Sorry Google, YouTube Captions Aren’t for the Deaf. They’re for Your Robots.

Google, who owns YouTube, rolled out auto-captioning for English language videos yesterday.  All videos with a clear English audio track will have automatically-generated captions.

At the press conference a deaf engineer did the product demo and students from the California School for the Deaf in Fremont came on stage.  Google is painting this development as a service to the hearing impaired.

But that’s not Google’s true motivator.

Changing audio into text lets Google spiders index the content of YouTube videos.

The spiders don’t understand audio so they can’t index videos for search.  But with YouTube audio available in text form, a huge and invisible chunk of the web is opened up to Google’s search technology.

This will be a watershed for Google.  It is a good time to buy Google stock.  The company will make a lot of money selling advertising on these newly-indexed videos. Here’s what to expect:

Better Search Results within YouTube

Now Google’s army of spiders can crawl, index, and return YouTube videos just like regular web pages.  The quality of YouTube search results will rise.

With better results comes better ad matching.  The YouTube search results page should have yellow “Sponsored Links” in the near future – just like normal Google searches do.

More Google Ads on YouTube

Matching relevant ads to YouTube videos was hard because the content was a mystery. Now that’s changed, so we should see more relevant text ads on video pages themselves.

For advertisers this is huge.  Many companies don’t buy ads on user-generated content sites for fear of appearing beside offensive material.  But now they can at least filter out any videos with offensive language.

More YouTube Videos in Normal Search Results

Finally, YouTube videos will show up much more often in regular search results. Google will apply its PageRank algorithm and keyword relevancy tests to videos just like HTML pages and feel more comfortable returning them.

They sometimes pull videos into regular search results now.  We will see many more.

There is no way Google’s main motivator was to help the estimated 0.3% of functionally deaf Americans watch online video.

But it is a feel-good story.

Tags: google, indexing, online video, youtube

  • brian

    Been looking around to see if google has actually indexed videos for search and it seems that most people haven’t found this to be the case just yet. Have you noticed text from videos showing up higher in google search results yet?

    Seems like the translation is pretty terrible from voice into captions. Can you say ALL YOUR BASE ARE BELONG TO US?!

  • Joseph McCullough

    I think it’s a great idea. Even if their is an “alternative motive” for the captioning, I think it will bring a better user experience from searches. I generally value video content more than text content for any phrase with “tutorial” in it for example. However, since there are SO many videos, I’m wondering how they would prevent a flood of youtube videos from filling up the first page of every search.

  • Xander Becket

    Haven’t noticed anything yet. It’s coming though, and all your base will soon belong to Google!

  • Pingback: Social Annotation « Beyond WebCT: Integrating Social Networking Tools Into Language & Culture Courses()

  • Pingback: Social Annotation « Beyond WebCT: Integrating Social Networking Tools Into Language & Culture Courses()

  • Pingback: The Surprising (Content) Future of Google+ | DigitalNext: A Blog on Emerging Media and Technology - Advertising Age | Internet Billboards()

  • voxleo

    There is a problem with this – But it DOES explain a lot.. Your article just clarified why the transcriptions are so nonsensical as to be utterly useless and just confusing to anyone trying to use that tool for actually understanding what’s being said in a clip. Google’s tool is going to look specifically for, and therefore “hear”, the words it’s commonly using as indexers. That will alter the product considerably when there are preferential interpretations of similar sounds. Almost all speech is nuance, and indexed words are anything but nuance. Its going to be horrifically unreliable and everything will be slanted to whatever buzzwords the indexer is deciding are hottest in the cloud at the moment. For example – “a bet” placed with “a fellow at a casino” or gambler is going to become some kind of crime if the indexing tool is built for police use- suddenly everyone will be “abetting felons facing treason”