The acceptance of speech as a primary user-interface in vehicles depends on how well speech recognition can overcome challenging conditions including high levels of noise, echo and competing speech, in which accuracy is known to degrade. To mitigate this, an acoustic front-end using the QNX Acoustics for Voice software library preprocesses multichannel microphone data from the vehicle and provides a cleaned signal to the recognizer. We demonstrate how three components of the front-end: beamforming, acoustic echo cancellation and zone interference cancellation, lead to significant improvements in word error rates.
(See document for exact affiliation information.)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
