Update app.py

app.py

@@ -1,22 +1,8 @@
 import gradio as gr
-import torch
 import os
-import subprocess
 from pytubefix import YouTube
 from moviepy.editor import VideoFileClip
 from transformers import pipeline
-import subprocess
-import sys
-
-# Ensure moviepy is installed
-try:
-    import moviepy.editor
-except ImportError:
-    subprocess.run([sys.executable, "-m", "pip", "install", "moviepy"], check=True)
-    import moviepy.editor  # Retry import after installation
-
-# Ensure required packages are installed inside Hugging Face Spaces
-subprocess.run(["pip", "install", "pytubefix", "moviepy", "transformers", "torchaudio"], check=True)
 
 # Load Whisper model for transcription
 asr = pipeline("automatic-speech-recognition", model="distil-whisper/distil-small.en")
@@ -24,32 +10,41 @@ asr = pipeline("automatic-speech-recognition", model="distil-whisper/distil-smal
 # Load Summarization model
 summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
 
-
 def process_youtube_link(youtube_url):
     try:
         # Download YouTube Video
         yt = YouTube(youtube_url)
-        video_stream = yt.streams.filter(only_audio=True).first()
-        video_path = video_stream.download(filename="video.mp4")
+        title = yt.title
+        print(f"Downloading: {title}")
+
+        video_stream = yt.streams.get_highest_resolution()
+        if not video_stream:
+            return "Error: No available video stream", ""
+
+        video_path = f"{title}.mp4"
+        video_stream.download(filename=video_path)
 
         # Extract Audio
-        audio_path = "audio.wav"
+        audio_path = f"{title}.wav"
         video = VideoFileClip(video_path)
-        video.audio.write_audiofile(audio_path)
+        video.audio.write_audiofile(audio_path, codec="pcm_s16le")
 
         # Transcribe Audio
-        transcription = asr(audio_path)
+        transcription = asr(audio_path, return_timestamps=True)
         transcribed_text = transcription["text"]
 
         # Summarize Transcription
         summary = summarizer(transcribed_text, max_length=150, min_length=50, do_sample=False)[0]["summary_text"]
 
+        # Clean up files after processing
+        os.remove(video_path)
+        os.remove(audio_path)
+
         return transcribed_text, summary
 
     except Exception as e:
         return f"Error: {str(e)}", ""
 
-
 # Create Gradio Interface
 iface = gr.Interface(
     fn=process_youtube_link,