Google just announced that YouTube can now automatically generate captions and subtitles for videos in English. For now, this feature is only enabled on a handful of partner channels, but Google plans to make this feature available for all users in the future.
In the meantime, YouTube now also offers a new ‘automatic caption timing’ feature for all new uploads that makes it easier to add captions manually. You simply upload a text file with a transcript of the video and Google’s speech recognition technology will figure out when those words are spoken and create captions based on this information.
As Google points out, YouTube’s users currently upload over 20 hours of video every minute – and most of this video isn’t accessible for users with hearing impairments. While uploaders could always add captions to their videos manually, only a very small minority of users ever did so.
YouTube’s speech recognition technology is based on the same speech-to-text algorithms that transcribe voicemails in Google Voice. You can also translate these captions into 51 languages.
As expected, these captions aren’t always perfect, but work surprisingly well on the videos that we have seen so far.
If you want to have a look at how these captions work, have a look at one of the videos in the UC Berkeley, Stanford, MIT, Yale, UCLA, Duke, UCTV, Columbia, PBS, National Geographic, Demand Media, UNSW and Google & YouTube channels.
