Rogerjs commited on
Commit
4f9d782
·
verified ·
1 Parent(s): 4d6d38d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -3
app.py CHANGED
@@ -9,6 +9,13 @@ from scipy.io import wavfile
9
  from bark import generate_audio, SAMPLE_RATE
10
  from bark.generation import preload_models, load_model, generate_text_semantic
11
 
 
 
 
 
 
 
 
12
  class VoiceCloningApp:
13
  def __init__(self):
14
  # Create working directory
@@ -19,8 +26,8 @@ class VoiceCloningApp:
19
  # Explicit model loading with error handling
20
  try:
21
  print("Attempting to load Bark models...")
22
- preload_models()
23
- print("Bark models loaded successfully.")
24
  except Exception as e:
25
  print(f"Error loading Bark models: {e}")
26
  raise RuntimeError(f"Could not load Bark models: {e}")
@@ -62,13 +69,15 @@ class VoiceCloningApp:
62
  history_prompt=None,
63
  temp=0.7,
64
  min_eos_p=0.05,
 
65
  )
66
 
67
  # Generate audio from semantic tokens
68
  audio_array = generate_audio(
69
  semantic_tokens,
70
  history_prompt=None,
71
- temp=0.7
 
72
  )
73
 
74
  # Save generated audio
 
9
  from bark import generate_audio, SAMPLE_RATE
10
  from bark.generation import preload_models, load_model, generate_text_semantic
11
 
12
+ # Add device detection
13
+ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
14
+ print("CUDA available:", torch.cuda.is_available())
15
+ print("CUDA device count:", torch.cuda.device_count())
16
+ if torch.cuda.is_available():
17
+ print("CUDA device name:", torch.cuda.get_device_name(0))
18
+
19
  class VoiceCloningApp:
20
  def __init__(self):
21
  # Create working directory
 
26
  # Explicit model loading with error handling
27
  try:
28
  print("Attempting to load Bark models...")
29
+ preload_models(device=device)
30
+ print(f"Bark models loaded successfully on {device}")
31
  except Exception as e:
32
  print(f"Error loading Bark models: {e}")
33
  raise RuntimeError(f"Could not load Bark models: {e}")
 
69
  history_prompt=None,
70
  temp=0.7,
71
  min_eos_p=0.05,
72
+ device=device
73
  )
74
 
75
  # Generate audio from semantic tokens
76
  audio_array = generate_audio(
77
  semantic_tokens,
78
  history_prompt=None,
79
+ temp=0.7,
80
+ device=device
81
  )
82
 
83
  # Save generated audio