Santa Clara, CA – November 4th, 2013 …New Sensory Research Facility to Open in Boulder, CO.
Sensory, Inc., a global leader in speech technology for consumer electronics, today announces a series of moves to provide mobile handset OEMs new technologies to improve the future of the user experience. Sensory is expanding its technology spectrum with investments in machine learning techniques, embedded Large Vocabulary Continuous Speech Recognition (LVCSR), and vision technologies through several strategic hires and the opening of a new research facility in Boulder, CO.
With its TrulyHandsfree™ voice control, Sensory has dominated the mobile handset platforms’ demand to have an ultra low power “always listening” technology for speech recognition, eliminating the need for button presses or touching. Mobile handset customers of this technology include Samsung, Pantech, LG, and Motorola, among others. To date, more than 40 mobile phones have shipped with Sensory’s TrulyHandsfree™, and many leading players are expected to release devices with Sensory technologies such as wake up triggers, command and control, speaker verification, and user defined triggers.
Machine Learning Focus in New Boulder Office
Sensory is expanding its workforce and access to top research institutes and PhDs by opening a new office in Boulder, CO that will specialize in machine learning techniques for spatiotemporal signal processing. As the VP Emerging Technologies, Dr. Bryan Pellom (former VP Speech Development at Rosetta Stone, Research Faculty at University of Colorado at Boulder) will guide the office’s research direction into new technologies, and will leverage relations with Sensory Co-Founder and University of Boulder Computer Science Professor Michael Mozer.
The Next Evolution of Search
In efforts to create the most powerful embedded recognizer available, Sensory is expanding into embedded Large Vocabulary Continuous Speech Recognition (LVCSR), allowing the company to take advantage of state of the art techniques including deep belief networks and weighted finite state transducer search. Dr. Kadri Hacioglu (formerly VP Speech Research at Rosetta Stone, Research Faculty at University of Colorado at Boulder) has been brought on board to lead the efforts as Sensory’s VP Spoken Language Systems. He joins Sensory’s Joe Wolfel (formerly with Talkhouse, MERL, and Dragon Systems) in the large vocabulary efforts based on the Talkhouse speech retrieval technology.
“The new LVCSR engine will enable a more natural interface between people and their products without the constraints of limited vocabulary engines,” said Todd Mozer, CEO and President, Sensory. “What’s more, the LVCSR technology is designed to run accurately on a device without needing to connect to the clouds every time a recognition task is required.”
Sensory plans to showcase its embedded LVCSR technology in early 2014.
Expanding our Senses to Vision
“Since its founding 20 years ago, Sensory has worked to improve the user experience in controlling consumer electronics through the use of Sensory functions. We are very excited to be adding computer vision technology to our portfolio to enable a new level of accuracy and convenience in intuitive user control,” Mozer said.
The vision team, led by vision industry veteran and Stanford PhD Dr. Gordon Haupt, will focus on mobile phone-based vision tasks that combine with speech technologies to improve user identification, speech recognition, demographic identification, and emotion detection.
Sensory’s first offerings combining vision and speech are expected to be released in 2014.
About Sensory, Inc.
Sensory, Inc. is the leader in speech technologies for consumer products, offering a complete line of IC and software-only solutions for speech recognition, speech synthesis, speaker verification, music synthesis and more. Sensory’s products are widely deployed in consumer electronics applications including mobile, automotive, Bluetooth™ devices, toys, and various home electronics.
TrulyHandsfree is a trademark of Sensory, Inc. All other trademarks or registered trademarks are those of their respective holders.
For more information, contact: