Abstract
The use of visual features to help acoustic speech recognition (ASR) is an appropriate tool to enhance ASR. In this paper, we propose a novel system integrates face detection, user identification and visual speech recognition. Here we use the self organizing map to achieve visual features extraction. Then, the extracted features are recognized using K-nearest neighbor classifier. Experimental results, using a database includes Arabic digits, show that the proposed system is promising and effectively comparable with other reported systems.