Transcribing Youtube videos or Podcasts

HT to https://github.com/Uberi/speech_recognition

Requirements

1. pip install SpeechRecognition
2. brew install coreutils
3. brew install ffmpeg
4. pip install --upgrade google-api-python-client
5. Install youtube-dl from here: https://github.com/rg3/youtube-dl/
   • FYI - a 1 hour video converts into a 600-700mb wav file
6. Create a Google API Service Account from https://cloud.google.com/docs/authentication/getting-started#creating_the_service_account
   • Enter the JSON key into transcribe.py where indicated.

Usage

3. For YOUTUBE ./transcribe.sh youtube YOUTUBE-URL

• For PODCASTS ./transcribe.sh file PODCAST's-MP3-URL

4. TimeStamp-RESULTS.txt is your result file.

Tests

• Elon Musk with Governors - https://www.youtube.com/watch?v=OYJ89vE-QfQ - see Jul-22-17-RESULT.TXT

  • Visualized the transcript here: https://voyant-tools.org/?corpus=2de4ddec8fe29d707afc40ece4f45d7b

• Bezos + Mossberg at Code '17 - https://www.youtube.com/watch?v=VAM6b0UkEYw - see BEZOS-CODE-17.TXT