YouTube Will Automatically Caption Your Video

Three years ago YouTube/Google added the ability for video authors to add captions to videos. Over time support for multiple caption tracks was included, the expansion of search to consider text in captions, and even machine translation support for the captions (see my other post about machine translation risks).

Even with hundreds of thousands of captioned videos on YouTube, new videos are posted at the rate of 20 hours of video per minute. For many companies (and not-for-profits and government agencies), YouTube provides the most cost-effective and ubiquitous method to distribute video content to users. So many of these organizations (particularly not-for-profits and government agencies) are required by law (US and elsewhere) to provide captions for video, but don’t have the experience or tools to do so. Users who are deaf are excluded from fully understanding this content as a result.

This is where the speech recognition features (ASR) of Google Voice come into play. This technology can parse the audio track of your videos and create captions automatically. Much like machine translation, the quality of these captions may not be the best, but it can at least provide enough information for a user who could not otherwise understand the video at all to glean some meaning and value.

In addition, Google is launching “automatic caption timing,” essentially allowing authors to easily make captions using a text file. As the video creator, an author will be able to create a text file with all the words in the video and Google’s speech recognition software will figure out where those words are spoken and take care of the timing. This technique can greatly increase the quality of captions on videos with very little effort (or cash outlay for tools) on the part of the video creator.

You can read more at the the YouTube Help Center article. You can also read the blog post announcing this feature at the Google Blog. The video below shows a short demo about the auto-captioning and auto-timing features.

Update (August 25, 2010): Paul Bukhovko of FatCow was kind enough to translate this entry into Belorussian: YouTube аўтаматычна захоплівае сваё відэа

Update: February 29, 2016

Take a look at Web Axe’s post Captions and YouTube for tips on how to correct YouTube’s auto-captions.