I keep praying for fewer buttons on my TV remote, yet we keep going in the opposite direction. Thankfully, AT&T seems to understand my pain and is looking to ease it by leveraging voice technologies. The carrier showed off some new concepts today for adding voice control and cloud-based natural language processing to the TV remote.
By connecting its Watson speech technology to an iOS remote control app, AT&T researchers are able to create a simple way to search for upcoming shows or look for specific content with recorded shows. One project allows you to speak into an iPhone and use normal sentences to find upcoming episodes or movies. You can ask for specific shows, which are pretty easy to find. But it’s surprisingly good at nailing down what genre, time or channel you’re looking for in broader searches. If you say, “What are some comedy movies on tonight,” or “Movies with Woody Allen and Diane Keaton” the system recognizes your request and pulls up the corresponding results. You still have to navigate a little through longer lists but it’s pretty cool to just say what you’re looking for and get back immediate results.
Another project allows you to apply voice search to find very specific information in your recorded content on your DVR. Say you remember a funny joke Jay Leno said about Obama last night, you can say, “Show me the joke about Obama from the Jay Leno show.” The system analyzes the metadata of your recorded shows and then examines the close captioned transcripts for shows that match that description before offering up shortened clips that feature your results. You can also view the results from an iPad or then sling the video to a TV with the push of a button like Apple’s AirPlay.
There are voice-activated remotes on the market now, but often, it’s just training your remote to respond to pre-set commands you record. Google TV is also supposed to get an Android remote app that allows you to conduct voice searches. AT&T’s voice capabilities appear to go further than just search with its deep data extraction from transcripts. But overall, I think voice is a perfect way to interact with the TV. When we watch TV, it’s still more of a lean back experience, so it helps to have a way to unlock TV data without having to haul out a keyboard. This also shows how cloud-based voice services are able to really change the way we interact with machines. It’s unclear when, or if, these AT&T projects will get deployed, though it’s easy to see them working with AT&T’s U-Verse TV system. It could also be adopted by other pay TV providers, which would be a great thing.
Related content on GigaOM Pro (sub req’d):