Sphinx-4 Multi-Recognizer Demo |
A simple Sphinx-4 application that demonstrates the parellel use use of several decoders. The basic idea is to embedd all additonal (or even all) recognizers into the frontend itself. Based on the pull-architecture implemented by the S4 FrontEnd, features pulled through the front-end will automatically trigger the decoding process in all embedded recognizers.
For sake of simplicity both Decoders in the configuration file share a common linguist. However this is neither necessary nor the common usecase. Normally both recognizers would decode on different search spaces.
To actually evaluate this system configuration enable the line 54 in WavFile, rebuild the jar, and run it.
Copyright 1999-2004 Carnegie Mellon University.