About the Corpus of Spontaneous Japanese:
The Corpus of Spontaneous Japanese (CSJ) is a database of spoken
Japanese developed by the Japan's national priority area research
project "Spontaneous Speech: Corpus and Processing Technology".
It contains about 650 hours of speech consisting of approximately
7.5 million words that were provided by more than 1,400 speakers.
For more details about the corpus, please visit the website of the
National Institute for Japanese Language (NINJAL). It is available
from the Institute.
http://www.ninjal.ac.jp/english/products/csj/
http://pj.ninjal.ac.jp/corpus_center/csj/

Meta-parameter tuning based on evolution strategy:
The meta-parameters of the system contained in conf/config_opt were
automatically tuned using evolution strategy. For the details,
please refer the following paper:
Takafumi Moriya, Tomohiro Tanaka, Takahiro Shinozaki, Shinji Watanabe,
and Kevin Duh, "Automation of System Building for State-of-the-art
Large Vocabulary Speech Recognition Using Evolution Strategy," Proc.
IEEE 2015 Automatic Speech Recognition and Understanding Workshop
(ASRU), 2015.


Each subdirectory of this directory contains the
scripts for a sequence of experiments.
s5: This is the current recommended recipe.
   The recipe supports the third and fourth editions of CSJ.