用这里的模型跑出现这个RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]).

Question

用这里的模型跑出现这个RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]).

wangkewk opened this issue 3 years ago · comments

张小智 commented 3 years ago

一样！

Dignite commented 3 years ago

+1

Xu GaoXiang commented 3 years ago

+1

wa007 commented 3 years ago

+1

wangkewk · Answer 1 · Mon Aug 23 2021 12:49:16 GMT+0800 (China Standard Time)

谁能解决

wangkewk · Answer 2 · Mon Aug 23 2021 12:51:29 GMT+0800 (China Standard Time)

wangkewk commented 3 years ago

Vega · Answer 3 · Mon Aug 23 2021 13:35:37 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的内容改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

wangkewk · Answer 4 · Mon Aug 23 2021 14:11:56 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

放心

softdi · Answer 5 · Mon Aug 23 2021 18:17:05 GMT+0800 (China Standard Time)

同样的问题！

Yu Fan · Answer 6 · Mon Aug 23 2021 18:38:42 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

谢谢，这是有效的。修改过之后，原来的纯杂音变成正常声音了

MiniAL · Answer 7 · Mon Aug 23 2021 22:19:17 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

谢谢，问题解决了

sanhuafeiluo · Answer 8 · Mon Aug 23 2021 22:21:57 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

谢谢，已经解决

vc815 · Answer 9 · Tue Aug 24 2021 13:16:53 GMT+0800 (China Standard Time)

同样问题

duolanda · Answer 10 · Wed Aug 25 2021 01:15:49 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

感谢！问题已顺利解决。

Yukikawa · Answer 11 · Thu Aug 26 2021 20:02:31 GMT+0800 (China Standard Time)

修改后完全正常，thanks~

sky · Answer 12 · Sat Aug 28 2021 17:58:43 GMT+0800 (China Standard Time)

修改后正常了，感谢

He Zheng · Answer 13 · Tue Aug 31 2021 14:29:37 GMT+0800 (China Standard Time)

问题确实解决了，但是声音质量没有哔哩哔哩的效果好，我特意找到的小说的录音，不知道是哪里有问题。
如果想要声音特别像某个人的声音，要怎么提高呢？

Alan · Answer 14 · Thu Sep 02 2021 16:09:13 GMT+0800 (China Standard Time)

同样的问题。

TwDnCTX · Answer 15 · Tue Sep 07 2021 02:04:47 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

如果使用自己训练的模型要把这个改回去才有效吗还是不用改也行我试了下新训练的没声音(也有可能是自己训练的问题)但是用给的模型是正常

Vega · Answer 16 · Tue Sep 07 2021 09:15:04 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

如果使用自己训练的模型要把这个改回去才有效吗还是不用改也行我试了下新训练的没声音(也有可能是自己训练的问题)但是用给的模型是正常

改回去效果会好一点但是不改也可以工作的

JeffCheung85 · Answer 17 · Tue Sep 07 2021 22:13:15 GMT+0800 (China Standard Time)

总算可以了，这个问题搞了好久，还以为本地安装的环境问题

QiYYZH · Answer 18 · Tue Sep 21 2021 14:19:26 GMT+0800 (China Standard Time)

出来的声音像机器人的声音，是因为不同的电脑环境出来的效果不一样么？那是否得自己重新训练模型？

Vega · Answer 19 · Tue Sep 21 2021 16:12:41 GMT+0800 (China Standard Time)

出来的声音像机器人的声音，是因为不同的电脑环境出来的效果不一样么？那是否得自己重新训练模型？

不是的，可能是vocoder或者输入音频不同导致的

chenyv118 · Answer 20 · Sun Oct 17 2021 14:05:00 GMT+0800 (China Standard Time)

唉，还是没有视频中的效果，听起来像刚来**的老外的塑料中文

chenyv118 · Answer 21 · Sun Oct 17 2021 14:46:27 GMT+0800 (China Standard Time)

