fumiama/Retrieval-based-Voice-Conversion-WebUI
1
0
Fork 0
mirror of https://github.com/fumiama/Retrieval-based-Voice-Conversion-WebUI.git synced 2026-06-09 20:40:48 +08:00
Code Activity

optimize(crepe): move crepe into rvc.f0

Browse Source
This commit is contained in:
源文雨
2024-06-14 14:29:36 +09:00
parent f79b925ee2
commit e298fde29c
7 changed files with 106 additions and 54 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
rvc/f0/rmvpe.py
View File

@@ -26,16 +26,18 @@ class RMVPE(F0Predictor):
f0_max = 8000
sampling_rate = 16000
super().__init__(hop_length, f0_min, f0_max, sampling_rate)
super().__init__(
hop_length,
f0_min,
f0_max,
sampling_rate,
device,
)
self.is_half = is_half
cents_mapping = 20 * np.arange(360) + 1997.3794084376191
self.cents_mapping = np.pad(cents_mapping, (4, 4)) # 368
if device is None:
device = "cuda:0" if torch.cuda.is_available() else "cpu"
self.device = device
self.mel_extractor = MelSpectrogram(
is_half=is_half,
n_mel_channels=128,
@@ -44,10 +46,10 @@ class RMVPE(F0Predictor):
hop_length=hop_length,
mel_fmin=f0_min,
mel_fmax=f0_max,
device=device,
).to(device)
device=self.device,
).to(self.device)
if "privateuseone" in str(device):
if "privateuseone" in str(self.device):
import onnxruntime as ort
self.model = ort.InferenceSession(
@@ -73,11 +75,11 @@ class RMVPE(F0Predictor):
mode="script",
inputs_path=None,
save_path=jit_model_path,
device=device,
device=self.device,
is_half=is_half,
)
model = torch.jit.load(BytesIO(ckpt["model"]), map_location=device)
model = torch.jit.load(BytesIO(ckpt["model"]), map_location=self.device)
return model
def get_default_model():
@@ -99,7 +101,7 @@ class RMVPE(F0Predictor):
else:
self.model = get_default_model()
self.model = self.model.to(device)
self.model = self.model.to(self.device)
def compute_f0(
self,