How MindMeld Is Letting Companies Add Voice Recognition to Any App The days of talking to all of your devices are closer than you think.
By Grant Davis
In the 2013 movie Her, Joaquin Phoenix plays a man who falls in love with his cloud-based operating system, which has been given Scarlett Johansson's voice. She anticipates his moods, arranges his schedule—all he has to do is talk to her. Director Spike Jonze set the action in the near future, but according to Tim Tuttle, founder and CEO of San Francisco-based Expect Labs, this type of anticipatory computing is already here.
"In the last two years, we've seen voice-recognition and artificial-intelligence research advance and become more accurate by 30 percent," he says. "That's more progress than we made in the last 10 years."
Tuttle—who has been working on the artificial intelligence behind voice-recognition technology since the '90s, first as a student at the Massachusetts Institute of Technology and then at Bell Labs—says Apple's Siri and Google Now have played a sizable part in that advancement. The more people use voice commands to activate apps and searches on their smartphones, the smarter the machine intelligence becomes. Indeed, Tuttle cites data showing that one in three searches on a smartphone are now done through voice.
"Most of the new devices coming out in the next five years won't have keyboards," he says, pointing to smartwatches and other wearable tech, infotainment systems in cars and the new class of appliances launching the Internet of Things.
If Tuttle has his way, Expect Labs' cloud-based voice-recognition and machine-learning software, MindMeld, will be an integral part of this transition. The program allows anyone with a mobile app to plug in voice recognition. "You open the app, press a mic button within the screen and speak your command," he says. Because you've contained your search to the data inside that particular app, your results will be faster and more accurate than if you used Siri or Google Now.
"People will have a better experience within the app, because Siri or Google Now can't access the app's granular and specialized data," Tuttle explains.
"We're at the point where, for American English at least, the accuracy issue has been solved, and speech technology can now better understand what's being said than most people in a room," he adds. "What's left is developing filters to deal with background noise and other voices, regional dialects and less-popular languages."
More than 1,200 companies are using MindMeld to power the voice-recognition feature on their user interfaces. While Tuttle wouldn't provide specifics, he claims that a major cable TV company, global automotive manufacturer and several government intelligence agencies are customers, in addition to mobile-app development shops, which can add it to their clients' apps. MindMeld is free of charge until it processes 1,000 voice queries per month; after that, volume-based monthly fees range from $49 (for up to 10,000 voice searches) to $1,999 (for up to 800,000 voice searches).
Tuttle believes that up to 40 percent of the apps launched for mobile devices will come with voice recognition by the end of this year. Three years from now, he says, voice will be the primary user interface with our devices.
"And we're still just scratching the surface," he adds. "Jarvis, the voice-activated computer used by Tony Stark in the Iron Man movies is only like five to 10 years away."
More tech brilliance
The Asus Chromebit plugs into the HDMI slot on any monitor or TV and (with a Bluetooth-connected keyboard and mouse) turns it into a Chromebook with 2 GB of RAM, 16 GB of storage, a USB port and Wi-Fi. Best of all, you can put your computer in your pocket when you're on the go.
It's the super-smartphone: The Android-powered V2 from Saygus has a wireless HD connection to stream films to a big screen, wireless charging, Harmon Kardon 3D sound, a 21-megapixel camera and 320 GB of storage. And it's waterproof.
Toss Panono's camera-covered ball into the air for a 360-degree panorama you can view from any angle using the company's app.
Sweden's Penclic has designed an ergonomically pleasing mouse that looks, feels and moves like a pen.
The Apple Watch is cool, but it needs an iPhone nearby to work. Atomic from Spur handles the basics—phone, text, email and social media—through Google Play, no smartphone necessary.
Biomeme turns an iPhone into a diagnostics tool that can run quick field tests on blood, urine or fecal matter for a variety of diseases or test DNA at a crime scene. It can even test for Ebola.
Loopwheels has literally reinvented the wheel with carbon-spring integral suspension that reduces vibration and absorbs shock, providing a lighter, smoother ride for bikes and wheelchairs.
Bluetooth technology in ilumi's LED bulbs lets users fine-tune color and brightness, set them to a timer, sync them to music and adjust throughout the day for a natural circadian rhythm.
It's Drone Central Station: Airware's Aerial Information Platform has everything needed to control a fleet of drones from a Windows laptop or tablet and keep them on the right side of evolving FAA regulations.
