x

app.py

@@ -8,19 +8,17 @@ def asr(url):
     # download audio
     # Options for youtube-dl
     ydl_opts = {
-        'outtmpl': 'my_video.mp4',
-        'overwrites': True, 
-        'format': 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best',
-        'merge_output_format': 'mp4',
+        'outtmpl': 'video_downloaded.%(ext)s',
     }
 
+
     # Create a youtube-dl object
     ydl = yt_dlp.YoutubeDL(ydl_opts)
 
     # Download the video
     info_dict = ydl.extract_info(url, download=True)
-
-    audio_file= open("my_video.mp4", "rb")
+    video_file = "video_downloaded.{}".format(info_dict["ext"])
+    audio_file= open(video_file, "rb")
     transcript = openai.Audio.transcribe("whisper-1", audio_file)
     output = openai.ChatCompletion.create(
     model="gpt-3.5-turbo",
@@ -28,6 +26,9 @@ def asr(url):
             {"role": "user", "content": "Transcript: {transcript}. \n Translate the video conversation transcript into fluent Chinese. Chinese: ".format(transcript=transcript["text"])},
         ]
     )
+
+    # delete the video
+    os.system("rm {}".format(video_file))
     return output['choices'][0]['message']['content'], transcript["text"]
 
 title = """