Acoustic Estimation of the Head Orientation for In-Car Communication Systems
* Presenting author
In order to overcome the communication difficulties among passengers due to a large amount of background noise in cars driven at high or even moderate speed, so-called In-car-communication (ICC) systems are recently used. Such ICC-systems improve the signal-to-noise ratio (SNR) within the car compartment by record, process, and playback the desired speech signal of the talking passenger over loudspeakers located close to the passengers. However, due to the acoustic directionality of human head, ICC-systems often record a distorted speech signal with degraded intelligibility and quality when the talking passenger turns his/her head. This contribution proposes an acoustic localization and head orientation estimation of the speaking passenger within a noisy car compartment. Estimating position and head orientation measures in an automotive environment is not trivial task especially with low SNR scenarios when the desired speech is buried in different background noises (engine, wind, tire noise, etc.) and with the presence of other speaking passengers within the car compartment. Those estimated measures would permit the improvement of speech technologies of the ICC-systems such as better microphone signal selection, gain control and equalization. Also, it would allow to decide who is the active speaker among passengers or determining who is talking to whom.