PublicMirrors/voice-changer
1
0
Fork 0
mirror of https://github.com/w-okada/voice-changer.git synced 2025-02-02 16:23:58 +03:00
Code Issues Actions Packages Projects Releases Wiki Activity

improve onnx

Browse Source
This commit is contained in:
w-okada 2023-09-10 05:48:34 +09:00
parent 8e822c16e3
commit 6f8710fd50
4 changed files with 11 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
server/restapi/MMVC_Rest_Fileuploader.py
View File

@ -104,6 +104,8 @@ class MMVC_Rest_Fileuploader:
return JSONResponse(content=json_compatible_item_data) return JSONResponse(content=json_compatible_item_data)
except Exception as e: except Exception as e:
print("[Voice Changer] get_onnx ex:", e) print("[Voice Changer] get_onnx ex:", e)
import traceback
traceback.print_exc()
def post_merge_models(self, request: str = Form(...)): def post_merge_models(self, request: str = Form(...)):
try: try:

1
server/voice_changer/RVC/inferencer/rvc_models/infer_pack/models.py
View File

@ -537,6 +537,7 @@ class GeneratorNSF(torch.nn.Module):
x = F.leaky_relu(x, LRELU_SLOPE) x = F.leaky_relu(x, LRELU_SLOPE)
x = self.ups[i](x) x = self.ups[i](x)
x_source = self.noise_convs[i](har_source) x_source = self.noise_convs[i](har_source)
x = x[:, :, : x_source.shape[2]]
x = x + x_source x = x + x_source
xs = None xs = None
for j in range(self.num_kernels): for j in range(self.num_kernels):

12
server/voice_changer/RVC/onnxExporter/export2onnx.py
View File

@ -88,17 +88,19 @@ def _export2onnx(input_model, output_model, output_model_simple, is_half, metada
net_g_onnx.load_state_dict(cpt["weight"], strict=False) net_g_onnx.load_state_dict(cpt["weight"], strict=False)
if is_half: if is_half:
net_g_onnx = net_g_onnx.half() net_g_onnx = net_g_onnx.half()
featsLength = 64
if is_half: if is_half:
feats = torch.HalfTensor(1, 2192, metadata["embChannels"]).to(dev) feats = torch.HalfTensor(1, featsLength, metadata["embChannels"]).to(dev)
else: else:
feats = torch.FloatTensor(1, 2192, metadata["embChannels"]).to(dev) feats = torch.FloatTensor(1, featsLength, metadata["embChannels"]).to(dev)
p_len = torch.LongTensor([2192]).to(dev) p_len = torch.LongTensor([featsLength]).to(dev)
sid = torch.LongTensor([0]).to(dev) sid = torch.LongTensor([0]).to(dev)
if metadata["f0"] is True: if metadata["f0"] is True:
pitch = torch.zeros(1, 2192, dtype=torch.int64).to(dev) pitch = torch.zeros(1, featsLength, dtype=torch.int64).to(dev)
pitchf = torch.FloatTensor(1, 2192).to(dev) pitchf = torch.FloatTensor(1, featsLength).to(dev)
input_names = ["feats", "p_len", "pitch", "pitchf", "sid"] input_names = ["feats", "p_len", "pitch", "pitchf", "sid"]
inputs = ( inputs = (
feats, feats,

2
server/voice_changer/RVC/pitchExtractor/RMVPOnnxEPitchExtractor.py → server/voice_changer/RVC/pitchExtractor/RMVPEOnnxPitchExtractor.py
View File

@ -5,7 +5,7 @@ from voice_changer.RVC.deviceManager.DeviceManager import DeviceManager
import onnxruntime import onnxruntime
class RMVPOnnxEPitchExtractor(PitchExtractor): class RMVPEOnnxPitchExtractor(PitchExtractor):
def __init__(self, file: str, gpu: int): def __init__(self, file: str, gpu: int):
super().__init__() super().__init__()