问题确实解决了，但是声音质量没有哔哩哔哩的效果好，我特意找到的小说的录音，不知道是哪里有问题。如果想要声音特别像某个人的声音，要怎么提高呢？

我也用的B站up主的模型，但是没有bilibili中的效果，我那边听起来像伏拉夫的调调，都不像中文

Vega · Answer 22 · Sun Oct 17 2021 17:09:13 GMT+0800 (China Standard Time)

问题确实解决了，但是声音质量没有哔哩哔哩的效果好，我特意找到的小说的录音，不知道是哪里有问题。如果想要声音特别像某个人的声音，要怎么提高呢？

我也用的B站up主的模型，但是没有bilibili中的效果，我那边听起来像伏拉夫的调调，都不像中文

如果录音清晰，平调情况下音色复制效果还是可以的，是不是哪里没运行好？

Jackxwb · Answer 23 · Tue Oct 26 2021 17:22:47 GMT+0800 (China Standard Time)

已修改synthesizer/utils/symbols.py，还是出现报错

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

Wu-Pretend · Answer 24 · Wed Oct 27 2021 18:39:14 GMT+0800 (China Standard Time)

已修改synthesizer/utils/symbols.py，还是出现报错

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

Me too

Vega · Answer 25 · Wed Oct 27 2021 20:16:45 GMT+0800 (China Standard Time)

试着用这个模型：
链接：https://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw
提取码：om7f
--来自百度网盘超级会员V3的分享

Jackxwb · Answer 26 · Thu Oct 28 2021 00:27:06 GMT+0800 (China Standard Time)

试着用这个模型：链接：https://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw 提取码：om7f --来自百度网盘超级会员V3的分享

可以运行起来了，但是生成的句子只有前半是读出来的，后半句都是杂音，多生成几次有时会好点有时又会倒退回去，而且生成的声音和原音频不像，差的有点远的那种，哈哈

Wu-Pretend · Answer 27 · Thu Oct 28 2021 21:21:09 GMT+0800 (China Standard Time)

试着用这个模型：链接：https://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw 提取码：om7f --来自百度网盘超级会员V3的分享

这个模型没问题，把_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!'(),-.:;? '改回原来的就行了

System-BXV · Answer 28 · Sat Nov 06 2021 09:14:32 GMT+0800 (China Standard Time)

蔓用这个模型：链接：https ://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw提取码：om7f --来自百度网盘超级会员V3的分享

这个可以解决了，但拿演示音频测试，生成的差了好多emmm

Vega · Answer 29 · Sat Nov 06 2021 09:51:45 GMT+0800 (China Standard Time)

蔓用这个模型：链接：https ://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw提取码：om7f --来自百度网盘超级会员V3的分享

这个可以解决了，但拿演示音频测试，生成的差了好多emmm

跑的步数很少，可以延续跑到100k+

Vega · Answer 30 · Sun Nov 07 2021 12:23:52 GMT+0800 (China Standard Time)

ceshi的模型需要将代码切换到10月20号左右的commit之后，再按issue #37 修改之后就可以用了
而作者的模型，需要将代码切换到10月20号左右的commit之后使用

KQDtianxiaK · Answer 31 · Tue Dec 07 2021 01:24:54 GMT+0800 (China Standard Time)

蔓用这个模型：链接：https ://pan.baidu.com/s/1fMh9IlgKJlL2PIiRTYDUvw提取码：om7f --来自百度网盘超级会员V3的分享

这个可以解决了，但拿演示音频测试，生成的差了好多emmm

跑的步数很少，可以延续跑到100k+
是不断的点synthesize only之后，输出的声音就会越来越好吗？

Happy fool · Answer 32 · Tue Dec 07 2021 21:29:25 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的内容改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

改了之后还是没用，，，，希望再看看

Icey-lin · Answer 33 · Tue Dec 07 2021 23:08:58 GMT+0800 (China Standard Time)

