Translates SRGS XML format into an AT&T described WFST.
It is thus useful for anyone who is trying to recognize speech using WFST-based recognizer (e.g. Kaldi), but has the set of acceptable utterances described in the SRGS format.
The algorithm applied is described in bachelor thesis of the first author: Finite State Grammars as Language Models for Automatic Speech Recognition.
Grammator is published under the terms of GPLv3.