AT&T yesterday showed off several IPTV concept applications at a press briefing in downtown Manhattan.
At first, it seemed like nothing especially new. The telco has trotted out many of these demos before — e.g., using the iPhone as a remote control for U-verse TV, and caller ID features on the TV. Peter Hill, vice president of video and converged services at AT&T Labs, demo’d both of those last fall at TelcoTV.
On Monday, Hill was there again, and he acknowledged at the outset that some of the apps were repeats.
But there was one new thing that caught my eye — or, I should say, my ear.
Off in one corner AT&T had set up a prototype of "multimodal voice search for IPTV," which combined speech recognition, natural-language queries and an interactive program guide.
You talk into a microphone built into the remote-control, and the set-top passes the digitized voice up to a server in the network, which analyzes the request and returns the search results. The system uses AT&T’s Watson speech-recognition algorithms.
The system worked; at least, the prototype did. It was very cool, and faster than keying through a series of IPG menus (or, needless to say, typing text using an on-screen keyboard). And it doesn’t require any speaker-dependent training, according to AT&T.
Cooler than cable? This feature, if AT&T can commercially launch it, actually is.