November 30, 2009

Google recently announced new technologies to make Closed Captioning easier on YouTube. The YouTube video player has been able to display Closed Captions for some time but only for those videos that have time-coded transcripts, which are rather tedious to construct from scratch. In their attempt to make this easier, Google developed “auto-time” that will match spoken words with written words to create a time-coded Closed Caption file. It relies on voice recognition technology to match sound to text and generates appropriately sized phrases for display at the bottom of the video frame.

I tried this out on the videos we have on Densho’s YouTube channel and found auto-time to work surprisingly well. The fact that I was able to quickly take advantage of Google’s new auto-time technology is a testament to Densho’s process to transcribe completely every video interview as they are captured. Because we have the full transcription available I could easily use Google’s auto-time to insert the time-coding data.

Auto-time doesn’t do a flawless job but it comes pretty close. Fortunately, YouTube allows you to download the generated time-coded file to make corrections or for use in other ways.
Google also announced “auto-cap” that will generate transcripts automatically. It hasn’t been generally released yet but looks like it may be useful for a rough first-pass approximation. It’s not quite up to the quality standards we aim for with Densho interviews so we’ll no doubt continue with manual transcriptions for some time.

Adding Closed Captions to Densho’s videos makes it possible for us to reach an even wider audience including the hearing impaired and even non-English speakers because Google is also working on automatic translation of Closed Captions to other languages. Just imagine, somewhere off in the not-too-distant future, people in every country will be able to enjoy Densho’s rich collection of stories in their own language.

In the meantime, I invite you to check out the videos on Densho’s YouTube channel, now with Closed Captioning!