Sensory Adds Multiple Wake Word and Voice Assistant Support
Speech recognition and natural language understanding technology developer Sensory has launched a new feature for its TrulyHandsfree wake word engine. Devices using the engine will be able to understand more than one wake word command from one brand so that a smart speaker could theoretically host more than one voice assistant at a time, for example.
Multiple Assistant Access
Once in place, Sensory’s new feature would allow a voice-enabled device to respond to standard and custom wake words, even allowing for the possibility of user-created terms. The broader range of terms Sensory’s tech can respond to is designed to fit into the Voice Interoperability Initiative (VII) that Amazon and other companies signed in September in an effort to create a platform flexible enough for more than one voice assistant.
Sensory has already embedded more than two dozen wake words into its models, which are used in a variety of Alexa, Google Assistant, Siri, Bixby, and other voice assistants. What’s changed is that TrulyHandsfree can differentiate among the potential wake words and ensure that the right one is activated. That’s usually been a source of compounding errors in speech recognition, but Sensory claims its software overcomes that problem. It may be partly due to Sensory’s acquisition of Vocalize and its testing technology back in February. The relatively low power cost that Sensory’s technology requires also makes the concept more feasible for the devices where TrulyHandsFree operates.
“We are lucky to have worked closely with most of the leaders in voice services to help create products with their wake words embedded,” Sensory CEO Todd Mozer said in a statement. “Because of this, we already have all of the data in place to combine multiple models. However, everyone in the industry has found that combining two or more wake word models into a product poses a significant challenge in maintaining optimal performance. The new capabilities we added to TrulyHandsfree overcomes some of the accuracy challenges without significantly boosting MIPS and memory requirements.”
The VII is still fairly new, but Sensory’s new offering is a good step to its stated goal of giving customers “choice and flexibility through multiple, interoperable voice services.” Amazon has already begun to do so, with the Facebook Portal and products by Orange and Deutsche Telekom. While smart speakers made by Sonos and other brands have access to multiple voice assistants, only one can be accessed at a time, requiring a change in the settings to switch between Alexa and Google Assistant, for instance. A device that can understand multiple wake words speeds up that switch considerably
The original VII signers didn’t include most of the major voice assistant companies like Google, Apple, Samsung, and Deutsche Telekom. Sensory is betting that that will change and those companies will get on board. The models for their wake words are already in place. Sensory’s new ability could theoretically be used in earbuds, appliances, and televisions, among other devices. With the lower energy and memory requirements the company claims, it would be relatively easy for future hardware to support multiple voice assistants, with users picking whichever is their favorite or has the voice app they wish to access at that moment.