已修改synthesizer/utils/symbols.py，还是出现报错

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

我觉得你这个估计是一开始你复制了模型到你的程序里面去了，重新解压一下那个程序的压缩包，然后重新来就可以了

Glider · Answer 34 · Tue Jan 04 2022 14:57:19 GMT+0800 (China Standard Time)

为什么我的源音频是黑色的，有大佬知道吗？

Glider · Answer 35 · Tue Jan 04 2022 14:57:52 GMT+0800 (China Standard Time)

源音频的Dataset和Speaker这些都是黑的，不能选择？

Vega · Answer 36 · Wed Jan 05 2022 00:39:22 GMT+0800 (China Standard Time)

没有被识别的数据集不训练的话就不用理会了

Glider · Answer 37 · Wed Jan 05 2022 15:22:12 GMT+0800 (China Standard Time)

大佬，是不是如果要克隆自己的声音的话，需要对自己做音源进行训练，而不能直接用community给的那些模型。昨天用给的模型（包括synthesizer和vector）克隆自己的录音，结果出来的梅尔频谱图是杂乱的，只有一堆电流声和噪声，求大佬指正错误

TwDnCTX · Answer 38 · Wed Jan 05 2022 15:23:54 GMT+0800 (China Standard Time)

我也一样啊.程序逻辑有声波频率上的错误.

…

------------------ 原始邮件 ------------------ 发件人: ***@***.***>; 发送时间: 2022年1月5日(星期三) 下午3:22 收件人: ***@***.***>; 抄送: ***@***.***>; ***@***.***>; 主题: Re: [babysor/MockingBird] 用这里的模型跑出现这个RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]). (#37) 大佬，是不是如果要克隆自己的声音的话，需要对自己做音源进行训练，而不能直接用community给的那些模型。昨天用给的模型（包括synthesizer和vector）克隆自己的录音，结果出来的梅尔频谱图是杂乱的，只有一堆电流声和噪声，求大佬指正错误 — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you commented.Message ID: ***@***.***>

fz00x0zf · Answer 39 · Sat Jan 08 2022 15:57:33 GMT+0800 (China Standard Time)

可以直接通过 quickstart （https://github.com/babysor/MockingBird/wiki/Quick-Start-(Newbie)）改用该模型，相关代码可以无需修改；环境 3.7.11

jingyin · Answer 40 · Sat Jan 08 2022 19:17:49 GMT+0800 (China Standard Time)

整篇评论都看了，raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(。生成的都是杂音，代码也照着改了都不行。换模型也不行。。。

tianming937 · Answer 41 · Sun Jan 16 2022 16:10:34 GMT+0800 (China Standard Time)

同样的错误copying a param with shape torch.Size([128, 512]) ，输出的声音全部是杂音

ChunMengXin · Answer 42 · Mon Jan 24 2022 17:17:21 GMT+0800 (China Standard Time)

纯萌新，请教一下切换到tag0.01怎么切换啊？完全没理解。
自己拿75k的训练了一阵目标语音，感觉模仿的声音还是不像，想换这个模型再训练试试

fengxiangyun · Answer 43 · Mon Jan 24 2022 18:51:38 GMT+0800 (China Standard Time)

已修改synthesizer/utils/symbols.py，还是出现报错

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

同样的报错你那个好了吗？

Glider · Answer 44 · Mon Jan 24 2022 21:16:16 GMT+0800 (China Standard Time)

还没xd

…

------------------ 原始邮件 ------------------ 发件人: ***@***.***>; 发送时间: 2022年1月24日(星期一) 晚上6:51 收件人: ***@***.***>; 抄送: ***@***.***>; ***@***.***>; 主题: Re: [babysor/MockingBird] 用这里的模型跑出现这个RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]). (#37) 已修改synthesizer/utils/symbols.py，还是出现报错 Synthesizer using device: cuda Trainable Parameters: 32.735M Traceback (most recent call last): File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda> func = lambda: self.synthesize() or self.vocode() File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token) File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms self.load() File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load self._model.load(self.model_fpath) File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load self.load_state_dict(checkpoint["model_state"], strict=False) File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict self.__class__.__name__, "\n\t".join(error_msgs))) RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]). size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]). size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]). size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]). 同样的报错你那个好了吗？ — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you commented.Message ID: ***@***.***>

