YouTube has announced it will offer automated video captioning for clips. It will serve as an accessibility tool for hard-of-hearing viewers, but could also make clips easier to find.
The site already allows manual captioning, but most clip uploaders don’t take advantage simply because of the relatively high cost of doing so. However, YouTube’s parent company Google has now decided to use the automated transcription technology from Google Voice to produce captions.
As Google Voice customers can attest, the system isn’t always perfect, and struggles with some accents: indeed, even Google’s own demonstration of the YouTube captioning includes a mistake (“salmon” in place of “sim card”). But it’s certainly a start and at least gives enough accurate detail to get the sense of what is being said.
As well as the automated captioning, the site is also allowing clip producers to upload a text file with a transcript of the spoken content in the clip. The text file doesn’t need any time codes as a Google system will attempt to match the text to the words as they are spoken.
The new tools have uses beyond simply making life easier for deaf users. Google notes that it will be able to offer automated translations of all captioning, which would make it possible for viewers from around the world (whether deaf or not) to understand the speech in a clip.
And while the firm hasn’t mentioned this point specifically, it seems inevitable that the text of the captioning will be added to Google’s search tools, making it easy to find clips which mention a topic but don’t include the phrase in the title, description or tags.
The automated transcription will initially only be available for a limited number of clip producers, including PBS and National Geographic. That’s mainly to allow more practical testing so that the system can be refined before being expanded. However, the option of uploading your own text file is available for all new clips right away.