Add modules

configs/.gitignore

@@ -0,0 +1 @@
+*.yaml

configs/s2.json

@@ -0,0 +1,91 @@
+{
+  "train": {
+    "log_interval": 100,
+    "eval_interval": 500,
+    "seed": 1234,
+    "epochs": 100,
+    "learning_rate": 0.0001,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 32,
+    "fp16_run": true,
+    "lr_decay": 0.999875,
+    "segment_size": 20480,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "text_low_lr_rate": 0.4, 
+    "grad_ckpt": false
+  },
+  "data": {
+    "max_wav_value": 32768.0,
+    "sampling_rate": 32000,
+    "filter_length": 2048,
+    "hop_length": 640,
+    "win_length": 2048,
+    "n_mel_channels": 128,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 300,
+    "cleaned_text": true
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      10,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      8,
+      2,
+      2
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 512,
+    "semantic_frame_rate": "25hz",
+    "freeze_quantizer": true
+  },
+  "s2_ckpt_dir": "logs/s2/big2k1",
+  "content_module": "cnhubert"
+}

configs/s2v2Pro.json

@@ -0,0 +1,91 @@
+{
+  "train": {
+    "log_interval": 100,
+    "eval_interval": 500,
+    "seed": 1234,
+    "epochs": 100,
+    "learning_rate": 0.0001,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 32,
+    "fp16_run": true,
+    "lr_decay": 0.999875,
+    "segment_size": 20480,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "text_low_lr_rate": 0.4, 
+    "grad_ckpt": false
+  },
+  "data": {
+    "max_wav_value": 32768.0,
+    "sampling_rate": 32000,
+    "filter_length": 2048,
+    "hop_length": 640,
+    "win_length": 2048,
+    "n_mel_channels": 128,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 300,
+    "cleaned_text": true
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.0,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      10,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      8,
+      2,
+      2
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 1024,
+    "semantic_frame_rate": "25hz",
+    "freeze_quantizer": true
+  },
+  "s2_ckpt_dir": "logs/s2/big2k1",
+  "content_module": "cnhubert"
+}

configs/s2v2ProPlus.json

@@ -0,0 +1,91 @@
+{
+  "train": {
+    "log_interval": 100,
+    "eval_interval": 500,
+    "seed": 1234,
+    "epochs": 100,
+    "learning_rate": 0.0001,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 32,
+    "fp16_run": true,
+    "lr_decay": 0.999875,
+    "segment_size": 20480,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "text_low_lr_rate": 0.4, 
+    "grad_ckpt": false
+  },
+  "data": {
+    "max_wav_value": 32768.0,
+    "sampling_rate": 32000,
+    "filter_length": 2048,
+    "hop_length": 640,
+    "win_length": 2048,
+    "n_mel_channels": 128,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 300,
+    "cleaned_text": true
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.0,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      10,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 768,
+    "upsample_kernel_sizes": [
+      20,
+      16,
+      8,
+      2,
+      2
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 1024,
+    "semantic_frame_rate": "25hz",
+    "freeze_quantizer": true
+  },
+  "s2_ckpt_dir": "logs/s2/big2k1",
+  "content_module": "cnhubert"
+}

eres2net/ERes2Net.py

@@ -0,0 +1,264 @@
+# Copyright 3D-Speaker (https://github.com/alibaba-damo-academy/3D-Speaker). All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""
+Res2Net implementation is adapted from https://github.com/wenet-e2e/wespeaker.
+ERes2Net incorporates both local and global feature fusion techniques to improve the performance.
+The local feature fusion (LFF) fuses the features within one single residual block to extract the local signal.
+The global feature fusion (GFF) takes acoustic features of different scales as input to aggregate global signal.
+"""
+
+import torch
+import math
+import torch.nn as nn
+import torch.nn.functional as F
+import pooling_layers as pooling_layers
+from fusion import AFF
+
+
+class ReLU(nn.Hardtanh):
+    def __init__(self, inplace=False):
+        super(ReLU, self).__init__(0, 20, inplace)
+
+    def __repr__(self):
+        inplace_str = "inplace" if self.inplace else ""
+        return self.__class__.__name__ + " (" + inplace_str + ")"
+
+
+class BasicBlockERes2Net(nn.Module):
+    expansion = 2
+
+    def __init__(self, in_planes, planes, stride=1, baseWidth=32, scale=2):
+        super(BasicBlockERes2Net, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+
+        convs = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = sp + spx[i]
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class BasicBlockERes2Net_diff_AFF(nn.Module):
+    expansion = 2
+
+    def __init__(self, in_planes, planes, stride=1, baseWidth=32, scale=2):
+        super(BasicBlockERes2Net_diff_AFF, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+
+        convs = []
+        fuse_models = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        for j in range(self.nums - 1):
+            fuse_models.append(AFF(channels=width))
+
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.fuse_models = nn.ModuleList(fuse_models)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = self.fuse_models[i - 1](sp, spx[i])
+
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class ERes2Net(nn.Module):
+    def __init__(
+        self,
+        block=BasicBlockERes2Net,
+        block_fuse=BasicBlockERes2Net_diff_AFF,
+        num_blocks=[3, 4, 6, 3],
+        m_channels=32,
+        feat_dim=80,
+        embedding_size=192,
+        pooling_func="TSTP",
+        two_emb_layer=False,
+    ):
+        super(ERes2Net, self).__init__()
+        self.in_planes = m_channels
+        self.feat_dim = feat_dim
+        self.embedding_size = embedding_size
+        self.stats_dim = int(feat_dim / 8) * m_channels * 8
+        self.two_emb_layer = two_emb_layer
+
+        self.conv1 = nn.Conv2d(1, m_channels, kernel_size=3, stride=1, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(m_channels)
+        self.layer1 = self._make_layer(block, m_channels, num_blocks[0], stride=1)
+        self.layer2 = self._make_layer(block, m_channels * 2, num_blocks[1], stride=2)
+        self.layer3 = self._make_layer(block_fuse, m_channels * 4, num_blocks[2], stride=2)
+        self.layer4 = self._make_layer(block_fuse, m_channels * 8, num_blocks[3], stride=2)
+
+        # Downsampling module for each layer
+        self.layer1_downsample = nn.Conv2d(
+            m_channels * 2, m_channels * 4, kernel_size=3, stride=2, padding=1, bias=False
+        )
+        self.layer2_downsample = nn.Conv2d(
+            m_channels * 4, m_channels * 8, kernel_size=3, padding=1, stride=2, bias=False
+        )
+        self.layer3_downsample = nn.Conv2d(
+            m_channels * 8, m_channels * 16, kernel_size=3, padding=1, stride=2, bias=False
+        )
+
+        # Bottom-up fusion module
+        self.fuse_mode12 = AFF(channels=m_channels * 4)
+        self.fuse_mode123 = AFF(channels=m_channels * 8)
+        self.fuse_mode1234 = AFF(channels=m_channels * 16)
+
+        self.n_stats = 1 if pooling_func == "TAP" or pooling_func == "TSDP" else 2
+        self.pool = getattr(pooling_layers, pooling_func)(in_dim=self.stats_dim * block.expansion)
+        self.seg_1 = nn.Linear(self.stats_dim * block.expansion * self.n_stats, embedding_size)
+        if self.two_emb_layer:
+            self.seg_bn_1 = nn.BatchNorm1d(embedding_size, affine=False)
+            self.seg_2 = nn.Linear(embedding_size, embedding_size)
+        else:
+            self.seg_bn_1 = nn.Identity()
+            self.seg_2 = nn.Identity()
+
+    def _make_layer(self, block, planes, num_blocks, stride):
+        strides = [stride] + [1] * (num_blocks - 1)
+        layers = []
+        for stride in strides:
+            layers.append(block(self.in_planes, planes, stride))
+            self.in_planes = planes * block.expansion
+        return nn.Sequential(*layers)
+
+    def forward(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out1_downsample = self.layer1_downsample(out1)
+        fuse_out12 = self.fuse_mode12(out2, out1_downsample)
+        out3 = self.layer3(out2)
+        fuse_out12_downsample = self.layer2_downsample(fuse_out12)
+        fuse_out123 = self.fuse_mode123(out3, fuse_out12_downsample)
+        out4 = self.layer4(out3)
+        fuse_out123_downsample = self.layer3_downsample(fuse_out123)
+        fuse_out1234 = self.fuse_mode1234(out4, fuse_out123_downsample)
+        stats = self.pool(fuse_out1234)
+
+        embed_a = self.seg_1(stats)
+        if self.two_emb_layer:
+            out = F.relu(embed_a)
+            out = self.seg_bn_1(out)
+            embed_b = self.seg_2(out)
+            return embed_b
+        else:
+            return embed_a
+
+    def forward3(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out1_downsample = self.layer1_downsample(out1)
+        fuse_out12 = self.fuse_mode12(out2, out1_downsample)
+        out3 = self.layer3(out2)
+        fuse_out12_downsample = self.layer2_downsample(fuse_out12)
+        fuse_out123 = self.fuse_mode123(out3, fuse_out12_downsample)
+        out4 = self.layer4(out3)
+        fuse_out123_downsample = self.layer3_downsample(fuse_out123)
+        fuse_out1234 = self.fuse_mode1234(out4, fuse_out123_downsample).flatten(start_dim=1, end_dim=2).mean(-1)
+        return fuse_out1234
+
+
+if __name__ == "__main__":
+    x = torch.zeros(10, 300, 80)
+    model = ERes2Net(feat_dim=80, embedding_size=192, pooling_func="TSTP")
+    model.eval()
+    out = model(x)
+    print(out.shape)  # torch.Size([10, 192])
+
+    num_params = sum(param.numel() for param in model.parameters())
+    print("{} M".format(num_params / 1e6))  # 6.61M

eres2net/ERes2NetV2.py

@@ -0,0 +1,272 @@
+# Copyright 3D-Speaker (https://github.com/alibaba-damo-academy/3D-Speaker). All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""
+To further improve the short-duration feature extraction capability of ERes2Net, we expand the channel dimension
+within each stage. However, this modification also increases the number of model parameters and computational complexity.
+To alleviate this problem, we propose an improved ERes2NetV2 by pruning redundant structures, ultimately reducing
+both the model parameters and its computational cost.
+"""
+
+import torch
+import math
+import torch.nn as nn
+import torch.nn.functional as F
+import pooling_layers as pooling_layers
+from fusion import AFF
+
+
+class ReLU(nn.Hardtanh):
+    def __init__(self, inplace=False):
+        super(ReLU, self).__init__(0, 20, inplace)
+
+    def __repr__(self):
+        inplace_str = "inplace" if self.inplace else ""
+        return self.__class__.__name__ + " (" + inplace_str + ")"
+
+
+class BasicBlockERes2NetV2(nn.Module):
+    def __init__(self, in_planes, planes, stride=1, baseWidth=26, scale=2, expansion=2):
+        super(BasicBlockERes2NetV2, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+        self.expansion = expansion
+
+        convs = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = sp + spx[i]
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class BasicBlockERes2NetV2AFF(nn.Module):
+    def __init__(self, in_planes, planes, stride=1, baseWidth=26, scale=2, expansion=2):
+        super(BasicBlockERes2NetV2AFF, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+        self.expansion = expansion
+
+        convs = []
+        fuse_models = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        for j in range(self.nums - 1):
+            fuse_models.append(AFF(channels=width, r=4))
+
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.fuse_models = nn.ModuleList(fuse_models)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = self.fuse_models[i - 1](sp, spx[i])
+
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class ERes2NetV2(nn.Module):
+    def __init__(
+        self,
+        block=BasicBlockERes2NetV2,
+        block_fuse=BasicBlockERes2NetV2AFF,
+        num_blocks=[3, 4, 6, 3],
+        m_channels=64,
+        feat_dim=80,
+        embedding_size=192,
+        baseWidth=26,
+        scale=2,
+        expansion=2,
+        pooling_func="TSTP",
+        two_emb_layer=False,
+    ):
+        super(ERes2NetV2, self).__init__()
+        self.in_planes = m_channels
+        self.feat_dim = feat_dim
+        self.embedding_size = embedding_size
+        self.stats_dim = int(feat_dim / 8) * m_channels * 8
+        self.two_emb_layer = two_emb_layer
+        self.baseWidth = baseWidth
+        self.scale = scale
+        self.expansion = expansion
+
+        self.conv1 = nn.Conv2d(1, m_channels, kernel_size=3, stride=1, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(m_channels)
+        self.layer1 = self._make_layer(block, m_channels, num_blocks[0], stride=1)
+        self.layer2 = self._make_layer(block, m_channels * 2, num_blocks[1], stride=2)
+        self.layer3 = self._make_layer(block_fuse, m_channels * 4, num_blocks[2], stride=2)
+        self.layer4 = self._make_layer(block_fuse, m_channels * 8, num_blocks[3], stride=2)
+
+        # Downsampling module
+        self.layer3_ds = nn.Conv2d(
+            m_channels * 4 * self.expansion,
+            m_channels * 8 * self.expansion,
+            kernel_size=3,
+            padding=1,
+            stride=2,
+            bias=False,
+        )
+
+        # Bottom-up fusion module
+        self.fuse34 = AFF(channels=m_channels * 8 * self.expansion, r=4)
+
+        self.n_stats = 1 if pooling_func == "TAP" or pooling_func == "TSDP" else 2
+        self.pool = getattr(pooling_layers, pooling_func)(in_dim=self.stats_dim * self.expansion)
+        self.seg_1 = nn.Linear(self.stats_dim * self.expansion * self.n_stats, embedding_size)
+        if self.two_emb_layer:
+            self.seg_bn_1 = nn.BatchNorm1d(embedding_size, affine=False)
+            self.seg_2 = nn.Linear(embedding_size, embedding_size)
+        else:
+            self.seg_bn_1 = nn.Identity()
+            self.seg_2 = nn.Identity()
+
+    def _make_layer(self, block, planes, num_blocks, stride):
+        strides = [stride] + [1] * (num_blocks - 1)
+        layers = []
+        for stride in strides:
+            layers.append(
+                block(
+                    self.in_planes, planes, stride, baseWidth=self.baseWidth, scale=self.scale, expansion=self.expansion
+                )
+            )
+            self.in_planes = planes * self.expansion
+        return nn.Sequential(*layers)
+
+    def forward(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out3 = self.layer3(out2)
+        out4 = self.layer4(out3)
+        out3_ds = self.layer3_ds(out3)
+        fuse_out34 = self.fuse34(out4, out3_ds)
+        stats = self.pool(fuse_out34)
+
+        embed_a = self.seg_1(stats)
+        if self.two_emb_layer:
+            out = F.relu(embed_a)
+            out = self.seg_bn_1(out)
+            embed_b = self.seg_2(out)
+            return embed_b
+        else:
+            return embed_a
+
+    def forward3(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out3 = self.layer3(out2)
+        out4 = self.layer4(out3)
+        out3_ds = self.layer3_ds(out3)
+        fuse_out34 = self.fuse34(out4, out3_ds)
+        # print(111111111,fuse_out34.shape)#111111111 torch.Size([16, 2048, 10, 72])
+        return fuse_out34.flatten(start_dim=1, end_dim=2).mean(-1)
+        # stats = self.pool(fuse_out34)
+        #
+        # embed_a = self.seg_1(stats)
+        # if self.two_emb_layer:
+        #     out = F.relu(embed_a)
+        #     out = self.seg_bn_1(out)
+        #     embed_b = self.seg_2(out)
+        #     return embed_b
+        # else:
+        #     return embed_a
+
+
+if __name__ == "__main__":
+    x = torch.randn(1, 300, 80)
+    model = ERes2NetV2(feat_dim=80, embedding_size=192, m_channels=64, baseWidth=26, scale=2, expansion=2)
+    model.eval()
+    y = model(x)
+    print(y.size())
+    macs, num_params = profile(model, inputs=(x,))
+    print("Params: {} M".format(num_params / 1e6))  # 17.86 M
+    print("MACs: {} G".format(macs / 1e9))  # 12.69 G

