2 Comments

Summary:

Nuance plans to integrate its growing number of speech command platforms into a universal assistant for the internet of things. It’s also trying to make its speech technology more dynamic, using crowdsourcing to capture the language zeitgeist.

Nuance Communications is sticking its voice recognition technology into a lot of products, from its own smartphone and PC personal assistants to connected TV and cars. Though all of those devices share the same natural language understanding (NLU) technology, none of them are interconnected. You’re car doesn’t know what you’re telling your TV, which is equally unaware of the movie search you just performed on your mobile phone.

Now Nuance, however, plans to break down those barriers between its separate platforms, and focus on what it’s calling “intelligent systems.” The idea is to create a personal assistant that will persist beyond the interface, that can be accessed anywhere from the cloud, said Matt Revis, VP and GM of Nuance’s handset division. “We want to unify the personal assistance experience across form factors,” Revis told GigaOM.

As part of that intelligent systems effort, Nuance is trying to delve deeper into language in addition to broadening spoken language’s scope in a technology. Its NLU technology already understands context and idiom to a certain degree. But Revis said it is launching an initiative called Living Language that will allow its servers to dynamically update and speech databases to reflect changing language patterns as they occur.

New Swype screenshot feature“What you get is a very rapidly evolving language model, one that almost moves ahead of the zeitgeist,” Revis said. Living Language will get its first test not on Nuance’s voice platforms, but in its evolving predictive-text interface Swype. Nuance will use crowdsourcing to start analyzing new words Swype users type into their texts and emails. As new words and phrases begin trending, Swype will update its lexicon and make that new vocabulary available to its Swype apps.

Nuance launched the new crowdsourcing capabilities at CES 2013 on Monday, making them available in the latest version of the Swype Android beta (users will need to opt in to send data as well as receive dictionary updates). But Revis said the company plans to expand those capabilities to voice technologies. “We’re just starting with the keyboard,” Revis said.

As for the rest of Nuance’s intelligent systems work, Revis said we’ll have to wait to see specific products, but as the year goes on we should start seeing closer integration of its various consumer-facing and white-label voice command platforms: Dragon TV, Dragon Go, Dragon Drive, Dragon Dictation and Dragon Mobile Assistant.

The building blocks are already in place for that integration, Revis said. Most of Nuance’s products rely on a hybrid approach, using software on the device for simple voice commands, but reaching out into the cloud for more complex speech recognition features. The key will be moving the assistant itself into the cloud to sit alongside the enabling technology, knocking down the walls separating its now-distinct voice assistants.

A common voice for a common platform

Each interface will remain specialized to a certain degree. Your TV doesn’t need to understand the command “turn down the AC,” for example, but there will be a lot of crossover between platforms. For instance, we could use our TVs to update our social networks, and tell our phones to tell our TVs to record programs while we’re away.

voice recognitionNuance also plans to do a lot of cosmetic work, creating unified themes for the app’s graphical user interfaces and a common set of voices and personalities for the assistants themselves. It’s imperative to create a common user experience to get consumers to embrace the idea of a universal voice assistant, Revis said.

Nuance will offer its universal assistant across its own Dragon products, but its primary business is licensing technology to hardware vendors – this will be no exception. Manufacturers like Apple, Samsung and LG sell across the consumer electronics landscape, making it possible for Nuance to bridge their TV and smartphone lines. Nuance also hopes to cross-license technology between vendors, so, for instance, a Samsung could partner with a Ford to ensure their TVs and cars use a common voice interface, Revis said.

As Nuance delves into more areas like the connected home and internet of things, there will be more opportunities to connect and interlink more devices. That’s where the potential of universal assistant could really get big. A talking washing machine or refrigerator might be useful, but a washing machine you could talk to through your smartphone would be even more useful.

Update: At CES, Nuance revealed two new customers for its Dragon Drive technology: Chrysler and ZTE. Chrysler will tap into Dragon Drive Messaging, speech recognition software that will allow driver to compose, send and listen to text messages using a voice interface. Chrysler is  implementing the service in its new Sprint-powered UConnect connected car platform, starting with the 2013 RAM 1500 and the 2013 STR Viper.

ZTE isn’t building a connected car platform so much as its developing a hands-free system. ZTE will begin embedding an app in its forthcoming Android phones called Car Mode, which allows users to control of variety of the phones functions — such as launching navigation apps, playing music and listening to dictated text messages — through simple voice prompts.

Update 2: The Next Web is reporting that Nuance has a code name for its cloud assistant, Project Wintermute.

Featured image courtesy of Shutterstock user brem stocker; Screaming image courtesy Flickr user Lazurite.

You’re subscribed! If you like, you can update your settings

  1. Reblogged this on Simon Hamer and commented:
    Voice assistant technology so those of us with big thumbs and clumsy typing can function more easily in the technological new world.

    1. Here here! Speaking as man with big fat sausage fingers…

Comments have been disabled for this post