pocketsphinx-hmm-zh-tdt 0.8.0+real-0ubuntu5 (i386 binary) in ubuntu saucy

 PocketSphinx is a lightweight, cross-platform, speaker-independent,
 large-vocabulary continuous speech recognition engine.
 .
 This package contains the default Mandarin acoustic model for
 PocketSphinx. It was trained from 50 hours of broadcast news and
 conversation speech downsampled to 8kHz.