eres2net/ERes2Net_huge.py

@@ -0,0 +1,289 @@
+# Copyright 3D-Speaker (https://github.com/alibaba-damo-academy/3D-Speaker). All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""Res2Net implementation is adapted from https://github.com/wenet-e2e/wespeaker.
+ERes2Net incorporates both local and global feature fusion techniques to improve the performance.
+The local feature fusion (LFF) fuses the features within one single residual block to extract the local signal.
+The global feature fusion (GFF) takes acoustic features of different scales as input to aggregate global signal.
+ERes2Net-huge is an upgraded version of ERes2Net that uses a larger number of parameters to achieve better
+recognition performance. Parameters expansion, baseWidth, and scale can be modified to obtain optimal performance.
+"""
+
+import torch
+import math
+import torch.nn as nn
+import torch.nn.functional as F
+import pooling_layers as pooling_layers
+from fusion import AFF
+
+
+class ReLU(nn.Hardtanh):
+    def __init__(self, inplace=False):
+        super(ReLU, self).__init__(0, 20, inplace)
+
+    def __repr__(self):
+        inplace_str = "inplace" if self.inplace else ""
+        return self.__class__.__name__ + " (" + inplace_str + ")"
+
+
+class BasicBlockERes2Net(nn.Module):
+    expansion = 4
+
+    def __init__(self, in_planes, planes, stride=1, baseWidth=24, scale=3):
+        super(BasicBlockERes2Net, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+
+        convs = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = sp + spx[i]
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class BasicBlockERes2Net_diff_AFF(nn.Module):
+    expansion = 4
+
+    def __init__(self, in_planes, planes, stride=1, baseWidth=24, scale=3):
+        super(BasicBlockERes2Net_diff_AFF, self).__init__()
+        width = int(math.floor(planes * (baseWidth / 64.0)))
+        self.conv1 = nn.Conv2d(in_planes, width * scale, kernel_size=1, stride=stride, bias=False)
+        self.bn1 = nn.BatchNorm2d(width * scale)
+        self.nums = scale
+
+        convs = []
+        fuse_models = []
+        bns = []
+        for i in range(self.nums):
+            convs.append(nn.Conv2d(width, width, kernel_size=3, padding=1, bias=False))
+            bns.append(nn.BatchNorm2d(width))
+        for j in range(self.nums - 1):
+            fuse_models.append(AFF(channels=width))
+
+        self.convs = nn.ModuleList(convs)
+        self.bns = nn.ModuleList(bns)
+        self.fuse_models = nn.ModuleList(fuse_models)
+        self.relu = ReLU(inplace=True)
+
+        self.conv3 = nn.Conv2d(width * scale, planes * self.expansion, kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+        self.shortcut = nn.Sequential()
+        if stride != 1 or in_planes != self.expansion * planes:
+            self.shortcut = nn.Sequential(
+                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
+                nn.BatchNorm2d(self.expansion * planes),
+            )
+        self.stride = stride
+        self.width = width
+        self.scale = scale
+
+    def forward(self, x):
+        residual = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+        spx = torch.split(out, self.width, 1)
+        for i in range(self.nums):
+            if i == 0:
+                sp = spx[i]
+            else:
+                sp = self.fuse_models[i - 1](sp, spx[i])
+
+            sp = self.convs[i](sp)
+            sp = self.relu(self.bns[i](sp))
+            if i == 0:
+                out = sp
+            else:
+                out = torch.cat((out, sp), 1)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        residual = self.shortcut(x)
+        out += residual
+        out = self.relu(out)
+
+        return out
+
+
+class ERes2Net(nn.Module):
+    def __init__(
+        self,
+        block=BasicBlockERes2Net,
+        block_fuse=BasicBlockERes2Net_diff_AFF,
+        num_blocks=[3, 4, 6, 3],
+        m_channels=64,
+        feat_dim=80,
+        embedding_size=192,
+        pooling_func="TSTP",
+        two_emb_layer=False,
+    ):
+        super(ERes2Net, self).__init__()
+        self.in_planes = m_channels
+        self.feat_dim = feat_dim
+        self.embedding_size = embedding_size
+        self.stats_dim = int(feat_dim / 8) * m_channels * 8
+        self.two_emb_layer = two_emb_layer
+
+        self.conv1 = nn.Conv2d(1, m_channels, kernel_size=3, stride=1, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(m_channels)
+
+        self.layer1 = self._make_layer(block, m_channels, num_blocks[0], stride=1)
+        self.layer2 = self._make_layer(block, m_channels * 2, num_blocks[1], stride=2)
+        self.layer3 = self._make_layer(block_fuse, m_channels * 4, num_blocks[2], stride=2)
+        self.layer4 = self._make_layer(block_fuse, m_channels * 8, num_blocks[3], stride=2)
+
+        self.layer1_downsample = nn.Conv2d(
+            m_channels * 4, m_channels * 8, kernel_size=3, padding=1, stride=2, bias=False
+        )
+        self.layer2_downsample = nn.Conv2d(
+            m_channels * 8, m_channels * 16, kernel_size=3, padding=1, stride=2, bias=False
+        )
+        self.layer3_downsample = nn.Conv2d(
+            m_channels * 16, m_channels * 32, kernel_size=3, padding=1, stride=2, bias=False
+        )
+
+        self.fuse_mode12 = AFF(channels=m_channels * 8)
+        self.fuse_mode123 = AFF(channels=m_channels * 16)
+        self.fuse_mode1234 = AFF(channels=m_channels * 32)
+
+        self.n_stats = 1 if pooling_func == "TAP" or pooling_func == "TSDP" else 2
+        self.pool = getattr(pooling_layers, pooling_func)(in_dim=self.stats_dim * block.expansion)
+        self.seg_1 = nn.Linear(self.stats_dim * block.expansion * self.n_stats, embedding_size)
+        if self.two_emb_layer:
+            self.seg_bn_1 = nn.BatchNorm1d(embedding_size, affine=False)
+            self.seg_2 = nn.Linear(embedding_size, embedding_size)
+        else:
+            self.seg_bn_1 = nn.Identity()
+            self.seg_2 = nn.Identity()
+
+    def _make_layer(self, block, planes, num_blocks, stride):
+        strides = [stride] + [1] * (num_blocks - 1)
+        layers = []
+        for stride in strides:
+            layers.append(block(self.in_planes, planes, stride))
+            self.in_planes = planes * block.expansion
+        return nn.Sequential(*layers)
+
+    def forward(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out1_downsample = self.layer1_downsample(out1)
+        fuse_out12 = self.fuse_mode12(out2, out1_downsample)
+        out3 = self.layer3(out2)
+        fuse_out12_downsample = self.layer2_downsample(fuse_out12)
+        fuse_out123 = self.fuse_mode123(out3, fuse_out12_downsample)
+        out4 = self.layer4(out3)
+        fuse_out123_downsample = self.layer3_downsample(fuse_out123)
+        fuse_out1234 = self.fuse_mode1234(out4, fuse_out123_downsample)
+        stats = self.pool(fuse_out1234)
+
+        embed_a = self.seg_1(stats)
+        if self.two_emb_layer:
+            out = F.relu(embed_a)
+            out = self.seg_bn_1(out)
+            embed_b = self.seg_2(out)
+            return embed_b
+        else:
+            return embed_a
+
+    def forward2(self, x, if_mean):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out1_downsample = self.layer1_downsample(out1)
+        fuse_out12 = self.fuse_mode12(out2, out1_downsample)
+        out3 = self.layer3(out2)
+        fuse_out12_downsample = self.layer2_downsample(fuse_out12)
+        fuse_out123 = self.fuse_mode123(out3, fuse_out12_downsample)
+        out4 = self.layer4(out3)
+        fuse_out123_downsample = self.layer3_downsample(fuse_out123)
+        fuse_out1234 = self.fuse_mode1234(out4, fuse_out123_downsample).flatten(start_dim=1, end_dim=2)  # bs,20480,T
+        if if_mean == False:
+            mean = fuse_out1234[0].transpose(1, 0)  # (T,20480),bs=T
+        else:
+            mean = fuse_out1234.mean(2)  # bs,20480
+        mean_std = torch.cat([mean, torch.zeros_like(mean)], 1)
+        return self.seg_1(mean_std)  # (T,192)
+
+        # stats = self.pool(fuse_out1234)
+        # if self.two_emb_layer:
+        #     out = F.relu(embed_a)
+        #     out = self.seg_bn_1(out)
+        #     embed_b = self.seg_2(out)
+        #     return embed_b
+        # else:
+        #     return embed_a
+
+    def forward3(self, x):
+        x = x.permute(0, 2, 1)  # (B,T,F) => (B,F,T)
+
+        x = x.unsqueeze_(1)
+        out = F.relu(self.bn1(self.conv1(x)))
+        out1 = self.layer1(out)
+        out2 = self.layer2(out1)
+        out1_downsample = self.layer1_downsample(out1)
+        fuse_out12 = self.fuse_mode12(out2, out1_downsample)
+        out3 = self.layer3(out2)
+        fuse_out12_downsample = self.layer2_downsample(fuse_out12)
+        fuse_out123 = self.fuse_mode123(out3, fuse_out12_downsample)
+        out4 = self.layer4(out3)
+        fuse_out123_downsample = self.layer3_downsample(fuse_out123)
+        fuse_out1234 = self.fuse_mode1234(out4, fuse_out123_downsample).flatten(start_dim=1, end_dim=2).mean(-1)
+        return fuse_out1234
+        # print(fuse_out1234.shape)
+        # print(fuse_out1234.flatten(start_dim=1,end_dim=2).shape)
+        # pdb.set_trace()

eres2net/fusion.py

@@ -0,0 +1,27 @@
+# Copyright 3D-Speaker (https://github.com/alibaba-damo-academy/3D-Speaker). All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+
+import torch
+import torch.nn as nn
+
+
+class AFF(nn.Module):
+    def __init__(self, channels=64, r=4):
+        super(AFF, self).__init__()
+        inter_channels = int(channels // r)
+
+        self.local_att = nn.Sequential(
+            nn.Conv2d(channels * 2, inter_channels, kernel_size=1, stride=1, padding=0),
+            nn.BatchNorm2d(inter_channels),
+            nn.SiLU(inplace=True),
+            nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+            nn.BatchNorm2d(channels),
+        )
+
+    def forward(self, x, ds_y):
+        xa = torch.cat((x, ds_y), dim=1)
+        x_att = self.local_att(xa)
+        x_att = 1.0 + torch.tanh(x_att)
+        xo = torch.mul(x, x_att) + torch.mul(ds_y, 2.0 - x_att)
+
+        return xo

