CEVA, Inc. and Sensory, Inc. have partnered to offer an advanced voice recognition solution based on CEVA-TeakLite-4 DSP and Sensory’s TrulyHandsfree always-on voice activation technology. The solution includes Sensory’s Low Power Speech Detector technology, which dramatically reduces power consumption for always-listening devices, and marks the first time this technology has been implemented on a DSP core, according to the companies. Sensory’s voice activation technology consumes less than 60uW on the DSP when implemented in a 28nm process.
Sensory's TrulyHandsfree Voice Control technology offers multiple phrase technology that recognizes, analyzes and responds to dozens of keywords. It consistently recognizes phrases even when embedded in sentences in a noisy environment. The fully programmable CEVA-TeakLite-4 DSP provides processing power to implement such advanced algorithms, while leaving plenty of headroom to support other audio/voice related functions in software.
The CEVA-TeakLite-4 is offered as an architecture framework, composed of four DSP cores, the TL410/TL411/TL420/TL421, all having a 32-bit native data width required for advanced pre and post-processing voice/audio applications. By taking advantage of a unified development infrastructure composed of code-compatible cores, a set of optimized software libraries and a unified tool chain, customers can significantly reduce software development costs while leveraging their software investment in future products. For more information, visit http://www.ceva-dsp.com/CEVA-TeakLite-4.html.
The companies will demonstrate Sensory’s TrulyHandsfree Voice Control running on CEVA silicon at the upcoming Mobile World Congress in CEVA’s booth (Stand 7i70) .