Update whisper_cs.py

whisper_cs.py

@@ -14,6 +14,7 @@ torch_dtype = torch.float32
 MODEL_PATH_V2 = "langtech-veu/whisper-timestamped-cs"
 MODEL_PATH_V2_FAST = "langtech-veu/faster-whisper-timestamped-cs"
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+print("[INFO] CUDA available:", torch.cuda.is_available())
 
 def clean_text(input_text):
     remove_chars = ['.', ',', ';', ':', '¿', '?', '«', '»', '-', '¡', '!', '@',
@@ -104,21 +105,20 @@ def cleanup_temp_files(*file_paths):
         if path and os.path.exists(path):
             os.remove(path)
 
-if torch.cuda.is_available():
+try:
     faster_model = WhisperModel(
         MODEL_PATH_V2_FAST,
-        device="cuda",
-        compute_type="float16"  #"int8_float16" 
+        device="cuda" if torch.cuda.is_available() else "cpu",
+        compute_type="float16" if torch.cuda.is_available() else "int8"
     )
-else:
+except RuntimeError as e:
+    print(f"[WARNING] Failed to load model on GPU: {e}")
     faster_model = WhisperModel(
         MODEL_PATH_V2_FAST,
         device="cpu",
         compute_type="int8"
     )
 
-faster_model = WhisperModel(MODEL_PATH_V2_FAST, device=DEVICE, compute_type="float16")
-
 def load_whisper_model(model_path: str):
     device = "cuda" if torch.cuda.is_available() else "cpu"
     model = whisper_ts.load_model(model_path, device=device)