eres2net/kaldi.py

@@ -0,0 +1,844 @@
+import math
+from typing import Tuple
+
+import torch
+import torchaudio
+from torch import Tensor
+
+__all__ = [
+    "get_mel_banks",
+    "inverse_mel_scale",
+    "inverse_mel_scale_scalar",
+    "mel_scale",
+    "mel_scale_scalar",
+    "spectrogram",
+    "fbank",
+    "mfcc",
+    "vtln_warp_freq",
+    "vtln_warp_mel_freq",
+]
+
+# numeric_limits<float>::epsilon() 1.1920928955078125e-07
+EPSILON = torch.tensor(torch.finfo(torch.float).eps)
+# 1 milliseconds = 0.001 seconds
+MILLISECONDS_TO_SECONDS = 0.001
+
+# window types
+HAMMING = "hamming"
+HANNING = "hanning"
+POVEY = "povey"
+RECTANGULAR = "rectangular"
+BLACKMAN = "blackman"
+WINDOWS = [HAMMING, HANNING, POVEY, RECTANGULAR, BLACKMAN]
+
+
+def _get_epsilon(device, dtype):
+    return EPSILON.to(device=device, dtype=dtype)
+
+
+def _next_power_of_2(x: int) -> int:
+    r"""Returns the smallest power of 2 that is greater than x"""
+    return 1 if x == 0 else 2 ** (x - 1).bit_length()
+
+
+def _get_strided(waveform: Tensor, window_size: int, window_shift: int, snip_edges: bool) -> Tensor:
+    r"""Given a waveform (1D tensor of size ``num_samples``), it returns a 2D tensor (m, ``window_size``)
+    representing how the window is shifted along the waveform. Each row is a frame.
+
+    Args:
+        waveform (Tensor): Tensor of size ``num_samples``
+        window_size (int): Frame length
+        window_shift (int): Frame shift
+        snip_edges (bool): If True, end effects will be handled by outputting only frames that completely fit
+            in the file, and the number of frames depends on the frame_length.  If False, the number of frames
+            depends only on the frame_shift, and we reflect the data at the ends.
+
+    Returns:
+        Tensor: 2D tensor of size (m, ``window_size``) where each row is a frame
+    """
+    assert waveform.dim() == 1
+    num_samples = waveform.size(0)
+    strides = (window_shift * waveform.stride(0), waveform.stride(0))
+
+    if snip_edges:
+        if num_samples < window_size:
+            return torch.empty((0, 0), dtype=waveform.dtype, device=waveform.device)
+        else:
+            m = 1 + (num_samples - window_size) // window_shift
+    else:
+        reversed_waveform = torch.flip(waveform, [0])
+        m = (num_samples + (window_shift // 2)) // window_shift
+        pad = window_size // 2 - window_shift // 2
+        pad_right = reversed_waveform
+        if pad > 0:
+            # torch.nn.functional.pad returns [2,1,0,1,2] for 'reflect'
+            # but we want [2, 1, 0, 0, 1, 2]
+            pad_left = reversed_waveform[-pad:]
+            waveform = torch.cat((pad_left, waveform, pad_right), dim=0)
+        else:
+            # pad is negative so we want to trim the waveform at the front
+            waveform = torch.cat((waveform[-pad:], pad_right), dim=0)
+
+    sizes = (m, window_size)
+    return waveform.as_strided(sizes, strides)
+
+
+def _feature_window_function(
+    window_type: str,
+    window_size: int,
+    blackman_coeff: float,
+    device: torch.device,
+    dtype: int,
+) -> Tensor:
+    r"""Returns a window function with the given type and size"""
+    if window_type == HANNING:
+        return torch.hann_window(window_size, periodic=False, device=device, dtype=dtype)
+    elif window_type == HAMMING:
+        return torch.hamming_window(window_size, periodic=False, alpha=0.54, beta=0.46, device=device, dtype=dtype)
+    elif window_type == POVEY:
+        # like hanning but goes to zero at edges
+        return torch.hann_window(window_size, periodic=False, device=device, dtype=dtype).pow(0.85)
+    elif window_type == RECTANGULAR:
+        return torch.ones(window_size, device=device, dtype=dtype)
+    elif window_type == BLACKMAN:
+        a = 2 * math.pi / (window_size - 1)
+        window_function = torch.arange(window_size, device=device, dtype=dtype)
+        # can't use torch.blackman_window as they use different coefficients
+        return (
+            blackman_coeff
+            - 0.5 * torch.cos(a * window_function)
+            + (0.5 - blackman_coeff) * torch.cos(2 * a * window_function)
+        ).to(device=device, dtype=dtype)
+    else:
+        raise Exception("Invalid window type " + window_type)
+
+
+def _get_log_energy(strided_input: Tensor, epsilon: Tensor, energy_floor: float) -> Tensor:
+    r"""Returns the log energy of size (m) for a strided_input (m,*)"""
+    device, dtype = strided_input.device, strided_input.dtype
+    log_energy = torch.max(strided_input.pow(2).sum(1), epsilon).log()  # size (m)
+    if energy_floor == 0.0:
+        return log_energy
+    return torch.max(log_energy, torch.tensor(math.log(energy_floor), device=device, dtype=dtype))
+
+
+def _get_waveform_and_window_properties(
+    waveform: Tensor,
+    channel: int,
+    sample_frequency: float,
+    frame_shift: float,
+    frame_length: float,
+    round_to_power_of_two: bool,
+    preemphasis_coefficient: float,
+) -> Tuple[Tensor, int, int, int]:
+    r"""Gets the waveform and window properties"""
+    channel = max(channel, 0)
+    assert channel < waveform.size(0), "Invalid channel {} for size {}".format(channel, waveform.size(0))
+    waveform = waveform[channel, :]  # size (n)
+    window_shift = int(sample_frequency * frame_shift * MILLISECONDS_TO_SECONDS)
+    window_size = int(sample_frequency * frame_length * MILLISECONDS_TO_SECONDS)
+    padded_window_size = _next_power_of_2(window_size) if round_to_power_of_two else window_size
+
+    assert 2 <= window_size <= len(waveform), "choose a window size {} that is [2, {}]".format(
+        window_size, len(waveform)
+    )
+    assert 0 < window_shift, "`window_shift` must be greater than 0"
+    assert padded_window_size % 2 == 0, (
+        "the padded `window_size` must be divisible by two. use `round_to_power_of_two` or change `frame_length`"
+    )
+    assert 0.0 <= preemphasis_coefficient <= 1.0, "`preemphasis_coefficient` must be between [0,1]"
+    assert sample_frequency > 0, "`sample_frequency` must be greater than zero"
+    return waveform, window_shift, window_size, padded_window_size
+
+
+def _get_window(
+    waveform: Tensor,
+    padded_window_size: int,
+    window_size: int,
+    window_shift: int,
+    window_type: str,
+    blackman_coeff: float,
+    snip_edges: bool,
+    raw_energy: bool,
+    energy_floor: float,
+    dither: float,
+    remove_dc_offset: bool,
+    preemphasis_coefficient: float,
+) -> Tuple[Tensor, Tensor]:
+    r"""Gets a window and its log energy
+
+    Returns:
+        (Tensor, Tensor): strided_input of size (m, ``padded_window_size``) and signal_log_energy of size (m)
+    """
+    device, dtype = waveform.device, waveform.dtype
+    epsilon = _get_epsilon(device, dtype)
+
+    # size (m, window_size)
+    strided_input = _get_strided(waveform, window_size, window_shift, snip_edges)
+
+    if dither != 0.0:
+        rand_gauss = torch.randn(strided_input.shape, device=device, dtype=dtype)
+        strided_input = strided_input + rand_gauss * dither
+
+    if remove_dc_offset:
+        # Subtract each row/frame by its mean
+        row_means = torch.mean(strided_input, dim=1).unsqueeze(1)  # size (m, 1)
+        strided_input = strided_input - row_means
+
+    if raw_energy:
+        # Compute the log energy of each row/frame before applying preemphasis and
+        # window function
+        signal_log_energy = _get_log_energy(strided_input, epsilon, energy_floor)  # size (m)
+
+    if preemphasis_coefficient != 0.0:
+        # strided_input[i,j] -= preemphasis_coefficient * strided_input[i, max(0, j-1)] for all i,j
+        offset_strided_input = torch.nn.functional.pad(strided_input.unsqueeze(0), (1, 0), mode="replicate").squeeze(
+            0
+        )  # size (m, window_size + 1)
+        strided_input = strided_input - preemphasis_coefficient * offset_strided_input[:, :-1]
+
+    # Apply window_function to each row/frame
+    window_function = _feature_window_function(window_type, window_size, blackman_coeff, device, dtype).unsqueeze(
+        0
+    )  # size (1, window_size)
+    strided_input = strided_input * window_function  # size (m, window_size)
+
+    # Pad columns with zero until we reach size (m, padded_window_size)
+    if padded_window_size != window_size:
+        padding_right = padded_window_size - window_size
+        strided_input = torch.nn.functional.pad(
+            strided_input.unsqueeze(0), (0, padding_right), mode="constant", value=0
+        ).squeeze(0)
+
+    # Compute energy after window function (not the raw one)
+    if not raw_energy:
+        signal_log_energy = _get_log_energy(strided_input, epsilon, energy_floor)  # size (m)
+
+    return strided_input, signal_log_energy
+
+
+def _subtract_column_mean(tensor: Tensor, subtract_mean: bool) -> Tensor:
+    # subtracts the column mean of the tensor size (m, n) if subtract_mean=True
+    # it returns size (m, n)
+    if subtract_mean:
+        col_means = torch.mean(tensor, dim=0).unsqueeze(0)
+        tensor = tensor - col_means
+    return tensor
+
+
+def spectrogram(
+    waveform: Tensor,
+    blackman_coeff: float = 0.42,
+    channel: int = -1,
+    dither: float = 0.0,
+    energy_floor: float = 1.0,
+    frame_length: float = 25.0,
+    frame_shift: float = 10.0,
+    min_duration: float = 0.0,
+    preemphasis_coefficient: float = 0.97,
+    raw_energy: bool = True,
+    remove_dc_offset: bool = True,
+    round_to_power_of_two: bool = True,
+    sample_frequency: float = 16000.0,
+    snip_edges: bool = True,
+    subtract_mean: bool = False,
+    window_type: str = POVEY,
+) -> Tensor:
+    r"""Create a spectrogram from a raw audio signal. This matches the input/output of Kaldi's
+    compute-spectrogram-feats.
+
+    Args:
+        waveform (Tensor): Tensor of audio of size (c, n) where c is in the range [0,2)
+        blackman_coeff (float, optional): Constant coefficient for generalized Blackman window. (Default: ``0.42``)
+        channel (int, optional): Channel to extract (-1 -> expect mono, 0 -> left, 1 -> right) (Default: ``-1``)
+        dither (float, optional): Dithering constant (0.0 means no dither). If you turn this off, you should set
+            the energy_floor option, e.g. to 1.0 or 0.1 (Default: ``0.0``)
+        energy_floor (float, optional): Floor on energy (absolute, not relative) in Spectrogram computation.  Caution:
+            this floor is applied to the zeroth component, representing the total signal energy.  The floor on the
+            individual spectrogram elements is fixed at std::numeric_limits<float>::epsilon(). (Default: ``1.0``)
+        frame_length (float, optional): Frame length in milliseconds (Default: ``25.0``)
+        frame_shift (float, optional): Frame shift in milliseconds (Default: ``10.0``)
+        min_duration (float, optional): Minimum duration of segments to process (in seconds). (Default: ``0.0``)
+        preemphasis_coefficient (float, optional): Coefficient for use in signal preemphasis (Default: ``0.97``)
+        raw_energy (bool, optional): If True, compute energy before preemphasis and windowing (Default: ``True``)
+        remove_dc_offset (bool, optional): Subtract mean from waveform on each frame (Default: ``True``)
+        round_to_power_of_two (bool, optional): If True, round window size to power of two by zero-padding input
+            to FFT. (Default: ``True``)
+        sample_frequency (float, optional): Waveform data sample frequency (must match the waveform file, if
+            specified there) (Default: ``16000.0``)
+        snip_edges (bool, optional): If True, end effects will be handled by outputting only frames that completely fit
+            in the file, and the number of frames depends on the frame_length.  If False, the number of frames
+            depends only on the frame_shift, and we reflect the data at the ends. (Default: ``True``)
+        subtract_mean (bool, optional): Subtract mean of each feature file [CMS]; not recommended to do
+            it this way.  (Default: ``False``)
+        window_type (str, optional): Type of window ('hamming'|'hanning'|'povey'|'rectangular'|'blackman')
+         (Default: ``'povey'``)
+
+    Returns:
+        Tensor: A spectrogram identical to what Kaldi would output. The shape is
+        (m, ``padded_window_size // 2 + 1``) where m is calculated in _get_strided
+    """
+    device, dtype = waveform.device, waveform.dtype
+    epsilon = _get_epsilon(device, dtype)
+
+    waveform, window_shift, window_size, padded_window_size = _get_waveform_and_window_properties(
+        waveform, channel, sample_frequency, frame_shift, frame_length, round_to_power_of_two, preemphasis_coefficient
+    )
+
+    if len(waveform) < min_duration * sample_frequency:
+        # signal is too short
+        return torch.empty(0)
+
+    strided_input, signal_log_energy = _get_window(
+        waveform,
+        padded_window_size,
+        window_size,
+        window_shift,
+        window_type,
+        blackman_coeff,
+        snip_edges,
+        raw_energy,
+        energy_floor,
+        dither,
+        remove_dc_offset,
+        preemphasis_coefficient,
+    )
+
+    # size (m, padded_window_size // 2 + 1, 2)
+    fft = torch.fft.rfft(strided_input)
+
+    # Convert the FFT into a power spectrum
+    power_spectrum = torch.max(fft.abs().pow(2.0), epsilon).log()  # size (m, padded_window_size // 2 + 1)
+    power_spectrum[:, 0] = signal_log_energy
+
+    power_spectrum = _subtract_column_mean(power_spectrum, subtract_mean)
+    return power_spectrum
+
+
+def inverse_mel_scale_scalar(mel_freq: float) -> float:
+    return 700.0 * (math.exp(mel_freq / 1127.0) - 1.0)
+
+
+def inverse_mel_scale(mel_freq: Tensor) -> Tensor:
+    return 700.0 * ((mel_freq / 1127.0).exp() - 1.0)
+
+
+def mel_scale_scalar(freq: float) -> float:
+    return 1127.0 * math.log(1.0 + freq / 700.0)
+
+
+def mel_scale(freq: Tensor) -> Tensor:
+    return 1127.0 * (1.0 + freq / 700.0).log()
+
+
+def vtln_warp_freq(
+    vtln_low_cutoff: float,
+    vtln_high_cutoff: float,
+    low_freq: float,
+    high_freq: float,
+    vtln_warp_factor: float,
+    freq: Tensor,
+) -> Tensor:
+    r"""This computes a VTLN warping function that is not the same as HTK's one,
+    but has similar inputs (this function has the advantage of never producing
+    empty bins).
+
+    This function computes a warp function F(freq), defined between low_freq
+    and high_freq inclusive, with the following properties:
+        F(low_freq) == low_freq
+        F(high_freq) == high_freq
+    The function is continuous and piecewise linear with two inflection
+        points.
+    The lower inflection point (measured in terms of the unwarped
+        frequency) is at frequency l, determined as described below.
+    The higher inflection point is at a frequency h, determined as
+        described below.
+    If l <= f <= h, then F(f) = f/vtln_warp_factor.
+    If the higher inflection point (measured in terms of the unwarped
+        frequency) is at h, then max(h, F(h)) == vtln_high_cutoff.
+        Since (by the last point) F(h) == h/vtln_warp_factor, then
+        max(h, h/vtln_warp_factor) == vtln_high_cutoff, so
+        h = vtln_high_cutoff / max(1, 1/vtln_warp_factor).
+          = vtln_high_cutoff * min(1, vtln_warp_factor).
+    If the lower inflection point (measured in terms of the unwarped
+        frequency) is at l, then min(l, F(l)) == vtln_low_cutoff
+        This implies that l = vtln_low_cutoff / min(1, 1/vtln_warp_factor)
+                            = vtln_low_cutoff * max(1, vtln_warp_factor)
+    Args:
+        vtln_low_cutoff (float): Lower frequency cutoffs for VTLN
+        vtln_high_cutoff (float): Upper frequency cutoffs for VTLN
+        low_freq (float): Lower frequency cutoffs in mel computation
+        high_freq (float): Upper frequency cutoffs in mel computation
+        vtln_warp_factor (float): Vtln warp factor
+        freq (Tensor): given frequency in Hz
+
+    Returns:
+        Tensor: Freq after vtln warp
+    """
+    assert vtln_low_cutoff > low_freq, "be sure to set the vtln_low option higher than low_freq"
+    assert vtln_high_cutoff < high_freq, "be sure to set the vtln_high option lower than high_freq [or negative]"
+    l = vtln_low_cutoff * max(1.0, vtln_warp_factor)
+    h = vtln_high_cutoff * min(1.0, vtln_warp_factor)
+    scale = 1.0 / vtln_warp_factor
+    Fl = scale * l  # F(l)
+    Fh = scale * h  # F(h)
+    assert l > low_freq and h < high_freq
+    # slope of left part of the 3-piece linear function
+    scale_left = (Fl - low_freq) / (l - low_freq)
+    # [slope of center part is just "scale"]
+
+    # slope of right part of the 3-piece linear function
+    scale_right = (high_freq - Fh) / (high_freq - h)
+
+    res = torch.empty_like(freq)
+
+    outside_low_high_freq = torch.lt(freq, low_freq) | torch.gt(freq, high_freq)  # freq < low_freq || freq > high_freq
+    before_l = torch.lt(freq, l)  # freq < l
+    before_h = torch.lt(freq, h)  # freq < h
+    after_h = torch.ge(freq, h)  # freq >= h
+
+    # order of operations matter here (since there is overlapping frequency regions)
+    res[after_h] = high_freq + scale_right * (freq[after_h] - high_freq)
+    res[before_h] = scale * freq[before_h]
+    res[before_l] = low_freq + scale_left * (freq[before_l] - low_freq)
+    res[outside_low_high_freq] = freq[outside_low_high_freq]
+
+    return res
+
+
+def vtln_warp_mel_freq(
+    vtln_low_cutoff: float,
+    vtln_high_cutoff: float,
+    low_freq,
+    high_freq: float,
+    vtln_warp_factor: float,
+    mel_freq: Tensor,
+) -> Tensor:
+    r"""
+    Args:
+        vtln_low_cutoff (float): Lower frequency cutoffs for VTLN
+        vtln_high_cutoff (float): Upper frequency cutoffs for VTLN
+        low_freq (float): Lower frequency cutoffs in mel computation
+        high_freq (float): Upper frequency cutoffs in mel computation
+        vtln_warp_factor (float): Vtln warp factor
+        mel_freq (Tensor): Given frequency in Mel
+
+    Returns:
+        Tensor: ``mel_freq`` after vtln warp
+    """
+    return mel_scale(
+        vtln_warp_freq(
+            vtln_low_cutoff, vtln_high_cutoff, low_freq, high_freq, vtln_warp_factor, inverse_mel_scale(mel_freq)
+        )
+    )
+
+
+def get_mel_banks(
+    num_bins: int,
+    window_length_padded: int,
+    sample_freq: float,
+    low_freq: float,
+    high_freq: float,
+    vtln_low: float,
+    vtln_high: float,
+    vtln_warp_factor: float,
+    device=None,
+    dtype=None,
+) -> Tuple[Tensor, Tensor]:
+    """
+    Returns:
+        (Tensor, Tensor): The tuple consists of ``bins`` (which is
+        melbank of size (``num_bins``, ``num_fft_bins``)) and ``center_freqs`` (which is
+        center frequencies of bins of size (``num_bins``)).
+    """
+    assert num_bins > 3, "Must have at least 3 mel bins"
+    assert window_length_padded % 2 == 0
+    num_fft_bins = window_length_padded / 2
+    nyquist = 0.5 * sample_freq
+
+    if high_freq <= 0.0:
+        high_freq += nyquist
