[GH-ISSUE #1830] onnx问题请教 #1222

New issue

Open

opened 2026-10-20 10:07:06 +00:00 by bsakaguchi · 9 comments

bsakaguchi commented 2026-10-20 10:07:06 +00:00

Owner

Copy link

Originally created by @Nian-Chen on GitHub (Feb 21, 2024).
Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1830

问题描述：我参考infer/modules/onnx/export.py尝试将net_g的torch模型转onnx，遇到一些问题，望大佬帮忙看看：

1. onnx推理时不支持动态长度，export.py中默认hubert长度为200，转成onnx后也只能推理hubert长度为200的输入，其他长度的会报错；虽然问题通过简单的方式解决了，但是不知道会不会影响结果，我的解决方式是：根据转onnx的warning内容，把infer/lib/infer_pack/attentions.py中用到的int删去。
2. 即使在不修改代码，并且保持onnx推理时hubert长度为200的情况下，对比torch模型的输出和onnx模型的输出，发现输出结果差别挺大的，不知道是为什么呢？

如下是运行代码：

import torch
import sys
sys.path.append("/czc/Retrieval-based-Voice-Conversion-WebUI-main")
from infer.lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
import time
def export_onnx(ModelPath, ExportedPath):
    cpt = torch.load(ModelPath, map_location="cpu")
    cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]
    vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768

    test_phone = torch.rand(1, 200, vec_channels)  # hidden unit
    test_phone_lengths = torch.tensor([200]).long()  # hidden unit 长度（貌似没啥用）
    test_pitch = torch.randint(size=(1, 200), low=5, high=255)  # 基频（单位赫兹）
    test_pitchf = torch.rand(1, 200)  # nsf基频
    test_ds = torch.LongTensor([0])  # 说话人ID
    test_rnd = torch.rand(1, 192, 200)  # 噪声（加入随机因子）
    device = "cpu"  # 导出时设备（不影响使用模型）
    net_g = SynthesizerTrnMsNSFsidM(
        *cpt["config"], is_half=False, version=cpt.get("version", "v1")
    )  # fp32导出（C++要支持fp16必须手动将内存重新排列所以暂时不用fp16）
    net_g.load_state_dict(cpt["weight"], strict=False)
    input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"]
    output_names = [
        "audio",
    ]
    # net_g.construct_spkmixmap(n_speaker) 多角色混合轨道导出
    torch.onnx.export(
        net_g,
        (
            test_phone.to(device),
            test_phone_lengths.to(device),
            test_pitch.to(device),
            test_pitchf.to(device),
            test_ds.to(device),
            test_rnd.to(device),
        ),
        ExportedPath,
        dynamic_axes={
            "phone": [1],
            "pitch": [1],
            "pitchf": [1],
            "rnd": [2],
        },
        do_constant_folding=False,
        opset_version=13,
        verbose=False,
        input_names=input_names,
        output_names=output_names,
    )
    return "Finished"
if __name__ == "__main__":
    # 有bug导致动态维度失效
    # infer/lib/infer_pack/attentions.py 将export中报错的地方的int去掉
    ModelPath = "/czc/Retrieval-based-Voice-Conversion-WebUI-main/assets/weights/nvzhubo.pth"
    model_path = "/czc/Retrieval-based-Voice-Conversion-WebUI-main/assets/weights/nvzhubo.onnx"
    # 转onnx
    export_onnx(ModelPath, model_path)
    cpt = torch.load(ModelPath, map_location="cpu")
    cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]
    vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768
    test_phone = torch.rand(1, 200, vec_channels)  # hidden unit
    test_phone_lengths = torch.tensor([200]).long()  # hidden unit 长度（貌似没啥用）
    test_pitch = torch.randint(size=(1, 200), low=5, high=255)  # 基频（单位赫兹）
    test_pitchf = torch.rand(1, 200)  # nsf基频
    test_ds = torch.LongTensor([0])  # 说话人ID
    test_rnd = torch.zeros(1, 192, 200)  # 噪声（加入随机因子）
    time1 = time.time()
    net_g = SynthesizerTrnMsNSFsidM(
        *cpt["config"], is_half=False, version=cpt.get("version", "v1")
    )  # fp32导出（C++要支持fp16必须手动将内存重新排列所以暂时不用fp16）
    net_g.eval()
    out1 = net_g(test_phone, test_phone_lengths, test_pitch, test_pitchf, test_ds, test_rnd)
    print(time.time()-time1)
    print(out1.shape)
    print(out1[0,0,:5])
    providers = ["CPUExecutionProvider"]
    # 保持输入一致
    test_phone = test_phone.numpy()  # hidden unit
    test_phone_lengths = test_phone_lengths.numpy()  # hidden unit 长度（貌似没啥用）
    test_pitch = test_pitch.numpy()  # 基频（单位赫兹）
    test_pitchf = test_pitchf.numpy()  # nsf基频
    test_ds = test_ds.numpy()  # 说话人ID
    test_rnd = test_rnd.numpy()  # 噪声（加入随机因子）
    input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"]
    onnx_input = {
        "phone": test_phone,
        "phone_lengths": test_phone_lengths,
        "pitch": test_pitch,
        "pitchf": test_pitchf,
        "ds": test_ds,
        "rnd": test_rnd,
    }
    import onnxruntime
    import numpy as np
    onnx = onnxruntime.InferenceSession(model_path, providers=providers)
    time2 = time.time()
    out2 = torch.tensor(onnx.run(None, onnx_input)[0])
    # out2 = torch.tensor((onnx.run(None, onnx_input)[0] * 32767).astype(np.int16))
    print(time.time()-time2)
    print(out2.shape)
    print(out2[0,0,:5])

