Voice Assistants Going Embedded

By Todd Mozer



January 13, 2020


Everyone knows that the use of voice assistants is exploding. Just returning from CES, it was amazing to see the number of booths that support Amazon?s Alexa and Google?s Assistant.

Everyone knows that the use of voice assistants is exploding. Just returning from CES, it was amazing to see the number of booths that support Amazon’s Alexa and Google’s Assistant. Nevertheless, recent research from Edison-NPR is showing that US household adoption of voice assistant applications is slowing down.

The survey reports that 157 million smart speakers are owned by U.S. households which reflects 135% growth over the past two years. However, that expansion is increasingly being driven by existing smart speaker households adding devices as opposed to new adoption.  I represent the household guy buying multiple device. I have  Google Home, Echo, Echo Show,  Sonos One, and an Apple HomePod. I gave away 6 Echo Shows as Christmas gifts. Admittedly, I’m probably an exception.

There are clearly a lot of other households that are not buying these devices. A variety of studies have all supported that 20-60% of households have serious concerns about privacy and “always listening” devices that send conversations to the cloud. Last year’s headlines were filled with stories of contractors that got private audio conversations from bedrooms. It’s not surprising that the household growth has slowed so substantially.

But it could get revived through a couple of new initiatives which will have a big effect on embedded voice assistants . With technology now available from Sensory, Microsoft, Amazon, and likely from Google, users will have a variety of options for domain specific embedded assistants that allow natural language communications from home appliances, without sending data to the cloud.  For CES Sensory introduced a video of the first such domain specific assistant, and gave live demos on the show floor.

The initiatives I mentioned are largely flying below the radar but are important for embedded voice assistants. Amazon has announced a Voice Interoperability Initiative, and the concept is that devices can have multiple voice assistants residing on them. I believe this will spur a number of devices to not just choose between cloud providers but to also have embedded assistants that control the device and support the manufacturer’s brand. For example, in the video above the phrase “Voice Genie” is used to activate the microwave this is a magical marketing moment for the brand manufacturer to have the consumer saying the brand name rather than Alexa, Siri, or some third party name. It also is an approach to give the consumer more direct device control by voice without having to go through awkward command sequences or privacy risks.

The other initiative comes from the Zigbee Alliance and has all the major assistant providers standardizing around ways to connect and communicate with home devices. With the emergence of such standards will come embedded voice assistants that control your home without ever having to risk your privacy by sending recordings to the cloud.  The past few years have been very interesting and dynamic for the emergence of voice assistants and AI in the home, and I expect the next few years will bring more interesting developments including embedded domain specific voice assistants.

Todd Mozer is the CEO of Sensory. He holds over a dozen patents in speech technology and has been involved in previous startups that reached IPO or were acquired by public companies. Todd holds an MBA from Stanford University, and has technical experience in machine learning, semiconductors, speech recognition, computer vision, and embedded software.

More from Todd