+
+    assert (0.0 <= low_freq < nyquist) and (0.0 < high_freq <= nyquist) and (low_freq < high_freq), (
+        "Bad values in options: low-freq {} and high-freq {} vs. nyquist {}".format(low_freq, high_freq, nyquist)
+    )
+
+    # fft-bin width [think of it as Nyquist-freq / half-window-length]
+    fft_bin_width = sample_freq / window_length_padded
+    mel_low_freq = mel_scale_scalar(low_freq)
+    mel_high_freq = mel_scale_scalar(high_freq)
+
+    # divide by num_bins+1 in next line because of end-effects where the bins
+    # spread out to the sides.
+    mel_freq_delta = (mel_high_freq - mel_low_freq) / (num_bins + 1)
+
+    if vtln_high < 0.0:
+        vtln_high += nyquist
+
+    assert vtln_warp_factor == 1.0 or (
+        (low_freq < vtln_low < high_freq) and (0.0 < vtln_high < high_freq) and (vtln_low < vtln_high)
+    ), "Bad values in options: vtln-low {} and vtln-high {}, versus low-freq {} and high-freq {}".format(
+        vtln_low, vtln_high, low_freq, high_freq
+    )
+
+    bin = torch.arange(num_bins).unsqueeze(1)
+    left_mel = mel_low_freq + bin * mel_freq_delta  # size(num_bins, 1)
+    center_mel = mel_low_freq + (bin + 1.0) * mel_freq_delta  # size(num_bins, 1)
+    right_mel = mel_low_freq + (bin + 2.0) * mel_freq_delta  # size(num_bins, 1)
+
+    if vtln_warp_factor != 1.0:
+        left_mel = vtln_warp_mel_freq(vtln_low, vtln_high, low_freq, high_freq, vtln_warp_factor, left_mel)
+        center_mel = vtln_warp_mel_freq(vtln_low, vtln_high, low_freq, high_freq, vtln_warp_factor, center_mel)
+        right_mel = vtln_warp_mel_freq(vtln_low, vtln_high, low_freq, high_freq, vtln_warp_factor, right_mel)
+
+    # center_freqs = inverse_mel_scale(center_mel)  # size (num_bins)
+    # size(1, num_fft_bins)
+    mel = mel_scale(fft_bin_width * torch.arange(num_fft_bins)).unsqueeze(0)
+
+    # size (num_bins, num_fft_bins)
+    up_slope = (mel - left_mel) / (center_mel - left_mel)
+    down_slope = (right_mel - mel) / (right_mel - center_mel)
+
+    if vtln_warp_factor == 1.0:
+        # left_mel < center_mel < right_mel so we can min the two slopes and clamp negative values
+        bins = torch.max(torch.zeros(1), torch.min(up_slope, down_slope))
+    else:
+        # warping can move the order of left_mel, center_mel, right_mel anywhere
+        bins = torch.zeros_like(up_slope)
+        up_idx = torch.gt(mel, left_mel) & torch.le(mel, center_mel)  # left_mel < mel <= center_mel
+        down_idx = torch.gt(mel, center_mel) & torch.lt(mel, right_mel)  # center_mel < mel < right_mel
+        bins[up_idx] = up_slope[up_idx]
+        bins[down_idx] = down_slope[down_idx]
+
+    return bins.to(device=device, dtype=dtype)  # , center_freqs
+
+
+cache = {}
+
+
+def fbank(
+    waveform: Tensor,
+    blackman_coeff: float = 0.42,
+    channel: int = -1,
+    dither: float = 0.0,
+    energy_floor: float = 1.0,
+    frame_length: float = 25.0,
+    frame_shift: float = 10.0,
+    high_freq: float = 0.0,
+    htk_compat: bool = False,
+    low_freq: float = 20.0,
+    min_duration: float = 0.0,
+    num_mel_bins: int = 23,
+    preemphasis_coefficient: float = 0.97,
+    raw_energy: bool = True,
+    remove_dc_offset: bool = True,
+    round_to_power_of_two: bool = True,
+    sample_frequency: float = 16000.0,
+    snip_edges: bool = True,
+    subtract_mean: bool = False,
+    use_energy: bool = False,
+    use_log_fbank: bool = True,
+    use_power: bool = True,
+    vtln_high: float = -500.0,
+    vtln_low: float = 100.0,
+    vtln_warp: float = 1.0,
+    window_type: str = POVEY,
+) -> Tensor:
+    r"""Create a fbank from a raw audio signal. This matches the input/output of Kaldi's
+    compute-fbank-feats.
+
+    Args:
+        waveform (Tensor): Tensor of audio of size (c, n) where c is in the range [0,2)
+        blackman_coeff (float, optional): Constant coefficient for generalized Blackman window. (Default: ``0.42``)
+        channel (int, optional): Channel to extract (-1 -> expect mono, 0 -> left, 1 -> right) (Default: ``-1``)
+        dither (float, optional): Dithering constant (0.0 means no dither). If you turn this off, you should set
+            the energy_floor option, e.g. to 1.0 or 0.1 (Default: ``0.0``)
+        energy_floor (float, optional): Floor on energy (absolute, not relative) in Spectrogram computation.  Caution:
+            this floor is applied to the zeroth component, representing the total signal energy.  The floor on the
+            individual spectrogram elements is fixed at std::numeric_limits<float>::epsilon(). (Default: ``1.0``)
+        frame_length (float, optional): Frame length in milliseconds (Default: ``25.0``)
+        frame_shift (float, optional): Frame shift in milliseconds (Default: ``10.0``)
+        high_freq (float, optional): High cutoff frequency for mel bins (if <= 0, offset from Nyquist)
+         (Default: ``0.0``)
+        htk_compat (bool, optional): If true, put energy last.  Warning: not sufficient to get HTK compatible features
+         (need to change other parameters). (Default: ``False``)
+        low_freq (float, optional): Low cutoff frequency for mel bins (Default: ``20.0``)
+        min_duration (float, optional): Minimum duration of segments to process (in seconds). (Default: ``0.0``)
+        num_mel_bins (int, optional): Number of triangular mel-frequency bins (Default: ``23``)
+        preemphasis_coefficient (float, optional): Coefficient for use in signal preemphasis (Default: ``0.97``)
+        raw_energy (bool, optional): If True, compute energy before preemphasis and windowing (Default: ``True``)
+        remove_dc_offset (bool, optional): Subtract mean from waveform on each frame (Default: ``True``)
+        round_to_power_of_two (bool, optional): If True, round window size to power of two by zero-padding input
+            to FFT. (Default: ``True``)
+        sample_frequency (float, optional): Waveform data sample frequency (must match the waveform file, if
+            specified there) (Default: ``16000.0``)
+        snip_edges (bool, optional): If True, end effects will be handled by outputting only frames that completely fit
+            in the file, and the number of frames depends on the frame_length.  If False, the number of frames
+            depends only on the frame_shift, and we reflect the data at the ends. (Default: ``True``)
+        subtract_mean (bool, optional): Subtract mean of each feature file [CMS]; not recommended to do
+            it this way.  (Default: ``False``)
+        use_energy (bool, optional): Add an extra dimension with energy to the FBANK output. (Default: ``False``)
+        use_log_fbank (bool, optional):If true, produce log-filterbank, else produce linear. (Default: ``True``)
+        use_power (bool, optional): If true, use power, else use magnitude. (Default: ``True``)
+        vtln_high (float, optional): High inflection point in piecewise linear VTLN warping function (if
+            negative, offset from high-mel-freq (Default: ``-500.0``)
+        vtln_low (float, optional): Low inflection point in piecewise linear VTLN warping function (Default: ``100.0``)
+        vtln_warp (float, optional): Vtln warp factor (only applicable if vtln_map not specified) (Default: ``1.0``)
+        window_type (str, optional): Type of window ('hamming'|'hanning'|'povey'|'rectangular'|'blackman')
+         (Default: ``'povey'``)
+
+    Returns:
+        Tensor: A fbank identical to what Kaldi would output. The shape is (m, ``num_mel_bins + use_energy``)
+        where m is calculated in _get_strided
+    """
+    device, dtype = waveform.device, waveform.dtype
+
+    waveform, window_shift, window_size, padded_window_size = _get_waveform_and_window_properties(
+        waveform, channel, sample_frequency, frame_shift, frame_length, round_to_power_of_two, preemphasis_coefficient
+    )
+
+    if len(waveform) < min_duration * sample_frequency:
+        # signal is too short
+        return torch.empty(0, device=device, dtype=dtype)
+
+    # strided_input, size (m, padded_window_size) and signal_log_energy, size (m)
+    strided_input, signal_log_energy = _get_window(
+        waveform,
+        padded_window_size,
+        window_size,
+        window_shift,
+        window_type,
+        blackman_coeff,
+        snip_edges,
+        raw_energy,
+        energy_floor,
+        dither,
+        remove_dc_offset,
+        preemphasis_coefficient,
+    )
+
+    # size (m, padded_window_size // 2 + 1)
+    spectrum = torch.fft.rfft(strided_input).abs()
+    if use_power:
+        spectrum = spectrum.pow(2.0)
+
+    # size (num_mel_bins, padded_window_size // 2)
+    # print(num_mel_bins, padded_window_size, sample_frequency, low_freq, high_freq, vtln_low, vtln_high, vtln_warp)
+
+    cache_key = "%s-%s-%s-%s-%s-%s-%s-%s-%s-%s" % (
+        num_mel_bins,
+        padded_window_size,
+        sample_frequency,
+        low_freq,
+        high_freq,
+        vtln_low,
+        vtln_high,
+        vtln_warp,
+        device,
+        dtype,
+    )
+    if cache_key not in cache:
+        mel_energies = get_mel_banks(
+            num_mel_bins,
+            padded_window_size,
+            sample_frequency,
+            low_freq,
+            high_freq,
+            vtln_low,
+            vtln_high,
+            vtln_warp,
+            device,
+            dtype,
+        )
+        cache[cache_key] = mel_energies
+    else:
+        mel_energies = cache[cache_key]
+
+    # pad right column with zeros and add dimension, size (num_mel_bins, padded_window_size // 2 + 1)
+    mel_energies = torch.nn.functional.pad(mel_energies, (0, 1), mode="constant", value=0)
+
+    # sum with mel fiterbanks over the power spectrum, size (m, num_mel_bins)
+    mel_energies = torch.mm(spectrum, mel_energies.T)
+    if use_log_fbank:
+        # avoid log of zero (which should be prevented anyway by dithering)
+        mel_energies = torch.max(mel_energies, _get_epsilon(device, dtype)).log()
+
+    # if use_energy then add it as the last column for htk_compat == true else first column
+    if use_energy:
+        signal_log_energy = signal_log_energy.unsqueeze(1)  # size (m, 1)
+        # returns size (m, num_mel_bins + 1)
+        if htk_compat:
+            mel_energies = torch.cat((mel_energies, signal_log_energy), dim=1)
+        else:
+            mel_energies = torch.cat((signal_log_energy, mel_energies), dim=1)
+
+    mel_energies = _subtract_column_mean(mel_energies, subtract_mean)
+    return mel_energies
+
+
+def _get_dct_matrix(num_ceps: int, num_mel_bins: int) -> Tensor:
+    # returns a dct matrix of size (num_mel_bins, num_ceps)
+    # size (num_mel_bins, num_mel_bins)
+    dct_matrix = torchaudio.functional.create_dct(num_mel_bins, num_mel_bins, "ortho")
+    # kaldi expects the first cepstral to be weighted sum of factor sqrt(1/num_mel_bins)
+    # this would be the first column in the dct_matrix for torchaudio as it expects a
+    # right multiply (which would be the first column of the kaldi's dct_matrix as kaldi
+    # expects a left multiply e.g. dct_matrix * vector).
+    dct_matrix[:, 0] = math.sqrt(1 / float(num_mel_bins))
+    dct_matrix = dct_matrix[:, :num_ceps]
+    return dct_matrix
+
+
+def _get_lifter_coeffs(num_ceps: int, cepstral_lifter: float) -> Tensor:
+    # returns size (num_ceps)
+    # Compute liftering coefficients (scaling on cepstral coeffs)
+    # coeffs are numbered slightly differently from HTK: the zeroth index is C0, which is not affected.
+    i = torch.arange(num_ceps)
+    return 1.0 + 0.5 * cepstral_lifter * torch.sin(math.pi * i / cepstral_lifter)
+
+
+def mfcc(
+    waveform: Tensor,
+    blackman_coeff: float = 0.42,
+    cepstral_lifter: float = 22.0,
+    channel: int = -1,
+    dither: float = 0.0,
+    energy_floor: float = 1.0,
+    frame_length: float = 25.0,
+    frame_shift: float = 10.0,
+    high_freq: float = 0.0,
+    htk_compat: bool = False,
+    low_freq: float = 20.0,
+    num_ceps: int = 13,
+    min_duration: float = 0.0,
+    num_mel_bins: int = 23,
+    preemphasis_coefficient: float = 0.97,
+    raw_energy: bool = True,
+    remove_dc_offset: bool = True,
+    round_to_power_of_two: bool = True,
+    sample_frequency: float = 16000.0,
+    snip_edges: bool = True,
+    subtract_mean: bool = False,
+    use_energy: bool = False,
+    vtln_high: float = -500.0,
+    vtln_low: float = 100.0,
+    vtln_warp: float = 1.0,
+    window_type: str = POVEY,
+) -> Tensor:
+    r"""Create a mfcc from a raw audio signal. This matches the input/output of Kaldi's
+    compute-mfcc-feats.
+
+    Args:
+        waveform (Tensor): Tensor of audio of size (c, n) where c is in the range [0,2)
+        blackman_coeff (float, optional): Constant coefficient for generalized Blackman window. (Default: ``0.42``)
+        cepstral_lifter (float, optional): Constant that controls scaling of MFCCs (Default: ``22.0``)
+        channel (int, optional): Channel to extract (-1 -> expect mono, 0 -> left, 1 -> right) (Default: ``-1``)
+        dither (float, optional): Dithering constant (0.0 means no dither). If you turn this off, you should set
+            the energy_floor option, e.g. to 1.0 or 0.1 (Default: ``0.0``)
+        energy_floor (float, optional): Floor on energy (absolute, not relative) in Spectrogram computation.  Caution:
+            this floor is applied to the zeroth component, representing the total signal energy.  The floor on the
+            individual spectrogram elements is fixed at std::numeric_limits<float>::epsilon(). (Default: ``1.0``)
+        frame_length (float, optional): Frame length in milliseconds (Default: ``25.0``)
+        frame_shift (float, optional): Frame shift in milliseconds (Default: ``10.0``)
+        high_freq (float, optional): High cutoff frequency for mel bins (if <= 0, offset from Nyquist)
+         (Default: ``0.0``)
+        htk_compat (bool, optional): If true, put energy last.  Warning: not sufficient to get HTK compatible
+         features (need to change other parameters). (Default: ``False``)
+        low_freq (float, optional): Low cutoff frequency for mel bins (Default: ``20.0``)
+        num_ceps (int, optional): Number of cepstra in MFCC computation (including C0) (Default: ``13``)
+        min_duration (float, optional): Minimum duration of segments to process (in seconds). (Default: ``0.0``)
+        num_mel_bins (int, optional): Number of triangular mel-frequency bins (Default: ``23``)
+        preemphasis_coefficient (float, optional): Coefficient for use in signal preemphasis (Default: ``0.97``)
+        raw_energy (bool, optional): If True, compute energy before preemphasis and windowing (Default: ``True``)
+        remove_dc_offset (bool, optional): Subtract mean from waveform on each frame (Default: ``True``)
+        round_to_power_of_two (bool, optional): If True, round window size to power of two by zero-padding input
+            to FFT. (Default: ``True``)
+        sample_frequency (float, optional): Waveform data sample frequency (must match the waveform file, if
+            specified there) (Default: ``16000.0``)
+        snip_edges (bool, optional): If True, end effects will be handled by outputting only frames that completely fit
+            in the file, and the number of frames depends on the frame_length.  If False, the number of frames
+            depends only on the frame_shift, and we reflect the data at the ends. (Default: ``True``)
+        subtract_mean (bool, optional): Subtract mean of each feature file [CMS]; not recommended to do
+            it this way.  (Default: ``False``)
+        use_energy (bool, optional): Add an extra dimension with energy to the FBANK output. (Default: ``False``)
+        vtln_high (float, optional): High inflection point in piecewise linear VTLN warping function (if
+            negative, offset from high-mel-freq (Default: ``-500.0``)
+        vtln_low (float, optional): Low inflection point in piecewise linear VTLN warping function (Default: ``100.0``)
+        vtln_warp (float, optional): Vtln warp factor (only applicable if vtln_map not specified) (Default: ``1.0``)
+        window_type (str, optional): Type of window ('hamming'|'hanning'|'povey'|'rectangular'|'blackman')
+         (Default: ``"povey"``)
+
+    Returns:
+        Tensor: A mfcc identical to what Kaldi would output. The shape is (m, ``num_ceps``)
+        where m is calculated in _get_strided
+    """
+    assert num_ceps <= num_mel_bins, "num_ceps cannot be larger than num_mel_bins: %d vs %d" % (num_ceps, num_mel_bins)
+
+    device, dtype = waveform.device, waveform.dtype
+
+    # The mel_energies should not be squared (use_power=True), not have mean subtracted
+    # (subtract_mean=False), and use log (use_log_fbank=True).
+    # size (m, num_mel_bins + use_energy)
+    feature = fbank(
+        waveform=waveform,
+        blackman_coeff=blackman_coeff,
+        channel=channel,
+        dither=dither,
+        energy_floor=energy_floor,
+        frame_length=frame_length,
+        frame_shift=frame_shift,
+        high_freq=high_freq,
+        htk_compat=htk_compat,
+        low_freq=low_freq,
+        min_duration=min_duration,
+        num_mel_bins=num_mel_bins,
+        preemphasis_coefficient=preemphasis_coefficient,
+        raw_energy=raw_energy,
+        remove_dc_offset=remove_dc_offset,
+        round_to_power_of_two=round_to_power_of_two,
+        sample_frequency=sample_frequency,
+        snip_edges=snip_edges,
+        subtract_mean=False,
+        use_energy=use_energy,
+        use_log_fbank=True,
+        use_power=True,
+        vtln_high=vtln_high,
+        vtln_low=vtln_low,
+        vtln_warp=vtln_warp,
+        window_type=window_type,
+    )
+
+    if use_energy:
+        # size (m)
+        signal_log_energy = feature[:, num_mel_bins if htk_compat else 0]
+        # offset is 0 if htk_compat==True else 1
+        mel_offset = int(not htk_compat)
+        feature = feature[:, mel_offset : (num_mel_bins + mel_offset)]
+
+    # size (num_mel_bins, num_ceps)
+    dct_matrix = _get_dct_matrix(num_ceps, num_mel_bins).to(dtype=dtype, device=device)
+
+    # size (m, num_ceps)
+    feature = feature.matmul(dct_matrix)
+
+    if cepstral_lifter != 0.0:
+        # size (1, num_ceps)
+        lifter_coeffs = _get_lifter_coeffs(num_ceps, cepstral_lifter).unsqueeze(0)
+        feature *= lifter_coeffs.to(device=device, dtype=dtype)
+
+    # if use_energy then replace the last column for htk_compat == true else first column
+    if use_energy:
+        feature[:, 0] = signal_log_energy
+
+    if htk_compat:
+        energy = feature[:, 0].unsqueeze(1)  # size (m, 1)
+        feature = feature[:, 1:]  # size (m, num_ceps - 1)
+        if not use_energy:
+            # scale on C0 (actually removing a scale we previously added that's
+            # part of one common definition of the cosine transform.)
+            energy *= math.sqrt(2)
+
+        feature = torch.cat((feature, energy), dim=1)
+
+    feature = _subtract_column_mean(feature, subtract_mean)
+    return feature

