Abstract
Context-dependent acoustic models have been applied inspeech recognition research for many years, and have beenshown to increase the recognition accuracy significantly.The most common approach is to use triphones. Recently,several speech recognition groups have started investigatingthe use of larger phonetic context windows when buildingacoustic models. In this paper we discuss some of thecomputational problems arising from wide context modeling(polyphonic modeling) and present methods...

This publication has 4 references indexed in Scilit: