YouTube now offers 1B videos with automatic captions


Google Inc. has hit a new milestone as its YouTube video hosting service passed 1 billion videos with automatic captions.

The automatic captioning service was launched in December 2009 as a way to make YouTube videos accessible for those with limited or no hearing. It combined Google’s automatic speech recognition technology with the YouTube caption system to offer automatic captions for videos.

Available for videos recorded in English, Dutch, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish, the automatic captioning service does not require any input from the user who has uploaded the video. The captions appear on the video shortly after the video is uploaded. Those who upload the video can edit the captions should they desire.

The service isn’t perfect in understanding everything said in a video, but Google is endeavoring to improve its accuracy. In a blog post published today, YouTube Product Manager Liat Kaver said that ultimately the challenge involved YouTube’s size and diversity, but using machine learning algorithms and expanding training data has resulted in a 50 percent leap in accuracy for automatic captions in English.

Kraver added that continuing to improve the accuracy of captions remains an important goal for YouTube going forward. along with increasing the number of languages they support. But Google appealed for help from YouTube users to assist in improving the captions provided, because corrections in captions made by users feed back into the machine learning algorithms to continually improve the process.

“We can’t do it alone,” Kaver said. “We count on the amazing YouTube community of creators and viewers everywhere. Ideally, every video would have an automatic caption track generated by our system and then reviewed and edited by the creator.”

Of the now more than 1 billion automatically captioned videos on YouTube, more than 15 million videos are watched each day.

Image courtesy of YouTube