Risk
3/4/2010
04:52 PM
Connect Directly
LinkedIn
Twitter
Google+
RSS
E-Mail
50%
50%

YouTube Gets Auto-Captioning From Google Speech Tech

The video sharing site now allows anyone to generate captions for videos with coherent spoken English.

In a move to make its massive store of video content more accessible, Google's YouTube is making automated caption generation available to all YouTube users.

YouTube initially added a caption feature in 2008. Last November, it introduced auto-captioning for a select group of partners.

Now, any video created with a clear audio track -- unless disallowed, an option for some of YouTube's content partners -- can be captioned automatically, thanks to the speech-to-text algorithms that power Google Voice Search.

What's more, those captions can be translated from English into one of 50 supported languages at the viewer's discretion.

At the moment, auto-captioning only works in videos with spoken English, but Google product manager Hiroto Tokusei says in a blog post that YouTube plans to support the captioning of more languages in the months ahead.

In a related effort, Google is also working to turn Android phones into universal translators through a combination of speech-to-text and translation technology.

"For content owners, the power of auto-captioning is significant," said Tokusei. "With just a few quick clicks your videos can be accessed by a whole new global audience. And captions can make is easier for users to discover content on YouTube."

Captions, as text content, are useful to Google as a way to improve search relevancy. And with the volume of information that Google has to manage -- over 20 hours of video are uploaded to YouTube every minute -- every improvement helps.

Although speech-to-text conversion isn't perfect, Tokusei says that Google's technology is getting better. Video owners can also improve caption files by downloading them, making corrections, and then uploading them back to YouTube.

Other Google accessibility projects include a talking RSS reader for Android devices, support for WAI-ARIA, the Accessible Rich Internet Applications Suite, in Google Chrome, and support for the AxsJAX framework.

About 650 million people live with a disability, according to the UN.

By 2015, Professor Adrian Davis of the British MRC Institute of Hearing Research estimates that more than 700 million people will be suffering from hearing loss of more than 25 dB, a consequence both of aging and of exposure to noise, among other causes.

Last October, Google consolidated its accessibility resources at a single Web address.

Comment  | 
Print  | 
More Insights
Register for Dark Reading Newsletters
White Papers
Cartoon
Current Issue
Flash Poll
Video
Slideshows
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2011-4403
Published: 2015-04-24
Multiple cross-site request forgery (CSRF) vulnerabilities in Zen Cart 1.3.9h allow remote attackers to hijack the authentication of administrators for requests that (1) delete a product via a delete_product_confirm action to product.php or (2) disable a product via a setflag action to categories.ph...

CVE-2012-2930
Published: 2015-04-24
Multiple cross-site request forgery (CSRF) vulnerabilities in TinyWebGallery (TWG) before 1.8.8 allow remote attackers to hijack the authentication of administrators for requests that (1) add a user via an adduser action to admin/index.php or (2) conduct static PHP code injection attacks in .htusers...

CVE-2012-2932
Published: 2015-04-24
Multiple cross-site scripting (XSS) vulnerabilities in TinyWebGallery (TWG) before 1.8.8 allow remote attackers to inject arbitrary web script or HTML via the (1) selitems[] parameter in a copy, (2) chmod, or (3) arch action to admin/index.php or (4) searchitem parameter in a search action to admin/...

CVE-2012-5451
Published: 2015-04-24
Multiple stack-based buffer overflows in HttpUtils.dll in TVMOBiLi before 2.1.0.3974 allow remote attackers to cause a denial of service (tvMobiliService service crash) via a long string in a (1) GET or (2) HEAD request to TCP port 30888.

CVE-2015-0297
Published: 2015-04-24
Red Hat JBoss Operations Network 3.3.1 does not properly restrict access to certain APIs, which allows remote attackers to execute arbitrary Java methos via the (1) ServerInvokerServlet or (2) SchedulerService or (3) cause a denial of service (disk consumption) via the ContentManager.

Dark Reading Radio
Archived Dark Reading Radio
Join security and risk expert John Pironti and Dark Reading Editor-in-Chief Tim Wilson for a live online discussion of the sea-changing shift in security strategy and the many ways it is affecting IT and business.