output_size = _prev_dim

    def forward(self, feat: torch.Tensor, feat_len: torch.Tensor):
        """

        Args:
            feat: (batch, feat_maxlen, feat_dim)
                The input feature tensors.
            feat_len: (batch,)
                The length of each feature tensor.

        Returns: feat, feat_len
            The embedded feature vectors with their lengths.

        """
        # forward the convolutional layers
        # (batch, feat_maxlen, feat_dim) -> (batch, feat_dim, feat_maxlen)
        feat = feat.transpose(1, 2)
        # (batch, feat_dim, feat_maxlen) -> (batch, conv_dim, feat_maxlen)
        feat = self.conv(feat)
        # (batch, conv_dim, feat_maxlen) -> (batch, feat_maxlen, conv_dim)
        return feat.transpose(1, 2)

forward(feat, feat_len)

Parameters:

Name Type Description Default
feat Tensor

(batch, feat_maxlen, feat_dim) The input feature tensors.

 required
feat_len Tensor

(batch,) The length of each feature tensor.

 required

feat, feat_len

Type Description

The embedded feature vectors with their lengths.
Source code in speechain/module/postnet/conv1d.py 
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
def forward(self, feat: torch.Tensor, feat_len: torch.Tensor):
    """

    Args:
        feat: (batch, feat_maxlen, feat_dim)
            The input feature tensors.
        feat_len: (batch,)
            The length of each feature tensor.

    Returns: feat, feat_len
        The embedded feature vectors with their lengths.

    """
    # forward the convolutional layers
    # (batch, feat_maxlen, feat_dim) -> (batch, feat_dim, feat_maxlen)
    feat = feat.transpose(1, 2)
    # (batch, feat_dim, feat_maxlen) -> (batch, conv_dim, feat_maxlen)
    feat = self.conv(feat)
    # (batch, conv_dim, feat_maxlen) -> (batch, feat_maxlen, conv_dim)
    return feat.transpose(1, 2)

module_init(feat_dim=None, conv_dims=[512, 512, 512, 512, 0], conv_kernel=5, conv_stride=1, conv_padding_mode='same', conv_batchnorm=True, conv_activation='Tanh', conv_dropout=None, zero_centered=False)

Parameters:

Name Type Description Default
feat_dim int

int The dimension of input acoustic feature tensors. Used for calculating the in_features of the first Linear layer.

 None
conv_dims int or List[int]

List[int] or int The values of out_channels of each Conv1d layer. If a list of integers is given, multiple Conv1d layers will be initialized. If an integer is given, there will be only one Conv1d layer -1: same size as the previous convolutional layer's dim 0: same size as the input feat_dim

 [512, 512, 512, 512, 0]
conv_kernel int

int The value of kernel_size of all Conv1d layers.

 5
conv_stride int

int The value of stride of all Conv1d layers.

 1
conv_padding_mode str

str The padding mode of convolutional layers. Must be one of ['valid', 'full', 'same', 'causal'].

 'same'
conv_batchnorm bool

bool Whether a BatchNorm1d layer is added right after a Conv1d layer

 True
conv_activation str

str The type of the activation function after all Conv1d layers. None means no activation function is needed.

 'Tanh'
conv_dropout float or List[float]

float or List[float] The values of p rate of the Dropout layer after each Linear layer.

 None
zero_centered bool

bool Whether the output of this module is centered at 0. If the specified activation function changes the centroid of the output distribution, e.g. ReLU and LeakyReLU, the activation function won't be attached to the final Linear layer if zer_centered is set to True.

