March 2, 2010
In an effort to make its videos more accessible, Google makes its speech-to-text tech mainstream
Google's YouTube announced Thursday morning that auto-captioning is now available to all users.
Both users and video owners can request for a video to be captioned, after which the video is added to a queue. Transcriptions could take an hour or a whole day. Though the captioning service had been available to a few small partners since 2008, this is the first time the feature is going live for the public.
The auto-captioning technology uses speech-to-text algorithms already employed by Google Voice, Google's phone service that transcribes voicemails for users. Though the service is far from perfect, the search company consistently requests if users would like to donate their messages so that the algorithms can constantly be improved. Of course, the clarity of the speaking voice affects the quality of the transcription a great deal.
"Twenty hours of video is uploaded to YouTube every minute," writes Product Manager Hiroto Tokusei. "Making some of these videos more accessible to people who have hearing disabilities or who speak different languages, not only represents a significant advancement in the democratization of information, it can also help foster greater collaboration and understanding."
In offering this new feature, Google has in mind not just the tens of millions of hearing-impaired people in the U.S., but the projected 700 million people across the globe who studies say will suffer from some form of hearing impairment by 2015.
An added bonus of the update is that users can opt to have the caption translated into any of 50 languages. While this likely increases the chance of an even more erroneous transcription, it means that auto-captioning will also be widely used by people who simply want to watch videos in a language they do not speak.
Google projects a huge spike in channels with auto-captions over the
next few months.
