By utilizing Sensory’s Low Power Speech Detection technology, Tensilica was able to further reduce power to an average of under 25 µW when implemented in a 28 nm HPL process, the lowest power implementation by far in the industry. This makes this joint solution ideal for always-on voice activated smartphones, tablets, appliances and automotive applications.
“The Low Power Speech Detection capability significantly extends battery life,” stated Bernard Brafman, vice president of business development at Sensory. “When the HiFi DSP detects speech, the full TrulyHandsfree 3.0 voice trigger features, including speaker verification passphrase, user-defined speaker verification and user-defined speaker identification, will be enabled.”
“Using Sensory’s TrulyHandsfree technology, we can offer our customers the lowest power total solution, with power ranging from below an average of 25 µW in basic detection mode,” said Larry Przywara, Tensilica’s senior director of mobile multimedia. “With this level of efficiency, Sensory’s voice trigger running on our HiFi DSP is the smallest part of the power budget for the entire voice signal chain.”
Sensory’s TrulyHandsfree 3.0 technology enables personalization of the user experience by calling up a specific user’s history, preferences and settings. Sensory’s speaker verification technology uses voice biometrics and analysis to authenticate a specific individual. Sensory’s TrulyHandsfree complete set of voice activation and recognition products allow users to access a device with just their voice, bypassing the typical 4-digit number manually entered on the screen for access.
Tensilica’s HiFi DSP family is the most widely used licensable audio/voice DSP family, with support for over 100 proven audio/voice software packages. More than 50 companies have licensed Tensilica’s HiFi DSP family, and they’ve shipped over 200 million HiFi DSP cores in their smartphones, tablets, computers, digital televisions, home entertainment systems and other devices. Tensilica’s HiFi Audio/Voice DSP family includes four cores. The recently introduced HiFi Mini is an ultra-low-power DSP optimized for always-on voice recognition. HiFi 2 and HiFi EP are the mainstream products for efficient audio