eres2net/pooling_layers.py

@@ -0,0 +1,101 @@
+# Copyright 3D-Speaker (https://github.com/alibaba-damo-academy/3D-Speaker). All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""This implementation is adapted from https://github.com/wenet-e2e/wespeaker."""
+
+import torch
+import torch.nn as nn
+
+
+class TAP(nn.Module):
+    """
+    Temporal average pooling, only first-order mean is considered
+    """
+
+    def __init__(self, **kwargs):
+        super(TAP, self).__init__()
+
+    def forward(self, x):
+        pooling_mean = x.mean(dim=-1)
+        # To be compatable with 2D input
+        pooling_mean = pooling_mean.flatten(start_dim=1)
+        return pooling_mean
+
+
+class TSDP(nn.Module):
+    """
+    Temporal standard deviation pooling, only second-order std is considered
+    """
+
+    def __init__(self, **kwargs):
+        super(TSDP, self).__init__()
+
+    def forward(self, x):
+        # The last dimension is the temporal axis
+        pooling_std = torch.sqrt(torch.var(x, dim=-1) + 1e-8)
+        pooling_std = pooling_std.flatten(start_dim=1)
+        return pooling_std
+
+
+class TSTP(nn.Module):
+    """
+    Temporal statistics pooling, concatenate mean and std, which is used in
+    x-vector
+    Comment: simple concatenation can not make full use of both statistics
+    """
+
+    def __init__(self, **kwargs):
+        super(TSTP, self).__init__()
+
+    def forward(self, x):
+        # The last dimension is the temporal axis
+        pooling_mean = x.mean(dim=-1)
+        pooling_std = torch.sqrt(torch.var(x, dim=-1) + 1e-8)
+        pooling_mean = pooling_mean.flatten(start_dim=1)
+        pooling_std = pooling_std.flatten(start_dim=1)
+
+        stats = torch.cat((pooling_mean, pooling_std), 1)
+        return stats
+
+
+class ASTP(nn.Module):
+    """Attentive statistics pooling: Channel- and context-dependent
+    statistics pooling, first used in ECAPA_TDNN.
+    """
+
+    def __init__(self, in_dim, bottleneck_dim=128, global_context_att=False):
+        super(ASTP, self).__init__()
+        self.global_context_att = global_context_att
+
+        # Use Conv1d with stride == 1 rather than Linear, then we don't
+        # need to transpose inputs.
+        if global_context_att:
+            self.linear1 = nn.Conv1d(in_dim * 3, bottleneck_dim, kernel_size=1)  # equals W and b in the paper
+        else:
+            self.linear1 = nn.Conv1d(in_dim, bottleneck_dim, kernel_size=1)  # equals W and b in the paper
+        self.linear2 = nn.Conv1d(bottleneck_dim, in_dim, kernel_size=1)  # equals V and k in the paper
+
+    def forward(self, x):
+        """
+        x: a 3-dimensional tensor in tdnn-based architecture (B,F,T)
+            or a 4-dimensional tensor in resnet architecture (B,C,F,T)
+            0-dim: batch-dimension, last-dim: time-dimension (frame-dimension)
+        """
+        if len(x.shape) == 4:
+            x = x.reshape(x.shape[0], x.shape[1] * x.shape[2], x.shape[3])
+        assert len(x.shape) == 3
+
+        if self.global_context_att:
+            context_mean = torch.mean(x, dim=-1, keepdim=True).expand_as(x)
+            context_std = torch.sqrt(torch.var(x, dim=-1, keepdim=True) + 1e-10).expand_as(x)
+            x_in = torch.cat((x, context_mean, context_std), dim=1)
+        else:
+            x_in = x
+
+        # DON'T use ReLU here! ReLU may be hard to converge.
+        alpha = torch.tanh(self.linear1(x_in))  # alpha = F.relu(self.linear1(x_in))
+        alpha = torch.softmax(self.linear2(alpha), dim=2)
+        mean = torch.sum(alpha * x, dim=2)
+        var = torch.sum(alpha * (x**2), dim=2) - mean**2
+        std = torch.sqrt(var.clamp(min=1e-10))
+        return torch.cat([mean, std], dim=1)

f5_tts/model/__init__.py

@@ -0,0 +1,13 @@
+# from f5_tts.model.cfm import CFM
+#
+# from f5_tts.model.backbones.unett import UNetT
+from GPT_SoVITS.f5_tts.model.backbones.dit import DiT
+# from f5_tts.model.backbones.dit import DiTNoCond
+# from f5_tts.model.backbones.dit import DiTNoCondNoT
+# from f5_tts.model.backbones.mmdit import MMDiT
+
+# from f5_tts.model.trainer import Trainer
+
+
+# __all__ = ["CFM", "UNetT", "DiT", "MMDiT", "Trainer"]
+# __all__ = ["CFM", "UNetT", "DiTNoCond","DiT", "MMDiT"]

f5_tts/model/backbones/README.md

@@ -0,0 +1,20 @@
+## Backbones quick introduction
+
+
+### unett.py
+- flat unet transformer
+- structure same as in e2-tts & voicebox paper except using rotary pos emb
+- update: allow possible abs pos emb & convnextv2 blocks for embedded text before concat
+
+### dit.py
+- adaln-zero dit
+- embedded timestep as condition
+- concatted noised_input + masked_cond + embedded_text, linear proj in
+- possible abs pos emb & convnextv2 blocks for embedded text before concat
+- possible long skip connection (first layer to last layer)
+
+### mmdit.py
+- sd3 structure
+- timestep as condition
+- left stream: text embedded and applied a abs pos emb
+- right stream: masked_cond & noised_input concatted and with same conv pos emb as unett

f5_tts/model/backbones/dit.py

@@ -0,0 +1,194 @@
+"""
+ein notation:
+b - batch
+n - sequence
+nt - text sequence
+nw - raw wave length
+d - dimension
+"""
+
+from __future__ import annotations
+
+import torch
+from torch import nn
+from torch.utils.checkpoint import checkpoint
+
+from x_transformers.x_transformers import RotaryEmbedding
+
+from GPT_SoVITS.f5_tts.model.modules import (
+    TimestepEmbedding,
+    ConvNeXtV2Block,
+    ConvPositionEmbedding,
+    DiTBlock,
+    AdaLayerNormZero_Final,
+    precompute_freqs_cis,
+    get_pos_embed_indices,
+)
+
+from module.commons import sequence_mask
+
+
+class TextEmbedding(nn.Module):
+    def __init__(self, text_dim, conv_layers=0, conv_mult=2):
+        super().__init__()
+        if conv_layers > 0:
+            self.extra_modeling = True
+            self.precompute_max_pos = 4096  # ~44s of 24khz audio
+            self.register_buffer("freqs_cis", precompute_freqs_cis(text_dim, self.precompute_max_pos), persistent=False)
+            self.text_blocks = nn.Sequential(
+                *[ConvNeXtV2Block(text_dim, text_dim * conv_mult) for _ in range(conv_layers)]
+            )
+        else:
+            self.extra_modeling = False
+
+    def forward(self, text: int["b nt"], seq_len, drop_text=False):  # noqa: F722
+        batch, text_len = text.shape[0], text.shape[1]
+
+        if drop_text:  # cfg for text
+            text = torch.zeros_like(text)
+
+        # possible extra modeling
+        if self.extra_modeling:
+            # sinus pos emb
+            batch_start = torch.zeros((batch,), dtype=torch.long)
+            pos_idx = get_pos_embed_indices(batch_start, seq_len, max_pos=self.precompute_max_pos)
+            text_pos_embed = self.freqs_cis[pos_idx]
+
+            # print(23333333,text.shape,text_pos_embed.shape)#torch.Size([7, 465, 256]) torch.Size([7, 465, 256])
+
+            text = text + text_pos_embed
+
+            # convnextv2 blocks
+            text = self.text_blocks(text)
+
+        return text
+
+
+# noised input audio and context mixing embedding
+
+
+class InputEmbedding(nn.Module):
+    def __init__(self, mel_dim, text_dim, out_dim):
+        super().__init__()
+        self.proj = nn.Linear(mel_dim * 2 + text_dim, out_dim)
+        self.conv_pos_embed = ConvPositionEmbedding(dim=out_dim)
+
+    def forward(self, x: float["b n d"], cond: float["b n d"], text_embed: float["b n d"], drop_audio_cond=False):  # noqa: F722
+        if drop_audio_cond:  # cfg for cond audio
+            cond = torch.zeros_like(cond)
+
+        x = self.proj(torch.cat((x, cond, text_embed), dim=-1))
+        x = self.conv_pos_embed(x) + x
+        return x
+
+
+# Transformer backbone using DiT blocks
+
+
+class DiT(nn.Module):
+    def __init__(
+        self,
+        *,
+        dim,
+        depth=8,
+        heads=8,
+        dim_head=64,
+        dropout=0.1,
+        ff_mult=4,
+        mel_dim=100,
+        text_dim=None,
+        conv_layers=0,
+        long_skip_connection=False,
+    ):
+        super().__init__()
+
+        self.time_embed = TimestepEmbedding(dim)
+        self.d_embed = TimestepEmbedding(dim)
+        if text_dim is None:
+            text_dim = mel_dim
+        self.text_embed = TextEmbedding(text_dim, conv_layers=conv_layers)
+        self.input_embed = InputEmbedding(mel_dim, text_dim, dim)
+
+        self.rotary_embed = RotaryEmbedding(dim_head)
+
+        self.dim = dim
+        self.depth = depth
+
+        self.transformer_blocks = nn.ModuleList(
+            [DiTBlock(dim=dim, heads=heads, dim_head=dim_head, ff_mult=ff_mult, dropout=dropout) for _ in range(depth)]
+        )
+        self.long_skip_connection = nn.Linear(dim * 2, dim, bias=False) if long_skip_connection else None
+
+        self.norm_out = AdaLayerNormZero_Final(dim)  # final modulation
+        self.proj_out = nn.Linear(dim, mel_dim)
+
+    def ckpt_wrapper(self, module):
+        # https://github.com/chuanyangjin/fast-DiT/blob/main/models.py
+        def ckpt_forward(*inputs):
+            outputs = module(*inputs)
+            return outputs
+
+        return ckpt_forward
+
+    def forward(  # x, prompt_x, x_lens, t, style,cond
+        self,  # d is channel,n is T
+        x0: float["b n d"],  # nosied input audio  # noqa: F722
+        cond0: float["b n d"],  # masked cond audio  # noqa: F722
+        x_lens,
+        time: float["b"] | float[""],  # time step  # noqa: F821 F722
+        dt_base_bootstrap,
+        text0,  # : int["b nt"]  # noqa: F722#####condition feature
+        use_grad_ckpt=False,  # bool
+        ###no-use
+        drop_audio_cond=False,  # cfg for cond audio
+        drop_text=False,  # cfg for text
+        # mask: bool["b n"] | None = None,  # noqa: F722
+        infer=False,  # bool
+        text_cache=None,  # torch tensor as text_embed
+        dt_cache=None,  # torch tensor as dt
+    ):
+        x = x0.transpose(2, 1)
+        cond = cond0.transpose(2, 1)
+        text = text0.transpose(2, 1)
+        mask = sequence_mask(x_lens, max_length=x.size(1)).to(x.device)
+
+        batch, seq_len = x.shape[0], x.shape[1]
+        if time.ndim == 0:
+            time = time.repeat(batch)
+
+        # t: conditioning time, c: context (text + masked cond audio), x: noised input audio
+        t = self.time_embed(time)
+        if infer and dt_cache is not None:
+            dt = dt_cache
+        else:
+            dt = self.d_embed(dt_base_bootstrap)
+        t += dt
+
+        if infer and text_cache is not None:
+            text_embed = text_cache
+        else:
+            text_embed = self.text_embed(text, seq_len, drop_text=drop_text)  ###need to change
+
+        x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)
+
+        rope = self.rotary_embed.forward_from_seq_len(seq_len)
+
+        if self.long_skip_connection is not None:
+            residual = x
+
+        for block in self.transformer_blocks:
+            if use_grad_ckpt:
+                x = checkpoint(self.ckpt_wrapper(block), x, t, mask, rope, use_reentrant=False)
+            else:
+                x = block(x, t, mask=mask, rope=rope)
+
+        if self.long_skip_connection is not None:
+            x = self.long_skip_connection(torch.cat((x, residual), dim=-1))
+
+        x = self.norm_out(x, t)
+        output = self.proj_out(x)
+
+        if infer:
+            return output, text_embed, dt
+        else:
+            return output

f5_tts/model/backbones/mmdit.py

@@ -0,0 +1,146 @@
+"""
+ein notation:
+b - batch
+n - sequence
+nt - text sequence
+nw - raw wave length
+d - dimension
+"""
+
+from __future__ import annotations
+
+import torch
+from torch import nn
+
+from x_transformers.x_transformers import RotaryEmbedding
+
+from f5_tts.model.modules import (
+    TimestepEmbedding,
+    ConvPositionEmbedding,
+    MMDiTBlock,
+    AdaLayerNormZero_Final,
+    precompute_freqs_cis,
+    get_pos_embed_indices,
+)
+
+
+# text embedding
+
+
+class TextEmbedding(nn.Module):
+    def __init__(self, out_dim, text_num_embeds):
+        super().__init__()
+        self.text_embed = nn.Embedding(text_num_embeds + 1, out_dim)  # will use 0 as filler token
+
+        self.precompute_max_pos = 1024
+        self.register_buffer("freqs_cis", precompute_freqs_cis(out_dim, self.precompute_max_pos), persistent=False)
+
+    def forward(self, text: int["b nt"], drop_text=False) -> int["b nt d"]:  # noqa: F722
+        text = text + 1
+        if drop_text:
+            text = torch.zeros_like(text)
+        text = self.text_embed(text)
+
+        # sinus pos emb
+        batch_start = torch.zeros((text.shape[0],), dtype=torch.long)
+        batch_text_len = text.shape[1]
+        pos_idx = get_pos_embed_indices(batch_start, batch_text_len, max_pos=self.precompute_max_pos)
+        text_pos_embed = self.freqs_cis[pos_idx]
+
+        text = text + text_pos_embed
+
+        return text
+
+
+# noised input & masked cond audio embedding
+
+
+class AudioEmbedding(nn.Module):
+    def __init__(self, in_dim, out_dim):
+        super().__init__()
+        self.linear = nn.Linear(2 * in_dim, out_dim)
+        self.conv_pos_embed = ConvPositionEmbedding(out_dim)
+
+    def forward(self, x: float["b n d"], cond: float["b n d"], drop_audio_cond=False):  # noqa: F722
+        if drop_audio_cond:
+            cond = torch.zeros_like(cond)
+        x = torch.cat((x, cond), dim=-1)
+        x = self.linear(x)
+        x = self.conv_pos_embed(x) + x
+        return x
+
+
+# Transformer backbone using MM-DiT blocks
+
+
+class MMDiT(nn.Module):
+    def __init__(
+        self,
+        *,
+        dim,
+        depth=8,
+        heads=8,
+        dim_head=64,
+        dropout=0.1,
+        ff_mult=4,
+        text_num_embeds=256,
+        mel_dim=100,
+    ):
+        super().__init__()
+
+        self.time_embed = TimestepEmbedding(dim)
+        self.text_embed = TextEmbedding(dim, text_num_embeds)
+        self.audio_embed = AudioEmbedding(mel_dim, dim)
+
+        self.rotary_embed = RotaryEmbedding(dim_head)
+
+        self.dim = dim
+        self.depth = depth
+
+        self.transformer_blocks = nn.ModuleList(
+            [
+                MMDiTBlock(
+                    dim=dim,
+                    heads=heads,
+                    dim_head=dim_head,
+                    dropout=dropout,
+                    ff_mult=ff_mult,
+                    context_pre_only=i == depth - 1,
+                )
+                for i in range(depth)
+            ]
+        )
+        self.norm_out = AdaLayerNormZero_Final(dim)  # final modulation
+        self.proj_out = nn.Linear(dim, mel_dim)
+
+    def forward(
+        self,
+        x: float["b n d"],  # nosied input audio  # noqa: F722
+        cond: float["b n d"],  # masked cond audio  # noqa: F722
+        text: int["b nt"],  # text  # noqa: F722
+        time: float["b"] | float[""],  # time step  # noqa: F821 F722
+        drop_audio_cond,  # cfg for cond audio
+        drop_text,  # cfg for text
+        mask: bool["b n"] | None = None,  # noqa: F722
+    ):
+        batch = x.shape[0]
+        if time.ndim == 0:
+            time = time.repeat(batch)
+
+        # t: conditioning (time), c: context (text + masked cond audio), x: noised input audio
+        t = self.time_embed(time)
+        c = self.text_embed(text, drop_text=drop_text)
+        x = self.audio_embed(x, cond, drop_audio_cond=drop_audio_cond)
+
+        seq_len = x.shape[1]
+        text_len = text.shape[1]
+        rope_audio = self.rotary_embed.forward_from_seq_len(seq_len)
+        rope_text = self.rotary_embed.forward_from_seq_len(text_len)
+
+        for block in self.transformer_blocks:
+            c, x = block(x, c, t, mask=mask, rope=rope_audio, c_rope=rope_text)
+
+        x = self.norm_out(x, t)
+        output = self.proj_out(x)
+
+        return output

