Update app.py

app.py

@@ -7,18 +7,12 @@ MAX_TOKENS = 32000
 device = "cuda" if torch.cuda.is_available() else "cpu"
 print(f"*** Device: {device}")
 
-# List models
-dict_models = {'Voxtral-Mini-3B-2507': 'mistralai/Voxtral-Mini-3B-2507',
-               'Voxtral-Small-24B-2507': 'mistralai/Voxtral-Small-24B-2507'}
-
-# Load models
-list_processor = []
-list_model = []
-for model_name in dict_models.values():
-    list_processor.append(AutoProcessor.from_pretrained(model_name))
-    list_model.append(VoxtralForConditionalGeneration.from_pretrained(model_name,
-                                                                      torch_dtype=torch.bfloat16,
-                                                                      device_map=device))
+model_name = 'mistralai/Voxtral-Mini-3B-2507'
+
+processor = AutoProcessor.from_pretrained(model_name)
+model = VoxtralForConditionalGeneration.from_pretrained(model_name,
+                                                        torch_dtype=torch.bfloat16,
+                                                        device_map=device)
 # Supported languages
 dict_languages = {"English": "en",
                   "French": "fr",
@@ -67,17 +61,12 @@ with gr.Blocks(title="Transcription") as transcript:
         with gr.Column():
             text_transcript = gr.Textbox(label="Generated Response", lines=10)
 
-    try:
-        model_index = list(dict_models.keys()).index(sel_model)
-        submit_transcript.click(
-            fn=process_transcript,
-            inputs=[dict_languages[sel_language], list_model[model_index],
-                    list_processor[model_index], sel_audio],
-            outputs=text_transcript
-        )
-    except:
-        text_transcript = 'Error'
 
+    submit_transcript.click(
+        fn=process_transcript,
+        inputs=[dict_languages[sel_language], model, processor, sel_audio],
+        outputs=text_transcript
+    )
 
 # Launch the app
 if __name__ == "__main__":