Techniques for Vocabulary Expansion in Hybrid Speech Recognition Systems

TL;DR

This paper reviews existing methods and introduces a novel technique for expanding vocabulary in hybrid speech recognition systems, addressing the out-of-vocabulary problem by using subword units and adaptable HMM modifications.

Contribution

It presents a new vocabulary expansion technique that improves recognition graph processing, enhancing the ability to recognize unseen words in hybrid speech recognition systems.

Findings

Effective subword-based recognition of OOV words

Enhanced HMM modification methods for vocabulary expansion

Improved recognition accuracy for unseen words

Abstract

The problem of out of vocabulary words (OOV) is typical for any speech recognition system, hybrid systems are usually constructed to recognize a fixed set of words and rarely can include all the words that will be encountered during exploitation of the system. One of the popular approach to cover OOVs is to use subword units rather then words. Such system can potentially recognize any previously unseen word if the word can be constructed from present subword units, but also non-existing words can be recognized. The other popular approach is to modify HMM part of the system so that it can be easily and effectively expanded with custom set of words we want to add to the system. In this paper we explore different existing methods of this solution on both graph construction and search method levels. We also present a novel vocabulary expansion techniques which solve some common internal subroutine problems regarding recognition graph processing.