Update app.py

app.py

@@ -7,6 +7,7 @@ import tempfile
 import uuid
 
 import torch
+import transformers
 
 from nemo.collections.asr.models import ASRModel
 from nemo.collections.asr.parts.utils.streaming_utils import FrameBatchMultiTaskAED
@@ -221,6 +222,12 @@ def on_src_or_tgt_lang_change(src_lang_value, tgt_lang_value, pnc_value):
 		)
 	return src_lang, tgt_lang, pnc
 
+llm_pipeline = transformers.pipeline(
+    "text-generation", 
+    model="meta-llama/Meta-Llama-3-8B", 
+    model_kwargs={"torch_dtype": torch.bfloat16}, 
+    device_map="auto"
+)
 
 with gr.Blocks(
 	title="NeMo Canary Model",
@@ -273,16 +280,28 @@ with gr.Blocks(
 			)
 
 			model_output_text_box = gr.Textbox(
-				label="Model Output",
+				label="Transcribed Text",
 				elem_id="model_output_text_box",
 			)
 
+            llm_output_text_box = gr.Textbox(
+				label="MyAlexa's Answer",
+				elem_id="llm_output_text_box",
+			)
+            
+
 	go_button.click(
 		fn=transcribe, 
 		inputs = [audio_file, src_lang, tgt_lang, pnc],
 		outputs = [model_output_text_box]
 	)
 
+    model_output_text_box.change(
+        fn=llm_pipeline.
+        inputs = [model_output_text_box.value],
+        outputs = [llm_output_text_box]
+    )
+
 	# call on_src_or_tgt_lang_change whenever src_lang or tgt_lang dropdown menus are changed	
 	src_lang.change(
 		fn=on_src_or_tgt_lang_change,
@@ -294,6 +313,8 @@ with gr.Blocks(
 		inputs=[src_lang, tgt_lang, pnc],
 		outputs=[src_lang, tgt_lang, pnc],
 	)
+    
+    
 
 
 demo.queue()