Updated: The latest version of Sensory’s TrulyHandsfree(TM) technology marks some radical leaps forward for embedded speech processing resources and secure device activation. The latest smartphones and mobile devices have raised user expectations surrounding how they can use their voices to take command and control of popular applications and functions. In June 2009, we described the first generation of Sensory’s hands-free technology and how it used “word-spotting” embedded in device firmware to let individuals users say pre-defined “trigger words” to initiate interactions with mobile phones or in-car devices.

During the ensuing years, the company refined the software and re-defined implementation scenarios. By July 2011, it introduced the next generation of TrulyHandsfree making it possible for its partners and customers tointroduce products and services that were more flexible (allowing a multiplicity of trigger words), robust (able to detect speech in noisy environments) and understanding (able to distinguish the meaning of phrases in order to carry out instructions in a way that feels natural).

TrulyHandsfree 3.0 builds on Sensory’s developments to date and more elegantly integrates the speaker identification and verification capabilities it introduced last May. Version 3.0 also allows device owners to combine the trigger phrase with search instructions so that a user can use a single utterance to wake up the device and carry out an order. The results are illustrated in this video:

I know this is a video demo, but I am especially impressed by the ability of the device to detect a specific individual’s voice and keywords in a moving car (which introduces wind and tire noise) with the radio on. The additional ability to distinguish between the voices multiple individuals who may be using a device should get a number of developers’ creative juices flowing. This is the basis for extreme personalization of consumer electronics at home, in a car or in your pocket.

Sensory’s Todd Mozer will be a featured speaker at Opus Research’s upcoming Voice Biometrics Conference in San Francisco May 8-9. He will discuss the advantages of integrating trigger words and speaker identification on consumer electronic devices. As rumors fly surrounding ways to activate smartphones or other mobile devices with fingerprint readers or front-facing cameras, the idea of secure, personalized access that requires “no touching” is extremely attractive.

Sensory has announced several go-to-market partnerships to ensure broad deployment of TrulyHandsfree 3.0. Wolfson Microelectronics is teaming with Sensory to incorporate ultra low-power speech detection and activation for its mobile audio processing platform. Santa Clara-based HiFi digital signal processing specialist Tensilica announced its partnership with Sensory.

Rounding out the troika of DSP partners licensing Truly Handsfree 3.0 is CEVA, whose IP was shipped in more than 1 billion devices, including smartphones from HTC, Huawei, Lenovo, LG, Nokia, Motorola, Samsung, Sony, TCL and ZTE.

Categories: Articles

Tags: Voice biometrics, mobile speech, speaker identification, Consumer Electronics, Sensory