With eyes busy on the road and hands busy on the wheel, speech recognition is a convenient alternative for controlling vehicle functions or accessing information while driving. The Infotainment Applications team is working on the next generation of speech recognition systems, seeking to understand not only what is said but also under what context it is said and by whom. We are building scalable solutions that can improve their recognition speed, accuracy and vocabulary throughout the vehicle’s lifetime.
By drawing acoustic cues picked up by the vehicle’s cockpit microphones and harnessing distributed processing architectures, our research is pushing towards simultaneous, conversational recognition of multiple speakers under a variety of vehicle noise environments. The team’s research is a joint effort with our Human Machine Interface colleagues at the ERL and the VW Group worldwide to ensure an intuitive speech dialog design that is adaptable to the varying cognitive workloads experienced while driving.