如下是两模型的输出结果：

torch.Size([1, 1, 80000])
tensor([0.0870, 0.0332, 0.1191, 0.2330, 0.2877], grad_fn=<SliceBackward0>)
torch.Size([1, 1, 80000])
tensor([0.0115, 0.0109, 0.0093, 0.0084, 0.0084])

如下是转onnx的输出日志：

/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input phone
  warnings.warn(
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input pitch
  warnings.warn(
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input pitchf
  warnings.warn(
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input rnd
  warnings.warn(
/czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:249: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  t_s == t_t
/czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:311: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  pad_length: int = max(length - (self.window_size + 1), 0)
/czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:312: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  slice_start_position = max((self.window_size + 1) - length, 0)
/czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:314: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if pad_length > 0:
/czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:366: TracerWarning: Converting a tensor to a Python integer might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  x_flat = x.view([batch, heads, int(length**2) + int(length * (length - 1))])
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/_internal/jit_utils.py:306: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.)
  _C._jit_pass_onnx_node_shape_type_inference(node, params_dict, opset_version)
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:689: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.)
  _C._jit_pass_onnx_graph_shape_type_inference(
/root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:1186: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.)
  _C._jit_pass_onnx_graph_shape_type_inference(
============= Diagnostic Run torch.onnx.export version 2.0.1+cu118 =============
verbose: False, log level: Level.ERROR
======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ========================
****

