A New Strategy of Direct Access for Speaker Identification System Based on Classification

Abstract

In this paper, we present a new direct access strategy for speaker identification system. DAMClass is a method for direct access strategy that speeds up the identification process without decreasing the identification rate drastically. This proposed method uses speaker classification strategy based on human voice’s original characteristics, such as pitch, flatness, brightness, and roll off. DAMClass decomposes available dataset into smaller sub-datasets in the form of classes or buckets based on the similarity of speaker’s original characteristics. DAMClass builds speaker dataset index based on range-based indexing of direct access facility and uses Nearest Neighbor Search, Range-Based Searching, and Multiclass-SVM Mapping as its access method. Experiments show that the direct access strategy with Multiclass-SVM algorithm outperforms the indexing accuracy of Range-Based Indexing and Nearest Neighbor for one to nine percent. DAMClass is shown to speed up the identification process 16 times faster than sequential access method with 91.05% indexing accuracy.