1
0
mirror of synced 2024-12-25 13:54:52 +01:00
Retrieval-based-Voice-Conve.../infer/modules/onnx/export.py

53 lines
1.8 KiB
Python
Raw Normal View History

2023-08-19 13:02:58 +02:00
import torch
from infer.lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
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"