Originally created by @Nian-Chen on GitHub (Feb 21, 2024). Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1830 问题描述：我参考infer/modules/onnx/export.py尝试将net_g的torch模型转onnx，遇到一些问题，望大佬帮忙看看： 1. onnx推理时不支持动态长度，export.py中默认hubert长度为200，转成onnx后也只能推理hubert长度为200的输入，其他长度的会报错；虽然问题通过简单的方式解决了，但是不知道会不会影响结果，我的解决方式是：根据转onnx的warning内容，把infer/lib/infer_pack/attentions.py中用到的int删去。 2. 即使在不修改代码，并且保持onnx推理时hubert长度为200的情况下，对比torch模型的输出和onnx模型的输出，发现输出结果差别挺大的，不知道是为什么呢？ 如下是运行代码： ``` import torch import sys sys.path.append("/czc/Retrieval-based-Voice-Conversion-WebUI-main") from infer.lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM import time def export_onnx(ModelPath, ExportedPath): cpt = torch.load(ModelPath, map_location="cpu") cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768 test_phone = torch.rand(1, 200, vec_channels) # hidden unit test_phone_lengths = torch.tensor([200]).long() # hidden unit 长度（貌似没啥用） test_pitch = torch.randint(size=(1, 200), low=5, high=255) # 基频（单位赫兹） test_pitchf = torch.rand(1, 200) # nsf基频 test_ds = torch.LongTensor([0]) # 说话人ID test_rnd = torch.rand(1, 192, 200) # 噪声（加入随机因子） device = "cpu" # 导出时设备（不影响使用模型） net_g = SynthesizerTrnMsNSFsidM( *cpt["config"], is_half=False, version=cpt.get("version", "v1") ) # fp32导出（C++要支持fp16必须手动将内存重新排列所以暂时不用fp16） net_g.load_state_dict(cpt["weight"], strict=False) input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"] output_names = [ "audio", ] # net_g.construct_spkmixmap(n_speaker) 多角色混合轨道导出 torch.onnx.export( net_g, ( test_phone.to(device), test_phone_lengths.to(device), test_pitch.to(device), test_pitchf.to(device), test_ds.to(device), test_rnd.to(device), ), ExportedPath, dynamic_axes={ "phone": [1], "pitch": [1], "pitchf": [1], "rnd": [2], }, do_constant_folding=False, opset_version=13, verbose=False, input_names=input_names, output_names=output_names, ) return "Finished" if __name__ == "__main__": # 有bug导致动态维度失效 # infer/lib/infer_pack/attentions.py 将export中报错的地方的int去掉 ModelPath = "/czc/Retrieval-based-Voice-Conversion-WebUI-main/assets/weights/nvzhubo.pth" model_path = "/czc/Retrieval-based-Voice-Conversion-WebUI-main/assets/weights/nvzhubo.onnx" # 转onnx export_onnx(ModelPath, model_path) cpt = torch.load(ModelPath, map_location="cpu") cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768 test_phone = torch.rand(1, 200, vec_channels) # hidden unit test_phone_lengths = torch.tensor([200]).long() # hidden unit 长度（貌似没啥用） test_pitch = torch.randint(size=(1, 200), low=5, high=255) # 基频（单位赫兹） test_pitchf = torch.rand(1, 200) # nsf基频 test_ds = torch.LongTensor([0]) # 说话人ID test_rnd = torch.zeros(1, 192, 200) # 噪声（加入随机因子） time1 = time.time() net_g = SynthesizerTrnMsNSFsidM( *cpt["config"], is_half=False, version=cpt.get("version", "v1") ) # fp32导出（C++要支持fp16必须手动将内存重新排列所以暂时不用fp16） net_g.eval() out1 = net_g(test_phone, test_phone_lengths, test_pitch, test_pitchf, test_ds, test_rnd) print(time.time()-time1) print(out1.shape) print(out1[0,0,:5]) providers = ["CPUExecutionProvider"] # 保持输入一致 test_phone = test_phone.numpy() # hidden unit test_phone_lengths = test_phone_lengths.numpy() # hidden unit 长度（貌似没啥用） test_pitch = test_pitch.numpy() # 基频（单位赫兹） test_pitchf = test_pitchf.numpy() # nsf基频 test_ds = test_ds.numpy() # 说话人ID test_rnd = test_rnd.numpy() # 噪声（加入随机因子） input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"] onnx_input = { "phone": test_phone, "phone_lengths": test_phone_lengths, "pitch": test_pitch, "pitchf": test_pitchf, "ds": test_ds, "rnd": test_rnd, } import onnxruntime import numpy as np onnx = onnxruntime.InferenceSession(model_path, providers=providers) time2 = time.time() out2 = torch.tensor(onnx.run(None, onnx_input)[0]) # out2 = torch.tensor((onnx.run(None, onnx_input)[0] * 32767).astype(np.int16)) print(time.time()-time2) print(out2.shape) print(out2[0,0,:5]) ``` 如下是两模型的输出结果： ``` torch.Size([1, 1, 80000]) tensor([0.0870, 0.0332, 0.1191, 0.2330, 0.2877], grad_fn=<SliceBackward0>) torch.Size([1, 1, 80000]) tensor([0.0115, 0.0109, 0.0093, 0.0084, 0.0084]) ``` 如下是转onnx的输出日志： ``` /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input phone warnings.warn( /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input pitch warnings.warn( /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input pitchf warnings.warn( /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:2033: UserWarning: No names were found for specified dynamic axes of provided input.Automatically generated names will be applied to each dynamic axes of input rnd warnings.warn( /czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:249: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! t_s == t_t /czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:311: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! pad_length: int = max(length - (self.window_size + 1), 0) /czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:312: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! slice_start_position = max((self.window_size + 1) - length, 0) /czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:314: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! if pad_length > 0: /czc/Retrieval-based-Voice-Conversion-WebUI-main/infer/lib/infer_pack/attentions.py:366: TracerWarning: Converting a tensor to a Python integer might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! x_flat = x.view([batch, heads, int(length**2) + int(length * (length - 1))]) /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/_internal/jit_utils.py:306: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.) _C._jit_pass_onnx_node_shape_type_inference(node, params_dict, opset_version) /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:689: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.) _C._jit_pass_onnx_graph_shape_type_inference( /root/miniconda3/envs/valle/lib/python3.10/site-packages/torch/onnx/utils.py:1186: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at ../torch/csrc/jit/passes/onnx/constant_fold.cpp:179.) _C._jit_pass_onnx_graph_shape_type_inference( ============= Diagnostic Run torch.onnx.export version 2.0.1+cu118 ============= verbose: False, log level: Level.ERROR ======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ======================== **** ```