 False
Source code in speechain/module/postnet/conv1d.py 
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
def module_init(
    self,
    feat_dim: int = None,
    conv_dims: int or List[int] = [512, 512, 512, 512, 0],
    conv_kernel: int = 5,
    conv_stride: int = 1,
    conv_padding_mode: str = "same",
    conv_batchnorm: bool = True,
    conv_activation: str = "Tanh",
    conv_dropout: float or List[float] = None,
    zero_centered: bool = False,
):
    """

    Args:
        feat_dim: int
            The dimension of input acoustic feature tensors.
            Used for calculating the in_features of the first Linear layer.
        conv_dims: List[int] or int
            The values of out_channels of each Conv1d layer.
            If a list of integers is given, multiple Conv1d layers will be initialized.
            If an integer is given, there will be only one Conv1d layer
            -1: same size as the previous convolutional layer's dim
            0: same size as the input feat_dim
        conv_kernel: int
            The value of kernel_size of all Conv1d layers.
        conv_stride: int
            The value of stride of all Conv1d layers.
        conv_padding_mode: str
            The padding mode of convolutional layers. Must be one of ['valid', 'full', 'same', 'causal'].
        conv_batchnorm: bool
            Whether a BatchNorm1d layer is added right after a Conv1d layer
        conv_activation: str
            The type of the activation function after all Conv1d layers.
            None means no activation function is needed.
        conv_dropout: float or List[float]
            The values of p rate of the Dropout layer after each Linear layer.
        zero_centered: bool
            Whether the output of this module is centered at 0.
            If the specified activation function changes the centroid of the output distribution, e.g. ReLU and
            LeakyReLU, the activation function won't be attached to the final Linear layer if zer_centered is set
            to True.

    """
    # Convolution arguments checking
    assert isinstance(
        conv_dims, (List, int)
    ), "The dimensions of convolutional layers must be given as a list of integers or an integer!"
    assert isinstance(
        conv_kernel, int
    ), "The sizes of convolutional kernels must be given as an integer!"
    assert isinstance(
        conv_stride, int
    ), "The lengths of convolutional strides must be given as an integer!"
    if conv_dropout is not None:
        assert isinstance(
            conv_dropout, (List, float)
        ), "The dropout rates of convolutional layers must be given as a list of integers or an integer!"

    # input_size initialization
    if self.input_size is not None:
        feat_dim = self.input_size
    else:
        assert feat_dim is not None

    # register convolution arguments
    self.conv_dims = conv_dims if isinstance(conv_dims, List) else [conv_dims]
    self.conv_kernel = conv_kernel
    self.conv_stride = conv_stride
    self.conv_padding_mode = conv_padding_mode
    self.conv_dropout = conv_dropout

    # Conv1d layers initialization
    _prev_dim = feat_dim
    _tmp_conv = []
    for i in range(len(self.conv_dims)):
        # 0 means go back to the input feat_dim
        if self.conv_dims[i] == 0:
            self.conv_dims[i] = feat_dim
        # -1 means equal to the previous layer
        elif self.conv_dims[i] == -1:
            self.conv_dims[i] = self.conv_dims[i - 1]
        _tmp_conv.append(
            # don't include bias in the convolutional layer if it is followed by a batchnorm layer
            # reference: https://stackoverflow.com/questions/46256747/can-not-use-both-bias-and-batch-normalization-in-convolution-layers
            Conv1dEv(
                in_channels=_prev_dim,
                out_channels=self.conv_dims[i],
                kernel_size=self.conv_kernel,
                stride=self.conv_stride,
                padding_mode=self.conv_padding_mode,
                bias=not conv_batchnorm,
            )
        )
        # BatchNorm is better to be placed before activation
        # reference: https://stackoverflow.com/questions/39691902/ordering-of-batch-normalization-and-dropout
        if conv_batchnorm:
            _tmp_conv.append(torch.nn.BatchNorm1d(self.conv_dims[i]))
        if conv_activation is not None:
            # no 'ReLU'-series activation is added for the last layer if zero_centered is specified
            if i != len(self.conv_dims) - 1 or not (
                zero_centered and "ReLU" in conv_activation
            ):
                _tmp_conv.append(getattr(torch.nn, conv_activation)())
        if conv_dropout is not None:
            _tmp_conv.append(
                torch.nn.Dropout(
                    p=(
                        self.conv_dropout
                        if not isinstance(self.conv_dropout, List)
                        else self.conv_dropout[i]
                    )
                )
            )
        _prev_dim = conv_dims[i]
    self.conv = torch.nn.Sequential(*_tmp_conv)
    self.output_size = _prev_dim