KasuganoSora-desu · Answer 45 · Wed Jan 26 2022 10:42:30 GMT+0800 (China Standard Time)

只输出杂音，按照评论来改了还是一样

Vega · Answer 46 · Wed Jan 26 2022 15:35:07 GMT+0800 (China Standard Time)

已修改synthesizer/utils/symbols.py，还是出现报错

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

同样的报错你那个好了吗？

版本先切换，再应用#37

falling-into-fall · Answer 47 · Tue Feb 15 2022 00:08:58 GMT+0800 (China Standard Time)

请问如何切换版本？

LiangDazhu · Answer 48 · Wed Mar 02 2022 15:38:48 GMT+0800 (China Standard Time)

LiangDazhu commented 2 years ago

cxgreat2014 · Answer 49 · Tue Apr 12 2022 13:38:03 GMT+0800 (China Standard Time)

Same issue

cxgreat2014 · Answer 50 · Tue Apr 12 2022 13:40:52 GMT+0800 (China Standard Time)

改完一样报错，看起来又有新的问题，，，，，，，，，

Htt-Qcty · Answer 51 · Thu Apr 28 2022 19:49:03 GMT+0800 (China Standard Time)

File "E:\语音克隆\MockingBird\synthesizer\models\tacotron.py", line 564, in load
self.load_state_dict(checkpoint["model_state"], strict=False)
File "E:\anaconda\envs\torch\lib\site-packages\torch\nn\modules\module.py", line 1497, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for Tacotron2:
size mismatch for embedding.weight: copying a param with shape torch.Size([148, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]).

Htt-Qcty · Answer 52 · Thu Apr 28 2022 19:50:02 GMT+0800 (China Standard Time)

怎么解决呀，我改了之后后面的70变化了

Htt-Qcty · Answer 53 · Thu Apr 28 2022 20:03:08 GMT+0800 (China Standard Time)

帮帮孩子吧

Htt-Qcty · Answer 54 · Thu Apr 28 2022 20:39:18 GMT+0800 (China Standard Time)

我这个是拿nVidia那个改了一点，为什么前面是148，怎么修改这个值

zhuhero2012 · Answer 55 · Fri Jan 13 2023 23:29:06 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的内容改为： _characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? ' 即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

改了没用啊

zhuhero2012 · Answer 56 · Thu Jan 26 2023 00:00:19 GMT+0800 (China Standard Time)

已修改，还是出现报错synthesizer/utils/symbols.py

Synthesizer using device: cuda
Trainable Parameters: 32.735M
Traceback (most recent call last):
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 123, in <lambda>
    func = lambda: self.synthesize() or self.vocode()
  File "D:\AI\sv2tts_china\MockingBird\toolbox\__init__.py", line 238, in synthesize
    specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 87, in synthesize_spectrograms
    self.load()
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\inference.py", line 65, in load
    self._model.load(self.model_fpath)
  File "D:\AI\sv2tts_china\MockingBird\synthesizer\models\tacotron.py", line 525, in load
    self.load_state_dict(checkpoint["model_state"], strict=False)
  File "D:\ProgramData\Anaconda3\envs\Real-Time-Voice-Cloning\lib\site-packages\torch\nn\modules\module.py", line 1483, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Tacotron:
        size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
        size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
        size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
        size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).

