Classification of Room Impulse Responses using Kohonen Neural Network

Abstract

In this paper a method for classifying acoustic room impulse responses is analyzed. Impulse responses represent basic source of information in room acoustics. The algorithm performs clustering using Kohonen neural network. The calculated multifractal parameters represent the input data in the Kohonen neural network. Clusters of impulse responses with similar acoustic characteristics are found. The experimental results verify the usability of the proposed algorithm.