Update app.py

app.py

@@ -25,22 +25,15 @@ dict_languages = {"English": "en",
                   "Hindi": "hi"}
 
 @spaces.GPU
-def process_transcript(model, processor, language, audio_path):
+def process_transcript(language, audio_path):
     """Process audio with selected Voxtral model and return the generated response"""
 
     if audio_path is None:
         return "Please provide some input audio: either upload an audio file or use the microphone."
     else:
         id_language = dict_languages[language]
-        raw_inputs = processor.apply_transcrition_request(language=id_language, audio=audio_path, model_id=model_name)
-#        inputs = inputs.to(device, dtype=torch.bfloat16)
-#        inputs = {k: v.to(device, dtype=torch.bfloat16) for k, v in inputs.items()}
-        inputs = {}
-        for k, v in raw_inputs.items():
-            if k == "input_ids":
-                inputs[k] = v.to(device=device, dtype=torch.long)
-            else:
-                inputs[k] = v.to(device=device, dtype=torch.bfloat16)
+        inputs = processor.apply_transcrition_request(language=id_language, audio=audio_path, model_id=model_name)
+        inputs = inputs.to(device, dtype=torch.bfloat16)
         outputs = model.generate(**inputs, max_new_tokens=MAX_TOKENS)
         decoded_outputs = processor.batch_decode(outputs[:, inputs.input_ids.shape[1]:], skip_special_tokens=True)
 
@@ -82,7 +75,7 @@ with gr.Blocks(title="Transcription") as transcript:
 
             
     submit_transcript.click(
-        fn=lambda v1, v2: process_transcript(model, processor, v1, v2),
+        fn=process_transcript,
         inputs=[sel_language, sel_audio],
         outputs=text_transcript
     )