An image based approach for speech perception

Classification of speech signal is one of the most vital problems in speech perception and spoken word recognition. Although, there have been many studies on the classification of speech signals but the results are still limited. In this paper, we propose an image based approach for speech signal classification based on the combination of Local Naïve Bayes Nearest Neighbor (LNBNN) and Scale-invariant Feature Transform (SIFT) features. The proposed approach allows training feature vectors to have different sizes and no re-training is needed for additional training data after training phase. With this approach, achieved classification results are very satisfactory. They are 72.8%, 100% and 95.0% on the ISOLET, Digits and Places databases, respectively.