NXP Semiconductors and Rubidium, a supplier of embedded speech processing technologies, have announced the integration of Rubidium”s multi-lingual Automatic Speech Recognition (ASR) and always-listening Voice Trigger engine into NXP’s CoolFlux DSP, an ultra-low power (ULP) licensable DSP core that has been optimised for digital audio processing.
The integration of Rubidium’s technology into the CoolFlux DSP provides NXP licensees with access to hands free voice trigger and later also speech recognition capability with beam forming, barge-in capability and a host of other voice user interface (VUI) functionalities.
“We are pleased to offer our CoolFlux customers Rubidium”s market proven technology,” says Johan Van Ginderdeuren, Director CoolFlux Licensing at NXP. “As an example, with only 0.5 MIPs in the quiet listening mode and 4 MIPs in the active trigger word recognition mode, we expect Rubidium’s Voice Trigger to typically consume only 28 microwatts on the CoolFlux DSP subsystem using an ULP 90nm library. This frees up resources for other activities on the DSP and minimises power consumption while being always on. In addition, the small memory footprint provides for extended Rubidium functionalities such as biometric speaker verification.”
“Voice trigger and deeply embedded speech recognition have become a highly-requested feature in many consumer products such as mobile phones, hands free car-kits, wearable devices, automotive multimedia and infotainment and smart TVs,” explained Shlomo Peller, CEO and Founder of Rubidium. “Rubidium’s ULP micro-embedded voice trigger and speech recognition draws considerably less power than other embedded product offerings, hence is a perfect fit for the NXP’s ULP CoolFlux DSP combining best of both low power worlds. One of the immediate benefits to CoolFlux DSP”s developer base is the availability of a truly hands free voice user interface combined with low power consumption and very small memory profile.”