In an advanced dialogue scenario it can be required that an Automatic Speech Recognition (ASR) system, working in the car environment, is able to react immediately to a driver`s command, while the car-audio is active or while a synthetic prompt is being reproduced (barge-in). In particular the aim of the work presented in this paper is to cancel the interfering car radio signal, addressing the well known problem of acoustic echo cancellation. To solve this problem, one can use the Normalized Least Mean Squared Algorithm (NLMS), which has some important advantages for the reference application, like robustness against disturbances and computational simplicity. The echo canceller shows good properties as acceptable adaptation speed and negligible speech distortion: in the best configuration the WER was lowered from 18.3% to 4.1%
In-car audio compensation based on NLMS for hands-free speech recognition
Matassoni, Marco;Omologo, Maurizio;Zieger, Christian
2004-01-01
Abstract
In an advanced dialogue scenario it can be required that an Automatic Speech Recognition (ASR) system, working in the car environment, is able to react immediately to a driver`s command, while the car-audio is active or while a synthetic prompt is being reproduced (barge-in). In particular the aim of the work presented in this paper is to cancel the interfering car radio signal, addressing the well known problem of acoustic echo cancellation. To solve this problem, one can use the Normalized Least Mean Squared Algorithm (NLMS), which has some important advantages for the reference application, like robustness against disturbances and computational simplicity. The echo canceller shows good properties as acceptable adaptation speed and negligible speech distortion: in the best configuration the WER was lowered from 18.3% to 4.1%I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
