A hierarchical framework is proposed to address the issues of modeling different type of words in keyword spotting (KWS). Keyword models are built at various levels according to the availability of training set resources for each individual word. The proposed approach improves the performance of KWS even when no training speech is available for the keywords. It also suggests an easier way to co...

boolean byte class double else extends final finally float if implements int interface long native new private public return short static super synchronized this throw throws transient try void volatile while ( ) [ ] { } ; , . = Figure 1: Java keywords.