For every video that I post to Planet Doug, I use an AI to clean up and correct the YouTube Captions. The YouTube Captions are created automatically by the YouTube computers and they are based on simple voice-to-text technology. I assume YouTube will eventually improve on this system, but for now, that’s what they use, and the results are quite bad. The spelling and punctuation is generally awful. And any special place names and proper names from around the world are generally completely wrong.
An AI can correct the spelling and punctuation, but it can also check the spelling of proper names by comparing them to official sources on the Internet. So, for example, it will take something like “Tittywanks Montreal” from the YouTube Captions, check Google Maps, and realize that I actually said “Titiwangsa Monorail.
I just did this for my most recent uploading video, and these are the notes that the AI provided for the corrections it made:
Notes on Corrections:
- Spelling Corrections:
- “Koala Lumpur” corrected to “Kuala Lumpur” (verified with Google Maps).
- “Raja boat” corrected to “Raja Bot” (based on context and common naming in Malaysia).
- “kitchenet” corrected to “kitchenette” (standard English spelling).
- “ringit” corrected to “ringgit” (standard Malaysian currency spelling).
- “Arresh” corrected to “Arish” (based on context and phonetic consistency).
- “Dvali” corrected to “Deepavali” (standard spelling for the festival, also known as Diwali).
- “dia” corrected to “diya” (standard spelling for the clay lamps used in Deepavali).
- “Bersa” corrected to “Bursa” (verified as the correct name for the Malaysian stock exchange).
- “Tittywanks Montreal” corrected to “Titiwangsa Monorail” (verified with Google Maps as the correct station name in Kuala Lumpur).
- “Aang Shri pedaling” corrected to “Ampang-Sri Petaling” (verified as the correct LRT line name).
- “Majamic” corrected to “Masjid Jamek” (verified with Google Maps).
- “Catholon” corrected to “Decathlon” (correct brand name).
- “P rack” corrected to “pannier rack” (standard term, consistent with context).
- “Pany” corrected to “pannier” (standard term for bicycle bags/racks).
- “Pet racks” corrected to “pannier racks” (consistent with standard terminology).
- “Janned” corrected to “Jandd” (verified as the correct brand name for Jandd Extreme racks).
- “Patronous” corrected to “Petronas” (verified as the correct name for Petronas Towers).
- “Single” corrected to “Suria” (verified as Suria KLCC, the mall near Petronas Towers).