7 WAYS YOU WILL SOON TALK TO YOUR TV, HOME, AND CAR

Soon You'll Be Chatting With Your Car, TV, and Home
By John Brandon, Popular Mechanics, 21 February 2013.

As impressive as Apple's Siri can be, we've seen plenty of examples of what the voice-activated personal assistant can't do. However, it and other voice recognition systems are getting smarter, understanding a new range of natural-language spoken commands. You'll be conversing with your car before you know it.

Siri, Apple's personal assistant, already understands complex voice commands. You can ask her to find nearby pizza places, and then see a list of options. Say "Tell my wife I'm running late," and Siri will prompt you to send a text. You can ask about the weather and she'll know you are in Chicago, not Miami.

Yet Siri is far from perfect. You can't dictate an email yet, and she doesn't always understand context. If you say "Play my favourite music," she will think you mean an artist named favourite music.

Changes are coming to speech tech in 2013. Several companies are developing next-gen speech technology that will understand everyday language, not just a small subset of voice commands. Using technology developed by Nuance, many modern cars can understand voice commands, and even let you dictate a text message. Google has added speech recognition to everything from smartphones to the Chrome browser. We'll be able to search for all George Clooney movies just by saying his name, control the climate in our car from the back seat by speaking the temperature, and control a PC with our voice in a way that actually works. Improved algorithms, better omnidirectional microphones, and home control of your TV are all coming.

1. Bing Voice Search

For the past year, Microsoft has steadily improved the Bing Voice Search feature on the Xbox 360 and on Windows 8. This year, voice search will become more fluid and even more intelligent.

Voice tech works a bit like mapping technology - it is always improving, says Stefan Weitz, the Microsoft director of search. According to Weitz, one recent improvement is in understanding exact vocabulary. Imagine saying "Xbox - Bing - What is the largest lake in Europe?" which asks Bing to search for the answer in Internet Explorer. In the past, Bing might have returned the answer Ladoga Lake in Russia. Now the answer is Caspian Sea because, while the name does not include the word lake, it is technically the largest lake.

"The system self-corrected based on a better understanding of the question," Weitz says, explaining that Voice Search is starting to understand more complexities of language. You can now say "Who played James Bond in Goldeneye," and the answer will come up in Wikipedia as Pierce Brosnan. In the past, the speech interpreter might not have understood you meant the actor who played James Bond.

Another improvement: Bing Voice Search is adapting to intent and constantly learning about what humans actually want. One example Weitz gives: If you say "What is the movie with Tom Cruise and unicorns," the speech engine might get confused and show you a list of Tom Cruise movies or information about unicorn myths, instead of the correct answer: Legend.

2. Google Voice Actions/Google Now

Google is not content to let Apple and Microsoft own the personal assistant and voice-search space on smartphones and tablets. Google Now, a service that appears on Android tablets and smartphones, can be both an assistant and voice-search aid. (To use it, pull down the top menu or press Home and swipe up. Make sure you have the latest Android 4.2.1 update for the latest features.)

Google Now responds to an amazing array of voice instructions. You can set an alarm, dictate an email, and instruct your phone or tablet to provide directions to a destination, just by using a navigate-to-main-street command. As with Bing, you can say a search string and have Google provide the results. Recently Google Now added a few interesting improvements to show where this is all going. You can tell your device to scan a barcode, post a status to Google Plus, and even listen to a song that is playing in the background and tell you the song and artist's name.

"A year ago, Google did not quite understand complex commands. We do a better job of extracting meaning and turning it into an answer," says Steve Cheng, Google product manager for mobile search. "You are not bound by typing only three words. The average query is less than three words. With voice it is easy." If you ask Google Now, "How tall is Barack Obama?" And then ask, "When was he born?" Google will understand the context.

Google is getting more personal and local too. If you ask for the location of Home Depot, Google knows you mean the one in your area. There's even an experimental search that works with voice that looks at your Gmail and knows your preferences. Ask about your next flight or upcoming meetings and the voice search will search your Gmail and find the answers.

3. Samsung Smart TV

Samsung released speech-enabled and gesture-controlled TVs last year, but they did not support natural-language search. Instead, voice control worked more like it does in a car or the Xbox 360 Kinect system, where you can say only specific commands.

In new models like the Samsung F8000 available this year, your TV will finally understand everyday language. For example, when you sit down in the living room, you can say "Show me any new programs starring Johnny Depp," and the TV will search both your DVR and streaming media sources.

