TrulyHandsfree 4.0 Uses Deep Learning to Filter Noise
Sensory, Inc. has launched a new version of its TrulyHandsfree speech recognition system. In addition to a number of other improvements, TrulyHandsfree 4.0 boasts of deep learning technology that allows it to be as much as 80 percent more accurate than the previous iteration of the system.
The system is designed to recognize speech for user command and voice interface purposes, and according to the company TrulyHandsfree has become the most widely used keyword identification system in the industry, with over a billion products using it. This latest system employs a neural network engine enabling deep learning processes for acoustic models, which means the system can learn to identify and filter out noise as it monitors user speech. It also features improved algorithms for harsh and reverb-heavy environments, improved scalability, and full compatibility with Sensory’s TrulyNatural, the company’s device-based natural speech recognition system.
In a statement, Sensory, Inc. CEO Todd Mozer pointed to the company’s pioneering record with previous iterations of TrulyHandsfree, calling it “an industry-first that literally changed the way people interact with devices of all kinds.” He added that the new iteration “takes performance to a whole new level with an accuracy, footprint and power consumption that others just can’t touch.” As more devices in the growing Internet of Things ecosystem implement voice-based user interfaces, Sensory, Inc. could indeed see many more integrations of its latest technology going forward.
August 7, 2015 – by Alex Perala
Via: Google Alert for Deep Learning