Update app.py

app.py

@@ -72,7 +72,17 @@ def transcribe_audio(audio_file):
         print(f"Audio duration: {len(audio) / 1000:.2f} seconds")
         print("Starting transcription...")
         input_features = whisper_processor(audio_array, sampling_rate=16000, return_tensors="pt").input_features.to(device)
-        predicted_ids = whisper_model.generate(input_features)
+        
+        # Create attention mask
+        attention_mask = torch.ones_like(input_features)
+        
+        # Generate with specific parameters
+        predicted_ids = whisper_model.generate(
+            input_features,
+            attention_mask=attention_mask,
+            language='en',
+            task='translate'
+        )
         transcription = whisper_processor.batch_decode(predicted_ids, skip_special_tokens=True)
         
         print(f"Transcription complete. Length: {len(transcription[0])} characters")