同样的报错你那个好了吗？

Felix zhao · Answer 57 · Sat Jan 28 2023 09:23:59 GMT+0800 (China Standard Time)

同样的报错你那个好了吗？

同样的报错你那个好了吗？

徐子睿 · Answer 58 · Fri Feb 03 2023 22:28:51 GMT+0800 (China Standard Time)

同问切换到v0.0.1依然不行（已加修复） pytorch是最新版 cuda11.7

Kefeng Zhou · Answer 59 · Mon May 08 2023 20:19:44 GMT+0800 (China Standard Time)

按说明修改后还是没用，一样报错。一定要自己训练吗？

MichaelToLearn · Answer 60 · Sat May 13 2023 14:48:24 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的内容改为： _characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? ' 即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

快两年了，这个还会兼容吗。。

zzllabcd · Answer 61 · Fri May 19 2023 23:57:46 GMT+0800 (China Standard Time)

RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]).
Traceback:
File "D:\ProgramData\Anaconda3\envs\voiceClone\lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 565, in _run_script
exec(code, module.dict)
File "C:\Users\zhanglong\AppData\Local\Temp\tmp342r_iv9.py", line 13, in
render_streamlit_ui()
File "H:\MockingBird\MockingBird\control\mkgui\base\ui\streamlit_ui.py", line 909, in render_streamlit_ui
session_state.output_data = opyrator(input=input_data_obj)
File "H:\MockingBird\MockingBird\control\mkgui\base\core.py", line 203, in call
return self.function(input_obj, **kwargs)
File "H:\MockingBird\MockingBird\control\mkgui\app.py", line 140, in synthesize
specs = current_synt.synthesize_spectrograms(texts, embeds)
File "H:\MockingBird\MockingBird\models\synthesizer\inference.py", line 91, in synthesize_spectrograms
self.load()
File "H:\MockingBird\MockingBird\models\synthesizer\inference.py", line 69, in load
self._model.load(self.model_fpath, self.device)
File "H:\MockingBird\MockingBird\models\synthesizer\models\base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "D:\ProgramData\Anaconda3\envs\voiceClone\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(

dreamcatcher · Answer 62 · Tue Jun 20 2023 00:29:05 GMT+0800 (China Standard Time)

Traceback (most recent call last):
File "D:\codeinstall\Python310\lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 552, in _run_script
exec(code, module.dict)
File "C:\Users\zy820\AppData\Local\Temp\tmpegto92vs.py", line 13, in
render_streamlit_ui()
File "D:\develop\workspace-project\MockingBird\control\mkgui\base\ui\streamlit_ui.py", line 909, in render_streamlit_ui
session_state.output_data = opyrator(input=input_data_obj)
File "D:\develop\workspace-project\MockingBird\control\mkgui\base\core.py", line 203, in call
return self.function(input_obj, **kwargs)
self._model.load(self.model_fpath, self.device)
File "D:\develop\workspace-project\MockingBird\models\synthesizer\models\base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "D:\codeinstall\Python310\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for Tacotron:
size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([75, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]).
size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]).

按照上面的解决方法改了还是报错，有新的解决方案没

吴汉晨 · Answer 63 · Sun Jul 23 2023 17:54:26 GMT+0800 (China Standard Time)

Traceback (most recent call last):
File "G:\PycharmProjects\MockingBird\control\toolbox_init_.py", line 260, in synthesize
specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token, steps=int(self.ui.length_slider.value())*200)
File "G:\PycharmProjects\MockingBird\models\synthesizer\inference.py", line 91, in synthesize_spectrograms
self.load()
File "G:\PycharmProjects\MockingBird\models\synthesizer\inference.py", line 69, in load
self._model.load(self.model_fpath, self.device)
File "G:\PycharmProjects\MockingBird\models\synthesizer\models\base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "C:\Users\Admin.conda\envs\pytorch\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for Tacotron:
size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([70, 512]) from checkpoint, the shape in current model is torch.Size([75, 512]).
size mismatch for encoder_proj.weight: copying a param with shape torch.Size([128, 512]) from checkpoint, the shape in current model is torch.Size([128, 1024]).
size mismatch for decoder.attn_rnn.weight_ih: copying a param with shape torch.Size([384, 768]) from checkpoint, the shape in current model is torch.Size([384, 1280]).
size mismatch for decoder.rnn_input.weight: copying a param with shape torch.Size([1024, 640]) from checkpoint, the shape in current model is torch.Size([1024, 1152]).
size mismatch for decoder.stop_proj.weight: copying a param with shape torch.Size([1, 1536]) from checkpoint, the shape in current model is torch.Size([1, 2048]).
QWindowsWindow::setGeometry: Unable to set geometry 1992x1001+0+29 (frame: 2010x1048-9-9) on QWidgetWindow/"UIClassWindow" on "\.\DISPLAY1". Resulting geometry: 1920x1001+0+29 (frame: 1938x1048-9-9) margins: 9, 38, 9, 9 minimum size: 1992x583 MINMAXINFO maxSize=0,0 maxpos=0,0 mintrack=2010,630 maxtrack=0,0)

