update and test twitch clip downloading

transcribe_video.py

@@ -1,23 +1,45 @@
 import os
+import subprocess
 import whisper
-from moviepy.editor import VideoFileClip
+from moviepy import VideoFileClip
 from whisper.utils import get_writer
 
-def extract_audio(video_path, audio_temp_path):
-    print("Extracting uncompressed WAV audio...")
-    video = VideoFileClip(video_path)
-    # Extract as WAV, strictly setting the sample rate to 16000Hz for Whisper
-    video.audio.write_audiofile(
-        audio_temp_path, 
-        codec="pcm_s16le", 
-        ffmpeg_params=["-ar", "16000", "-ac", "1"]
-    )
-    video.close()
+model = whisper.load_model("base")
 
-def transcribe_to_srt(audio_path, output_directory, output_filename):
-    print("Loading Whisper model...")
-    model = whisper.load_model("base")
+def extract_audio(video_path, audio_temp_path):
+    # Create a temporary path for a sanitized copy of the video
+    sanitized_video_path = video_path.replace(".mp4", "_clean.mp4")
     
+    print("Sanitizing video metadata for MoviePy parser...")
+    # -map_chapters -1 removes chapter layouts that break the parser.
+    # -sn strips text/subtitle streams that crash MoviePy.
+    # -c copy copies video and audio instantly without quality loss.
+    cleanup_cmd = [
+        "ffmpeg", "-y", "-i", video_path, 
+        "-map_chapters", "-1", "-sn", 
+        "-c", "copy", sanitized_video_path
+    ]
+    
+    # Run the sanitization process silently
+    subprocess.run(cleanup_cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+
+    print("Extracting uncompressed WAV audio...")
+    try:
+        # Load the sanitized file instead of the raw Twitch clip
+        with VideoFileClip(sanitized_video_path) as video:
+            video.audio.write_audiofile(
+                audio_temp_path, 
+                fps=16000,
+                codec="pcm_s16le", 
+                ffmpeg_params=["-ac", "1"]
+            )
+    finally:
+        # Always clean up the temporary sanitized video on Windows 11
+        if os.path.exists(sanitized_video_path):
+            os.remove(sanitized_video_path)
+
+
+def transcribe_to_srt(audio_path, output_directory, output_filename):    
     print("Transcribing audio...")
     result = model.transcribe(audio_path)
     
@@ -26,6 +48,9 @@ def transcribe_to_srt(audio_path, output_directory, output_filename):
     srt_writer(result, output_filename, {})
     
     print(f"SRT subtitle file saved in: {output_directory}")
+    if os.path.exists(audio_path):
+        os.remove(audio_path)
+
 
 if __name__ == "__main__":
     video_path = "my_video.mp4"