bsakaguchi added the

stale

label 2026-10-20 10:07:06 +00:00

bsakaguchi commented 2026-10-20 10:07:07 +00:00

Author

Owner

Copy link

@Nian-Chen commented on GitHub (Mar 13, 2024):

破案了 问题出在SineGen

<!-- gh-comment-id:1994090338 --> @Nian-Chen commented on GitHub (Mar 13, 2024): 破案了 问题出在SineGen

bsakaguchi commented 2026-10-20 10:07:07 +00:00

Author

Owner

Copy link

@cong7i commented on GitHub (Mar 13, 2024):

@Nian-Chen 请问结果差异大的问题怎么解决

<!-- gh-comment-id:1994297369 --> @cong7i commented on GitHub (Mar 13, 2024): @Nian-Chen 请问结果差异大的问题怎么解决

bsakaguchi commented 2026-10-20 10:07:07 +00:00

Author

Owner

Copy link

@thisisjacky6 commented on GitHub (Mar 16, 2024):

破案了 问题出在SineGen

请问如何解决sinegen的问题

<!-- gh-comment-id:2001914158 --> @thisisjacky6 commented on GitHub (Mar 16, 2024): > 破案了 问题出在SineGen 请问如何解决sinegen的问题

bsakaguchi commented 2026-10-20 10:07:07 +00:00

Author

Owner

Copy link

@Nian-Chen commented on GitHub (Mar 19, 2024):

破案了 问题出在SineGen

请问如何解决sinegen的问题

我的方案是把这部分剥离出来，用torch计算

<!-- gh-comment-id:2005558475 --> @Nian-Chen commented on GitHub (Mar 19, 2024): > > 破案了 问题出在SineGen > > 请问如何解决sinegen的问题 我的方案是把这部分剥离出来，用torch计算

bsakaguchi commented 2026-10-20 10:07:07 +00:00

Author

Owner

Copy link

@Dazai0x commented on GitHub (Apr 5, 2024):

请问sinegen的问题具体怎么改 @Nian-Chen

<!-- gh-comment-id:2039928088 --> @Dazai0x commented on GitHub (Apr 5, 2024): 请问sinegen的问题具体怎么改 @Nian-Chen

bsakaguchi commented 2026-10-20 10:07:08 +00:00

Author

Owner

Copy link

@Bella-Tim commented on GitHub (Apr 23, 2024):

请问底膜hubert_base.pt转成hubert_base.onnx,你们实现了吗？

<!-- gh-comment-id:2071310133 --> @Bella-Tim commented on GitHub (Apr 23, 2024): 请问底膜hubert_base.pt转成hubert_base.onnx,你们实现了吗？

bsakaguchi commented 2026-10-20 10:07:08 +00:00

Author

Owner

Copy link

@Dazai0x commented on GitHub (Oct 2, 2024):