你好,除了一个75k steps 的合成器正常运行了，25k 150k 200k 的均出现类似的错误，这个是加载mandarin_200k.pt的合成器时候的报错，到现在还有解决方案吗？谢谢

吴汉晨 · Answer 64 · Sun Jul 23 2023 17:56:02 GMT+0800 (China Standard Time)

上面提供改的所有方案都试过了全部没用,不知道真正导致数据不同步的错误在运行环境哪里

feizi · Answer 65 · Thu Aug 24 2023 18:34:04 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

感谢！问题已顺利解决。

RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([75, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]). size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]).
Traceback:
File "C:\Users\Admin\AppData\Local\Programs\Python\Python310\lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 552, in _run_script
exec(code, module.dict)
File "C:\Users\Admin\AppData\Local\Temp\tmpsj6156uv.py", line 13, in
render_streamlit_ui()
File "E:\GithubProjects\MockingBird-main\control\mkgui\base\ui\streamlit_ui.py", line 909, in render_streamlit_ui
session_state.output_data = opyrator(input=input_data_obj)
File "E:\GithubProjects\MockingBird-main\control\mkgui\base\core.py", line 203, in call
return self.function(input_obj, **kwargs)
File "E:\GithubProjects\MockingBird-main\control\mkgui\app.py", line 140, in synthesize
specs = current_synt.synthesize_spectrograms(texts, embeds)
File "E:\GithubProjects\MockingBird-main\models\synthesizer\inference.py", line 91, in synthesize_spectrograms
self.load()
File "E:\GithubProjects\MockingBird-main\models\synthesizer\inference.py", line 69, in load
self._model.load(self.model_fpath, self.device)
File "E:\GithubProjects\MockingBird-main\models\synthesizer\models\base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "C:\Users\Admin\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(

按照这个修改了MockingBird-main\models\synthesizer\utils目录下面的symbols.py文件里面的第11行代码，但是依旧还是报错，不知道什么原因？

MickeyCHAN · Answer 66 · Tue Oct 10 2023 14:33:40 GMT+0800 (China Standard Time)

在我实际使用中发现，如果出现尺寸不匹配的问题，有说是输入框文字切割的问题，原始仓库Real-Time-Voice-Cloning也会出现这个问题。
但是多点击几次好像就不报这个错误，但是输出的音频还是以杂音为主

Centralcosmic · Answer 67 · Sat Nov 18 2023 01:00:28 GMT+0800 (China Standard Time)

这个是我最近一个修复导致的不兼容问题，你可以把文件中：synthesizer/utils/symbols.py 第11行的symbols 改为：
_characters = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz12340!\'(),-.:;? '
即可。暂时先不要关闭这个issue吧。我看下遇到的人太多的话我做个兼容

