The Internet of Things: A New Evolution for Speech Technology

“The answer,” she says, “is speech,” but she agrees that the speech-interface landscape for the Internet of Things is still largely proprietary, and that has limited the market for IoT products. Currently, the majority of speech-enabled IoT devices operate in one of four ecosystems: Amazon, Apple, Microsoft, and Google.

Ahmed Bouzid, former product head for the connected home project at Amazon and current CEO of Witlingo, notes that Amazon continues to be a leader in controlling the Internet of Things in the home. Part of this success comes from Amazon’s relatively low investment in proprietary hardware.

According to Consumer Intelligence Research Partners, Amazon’s IoT user interface and hub, the Echo, had sold three million units as of this April, after only a year on the market, with an expansion of the brand into a second device, the Echo Dot. The majority of Echo owners use the device for information, taking advantage of the Echo’s speech-recognizing intelligent assistant, Alexa. This is Amazon’s premier IoT hardware node, whose market penetration comes from its less-proprietary approach to the IoT. In June of last year, Amazon debuted the Alexa Skills Kit, which gave developers and device makers the ability to make their products compatible with the Amazon voice-controlled intelligent assistant. As a result, the Alexa Smart Home ecosystem has grown to include IoT devices ranging from Samsung’s SmartThings and Phillips’s Hue line to offerings from Wemo, Insteo, and Wink, altogether comprising hundreds of home fixtures that can be controlled directly by Alexa.

This open-source approach to market penetration has caught the attention of more proprietary companies looking to market their solutions as IoT user interfaces (UIs). Apple, traditionally very closed to third-party hardware as a platform for its software, has uncharacteristically opened itself to IoT hardware partnerships. Apple iOS 10, released in September, includes a new, baked-in Home app designed to manage certified Apple HomeKit-enabled products for the Internet of Things. This update is included on all of Apple’s popular internet-enabled stable of iOS devices (the iPad Pro, Air, and Minis 2 through 4; the ubiquitous iPhone, models 5 through 7) as well as Apple’s latest wearable technology, the Apple Watch Series 2, which does not utilize iOS. Also jumping into the IoT is Google, which this year announced the Google Home, with an expected suite of features that closely mirrors the Echo, and with it a speech-enabled intelligent assistant—aptly named Assistant—as its controlling interface.

Going Further with Speech

The hope, in all likelihood, is that Google Home will be able to provide a superior type of intelligent assistant under the hood of its voice UI to set it apart from the Echo and Apple Home. Both Bouzid and Victor Melfi, CSO and senior vice president of marketing at VoiceBox Technologies, agree that the key to speech’s future is a better user experience.

“Context is key,” says Melfi, whose speech solutions at VoiceBox Technologies supply intelligent assistance and voice command to the portion of the market that is wary of partnership with the proprietary leaders. “Most intelligent assistants have accurate speech recognition and are good with single commands, but have a lot of difficulty tracking the context of utterances through a conversation,” Melfi says.

Google has promised to provide a context-sensitive speech UI similar to the kind that Melfi’s VoiceBox Technologies currently offers. VoiceBox touts “context management,” which recognizes pronouns and attributes them to their antecedents, considers the context of homophones and proper nouns, and can track multiple lines of inquiry in a single conversation with the UI. No doubt Google will leverage its vast experience in natural language understanding stemming from its flagship search engine, along with a massive data set on which to perform machine learning to train Assistant.