Automated closed captioning is something of a holy grail for video producers. While few would argue the value of adding captions for accessibility reasons, it is frequently cost prohibitive, especially for low-budget productions. The cost of transcribing and encoding captions on a finished program can easily run more than the entire production budget. So I was intrigued when I heard a recent NPR piece on Google’s new automatic caption feature in YouTube.
Automatic captioning is currently in beta and available on educational sites, including UMBCtube. Automatic processing is taking place on a seemingly random selection of recently uploaded videos. Site admins can also request processing of individual older videos as well. Once processed, viewers can choose to view captions by clicking on the CC icon on the movie control bar. Note that in some movie views, such as the one below, the CC icon will not appear until the movie is started. Viewers then need to click the “Transcribe Audio” option on the menu. This is a little confusing as it implies that the transcription is occurring on the fly. If this is the case, why is there a need to initially process the video? After transcribing captions appear on the movie. Viewers also have the option of changing the font and color of the captions for better viewing.
Perhaps much more useful is the option to translate captions into the full range of languages supported by Google’s translation service. This option is only available to the viewer after they activate “Transcribe Audio.” So not only can you watch the latest episode of “UMBC In the Loop” with captions, you can enjoy it in Yiddish!
Google admits to only an 80% accuracy rate in transcription, yet for a free service this is pretty impressive. Content providers have the option of downloading a text version of the transcript that can be corrected and uploaded back to YouTube. This could be a very powerful tool for producers who could use this transcript in any number of ways beyond correcting the transcription.
A few years ago a less accurate server-based solution for transcription would run more than $30,000, placing this service out of the range of most small operation producers. YouTube’s new auto-caption feature will make available content that would never otherwise be transcribed.