感谢！问题已顺利解决。

RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([75, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]). size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]). Traceback: File "C:\Users\Admin\AppData\Local\Programs\Python\Python310\lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 552, in _run_script exec(code, module.dict) File "C:\Users\Admin\AppData\Local\Temp\tmpsj6156uv.py", line 13, in render_streamlit_ui() File "E:\GithubProjects\MockingBird-main\control\mkgui\base\ui\streamlit_ui.py", line 909, in render_streamlit_ui session_state.output_data = opyrator(input=input_data_obj) File "E:\GithubProjects\MockingBird-main\control\mkgui\base\core.py", line 203, in call return self.function(input_obj, **kwargs) File "E:\GithubProjects\MockingBird-main\control\mkgui\app.py", line 140, in synthesize specs = current_synt.synthesize_spectrograms(texts, embeds) File "E:\GithubProjects\MockingBird-main\models\synthesizer\inference.py", line 91, in synthesize_spectrograms self.load() File "E:\GithubProjects\MockingBird-main\models\synthesizer\inference.py", line 69, in load self._model.load(self.model_fpath, self.device) File "E:\GithubProjects\MockingBird-main\models\synthesizer\models\base.py", line 55, in load self.load_state_dict(state, strict=False) File "C:\Users\Admin\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(

按照这个修改了MockingBird-main\models\synthesizer\utils目录下面的symbols.py文件里面的第11行代码，但是依旧还是报错，不知道什么原因？

同不知道解决了吗

SpaceRexxx · Answer 68 · Wed Jan 24 2024 10:55:06 GMT+0800 (China Standard Time)

修改后还是报错:
RuntimeError: Error(s) in loading state_dict for Tacotron: size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([75, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]). size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]).
Traceback:
File "/Users/ywy/Library/Python/3.11/lib/python/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 535, in _run_script
exec(code, module.dict)
File "/private/var/folders/53/3r03mt7d4v9bsvhljvnd_zs80000gn/T/tmpo53ek00n.py", line 13, in
render_streamlit_ui()
File "/Users/ywy/MockingBird/control/mkgui/base/ui/streamlit_ui.py", line 909, in render_streamlit_ui
session_state.output_data = opyrator(input=input_data_obj)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ywy/MockingBird/control/mkgui/base/core.py", line 203, in call
return self.function(input_obj, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ywy/MockingBird/control/mkgui/app.py", line 140, in synthesize
specs = current_synt.synthesize_spectrograms(texts, embeds)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ywy/MockingBird/models/synthesizer/inference.py", line 91, in synthesize_spectrograms
self.load()
File "/Users/ywy/MockingBird/models/synthesizer/inference.py", line 69, in load
self._model.load(self.model_fpath, self.device)
File "/Users/ywy/MockingBird/models/synthesizer/models/base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(

Raksha · Answer 69 · Thu Feb 08 2024 10:22:46 GMT+0800 (China Standard Time)

Traceback (most recent call last):
File "D:\python\tts\chatgpt\tts\MockingBird-main\control\toolbox_init_.py", line 144, in
func = lambda: self.synthesize() or self.vocode()
^^^^^^^^^^^^^^^^^
File "D:\python\tts\chatgpt\tts\MockingBird-main\control\toolbox_init_.py", line 260, in synthesize
specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token, steps=int(self.ui.length_slider.value())*200)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\python\tts\chatgpt\tts\MockingBird-main\models\synthesizer\inference.py", line 91, in synthesize_spectrograms
self.load()
File "D:\python\tts\chatgpt\tts\MockingBird-main\models\synthesizer\inference.py", line 69, in load
self._model.load(self.model_fpath, self.device)
File "D:\python\tts\chatgpt\tts\MockingBird-main\models\synthesizer\models\base.py", line 55, in load
self.load_state_dict(state, strict=False)
File "C:\ProgramData\anaconda3\envs\pytorch\Lib\site-packages\torch\nn\modules\module.py", line 2152, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for Tacotron:
size mismatch for encoder.embedding.weight: copying a param with shape torch.Size([75, 512]) from checkpoint, the shape in current model is torch.Size([70, 512]).
size mismatch for gst.stl.attention.W_query.weight: copying a param with shape torch.Size([512, 256]) from checkpoint, the shape in current model is torch.Size([512, 512]).