try yt-dlp

hls_download.py

@@ -1,35 +1,59 @@
-import subprocess
+import yt_dlp
 import os
+import subprocess
 
-def download_clips(stream_url, out_dir, start_time, end_time, resize=True):
+def download_clips_yt_dlp(stream_url, out_dir, start_time, end_time, resize=True):
     output_file = os.path.join(out_dir, f"train_{len(os.listdir(out_dir))}.mp4")
     
-    ffmpeg_cmd = [
-        'ffmpeg',
-        '-f', 'hls',
-        '-i', stream_url,
-        '-ss', str(max(0, start_time - 2)),  # Ensure start time is not negative
-        '-to', str(end_time + 2),
-        '-r', '30',
-        '-c:v', 'libx264',
-        '-preset', 'ultrafast',  # Use fastest encoding preset
-        '-crf', '23',  # Slightly lower CRF for better quality
-        '-maxrate', '1M',  # Limit bitrate
-        '-bufsize', '2M',
-        '-vf', f"scale=-2:{360 if resize else 720}",
-        '-c:a', 'aac',  # Re-encode audio for better compatibility
-        '-b:a', '128k',
-        '-max_muxing_queue_size', '1024',  # Increase muxing queue size
-        '-y',  # Overwrite output file if it exists
-        output_file
-    ]
+    # Calculate the duration we want to download
+    duration = end_time - start_time + 4  # Add 4 seconds for 2 second padding on each side
+
+    # Configure yt-dlp options
+    ydl_opts = {
+        'format': 'best',
+        'outtmpl': os.path.join(out_dir, 'temp_download.%(ext)s'),
+        'nocheckcertificate': True,
+        'no_warnings': True,
+        'external_downloader': 'ffmpeg',
+        'external_downloader_args': [
+            'ffmpeg_i', 
+            f'-ss {max(0, start_time - 2)}',  # Start 2 seconds earlier, but not before 0
+            f'-t {duration}'  # Download only the duration we need
+        ],
+    }
 
-    print([f'{x} ' for x in ffmpeg_cmd])
-    
     try:
-        subprocess.run(ffmpeg_cmd, check=True, capture_output=True, text=True)
+        # Download the specific portion of the video
+        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+            info = ydl.extract_info(stream_url, download=True)
+            downloaded_file = ydl.prepare_filename(info)
+
+        # Use FFmpeg to process the downloaded file if resizing is needed
+        if resize:
+            processed_file = os.path.join(out_dir, 'temp_processed.mp4')
+            ffmpeg_cmd = [
+                'ffmpeg',
+                '-i', downloaded_file,
+                '-c:v', 'libx264',
+                '-preset', 'veryfast',
+                '-crf', '23',
+                '-maxrate', '2M',
+                '-bufsize', '4M',
+                '-vf', f"scale=-2:360",
+                '-c:a', 'aac',
+                '-b:a', '128k',
+                '-y',
+                processed_file
+            ]
+
+            subprocess.run(ffmpeg_cmd, check=True, capture_output=True, text=True)
+            os.remove(downloaded_file)
+            os.rename(processed_file, output_file)
+        else:
+            os.rename(downloaded_file, output_file)
+
         return output_file
-    except subprocess.CalledProcessError as e:
-        print(f"Error occurred: {e}")
-        print(f"FFmpeg output: {e.output}")
+
+    except Exception as e:
+        print(f"An error occurred: {e}")
         return None

requirements.txt

@@ -9,4 +9,5 @@ opencv-python-headless==4.7.0.68
 # openvino-dev==2022.3.0
 torch
 torchvision
-onnxruntime-gpu
+onnxruntime-gpu
+yt-dlp