Speech recognition as a distraction-mitigation solution for automotive features has emerged as the classic love-hate situation for new technology. Speech-recognition’s promise is undeniable—but its to-now-frustrating execution has been equally undeniable.
At the 2017 CES (formerly the Consumer Electronics Show), Honda said it plans to do something about it, announcing a collaboration—through its startup company-advancing Xcelerator program—to develop for automotive use the “optical microphone” of Isreal-based VocalZoom to markedly enhance the accuracy of speech recognition. Honda said VocalZoom’s optical sensor can deliver a “near-perfect reference signal that automotive voice-control systems can understand and quickly respond to, regardless of noise levels. The result is clean, isolated driver commands that are significantly easier for automotive voice-recognition systems to understand and obey than was previously possible with traditional voice-control solutions.”
The VocalZoom module incorporates a lens, laser and application-specific integrated circuit (ASIC) chip, using the laser to measure and tiny vibrations in the throat and face when speaking, greatly augmenting the system’s accompanying acoustic microphone signal by providing “an isolated, near-perfect reference signal that automotive voice control systems can understand and quickly respond to, regardless of noise levels,” said Honda in a release. The company said testing has shown at least a 50% improvement over standard acoustic voice recognition in a quiet vehicular environment and better results in noisy environments.
Eitan David, VocalZoom vice-president of products, told Automotive Engineering at CES 2017 that although the optical microphone componentry could be incorporated into a vehicle’s existing camera system, the VocalZoom technology does require its own sensor. Ideally, the VocalZoom sensor would be placed in the rearview mirror, dashboard or headliner to enable a clear line of sight to the driver’s face.
David said the company’s optical microphone already is used in aviation and industrial headsets and wearable applications and that VocalZoom now is working with Honda and speech-recognition specialist Nuance to adapt the system for automotive use. He said it is expected to be ready for production-vehicle application in roughly 12-16 months. Honda is keen to advance the accuracy of voice recognition as part of its new Cooperative Mobility Ecosystem with which Honda hopes to ultimately eliminate traffic fatalities.
Meanwhile, David also said in addition to voice control for navigation, phone and other driver-assistance features, VocalZoom is expected to enable a range of biometric features that include heart-rate monitoring (again by measuring minute pulsations in the facial skin) and voice- or facial-recognition security functions such as confirmation of identity for mobile-sourced transactions.
VocalZoom also is working with automotive supplier Magna to incorporate optical-sensor technology into Magna’s rearview-mirror product line.