Update djcm_module.py
Browse files- djcm_module.py +12 -5
djcm_module.py
CHANGED
|
@@ -1,16 +1,23 @@
|
|
| 1 |
import torch
|
|
|
|
|
|
|
| 2 |
|
| 3 |
class DJCM:
|
| 4 |
-
def __init__(self, model_path="models/djcm.pt", device="cuda"):
|
| 5 |
self.device = device
|
|
|
|
|
|
|
|
|
|
| 6 |
self.model = torch.load(model_path, map_location=device)
|
| 7 |
self.model.eval()
|
| 8 |
|
| 9 |
-
def
|
| 10 |
"""
|
| 11 |
-
|
| 12 |
-
return: f0
|
| 13 |
"""
|
|
|
|
| 14 |
with torch.no_grad():
|
| 15 |
f0 = self.model(audio_tensor)
|
| 16 |
-
return f0
|
|
|
|
|
|
| 1 |
import torch
|
| 2 |
+
import os
|
| 3 |
+
import numpy as np
|
| 4 |
|
| 5 |
class DJCM:
|
| 6 |
+
def __init__(self, model_path="/content/HRVC/models/rvc/predictors/djcm.pt", device="cuda"):
|
| 7 |
self.device = device
|
| 8 |
+
self.model_path = model_path
|
| 9 |
+
if not os.path.exists(model_path):
|
| 10 |
+
raise FileNotFoundError(f"DJCM model not found at {model_path}")
|
| 11 |
self.model = torch.load(model_path, map_location=device)
|
| 12 |
self.model.eval()
|
| 13 |
|
| 14 |
+
def infer_from_audio(self, audio_np):
|
| 15 |
"""
|
| 16 |
+
audio_np: np.ndarray, shape [T]
|
| 17 |
+
return: f0 np.ndarray
|
| 18 |
"""
|
| 19 |
+
audio_tensor = torch.from_numpy(audio_np).float().unsqueeze(0).to(self.device)
|
| 20 |
with torch.no_grad():
|
| 21 |
f0 = self.model(audio_tensor)
|
| 22 |
+
return f0.squeeze().cpu().numpy()
|
| 23 |
+
|