Comparison of Distance Metrics for Phoneme Classification based on Deep Neural Network Features and Weighted k-NN Classifier

K-nearest neighbor (k-NN) classification is a powerful and simple method for classification. k-NN classifiers approximate a Bayesian classifier for a large number of data samples. The accuracy of k-NN classifier relies on the distance metric used for calculating nearest neighbor and features used for instances in training and testing data. In this paper we use deep neural networks (DNNs) as a feature extractor to learn discriminative internal structure of the data. We compared different distance metrics for calculating nearest neighbor in our speaker similarity score algorithm for phoneme classification and their execution time. The city block distance metric is computationally efficient and provides good phoneme classification accuracy.