f5_tts/model/backbones/unett.py

@@ -0,0 +1,219 @@
+"""
+ein notation:
+b - batch
+n - sequence
+nt - text sequence
+nw - raw wave length
+d - dimension
+"""
+
+from __future__ import annotations
+from typing import Literal
+
+import torch
+from torch import nn
+import torch.nn.functional as F
+
+from x_transformers import RMSNorm
+from x_transformers.x_transformers import RotaryEmbedding
+
+from f5_tts.model.modules import (
+    TimestepEmbedding,
+    ConvNeXtV2Block,
+    ConvPositionEmbedding,
+    Attention,
+    AttnProcessor,
+    FeedForward,
+    precompute_freqs_cis,
+    get_pos_embed_indices,
+)
+
+
+# Text embedding
+
+
+class TextEmbedding(nn.Module):
+    def __init__(self, text_num_embeds, text_dim, conv_layers=0, conv_mult=2):
+        super().__init__()
+        self.text_embed = nn.Embedding(text_num_embeds + 1, text_dim)  # use 0 as filler token
+
+        if conv_layers > 0:
+            self.extra_modeling = True
+            self.precompute_max_pos = 4096  # ~44s of 24khz audio
+            self.register_buffer("freqs_cis", precompute_freqs_cis(text_dim, self.precompute_max_pos), persistent=False)
+            self.text_blocks = nn.Sequential(
+                *[ConvNeXtV2Block(text_dim, text_dim * conv_mult) for _ in range(conv_layers)]
+            )
+        else:
+            self.extra_modeling = False
+
+    def forward(self, text: int["b nt"], seq_len, drop_text=False):  # noqa: F722
+        text = text + 1  # use 0 as filler token. preprocess of batch pad -1, see list_str_to_idx()
+        text = text[:, :seq_len]  # curtail if character tokens are more than the mel spec tokens
+        batch, text_len = text.shape[0], text.shape[1]
+        text = F.pad(text, (0, seq_len - text_len), value=0)
+
+        if drop_text:  # cfg for text
+            text = torch.zeros_like(text)
+
+        text = self.text_embed(text)  # b n -> b n d
+
+        # possible extra modeling
+        if self.extra_modeling:
+            # sinus pos emb
+            batch_start = torch.zeros((batch,), dtype=torch.long)
+            pos_idx = get_pos_embed_indices(batch_start, seq_len, max_pos=self.precompute_max_pos)
+            text_pos_embed = self.freqs_cis[pos_idx]
+            text = text + text_pos_embed
+
+            # convnextv2 blocks
+            text = self.text_blocks(text)
+
+        return text
+
+
+# noised input audio and context mixing embedding
+
+
+class InputEmbedding(nn.Module):
+    def __init__(self, mel_dim, text_dim, out_dim):
+        super().__init__()
+        self.proj = nn.Linear(mel_dim * 2 + text_dim, out_dim)
+        self.conv_pos_embed = ConvPositionEmbedding(dim=out_dim)
+
+    def forward(self, x: float["b n d"], cond: float["b n d"], text_embed: float["b n d"], drop_audio_cond=False):  # noqa: F722
+        if drop_audio_cond:  # cfg for cond audio
+            cond = torch.zeros_like(cond)
+
+        x = self.proj(torch.cat((x, cond, text_embed), dim=-1))
+        x = self.conv_pos_embed(x) + x
+        return x
+
+
+# Flat UNet Transformer backbone
+
+
+class UNetT(nn.Module):
+    def __init__(
+        self,
+        *,
+        dim,
+        depth=8,
+        heads=8,
+        dim_head=64,
+        dropout=0.1,
+        ff_mult=4,
+        mel_dim=100,
+        text_num_embeds=256,
+        text_dim=None,
+        conv_layers=0,
+        skip_connect_type: Literal["add", "concat", "none"] = "concat",
+    ):
+        super().__init__()
+        assert depth % 2 == 0, "UNet-Transformer's depth should be even."
+
+        self.time_embed = TimestepEmbedding(dim)
+        if text_dim is None:
+            text_dim = mel_dim
+        self.text_embed = TextEmbedding(text_num_embeds, text_dim, conv_layers=conv_layers)
+        self.input_embed = InputEmbedding(mel_dim, text_dim, dim)
+
+        self.rotary_embed = RotaryEmbedding(dim_head)
+
+        # transformer layers & skip connections
+
+        self.dim = dim
+        self.skip_connect_type = skip_connect_type
+        needs_skip_proj = skip_connect_type == "concat"
+
+        self.depth = depth
+        self.layers = nn.ModuleList([])
+
+        for idx in range(depth):
+            is_later_half = idx >= (depth // 2)
+
+            attn_norm = RMSNorm(dim)
+            attn = Attention(
+                processor=AttnProcessor(),
+                dim=dim,
+                heads=heads,
+                dim_head=dim_head,
+                dropout=dropout,
+            )
+
+            ff_norm = RMSNorm(dim)
+            ff = FeedForward(dim=dim, mult=ff_mult, dropout=dropout, approximate="tanh")
+
+            skip_proj = nn.Linear(dim * 2, dim, bias=False) if needs_skip_proj and is_later_half else None
+
+            self.layers.append(
+                nn.ModuleList(
+                    [
+                        skip_proj,
+                        attn_norm,
+                        attn,
+                        ff_norm,
+                        ff,
+                    ]
+                )
+            )
+
+        self.norm_out = RMSNorm(dim)
+        self.proj_out = nn.Linear(dim, mel_dim)
+
+    def forward(
+        self,
+        x: float["b n d"],  # nosied input audio  # noqa: F722
+        cond: float["b n d"],  # masked cond audio  # noqa: F722
+        text: int["b nt"],  # text  # noqa: F722
+        time: float["b"] | float[""],  # time step  # noqa: F821 F722
+        drop_audio_cond,  # cfg for cond audio
+        drop_text,  # cfg for text
+        mask: bool["b n"] | None = None,  # noqa: F722
+    ):
+        batch, seq_len = x.shape[0], x.shape[1]
+        if time.ndim == 0:
+            time = time.repeat(batch)
+
+        # t: conditioning time, c: context (text + masked cond audio), x: noised input audio
+        t = self.time_embed(time)
+        text_embed = self.text_embed(text, seq_len, drop_text=drop_text)
+        x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)
+
+        # postfix time t to input x, [b n d] -> [b n+1 d]
+        x = torch.cat([t.unsqueeze(1), x], dim=1)  # pack t to x
+        if mask is not None:
+            mask = F.pad(mask, (1, 0), value=1)
+
+        rope = self.rotary_embed.forward_from_seq_len(seq_len + 1)
+
+        # flat unet transformer
+        skip_connect_type = self.skip_connect_type
+        skips = []
+        for idx, (maybe_skip_proj, attn_norm, attn, ff_norm, ff) in enumerate(self.layers):
+            layer = idx + 1
+
+            # skip connection logic
+            is_first_half = layer <= (self.depth // 2)
+            is_later_half = not is_first_half
+
+            if is_first_half:
+                skips.append(x)
+
+            if is_later_half:
+                skip = skips.pop()
+                if skip_connect_type == "concat":
+                    x = torch.cat((x, skip), dim=-1)
+                    x = maybe_skip_proj(x)
+                elif skip_connect_type == "add":
+                    x = x + skip
+
+            # attention and feedforward blocks
+            x = attn(attn_norm(x), rope=rope, mask=mask) + x
+            x = ff(ff_norm(x)) + x
+
+        assert len(skips) == 0
+
+        x = self.norm_out(x)[:, 1:, :]  # unpack t from x
+
+        return self.proj_out(x)

