RVC-Project/Retrieval-based-Voice-Conversion-WebUI

Fork 0

mirror of https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI.git synced 2026-11-05 00:27:30 +00:00

[GH-ISSUE #1983] ONNX Inference Problem #1344

New issue

Open

opened 2026-10-20 10:08:14 +00:00 by bsakaguchi · 5 comments

bsakaguchi commented

2026-10-20 10:08:14 +00:00

Owner

Originally created by @LopSdir on GitHub (Apr 20, 2024).
Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1983

I use the demo of inferencing by onnx, but I had this problem ,I try many times to solve this problem, but still can't solve it.
I also use the old onnx, but still can't.

So may I know how to solve this problem or can I get the testing onnx file

Originally created by @LopSdir on GitHub (Apr 20, 2024). Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1983 ![image](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/assets/49876859/5a53ab97-f648-4b38-b0be-f2e4d3566039) I use the demo of inferencing by onnx, but I had this problem ,I try many times to solve this problem, but still can't solve it. I also use the old onnx, but still can't. So may I know how to solve this problem or can I get the testing onnx file

bsakaguchi added the

bug

stale

labels

2026-10-20 10:08:14 +00:00

bsakaguchi commented

2026-10-20 10:08:15 +00:00

Author

Owner

@fumiama commented on GitHub (Apr 23, 2024):

Try the latest commit.

@fumiama commented on GitHub (Apr 23, 2024): Try the latest commit.

bsakaguchi commented

2026-10-20 10:08:15 +00:00

Author

Owner

@yc930401 commented on GitHub (Aug 16, 2024):

最新版本代码，还是同样的问题。

@yc930401 commented on GitHub (Aug 16, 2024): ![image](https://github.com/user-attachments/assets/fae0612b-dbb2-43ad-b2a2-4f8373c6f5d8) 最新版本代码，还是同样的问题。

bsakaguchi commented

2026-10-20 10:08:15 +00:00

Author

Owner

@samolego commented on GitHub (Sep 11, 2024):

I too get this error (with different expected and requested sizes) 😢
Were you able to solve it? (#2298)

@samolego commented on GitHub (Sep 11, 2024): I too get this error (with different expected and requested sizes) :cry: Were you able to solve it? (#2298)

bsakaguchi commented

2026-10-20 10:08:15 +00:00

Author

Owner

@zhixingheyixsh commented on GitHub (Jan 8, 2025):

我也遇到这个问题了，看报错是在 /enc_p/encoder/attn_layers.0/Reshape_10，应该是infer/lib/infer_pack/attentions.py 中Encoder 中的 Mutiheadattention 里有个reshape,但是我看forward里是没有的，反而上一层 TextEncoder 的forward里有几个reshape.

另外，请问你的vec-768-layer-12.onnx 是在哪儿块下载的？是这个地址吗 https://huggingface.co/NaruseMioShirakana/MoeSS-SUBModel

其它地方也有这个问题，可以参考#2298
有个地方说了解决方案
#1830

@zhixingheyixsh commented on GitHub (Jan 8, 2025): 我也遇到这个问题了，看报错是在 /enc_p/encoder/attn_layers.0/Reshape_10，应该是infer/lib/infer_pack/attentions.py 中Encoder 中的 Mutiheadattention 里有个reshape,但是我看forward里是没有的，反而上一层 TextEncoder 的forward里有几个reshape. 另外，请问你的vec-768-layer-12.onnx 是在哪儿块下载的？是这个地址吗 https://huggingface.co/NaruseMioShirakana/MoeSS-SUBModel 其它地方也有这个问题，可以参考#2298 有个地方说了解决方案 #1830

bsakaguchi commented

2026-10-20 10:08:15 +00:00

Author

Owner

@chuckcxr commented on GitHub (May 15, 2025):

I also encountered this problem. The error is in /enc_p/encoder/attn_layers.0/Reshape_10. There should be a reshape in Mutiheadattention in Encoder in infer/lib/infer_pack/attentions.py, but I don't see it in forward. Instead, there are several reshapes in the forward of the previous layer TextEncoder.

Also, where did you download vec-768-layer-12.onnx? Is it this address https://huggingface.co/NaruseMioShirakana/MoeSS-SUBModel

This problem also occurs in other places. You can refer to #2298. There is a solution mentioned in #1830.

False....that link they point to this page.....this page point to that page.....circles and circles and no answer >_<

@chuckcxr commented on GitHub (May 15, 2025): > I also encountered this problem. The error is in /enc_p/encoder/attn_layers.0/Reshape_10. There should be a reshape in Mutiheadattention in Encoder in infer/lib/infer_pack/attentions.py, but I don't see it in forward. Instead, there are several reshapes in the forward of the previous layer TextEncoder. > > Also, where did you download vec-768-layer-12.onnx? Is it this address https://huggingface.co/NaruseMioShirakana/MoeSS-SUBModel > > This problem also occurs in other places. You can refer to [#2298.](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/2298) There is a solution mentioned in [#1830.](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1830) False....that link they point to this page.....this page point to that page.....circles and circles and no answer >_<