Combination of Subtractive Clustering and Radial Basis Function in Speaker Identification

TL;DR

This paper introduces a novel speaker identification method combining subtractive clustering with Radial Basis Function neural networks to improve training speed and accuracy in identifying speakers from a database.

Contribution

It proposes a new approach that integrates subtractive clustering with RBF networks, enhancing training efficiency and prediction accuracy in speaker identification.

Findings

Faster training speed due to subtractive clustering for hidden node selection.

Improved prediction accuracy with regularized RBF network.

Effective identification of speakers from a trained database.

Abstract

Speaker identification is the process of determining which registered speaker provides a given utterance. Speaker identification required to make a claim on the identity of speaker from the Ns trained speaker in its user database. In this study, we propose the combination of clustering algorithm and the classification technique - subtractive and Radial Basis Function (RBF). The proposed technique is chosen because RBF is a simpler network structures and faster learning algorithm. RBF finds the input to output map using the local approximators which will combine the linear of the approximators and cause the linear combiner have few weights. Besides that, RBF neural network model using subtractive clustering algorithm for selecting the hidden node centers, which can achieve faster training speed. In the meantime, the RBF network was trained with a regularization term so as to minimize the variances of the nodes in the hidden layer and perform more accu-rate prediction.