Update app.py

app.py

@@ -67,11 +67,11 @@ def transcribe_audio(audio_file):
         print("Loading audio file...")
         audio = AudioSegment.from_file(audio_file)
         audio = audio.set_channels(1).set_frame_rate(16000)
-        audio_array = audio.get_array_of_samples()
+        audio_array = torch.tensor(audio.get_array_of_samples()).float()
         
         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)
+        input_features = whisper_processor(audio_array, sampling_rate=16000, return_tensors="pt").input_features.to(device).to(torch.float16)
+        predicted_ids = whisper_model.generate(input_features, language='en', task='translate')
         transcription = whisper_processor.batch_decode(predicted_ids, skip_special_tokens=True)
         
         print(f"Transcription complete. Length: {len(transcription[0])} characters")