Description
------------
Given a long conference stream of multiple talks, we use silence detection of the audio to determine the timecode of each talk.


Requirement
------------
- python3
- ffmpeg
- [youtube-dl](https://github.com/ytdl-org/youtube-dl)
- [pydub](https://github.com/jiaaro/pydub)
- [auditok 0.1.8](https://github.com/amsehili/auditok/)


CLI
-----------
./ep_yt2talks.sh youtubeURL


Example
-----------
./ep_yt2talks.sh https://www.youtube.com/watch?v=hgry3vkuI5Y

Result: ['00:22:45 - 00:28:37', '00:34:58 - 01:18:16', '01:50:47 - 02:36:27', '02:41:06 - 03:22:00', '03:30:44 - 04:16:32', '05:20:40 - 05:47:48', '05:55:31 - 06:26:51', '06:50:34 - 07:20:44', '07:25:23 - 07:55:26', '08:07:27 - 08:51:30', '08:55:33 - 10:01:14']
Each item in the list corresponds to the start and end time of each talk in the video stream.