Speech recognition using the metric defined by integra-normalizer
Sung-Soo Kim, Dae-Jong Lee, Keun-Chang Kwark, Ju‐Sik Kim, Jeong-Woong Ryu, Sanghyuk Lee · 2002
This paper represents a new method of recognizing speech using the metric defined by integra-normalizer (IN). A neuro-fuzzy method is also demonstrated as a comparison to the proposed method. A codebook is constructed with a set of feature vectors extracted from the raw speech data. There are various schemes for measuring the distance between a set of information. In this paper, the distance between feature vectors is obtained by using the new metric defined by IN. The metric by IN possesses an advantage to the conventional metrics such as the metric defined by the least square error in L/sup 2/ or in l/sup 2/ spaces. With the approach proposed, the information on the patterns of the speech features is mapped to the feature vectors and the metric measures the difference between speech patterns considering the shape of patterns. The results of the computer simulation are shown for the validity of this proposed method.