Word recognition through mapping of lip movements from speech utterance using audiovisual fusion and MLP

Speech has information more than text, but under noisy environment speech sufferance from disadvantage of not properly decoded by humans and same is true with machines. being bimodal along audio features if we augment visual specifically related to lip movements. the degree recognition can be improved. The objective this work use aid word recognition. In extracted MFCC for Geometrical movements together used in machine learning algorithm predict utterances. Videos utterances are TIMID database. With statistical corresponding form input feature vector (Multi-layer perceptron). experimental results show that using MLP have obtained a accuracy 91% KNN Classifier attained 61%. presented here important implications applications HMI communication helps hearing impaired.