Update app.py

app.py

@@ -7,6 +7,7 @@ import time
 import torch
 from transformers import pipeline
 import re
+from whisperplus import download_youtube_to_mp3
 
 # Initialize the LLM and other components
 llm = HuggingFaceEndpoint(
@@ -85,7 +86,15 @@ def process_audio(audio_path):
         classifications.append(f"Sentence: {sentence}\nTopic: {classification}\nLanguage: {lang}\nTime: {duration:.2f}s")
     
     return "\n\n".join(classifications)
-
+def handle_audio_input(audio_path=None, youtube_url=None):
+    if youtube_url:
+        audio_path = download_youtube_to_mp3(youtube_url, output_dir="downloads", filename="youtube_audio")
+    
+    if audio_path:
+        return process_audio(audio_path)
+    else:
+        return "No audio input provided."
+    
 # Create the Gradio interface
 def create_gradio_interface():
     with gr.Blocks() as iface:
@@ -104,10 +113,14 @@ def create_gradio_interface():
 
         with gr.Tab("Audio Input"):
             audio_input = gr.Audio(label="Upload Audio", type="filepath")
+            youtube_input = gr.Textbox(label="YT URL (optional)")
             audio_output = gr.Textbox(label="Detected Topics from Audio")
             audio_submit_btn = gr.Button("Process Audio")
 
-            audio_submit_btn.click(fn=process_audio, inputs=audio_input, outputs=audio_output)
+            def on_audio_submit(audio, youtube_url):
+                return handle_audio_input(audio_path=audio, youtube_url=youtube_url)
+
+            audio_submit_btn.click(fn=on_audio_submit, inputs=[audio_input, youtube_input], outputs=audio_output)
 
     iface.launch()