October 19, 2016 -
VocalZoom, a supplier of Human-to-Machine Communication (HMC) optical sensors, has completed the final stage of design work with the iFLYTEK speech recognition platform.
iFLYTEK is preparing to launch a headset product by the end of the year featuring the VocalZoom HMC solution.
Based on recent tests of iFLYTEK’s pre-production headset product, the VocalZoom HMC solution can deliver near-perfect speech recognition in noisy voice-control environments, as well as deliver an even better performance than in early iFLYTEK lab tests.
“Our upcoming headset featuring the VocalZoom sensor will redefine ASR performance in its product class and open up new applications in noisy environments where voice control products previously didn’t work,” said Haikun Wang, senior researcher at iFLYTEK. “We also have entered the development stage with VocalZoom on another product for voice control in the connected car using a forthcoming longer-range version of their sensor.”
Early tests of the VocalZoom sensor with iFLYTEK’s voice cloud intelligent speech technology platform showed performance improvements of at least 50 percent over traditional speech-recognition technology in a quiet environment.
The sensor performed even better in noisy environments such as outdoors with street sounds, and in vehicles with windows open, engine noise, music playing and other interference.
Testing of iFLYTEK’s pre-production headset featuring VocalZoom’s sensor are delivering even greater improvements than those early results in quiet environments, while continuing to deliver near-perfect word recognition performance in noisy environments.
This compares with traditional alternative solutions using one, two or an array of acoustic microphones with noise reduction technology, which cannot provide clean enough, isolated speaker input to operate satisfactorily in noisy environments.
VocalZoom’s compact, low-profile optical sensor solution improves voice recognition and voice biometric, initially in headsets and wearables, and in the near future for the connected car and other longer-range sensor applications.
By integrating the VocalZoom optical HMC sensor into a voice-control solution with a focus on facial skin around the cheeks, ears, neck and throat, tiny vibrations during speech can be collected, measured and converted to an isolated, near-perfect reference signal that machines can decipher – regardless of noise levels.
“We believe that iFLYTEK is poised to revolutionize how users communicate with devices, applications and machines, leveraging the breakthrough capabilities of our optical HMC sensors,” said Rammy Bahalul, vice president of sales and business development for VocalZoom. “We also look forward to working with iFLYTEK on additional projects including a reference design that will simplify VocalZoom sensor design-in for product developers using the iFLYTEK ASR, and iFLYTEK’s next initiative for voice control and voice biometric solutions in the connected car.”