Efficient generation of pronunciation dictionaries: machine learning factors during bootstrapping

Abstract

The authors focus on factors related to the underlying rule-extraction algorithms, and demonstrate variants of the Dynamically Expanding Context algorithm, which are beneficial for this application. They show that continuous updating of the learned rules, coupled with a new approach to grapheme-to-phoneme alignment and a sliding-window approach to choosing the content window, leads to an efficient and accurate bootstrapping mechanism. In this paper the authors describe the techniques implemented to optimise the process from a machine learning perspective and report on the results achieved.