你好，这个问题如何解决的我记的不太清了，但我记得在解决以后，将onnx模型导入进行预测，模型的效果将大幅度下降，因此不建议使用onnx模型

---原始邮件---
发件人: @.>
发送时间: 2024年10月2日(周三) 凌晨3:05
收件人: @.>;
抄送: @.@.>;
主题: Re: [RVC-Project/Retrieval-based-Voice-Conversion-WebUI] onnx问题请教 (Issue #1830)

@Nian-Chen 我在使用原本的web导出onnx的时候并没有报错，但是onnx仍然不支持动态长度，是固定的200，这个该如何解决，你说的int是哪一个地方的int，我尝试修改之后仍然固定长度

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: @.***>

<!-- gh-comment-id:2387302985 --> @Dazai0x commented on GitHub (Oct 2, 2024): 你好，这个问题如何解决的我记的不太清了，但我记得在解决以后，将onnx模型导入进行预测，模型的效果将大幅度下降，因此不建议使用onnx模型 ---原始邮件--- 发件人: ***@***.***&gt; 发送时间: 2024年10月2日(周三) 凌晨3:05 收件人: ***@***.***&gt;; 抄送: ***@***.******@***.***&gt;; 主题: Re: [RVC-Project/Retrieval-based-Voice-Conversion-WebUI] onnx问题请教 (Issue #1830) @Nian-Chen 我在使用原本的web导出onnx的时候并没有报错，但是onnx仍然不支持动态长度，是固定的200，这个该如何解决，你说的int是哪一个地方的int，我尝试修改之后仍然固定长度 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***&gt;

bsakaguchi commented 2026-10-20 10:07:08 +00:00

Author

Owner

Copy link

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

@Nian-Chen 如果采用动态控制流的方式导出是不是没这些问题了

<!-- gh-comment-id:2576841896 --> @zhixingheyixsh commented on GitHub (Jan 8, 2025): @Nian-Chen 如果采用动态控制流的方式导出是不是没这些问题了

bsakaguchi commented 2026-10-20 10:07:08 +00:00

Author

Owner

Copy link

@Dazai0x commented on GitHub (Jan 8, 2025):

我不清楚，我并没有尝试

---原始邮件---
发件人: @.>
发送时间: 2025年1月8日(周三) 下午2:22
收件人: @.>;
抄送: @.@.>;
主题: Re: [RVC-Project/Retrieval-based-Voice-Conversion-WebUI] onnx问题请教 (Issue #1830)

@Nian-Chen 如果采用动态控制流的方式导出是不是没这些问题了

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: @.***>

<!-- gh-comment-id:2576850359 --> @Dazai0x commented on GitHub (Jan 8, 2025): 我不清楚，我并没有尝试 ---原始邮件--- 发件人: ***@***.***&gt; 发送时间: 2025年1月8日(周三) 下午2:22 收件人: ***@***.***&gt;; 抄送: ***@***.******@***.***&gt;; 主题: Re: [RVC-Project/Retrieval-based-Voice-Conversion-WebUI] onnx问题请教 (Issue #1830) @Nian-Chen 如果采用动态控制流的方式导出是不是没这些问题了 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***&gt;

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

main

20240604

v1

2.3.260718

2.2.231006

2.1.230814

2.0.230618

2.0.230528

1.2.230428

1.1.230416

1.0.230410

Labels

Clear labels   

AMD/Intel

  

bug

  

bug

  

bug

  

colab

  

dependencies

  

documentation

  

duplicate

  

enhancement

  

following up

  

good first issue

  

help wanted

  

invalid

  

question

  

stale

  

stale

  

stale

  

todo list

  

wontfix

No labels

AMD/Intel

bug

bug

bug

colab

dependencies

documentation

duplicate

enhancement

following up

good first issue

help wanted

invalid

question

stale

stale

stale

todo list

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

RVC-Project/Retrieval-based-Voice-Conversion-WebUI#1222

No description provided.