Initial commit

.env_example

@@ -0,0 +1,2 @@
+OPENAI_API_KEY=
+OPENAI_BASE_URL=

.gitignore

@@ -0,0 +1,2 @@
+venv/*
+.env

app.py

@@ -0,0 +1,40 @@
+import argparse
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from moviepy import VideoFileClip
+from openai import OpenAI
+from pydub import AudioSegment
+from pydub.playback import play
+
+DEFAULT_RESPONSES_INSTRUCTIONS = "You will be provided a video transcription for which you are to generate a blog post in Markdown format summarizing the video's contents."
+TMP_AUDIO_PATH = "/tmp/video-summary-bot-tmp-audio.wav"
+
+async def main():
+    parser = argparse.ArgumentParser(description="Use AI models to summarize videos")
+    parser.add_argument('--video-file', type=str, help="Path to the video to be summarized")
+    parser.add_argument('--segment-duration', type=int, help="Lenght of audio segments")
+    args = parser.parse_args()
+    load_dotenv()
+    VideoFileClip(args.video_file).audio.write_audiofile(TMP_AUDIO_PATH)
+    audio = AudioSegment.from_wav(TMP_AUDIO_PATH)
+    segments = [audio[i:i + args.segment_duration] for i in range(0, len(audio), args.segment_duration)]
+    # TODO: Test OpenAI
+    client = OpenAI()
+    transcription = ' '.join([
+        await client.audio.transcriptions.create(
+            model=args.transcription_model,
+            file=each
+            ).text for each in segments
+        ])
+    summary = client.responses.create(
+            model=args.responses_model,
+            instructions=DEFAULT_RESPONSES_INSTRUCTIONS,
+            input=transcription
+            )
+    return summary
+
+if __name__ == '__main__':
+    summary = main()
+    print(summary)

requirements.txt

@@ -0,0 +1,2 @@
+moviepy
+pydub