Update app.py

app.py

@@ -14,12 +14,15 @@ from unstructured.partition.auto import partition
 
 # Imports for advanced file processing
 import speech_recognition as sr
-from moviepy.editor import VideoFileClip
+from pydub import AudioSegment
 
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# --- Tool Definition (Upgraded for Full Multimodality) ---
+# --- Constants ---
+DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+
+# --- Tool Definition (Upgraded for Full Multimodality with pydub) ---
 @tool
 def file_reader(file_path: str) -> str:
     """
@@ -37,6 +40,7 @@ def file_reader(file_path: str) -> str:
         str: Extracted or transcribed content as text.
     """
     temp_file_path = None
+    audio_temp_path = None
     try:
         # Download the file if it's a URL
         if file_path.startswith("http://") or file_path.startswith("https://"):
@@ -57,25 +61,20 @@ def file_reader(file_path: str) -> str:
             if mime_type.startswith("audio/"):
                 with sr.AudioFile(local_path) as source:
                     audio = recognizer.record(source)
-                    # Using whisper for robust speech recognition
                     return recognizer.recognize_whisper(audio)
 
-            # Handle video files by extracting audio
+            # Handle video files by extracting audio with pydub
             elif mime_type.startswith("video/"):
-                # Use a temporary file for the extracted audio
                 with NamedTemporaryFile(suffix=".wav", delete=False) as audio_temp:
                     audio_temp_path = audio_temp.name
                 
-                clip = VideoFileClip(local_path)
-                clip.audio.write_audiofile(audio_temp_path, codec='pcm_s16le')
+                # Extract audio using pydub
+                video_audio = AudioSegment.from_file(local_path, format=mime_type.split('/')[1])
+                video_audio.export(audio_temp_path, format="wav")
                 
                 with sr.AudioFile(audio_temp_path) as source:
                     audio = recognizer.record(source)
                 
-                # Clean up the temporary audio file
-                os.remove(audio_temp_path)
-
-                # Using whisper for robust speech recognition
                 return recognizer.recognize_whisper(audio)
 
         # Default to handling text and images with OCR if not audio/video
@@ -88,6 +87,10 @@ def file_reader(file_path: str) -> str:
         # Clean up the downloaded file if it exists
         if temp_file_path and os.path.exists(temp_file_path):
             os.remove(temp_file_path)
+        # Clean up the temporary audio file
+        if audio_temp_path and os.path.exists(audio_temp_path):
+            os.remove(audio_temp_path)
+
 
 
 # --- Agent Class (Updated with More Powerful Model and Tools) ---