f5_tts/model/modules.py

@@ -0,0 +1,666 @@
+"""
+ein notation:
+b - batch
+n - sequence
+nt - text sequence
+nw - raw wave length
+d - dimension
+"""
+
+from __future__ import annotations
+
+import math
+from typing import Optional
+
+import torch
+import torch.nn.functional as F
+import torchaudio
+from librosa.filters import mel as librosa_mel_fn
+from torch import nn
+from x_transformers.x_transformers import apply_rotary_pos_emb
+
+
+# raw wav to mel spec
+
+
+mel_basis_cache = {}
+hann_window_cache = {}
+
+
+def get_bigvgan_mel_spectrogram(
+    waveform,
+    n_fft=1024,
+    n_mel_channels=100,
+    target_sample_rate=24000,
+    hop_length=256,
+    win_length=1024,
+    fmin=0,
+    fmax=None,
+    center=False,
+):  # Copy from https://github.com/NVIDIA/BigVGAN/tree/main
+    device = waveform.device
+    key = f"{n_fft}_{n_mel_channels}_{target_sample_rate}_{hop_length}_{win_length}_{fmin}_{fmax}_{device}"
+
+    if key not in mel_basis_cache:
+        mel = librosa_mel_fn(sr=target_sample_rate, n_fft=n_fft, n_mels=n_mel_channels, fmin=fmin, fmax=fmax)
+        mel_basis_cache[key] = torch.from_numpy(mel).float().to(device)  # TODO: why they need .float()?
+        hann_window_cache[key] = torch.hann_window(win_length).to(device)
+
+    mel_basis = mel_basis_cache[key]
+    hann_window = hann_window_cache[key]
+
+    padding = (n_fft - hop_length) // 2
+    waveform = torch.nn.functional.pad(waveform.unsqueeze(1), (padding, padding), mode="reflect").squeeze(1)
+
+    spec = torch.stft(
+        waveform,
+        n_fft,
+        hop_length=hop_length,
+        win_length=win_length,
+        window=hann_window,
+        center=center,
+        pad_mode="reflect",
+        normalized=False,
+        onesided=True,
+        return_complex=True,
+    )
+    spec = torch.sqrt(torch.view_as_real(spec).pow(2).sum(-1) + 1e-9)
+
+    mel_spec = torch.matmul(mel_basis, spec)
+    mel_spec = torch.log(torch.clamp(mel_spec, min=1e-5))
+
+    return mel_spec
+
+
+def get_vocos_mel_spectrogram(
+    waveform,
+    n_fft=1024,
+    n_mel_channels=100,
+    target_sample_rate=24000,
+    hop_length=256,
+    win_length=1024,
+):
+    mel_stft = torchaudio.transforms.MelSpectrogram(
+        sample_rate=target_sample_rate,
+        n_fft=n_fft,
+        win_length=win_length,
+        hop_length=hop_length,
+        n_mels=n_mel_channels,
+        power=1,
+        center=True,
+        normalized=False,
+        norm=None,
+    ).to(waveform.device)
+    if len(waveform.shape) == 3:
+        waveform = waveform.squeeze(1)  # 'b 1 nw -> b nw'
+
+    assert len(waveform.shape) == 2
+
+    mel = mel_stft(waveform)
+    mel = mel.clamp(min=1e-5).log()
+    return mel
+
+
+class MelSpec(nn.Module):
+    def __init__(
+        self,
+        n_fft=1024,
+        hop_length=256,
+        win_length=1024,
+        n_mel_channels=100,
+        target_sample_rate=24_000,
+        mel_spec_type="vocos",
+    ):
+        super().__init__()
+        assert mel_spec_type in ["vocos", "bigvgan"], print("We only support two extract mel backend: vocos or bigvgan")
+
+        self.n_fft = n_fft
+        self.hop_length = hop_length
+        self.win_length = win_length
+        self.n_mel_channels = n_mel_channels
+        self.target_sample_rate = target_sample_rate
+
+        if mel_spec_type == "vocos":
+            self.extractor = get_vocos_mel_spectrogram
+        elif mel_spec_type == "bigvgan":
+            self.extractor = get_bigvgan_mel_spectrogram
+
+        self.register_buffer("dummy", torch.tensor(0), persistent=False)
+
+    def forward(self, wav):
+        if self.dummy.device != wav.device:
+            self.to(wav.device)
+
+        mel = self.extractor(
+            waveform=wav,
+            n_fft=self.n_fft,
+            n_mel_channels=self.n_mel_channels,
+            target_sample_rate=self.target_sample_rate,
+            hop_length=self.hop_length,
+            win_length=self.win_length,
+        )
+
+        return mel
+
+
+# sinusoidal position embedding
+
+
+class SinusPositionEmbedding(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.dim = dim
+
+    def forward(self, x, scale=1000):
+        device = x.device
+        half_dim = self.dim // 2
+        emb = math.log(10000) / (half_dim - 1)
+        emb = torch.exp(torch.arange(half_dim, device=device).float() * -emb)
+        emb = scale * x.unsqueeze(1) * emb.unsqueeze(0)
+        emb = torch.cat((emb.sin(), emb.cos()), dim=-1)
+        return emb
+
+
+# convolutional position embedding
+
+
+class ConvPositionEmbedding(nn.Module):
+    def __init__(self, dim, kernel_size=31, groups=16):
+        super().__init__()
+        assert kernel_size % 2 != 0
+        self.conv1d = nn.Sequential(
+            nn.Conv1d(dim, dim, kernel_size, groups=groups, padding=kernel_size // 2),
+            nn.Mish(),
+            nn.Conv1d(dim, dim, kernel_size, groups=groups, padding=kernel_size // 2),
+            nn.Mish(),
+        )
+
+    def forward(self, x: float["b n d"], mask: bool["b n"] | None = None):  # noqa: F722
+        if mask is not None:
+            mask = mask[..., None]
+            x = x.masked_fill(~mask, 0.0)
+
+        x = x.permute(0, 2, 1)
+        x = self.conv1d(x)
+        out = x.permute(0, 2, 1)
+
+        if mask is not None:
+            out = out.masked_fill(~mask, 0.0)
+
+        return out
+
+
+# rotary positional embedding related
+
+
+def precompute_freqs_cis(dim: int, end: int, theta: float = 10000.0, theta_rescale_factor=1.0):
+    # proposed by reddit user bloc97, to rescale rotary embeddings to longer sequence length without fine-tuning
+    # has some connection to NTK literature
+    # https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_scaled_rope_allows_llama_models_to_have/
+    # https://github.com/lucidrains/rotary-embedding-torch/blob/main/rotary_embedding_torch/rotary_embedding_torch.py
+    theta *= theta_rescale_factor ** (dim / (dim - 2))
+    freqs = 1.0 / (theta ** (torch.arange(0, dim, 2)[: (dim // 2)].float() / dim))
+    t = torch.arange(end, device=freqs.device)  # type: ignore
+    freqs = torch.outer(t, freqs).float()  # type: ignore
+    freqs_cos = torch.cos(freqs)  # real part
+    freqs_sin = torch.sin(freqs)  # imaginary part
+    return torch.cat([freqs_cos, freqs_sin], dim=-1)
+
+
+def get_pos_embed_indices(start, length, max_pos, scale=1.0):
+    # length = length if isinstance(length, int) else length.max()
+    scale = scale * torch.ones_like(start, dtype=torch.float32)  # in case scale is a scalar
+    pos = (
+        start.unsqueeze(1)
+        + (torch.arange(length, device=start.device, dtype=torch.float32).unsqueeze(0) * scale.unsqueeze(1)).long()
+    )
+    # avoid extra long error.
+    pos = torch.where(pos < max_pos, pos, max_pos - 1)
+    return pos
+
+
+# Global Response Normalization layer (Instance Normalization ?)
+
+
+class GRN(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.gamma = nn.Parameter(torch.zeros(1, 1, dim))
+        self.beta = nn.Parameter(torch.zeros(1, 1, dim))
+
+    def forward(self, x):
+        Gx = torch.norm(x, p=2, dim=1, keepdim=True)
+        Nx = Gx / (Gx.mean(dim=-1, keepdim=True) + 1e-6)
+        return self.gamma * (x * Nx) + self.beta + x
+
+
+# ConvNeXt-V2 Block https://github.com/facebookresearch/ConvNeXt-V2/blob/main/models/convnextv2.py
+# ref: https://github.com/bfs18/e2_tts/blob/main/rfwave/modules.py#L108
+
+
+class ConvNeXtV2Block(nn.Module):
+    def __init__(
+        self,
+        dim: int,
+        intermediate_dim: int,
+        dilation: int = 1,
+    ):
+        super().__init__()
+        padding = (dilation * (7 - 1)) // 2
+        self.dwconv = nn.Conv1d(
+            dim, dim, kernel_size=7, padding=padding, groups=dim, dilation=dilation
+        )  # depthwise conv
+        self.norm = nn.LayerNorm(dim, eps=1e-6)
+        self.pwconv1 = nn.Linear(dim, intermediate_dim)  # pointwise/1x1 convs, implemented with linear layers
+        self.act = nn.GELU()
+        self.grn = GRN(intermediate_dim)
+        self.pwconv2 = nn.Linear(intermediate_dim, dim)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        residual = x
+        x = x.transpose(1, 2)  # b n d -> b d n
+        x = self.dwconv(x)
+        x = x.transpose(1, 2)  # b d n -> b n d
+        x = self.norm(x)
+        x = self.pwconv1(x)
+        x = self.act(x)
+        x = self.grn(x)
+        x = self.pwconv2(x)
+        return residual + x
+
+
+# AdaLayerNormZero
+# return with modulated x for attn input, and params for later mlp modulation
+
+
+class AdaLayerNormZero(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+
+        self.silu = nn.SiLU()
+        self.linear = nn.Linear(dim, dim * 6)
+
+        self.norm = nn.LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+
+    def forward(self, x, emb=None):
+        emb = self.linear(self.silu(emb))
+        shift_msa, scale_msa, gate_msa, shift_mlp, scale_mlp, gate_mlp = torch.chunk(emb, 6, dim=1)
+
+        x = self.norm(x) * (1 + scale_msa[:, None]) + shift_msa[:, None]
+        return x, gate_msa, shift_mlp, scale_mlp, gate_mlp
+
+
+# AdaLayerNormZero for final layer
+# return only with modulated x for attn input, cuz no more mlp modulation
+
+
+class AdaLayerNormZero_Final(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+
+        self.silu = nn.SiLU()
+        self.linear = nn.Linear(dim, dim * 2)
+
+        self.norm = nn.LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+
+    def forward(self, x, emb):
+        emb = self.linear(self.silu(emb))
+        scale, shift = torch.chunk(emb, 2, dim=1)
+
+        x = self.norm(x) * (1 + scale)[:, None, :] + shift[:, None, :]
+        return x
+
+
+# FeedForward
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, dim_out=None, mult=4, dropout=0.0, approximate: str = "none"):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        dim_out = dim_out if dim_out is not None else dim
+
+        activation = nn.GELU(approximate=approximate)
+        project_in = nn.Sequential(nn.Linear(dim, inner_dim), activation)
+        self.ff = nn.Sequential(project_in, nn.Dropout(dropout), nn.Linear(inner_dim, dim_out))
+
+    def forward(self, x):
+        return self.ff(x)
+
+
+# Attention with possible joint part
+# modified from diffusers/src/diffusers/models/attention_processor.py
+
+
+class Attention(nn.Module):
+    def __init__(
+        self,
+        processor: JointAttnProcessor | AttnProcessor,
+        dim: int,
+        heads: int = 8,
+        dim_head: int = 64,
+        dropout: float = 0.0,
+        context_dim: Optional[int] = None,  # if not None -> joint attention
+        context_pre_only=None,
+    ):
+        super().__init__()
+
+        if not hasattr(F, "scaled_dot_product_attention"):
+            raise ImportError("Attention equires PyTorch 2.0, to use it, please upgrade PyTorch to 2.0.")
+
+        self.processor = processor
+
+        self.dim = dim
+        self.heads = heads
+        self.inner_dim = dim_head * heads
+        self.dropout = dropout
+
+        self.context_dim = context_dim
+        self.context_pre_only = context_pre_only
+
+        self.to_q = nn.Linear(dim, self.inner_dim)
+        self.to_k = nn.Linear(dim, self.inner_dim)
+        self.to_v = nn.Linear(dim, self.inner_dim)
+
+        if self.context_dim is not None:
+            self.to_k_c = nn.Linear(context_dim, self.inner_dim)
+            self.to_v_c = nn.Linear(context_dim, self.inner_dim)
+            if self.context_pre_only is not None:
+                self.to_q_c = nn.Linear(context_dim, self.inner_dim)
+
+        self.to_out = nn.ModuleList([])
+        self.to_out.append(nn.Linear(self.inner_dim, dim))
+        self.to_out.append(nn.Dropout(dropout))
+
+        if self.context_pre_only is not None and not self.context_pre_only:
+            self.to_out_c = nn.Linear(self.inner_dim, dim)
+
+    def forward(
+        self,
+        x: float["b n d"],  # noised input x  # noqa: F722
+        c: float["b n d"] = None,  # context c  # noqa: F722
+        mask: bool["b n"] | None = None,  # noqa: F722
+        rope=None,  # rotary position embedding for x
+        c_rope=None,  # rotary position embedding for c
+    ) -> torch.Tensor:
+        if c is not None:
+            return self.processor(self, x, c=c, mask=mask, rope=rope, c_rope=c_rope)
+        else:
+            return self.processor(self, x, mask=mask, rope=rope)
+
+
+# Attention processor
+
+
+# from torch.nn.attention import SDPBackend
+# torch.backends.cuda.enable_flash_sdp(True)
+class AttnProcessor:
+    def __init__(self):
+        pass
+
+    def __call__(
+        self,
+        attn: Attention,
+        x: float["b n d"],  # noised input x  # noqa: F722
+        mask: bool["b n"] | None = None,  # noqa: F722
+        rope=None,  # rotary position embedding
+    ) -> torch.FloatTensor:
+        batch_size = x.shape[0]
+
+        # `sample` projections.
+        query = attn.to_q(x)
+        key = attn.to_k(x)
+        value = attn.to_v(x)
+
+        # apply rotary position embedding
+        if rope is not None:
+            freqs, xpos_scale = rope
+            q_xpos_scale, k_xpos_scale = (xpos_scale, xpos_scale**-1.0) if xpos_scale is not None else (1.0, 1.0)
+
+            query = apply_rotary_pos_emb(query, freqs, q_xpos_scale)
+            key = apply_rotary_pos_emb(key, freqs, k_xpos_scale)
+
+        # attention
+        inner_dim = key.shape[-1]
+        head_dim = inner_dim // attn.heads
+        query = query.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+        key = key.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+        value = value.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+
+        # mask. e.g. inference got a batch with different target durations, mask out the padding
+        if mask is not None:
+            attn_mask = mask
+            attn_mask = attn_mask.unsqueeze(1).unsqueeze(1)  # 'b n -> b 1 1 n'
+            # print(3433333333,attn_mask.shape)
+            attn_mask = attn_mask.expand(batch_size, attn.heads, query.shape[-2], key.shape[-2])
+        else:
+            attn_mask = None
+        # with torch.nn.attention.sdpa_kernel(backends=[SDPBackend.EFFICIENT_ATTENTION]):
+        # with torch.backends.cuda.sdp_kernel(enable_flash=True, enable_math=False, enable_mem_efficient=True):
+        # with torch.backends.cuda.sdp_kernel(enable_flash=True, enable_math=True, enable_mem_efficient=False):
+        #     print(torch.backends.cuda.flash_sdp_enabled())
+        #     print(torch.backends.cuda.mem_efficient_sdp_enabled())
+        #     print(torch.backends.cuda.math_sdp_enabled())
+        x = F.scaled_dot_product_attention(query, key, value, attn_mask=attn_mask, dropout_p=0.0, is_causal=False)
+        x = x.transpose(1, 2).reshape(batch_size, -1, attn.heads * head_dim)
+        x = x.to(query.dtype)
+
+        # linear proj
+        x = attn.to_out[0](x)
+        # dropout
+        x = attn.to_out[1](x)
+
+        if mask is not None:
+            mask = mask.unsqueeze(-1)
+            x = x.masked_fill(~mask, 0.0)
+
+        return x
+
+
+# Joint Attention processor for MM-DiT
+# modified from diffusers/src/diffusers/models/attention_processor.py
+
+
+class JointAttnProcessor:
+    def __init__(self):
+        pass
+
+    def __call__(
+        self,
+        attn: Attention,
+        x: float["b n d"],  # noised input x  # noqa: F722
+        c: float["b nt d"] = None,  # context c, here text # noqa: F722
+        mask: bool["b n"] | None = None,  # noqa: F722
+        rope=None,  # rotary position embedding for x
+        c_rope=None,  # rotary position embedding for c
+    ) -> torch.FloatTensor:
+        residual = x
+
+        batch_size = c.shape[0]
+
+        # `sample` projections.
+        query = attn.to_q(x)
+        key = attn.to_k(x)
+        value = attn.to_v(x)
+
+        # `context` projections.
+        c_query = attn.to_q_c(c)
+        c_key = attn.to_k_c(c)
+        c_value = attn.to_v_c(c)
+
+        # apply rope for context and noised input independently
+        if rope is not None:
+            freqs, xpos_scale = rope
+            q_xpos_scale, k_xpos_scale = (xpos_scale, xpos_scale**-1.0) if xpos_scale is not None else (1.0, 1.0)
+            query = apply_rotary_pos_emb(query, freqs, q_xpos_scale)
+            key = apply_rotary_pos_emb(key, freqs, k_xpos_scale)
+        if c_rope is not None:
+            freqs, xpos_scale = c_rope
+            q_xpos_scale, k_xpos_scale = (xpos_scale, xpos_scale**-1.0) if xpos_scale is not None else (1.0, 1.0)
+            c_query = apply_rotary_pos_emb(c_query, freqs, q_xpos_scale)
+            c_key = apply_rotary_pos_emb(c_key, freqs, k_xpos_scale)
+
+        # attention
+        query = torch.cat([query, c_query], dim=1)
+        key = torch.cat([key, c_key], dim=1)
+        value = torch.cat([value, c_value], dim=1)
+
+        inner_dim = key.shape[-1]
+        head_dim = inner_dim // attn.heads
+        query = query.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+        key = key.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+        value = value.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
+
+        # mask. e.g. inference got a batch with different target durations, mask out the padding
+        if mask is not None:
+            attn_mask = F.pad(mask, (0, c.shape[1]), value=True)  # no mask for c (text)
+            attn_mask = attn_mask.unsqueeze(1).unsqueeze(1)  # 'b n -> b 1 1 n'
+            attn_mask = attn_mask.expand(batch_size, attn.heads, query.shape[-2], key.shape[-2])
+        else:
+            attn_mask = None
+
+        x = F.scaled_dot_product_attention(query, key, value, attn_mask=attn_mask, dropout_p=0.0, is_causal=False)
+        x = x.transpose(1, 2).reshape(batch_size, -1, attn.heads * head_dim)
+        x = x.to(query.dtype)
+
+        # Split the attention outputs.
+        x, c = (
+            x[:, : residual.shape[1]],
+            x[:, residual.shape[1] :],
+        )
+
+        # linear proj
+        x = attn.to_out[0](x)
+        # dropout
+        x = attn.to_out[1](x)
+        if not attn.context_pre_only:
+            c = attn.to_out_c(c)
+
+        if mask is not None:
+            mask = mask.unsqueeze(-1)
+            x = x.masked_fill(~mask, 0.0)
+            # c = c.masked_fill(~mask, 0.)  # no mask for c (text)
+
+        return x, c
+
+
+# DiT Block
+
+
+class DiTBlock(nn.Module):
+    def __init__(self, dim, heads, dim_head, ff_mult=4, dropout=0.1):
+        super().__init__()
+
+        self.attn_norm = AdaLayerNormZero(dim)
+        self.attn = Attention(
+            processor=AttnProcessor(),
+            dim=dim,
+            heads=heads,
+            dim_head=dim_head,
+            dropout=dropout,
+        )
+
+        self.ff_norm = nn.LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+        self.ff = FeedForward(dim=dim, mult=ff_mult, dropout=dropout, approximate="tanh")
+
+    def forward(self, x, t, mask=None, rope=None):  # x: noised input, t: time embedding
+        # pre-norm & modulation for attention input
+        norm, gate_msa, shift_mlp, scale_mlp, gate_mlp = self.attn_norm(x, emb=t)
+
+        # attention
+        attn_output = self.attn(x=norm, mask=mask, rope=rope)
+
+        # process attention output for input x
+        x = x + gate_msa.unsqueeze(1) * attn_output
+
+        norm = self.ff_norm(x) * (1 + scale_mlp[:, None]) + shift_mlp[:, None]
+        ff_output = self.ff(norm)
+        x = x + gate_mlp.unsqueeze(1) * ff_output
+
+        return x
+
+
+# MMDiT Block https://arxiv.org/abs/2403.03206
+
+
+class MMDiTBlock(nn.Module):
+    r"""
+    modified from diffusers/src/diffusers/models/attention.py
+
+    notes.
+    _c: context related. text, cond, etc. (left part in sd3 fig2.b)
+    _x: noised input related. (right part)
+    context_pre_only: last layer only do prenorm + modulation cuz no more ffn
+    """
+
+    def __init__(self, dim, heads, dim_head, ff_mult=4, dropout=0.1, context_pre_only=False):
+        super().__init__()
+
+        self.context_pre_only = context_pre_only
+
+        self.attn_norm_c = AdaLayerNormZero_Final(dim) if context_pre_only else AdaLayerNormZero(dim)
+        self.attn_norm_x = AdaLayerNormZero(dim)
+        self.attn = Attention(
+            processor=JointAttnProcessor(),
+            dim=dim,
+            heads=heads,
+            dim_head=dim_head,
+            dropout=dropout,
+            context_dim=dim,
+            context_pre_only=context_pre_only,
+        )
+
+        if not context_pre_only:
+            self.ff_norm_c = nn.LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+            self.ff_c = FeedForward(dim=dim, mult=ff_mult, dropout=dropout, approximate="tanh")
+        else:
+            self.ff_norm_c = None
+            self.ff_c = None
+        self.ff_norm_x = nn.LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+        self.ff_x = FeedForward(dim=dim, mult=ff_mult, dropout=dropout, approximate="tanh")
+
+    def forward(self, x, c, t, mask=None, rope=None, c_rope=None):  # x: noised input, c: context, t: time embedding
+        # pre-norm & modulation for attention input
+        if self.context_pre_only:
+            norm_c = self.attn_norm_c(c, t)
+        else:
+            norm_c, c_gate_msa, c_shift_mlp, c_scale_mlp, c_gate_mlp = self.attn_norm_c(c, emb=t)
+        norm_x, x_gate_msa, x_shift_mlp, x_scale_mlp, x_gate_mlp = self.attn_norm_x(x, emb=t)
+
+        # attention
+        x_attn_output, c_attn_output = self.attn(x=norm_x, c=norm_c, mask=mask, rope=rope, c_rope=c_rope)
+
+        # process attention output for context c
+        if self.context_pre_only:
+            c = None
+        else:  # if not last layer
+            c = c + c_gate_msa.unsqueeze(1) * c_attn_output
+
+            norm_c = self.ff_norm_c(c) * (1 + c_scale_mlp[:, None]) + c_shift_mlp[:, None]
+            c_ff_output = self.ff_c(norm_c)
+            c = c + c_gate_mlp.unsqueeze(1) * c_ff_output
+
+        # process attention output for input x
+        x = x + x_gate_msa.unsqueeze(1) * x_attn_output
+
+        norm_x = self.ff_norm_x(x) * (1 + x_scale_mlp[:, None]) + x_shift_mlp[:, None]
+        x_ff_output = self.ff_x(norm_x)
+        x = x + x_gate_mlp.unsqueeze(1) * x_ff_output
+
+        return c, x
+
+
+# time step conditioning embedding
+
+
+class TimestepEmbedding(nn.Module):
+    def __init__(self, dim, freq_embed_dim=256):
+        super().__init__()
+        self.time_embed = SinusPositionEmbedding(freq_embed_dim)
+        self.time_mlp = nn.Sequential(nn.Linear(freq_embed_dim, dim), nn.SiLU(), nn.Linear(dim, dim))
+
+    def forward(self, timestep: float["b"]):  # noqa: F821
+        time_hidden = self.time_embed(timestep)
+        time_hidden = time_hidden.to(timestep.dtype)
+        time = self.time_mlp(time_hidden)  # b d
+        return time