Better microphones are one big driving factor in advancing this tech. Samsung says the new models use a unidirectional microphone with a 50-decibel sensitivity. The company also says the 2013 models like the F8000 use an improved algorithm that analyzes what you say and can show your spoken command on the screen as a visual verification that the TV understood you correctly.

4. QNX CAR Application Platform 2.0

QNX is best known for its BlackBerry operating system, including what's on the upcoming BlackBerry 10 devices. Soon major automakers will use the QNX development platform, which supports natural language queries similar to how Apple Siri or Google Voice Commands do. For example, you can say to your car "I'm hungry," and the navigation system will respond by offering a list of nearby eateries. The system will also let you schedule an appointment on a calendar, dictate a message, and do an Internet search by voice.

Kerry Johnson, the automotive product manager, tells PM that the new development platform will also provide some added features to deal with times when people are driving with several passengers. Using multiple microphones in the car, the system will be able to detect who is speaking. If the music is blaring and a passenger in the back wants to control the heat level, the car would recognize those commands.

The speech technology for CAR 2.0 relies on AT&T Watson, a speech recognition engine that can understand natural language commands. QNX and AT&T demonstrated how this would work at CES 2013 with the Bentley Continental GT. The car could understand who was speaking, and from where, using multiple microphones. Mazin Gilbert, the Assistant Vice President of Technical Research for AT&T Labs, says your car will become a virtual assistant. You'll be able to connect to your home to bump up the thermostat, call the boss (just by saying "Call the boss"), and arrange a golf outing using an AI bot that knows your schedule.

5. AT&T Watson

AT&T Watson (no relation to the IBM Jeopardy! champ) can do more than just control the temp in your car. This year, the technology might even be able to translate what you say into another language on the fly. If you are speaking to someone from China, you might be able to say a phrase, wait only a few seconds, and have the person on the other line hear what you say in his or her own language (and vice versa).

Another cool example: The IVEE talking alarm clock, available this May, uses Watson technology to engage in natural conversation. You can say "It's a bit cold, can you raise the temperature?" and the clock will raise the thermostat and tell you the outside temperature; it might even suggest that you wear a coat. You can say "Remind me to take the pizza out of the oven," and the clock will set an audible reminder. IVEE will be able to talk to you about weather conditions, traffic problems, and other events instead of waiting for you to issue a command. In the future the clock might connect to the iRobot vacuum cleaner and other gadgets in your home.

6. Nuance Project Wintermute

Nuance Project Wintermute is a personal assistant that follows you from one gadget to another and even into your home and car. Once the system learns how you speak and what you ask for most often, the assistant will learn and respond more accurately.

Mike Thompson, Nuance vice president and general manager, says the personal assistant of today helps with scheduling and reminders. The next iteration will understand preferences and behaviours, then make suggestions, solve problems, and predict what you might want to do.

At CES 2013, Nuance showed how Wintermute (the name comes from the novel Neuromancer by William Gibson) follows you from one device to another. The contextual awareness is amazing: In your car, you can say "Listen to Coldplay," and the system will play the latest album. When you get home, you can say "Turn music on again," and the same song will resume playing. You might be watching basketball on a tablet during your train ride home; when you get home to your living room, you can say "Put that game back on," and it would resume on your TV.

7. Chevrolet Siri Eyes Free

Owners of the new 2013 Chevy Spark can hold a conversation with their car. But existing owners won't have to wait for a dealer upgrade or a 2014 model to experience the same tech: The MyLink upgrade will come out next month.

With Siri integration on your iPhone, you can make a call to a contact, play a song, switch audio sources, listen to and create a new text message, and create a calendar appointment. All of the typical functions of Siri will also be available as you drive, so you can check a baseball box score or search for a research paper as you drive. The phone dims automatically when you use the Eyes Free feature from the car's console.

“Eyes Free is the term for what Siri allows in the vehicle while it is moving,” says David Szczepanski, lead connected costumer specialist for General Motors, North Central Region. “Our goal is to make you are safe in the vehicle. When you ask a query about, say, what is the largest ocean in the world, it will only show the info on the phone and will tell you you can’t access the info in the car.”

Down the road, you might be able to book a restaurant appointment through the OpenTable app by asking Siri to set it up for you, he says. This opens up a world of new voice commands, such as ordering a pizza or asking about the cheapest place for gas nearby.

[Source: Popular Mechanics. Edited. Some links added.]