Sensory Extends Voice and Visual AI Platform to the Cloud
Voice and vision AI tech firm Sensory has introduced a new cloud-based platform for companies seeking its biometric and speech recognition tools developed. The beta version SensoryCloud.ai released this week is a notable departure from the company’s history of on-edge processing, without the use of cloud servers.
SensoryCloud offers a handful of voice and visual AI services as APIs or SDKs. Customers can add speech-to-text, sound identification, wake word confirmation, and biometric verification by face or voice, with others to follow. The solutions are not enormously different from what Sensory has provided before, except for where and how the data is processed.
Sensory is known for its VoiceHub platform where clients could create custom wakewords and voice commands to install on smart devices. Sensory’s TrulyHandsfree technology powers the devices and software built by those choosing to avoid Amazon Alexa or Google Assistant in favor of a unique option of their own. There has been plenty of demand for Sensory’s tech and the company has pushed out new features and capabilities to entice even more over the last couple of years. Zoom tapped Sensory in July to help shift its voice command system away from the cloud. Sensory further widened VoiceHub’s abilities by adding models designed to understand children in October. Sensory even developed voice and facial biometric tools for those with a cold and wearing a mask early on in the COVID-19 pandemic.
The cloud platform doesn’t have some of the latency and privacy benefits of on-edge processing, but it offers other compensations that Sensory’s clients may prefer. By running the data processing in the cloud, the clients have far more control over how to deploy its AI features and can access and analyze the resulting data to learn what customers want or discover gaps in its services. The API system is also more familiar to many programmers. And Sensory claims it doesn’t diminish the capabilities of its software. The speech-to-text program claims a word error rate under 5% and the wake word verification wipes out up to 90% of false awakenings compared to rival options.
“We have a history of building fast and accurate AI models, and we paired this capability with some of the brightest and freshest minds in the cloud industry,” Sensory CEO Todd Mozer said. “The result is a hybrid cloud platform that uses state-of-the-art AI to address customers unique needs for control, flexibility, cost, accuracy, reliability, features, latency, and privacy.”