prepare_datasets/1-get-text.py

@@ -0,0 +1,143 @@
+# -*- coding: utf-8 -*-
+
+import os
+
+inp_text = os.environ.get("inp_text")
+inp_wav_dir = os.environ.get("inp_wav_dir")
+exp_name = os.environ.get("exp_name")
+i_part = os.environ.get("i_part")
+all_parts = os.environ.get("all_parts")
+if "_CUDA_VISIBLE_DEVICES" in os.environ:
+    os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
+opt_dir = os.environ.get("opt_dir")
+bert_pretrained_dir = os.environ.get("bert_pretrained_dir")
+import torch
+
+is_half = eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
+version = os.environ.get("version", None)
+import traceback
+import os.path
+from text.cleaner import clean_text
+from transformers import AutoModelForMaskedLM, AutoTokenizer
+from tools.my_utils import clean_path
+
+# inp_text=sys.argv[1]
+# inp_wav_dir=sys.argv[2]
+# exp_name=sys.argv[3]
+# i_part=sys.argv[4]
+# all_parts=sys.argv[5]
+# os.environ["CUDA_VISIBLE_DEVICES"]=sys.argv[6]#i_gpu
+# opt_dir="/data/docker/liujing04/gpt-vits/fine_tune_dataset/%s"%exp_name
+# bert_pretrained_dir="/data/docker/liujing04/bert-vits2/Bert-VITS2-master20231106/bert/chinese-roberta-wwm-ext-large"
+
+from time import time as ttime
+import shutil
+
+
+def my_save(fea, path):  #####fix issue: torch.save doesn't support chinese path
+    dir = os.path.dirname(path)
+    name = os.path.basename(path)
+    # tmp_path="%s/%s%s.pth"%(dir,ttime(),i_part)
+    tmp_path = "%s%s.pth" % (ttime(), i_part)
+    torch.save(fea, tmp_path)
+    shutil.move(tmp_path, "%s/%s" % (dir, name))
+
+
+txt_path = "%s/2-name2text-%s.txt" % (opt_dir, i_part)
+if os.path.exists(txt_path) == False:
+    bert_dir = "%s/3-bert" % (opt_dir)
+    os.makedirs(opt_dir, exist_ok=True)
+    os.makedirs(bert_dir, exist_ok=True)
+    if torch.cuda.is_available():
+        device = "cuda:0"
+    # elif torch.backends.mps.is_available():
+    #     device = "mps"
+    else:
+        device = "cpu"
+    if os.path.exists(bert_pretrained_dir):
+        ...
+    else:
+        raise FileNotFoundError(bert_pretrained_dir)
+    tokenizer = AutoTokenizer.from_pretrained(bert_pretrained_dir)
+    bert_model = AutoModelForMaskedLM.from_pretrained(bert_pretrained_dir)
+    if is_half == True:
+        bert_model = bert_model.half().to(device)
+    else:
+        bert_model = bert_model.to(device)
+
+    def get_bert_feature(text, word2ph):
+        with torch.no_grad():
+            inputs = tokenizer(text, return_tensors="pt")
+            for i in inputs:
+                inputs[i] = inputs[i].to(device)
+            res = bert_model(**inputs, output_hidden_states=True)
+            res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()[1:-1]
+
+        assert len(word2ph) == len(text)
+        phone_level_feature = []
+        for i in range(len(word2ph)):
+            repeat_feature = res[i].repeat(word2ph[i], 1)
+            phone_level_feature.append(repeat_feature)
+
+        phone_level_feature = torch.cat(phone_level_feature, dim=0)
+
+        return phone_level_feature.T
+
+    def process(data, res):
+        for name, text, lan in data:
+            try:
+                name = clean_path(name)
+                name = os.path.basename(name)
+                print(name)
+                phones, word2ph, norm_text = clean_text(text.replace("%", "-").replace("￥", ","), lan, version)
+                path_bert = "%s/%s.pt" % (bert_dir, name)
+                if os.path.exists(path_bert) == False and lan == "zh":
+                    bert_feature = get_bert_feature(norm_text, word2ph)
+                    assert bert_feature.shape[-1] == len(phones)
+                    # torch.save(bert_feature, path_bert)
+                    my_save(bert_feature, path_bert)
+                phones = " ".join(phones)
+                # res.append([name,phones])
+                res.append([name, phones, word2ph, norm_text])
+            except:
+                print(name, text, traceback.format_exc())
+
+    todo = []
+    res = []
+    with open(inp_text, "r", encoding="utf8") as f:
+        lines = f.read().strip("\n").split("\n")
+
+    language_v1_to_language_v2 = {
+        "ZH": "zh",
+        "zh": "zh",
+        "JP": "ja",
+        "jp": "ja",
+        "JA": "ja",
+        "ja": "ja",
+        "EN": "en",
+        "en": "en",
+        "En": "en",
+        "KO": "ko",
+        "Ko": "ko",
+        "ko": "ko",
+        "yue": "yue",
+        "YUE": "yue",
+        "Yue": "yue",
+    }
+    for line in lines[int(i_part) :: int(all_parts)]:
+        try:
+            wav_name, spk_name, language, text = line.split("|")
+            # todo.append([name,text,"zh"])
+            if language in language_v1_to_language_v2.keys():
+                todo.append([wav_name, text, language_v1_to_language_v2.get(language, language)])
+            else:
+                print(f"\033[33m[Waring] The {language = } of {wav_name} is not supported for training.\033[0m")
+        except:
+            print(line, traceback.format_exc())
+
+    process(todo, res)
+    opt = []
+    for name, phones, word2ph, norm_text in res:
+        opt.append("%s\t%s\t%s\t%s" % (name, phones, word2ph, norm_text))
+    with open(txt_path, "w", encoding="utf8") as f:
+        f.write("\n".join(opt) + "\n")

prepare_datasets/2-get-hubert-wav32k.py

@@ -0,0 +1,134 @@
+# -*- coding: utf-8 -*-
+
+import sys
+import os
+
+inp_text = os.environ.get("inp_text")
+inp_wav_dir = os.environ.get("inp_wav_dir")
+exp_name = os.environ.get("exp_name")
+i_part = os.environ.get("i_part")
+all_parts = os.environ.get("all_parts")
+if "_CUDA_VISIBLE_DEVICES" in os.environ:
+    os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
+from feature_extractor import cnhubert
+
+opt_dir = os.environ.get("opt_dir")
+cnhubert.cnhubert_base_path = os.environ.get("cnhubert_base_dir")
+import torch
+
+is_half = eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
+
+import traceback
+import numpy as np
+from scipy.io import wavfile
+import librosa
+
+now_dir = os.getcwd()
+sys.path.append(now_dir)
+from tools.my_utils import load_audio, clean_path
+
+# from config import cnhubert_base_path
+# cnhubert.cnhubert_base_path=cnhubert_base_path
+# inp_text=sys.argv[1]
+# inp_wav_dir=sys.argv[2]
+# exp_name=sys.argv[3]
+# i_part=sys.argv[4]
+# all_parts=sys.argv[5]
+# os.environ["CUDA_VISIBLE_DEVICES"]=sys.argv[6]
+# cnhubert.cnhubert_base_path=sys.argv[7]
+# opt_dir="/data/docker/liujing04/gpt-vits/fine_tune_dataset/%s"%exp_name
+
+from time import time as ttime
+import shutil
+
+
+def my_save(fea, path):  #####fix issue: torch.save doesn't support chinese path
+    dir = os.path.dirname(path)
+    name = os.path.basename(path)
+    # tmp_path="%s/%s%s.pth"%(dir,ttime(),i_part)
+    tmp_path = "%s%s.pth" % (ttime(), i_part)
+    torch.save(fea, tmp_path)
+    shutil.move(tmp_path, "%s/%s" % (dir, name))
+
+
+hubert_dir = "%s/4-cnhubert" % (opt_dir)
+wav32dir = "%s/5-wav32k" % (opt_dir)
+os.makedirs(opt_dir, exist_ok=True)
+os.makedirs(hubert_dir, exist_ok=True)
+os.makedirs(wav32dir, exist_ok=True)
+
+maxx = 0.95
+alpha = 0.5
+if torch.cuda.is_available():
+    device = "cuda:0"
+# elif torch.backends.mps.is_available():
+#     device = "mps"
+else:
+    device = "cpu"
+model = cnhubert.get_model()
+# is_half=False
+if is_half == True:
+    model = model.half().to(device)
+else:
+    model = model.to(device)
+
+nan_fails = []
+
+
+def name2go(wav_name, wav_path):
+    hubert_path = "%s/%s.pt" % (hubert_dir, wav_name)
+    if os.path.exists(hubert_path):
+        return
+    tmp_audio = load_audio(wav_path, 32000)
+    tmp_max = np.abs(tmp_audio).max()
+    if tmp_max > 2.2:
+        print("%s-filtered,%s" % (wav_name, tmp_max))
+        return
+    tmp_audio32 = (tmp_audio / tmp_max * (maxx * alpha * 32768)) + ((1 - alpha) * 32768) * tmp_audio
+    tmp_audio32b = (tmp_audio / tmp_max * (maxx * alpha * 1145.14)) + ((1 - alpha) * 1145.14) * tmp_audio
+    tmp_audio = librosa.resample(tmp_audio32b, orig_sr=32000, target_sr=16000)  # 不是重采样问题
+    tensor_wav16 = torch.from_numpy(tmp_audio)
+    if is_half == True:
+        tensor_wav16 = tensor_wav16.half().to(device)
+    else:
+        tensor_wav16 = tensor_wav16.to(device)
+    ssl = model.model(tensor_wav16.unsqueeze(0))["last_hidden_state"].transpose(1, 2).cpu()  # torch.Size([1, 768, 215])
+    if np.isnan(ssl.detach().numpy()).sum() != 0:
+        nan_fails.append((wav_name, wav_path))
+        print("nan filtered:%s" % wav_name)
+        return
+    wavfile.write(
+        "%s/%s" % (wav32dir, wav_name),
+        32000,
+        tmp_audio32.astype("int16"),
+    )
+    my_save(ssl, hubert_path)
+
+
+with open(inp_text, "r", encoding="utf8") as f:
+    lines = f.read().strip("\n").split("\n")
+
+for line in lines[int(i_part) :: int(all_parts)]:
+    try:
+        # wav_name,text=line.split("\t")
+        wav_name, spk_name, language, text = line.split("|")
+        wav_name = clean_path(wav_name)
+        if inp_wav_dir != "" and inp_wav_dir != None:
+            wav_name = os.path.basename(wav_name)
+            wav_path = "%s/%s" % (inp_wav_dir, wav_name)
+
+        else:
+            wav_path = wav_name
+            wav_name = os.path.basename(wav_name)
+        name2go(wav_name, wav_path)
+    except:
+        print(line, traceback.format_exc())
+
+if len(nan_fails) > 0 and is_half == True:
+    is_half = False
+    model = model.float()
+    for wav in nan_fails:
+        try:
+            name2go(wav[0], wav[1])
+        except:
+            print(wav_name, traceback.format_exc())

prepare_datasets/2-get-sv.py

@@ -0,0 +1,115 @@
+# -*- coding: utf-8 -*-
+
+import sys
+import os
+
+inp_text = os.environ.get("inp_text")
+inp_wav_dir = os.environ.get("inp_wav_dir")
+exp_name = os.environ.get("exp_name")
+i_part = os.environ.get("i_part")
+all_parts = os.environ.get("all_parts")
+if "_CUDA_VISIBLE_DEVICES" in os.environ:
+    os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
+
+opt_dir = os.environ.get("opt_dir")
+sv_path = os.environ.get("sv_path")
+import torch
+
+is_half = eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
+
+import traceback
+import torchaudio
+
+now_dir = os.getcwd()
+sys.path.append(now_dir)
+sys.path.append(f"{now_dir}/GPT_SoVITS/eres2net")
+from tools.my_utils import clean_path
+from time import time as ttime
+import shutil
+from ERes2NetV2 import ERes2NetV2
+import kaldi as Kaldi
+
+
+def my_save(fea, path):  #####fix issue: torch.save doesn't support chinese path
+    dir = os.path.dirname(path)
+    name = os.path.basename(path)
+    # tmp_path="%s/%s%s.pth"%(dir,ttime(),i_part)
+    tmp_path = "%s%s.pth" % (ttime(), i_part)
+    torch.save(fea, tmp_path)
+    shutil.move(tmp_path, "%s/%s" % (dir, name))
+
+
+sv_cn_dir = "%s/7-sv_cn" % (opt_dir)
+wav32dir = "%s/5-wav32k" % (opt_dir)
+os.makedirs(opt_dir, exist_ok=True)
+os.makedirs(sv_cn_dir, exist_ok=True)
+os.makedirs(wav32dir, exist_ok=True)
+
+maxx = 0.95
+alpha = 0.5
+if torch.cuda.is_available():
+    device = "cuda:0"
+# elif torch.backends.mps.is_available():
+#     device = "mps"
+else:
+    device = "cpu"
+
+
+class SV:
+    def __init__(self, device, is_half):
+        pretrained_state = torch.load(sv_path, map_location="cpu")
+        embedding_model = ERes2NetV2(baseWidth=24, scale=4, expansion=4)
+        embedding_model.load_state_dict(pretrained_state)
+        embedding_model.eval()
+        self.embedding_model = embedding_model
+        self.res = torchaudio.transforms.Resample(32000, 16000).to(device)
+        if is_half == False:
+            self.embedding_model = self.embedding_model.to(device)
+        else:
+            self.embedding_model = self.embedding_model.half().to(device)
+        self.is_half = is_half
+
+    def compute_embedding3(self, wav):  # (1,x)#-1~1
+        with torch.no_grad():
+            wav = self.res(wav)
+            if self.is_half == True:
+                wav = wav.half()
+            feat = torch.stack(
+                [Kaldi.fbank(wav0.unsqueeze(0), num_mel_bins=80, sample_frequency=16000, dither=0) for wav0 in wav]
+            )
+            sv_emb = self.embedding_model.forward3(feat)
+        return sv_emb
+
+
+sv = SV(device, is_half)
+
+
+def name2go(wav_name, wav_path):
+    sv_cn_path = "%s/%s.pt" % (sv_cn_dir, wav_name)
+    if os.path.exists(sv_cn_path):
+        return
+    wav_path = "%s/%s" % (wav32dir, wav_name)
+    wav32k, sr0 = torchaudio.load(wav_path)
+    assert sr0 == 32000
+    wav32k = wav32k.to(device)
+    emb = sv.compute_embedding3(wav32k).cpu()  # torch.Size([1, 20480])
+    my_save(emb, sv_cn_path)
+
+
+with open(inp_text, "r", encoding="utf8") as f:
+    lines = f.read().strip("\n").split("\n")
+
+for line in lines[int(i_part) :: int(all_parts)]:
+    try:
+        wav_name, spk_name, language, text = line.split("|")
+        wav_name = clean_path(wav_name)
+        if inp_wav_dir != "" and inp_wav_dir != None:
+            wav_name = os.path.basename(wav_name)
+            wav_path = "%s/%s" % (inp_wav_dir, wav_name)
+
+        else:
+            wav_path = wav_name
+            wav_name = os.path.basename(wav_name)
+        name2go(wav_name, wav_path)
+    except:
+        print(line, traceback.format_exc())

prepare_datasets/3-get-semantic.py

@@ -0,0 +1,118 @@
+import os
+
+inp_text = os.environ.get("inp_text")
+exp_name = os.environ.get("exp_name")
+i_part = os.environ.get("i_part")
+all_parts = os.environ.get("all_parts")
+if "_CUDA_VISIBLE_DEVICES" in os.environ:
+    os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
+opt_dir = os.environ.get("opt_dir")
+pretrained_s2G = os.environ.get("pretrained_s2G")
+s2config_path = os.environ.get("s2config_path")
+
+if os.path.exists(pretrained_s2G):
+    ...
+else:
+    raise FileNotFoundError(pretrained_s2G)
+# version=os.environ.get("version","v2")
+size = os.path.getsize(pretrained_s2G)
+if size < 82978 * 1024:
+    version = "v1"
+elif size < 100 * 1024 * 1024:
+    version = "v2"
+elif size < 103520 * 1024:
+    version = "v1"
+elif size < 700 * 1024 * 1024:
+    version = "v2"
+else:
+    version = "v3"
+import torch
+
+is_half = eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
+import traceback
+import sys
+
+now_dir = os.getcwd()
+sys.path.append(now_dir)
+import logging
+import utils
+
+if version != "v3":
+    from module.models import SynthesizerTrn
+else:
+    from module.models import SynthesizerTrnV3 as SynthesizerTrn
+from tools.my_utils import clean_path
+
+logging.getLogger("numba").setLevel(logging.WARNING)
+# from config import pretrained_s2G
+
+# inp_text=sys.argv[1]
+# exp_name=sys.argv[2]
+# i_part=sys.argv[3]
+# all_parts=sys.argv[4]
+# os.environ["CUDA_VISIBLE_DEVICES"]=sys.argv[5]
+# opt_dir="/data/docker/liujing04/gpt-vits/fine_tune_dataset/%s"%exp_name
+
+
+hubert_dir = "%s/4-cnhubert" % (opt_dir)
+semantic_path = "%s/6-name2semantic-%s.tsv" % (opt_dir, i_part)
+if os.path.exists(semantic_path) == False:
+    os.makedirs(opt_dir, exist_ok=True)
+
+    if torch.cuda.is_available():
+        device = "cuda"
+    # elif torch.backends.mps.is_available():
+    #     device = "mps"
+    else:
+        device = "cpu"
+    hps = utils.get_hparams_from_file(s2config_path)
+    vq_model = SynthesizerTrn(
+        hps.data.filter_length // 2 + 1,
+        hps.train.segment_size // hps.data.hop_length,
+        n_speakers=hps.data.n_speakers,
+        version=version,
+        **hps.model,
+    )
+    if is_half == True:
+        vq_model = vq_model.half().to(device)
+    else:
+        vq_model = vq_model.to(device)
+    vq_model.eval()
+    # utils.load_checkpoint(utils.latest_checkpoint_path(hps.s2_ckpt_dir, "G_*.pth"), vq_model, None, True)
+    # utils.load_checkpoint(pretrained_s2G, vq_model, None, True)
+    print(
+        vq_model.load_state_dict(
+            torch.load(pretrained_s2G, map_location="cpu", weights_only=False)["weight"], strict=False
+        )
+    )
+
+    def name2go(wav_name, lines):
+        hubert_path = "%s/%s.pt" % (hubert_dir, wav_name)
+        if os.path.exists(hubert_path) == False:
+            return
+        ssl_content = torch.load(hubert_path, map_location="cpu")
+        if is_half == True:
+            ssl_content = ssl_content.half().to(device)
+        else:
+            ssl_content = ssl_content.to(device)
+        codes = vq_model.extract_latent(ssl_content)
+        semantic = " ".join([str(i) for i in codes[0, 0, :].tolist()])
+        lines.append("%s\t%s" % (wav_name, semantic))
+
+    with open(inp_text, "r", encoding="utf8") as f:
+        lines = f.read().strip("\n").split("\n")
+
+    lines1 = []
+    for line in lines[int(i_part) :: int(all_parts)]:
+        # print(line)
+        try:
+            # wav_name,text=line.split("\t")
+            wav_name, spk_name, language, text = line.split("|")
+            wav_name = clean_path(wav_name)
+            wav_name = os.path.basename(wav_name)
+            # name2go(name,lines1)
+            name2go(wav_name, lines1)
+        except:
+            print(line, traceback.format_exc())
+    with open(semantic_path, "w", encoding="utf8") as f:
+        f.write("\n".join(lines1))