Phone clustering using the bhattacharyya distance

Abstract

In this paper we study using the classication-based Bhattacharyya distance measure to guide biphone clustering. The Bhattacharyya distance is a theoretical distance measure between two Gaussian distributions which is equivalent t o a n upper bound on the optimal Bayesian classication error probability. It also has the desirable properties of being computationally simple and extensible to more Gaussian mixtures. Using the Bhattacharyya distance measure in a datadriven approach together with a novel 2-Level Agglomerative Hierarchical Biphone Clustering algorithm, generalized left/right biphones (BGBs) are derived. A neural-net based phone recognizer trained on the BGBs is found to have better frame-level phone recognition than one trained on generalized biphones (BCGBs) derived from a set of commonlyused broad categories. We further evaluate the new BGBs on an isolated-word recognition task of perplexity 40 and obtain a 16.2% error reduction over the broad-category generalized biphones (BCGBs) and a 41.8% error reduction over the monophones.

Using accent information in ASR models for Swedish

“…If we augment the space with a metric we can measure the pairwise acoustic (dis)similarities in the pronunciation of a phoneme between two of accent areas. The distance measure chosen for this study is the Bhattacharyya distance [4] defined as:…”