readme

It is the Speech recognition software. It is platform independent. To execute the source code,

# Copyright 1999-2002 Carnegie Mellon University.  
# Portions Copyright 2002 Sun Microsystems, Inc.  
# Portions Copyright 2002 Mitsubishi Electronic Research Laboratories.
# All Rights Reserved.  Use is subject to license terms.
# 
# See the file "license.terms" for information on usage and
# redistribution of this file, and for a DISCLAIMER OF ALL 
# WARRANTIES.

The directories here contain various models trained from
the WSJ data

Features used to train the models were computed from
broadband 16 bit sampled data sampled at 16Khz.
The features were 13 dimensional cepstra computed
from 40 mel filters covering the frequency range
130Hz to 6800Hz.
The feature type used was 1s_c_d_dd


Contents of this directory:

am.props
--------

Properties file containing the parameters used to train the acoustic model.

alpha.dict
----------

Dictionary containing only the alphabet and their phoneme mapping(s).

cd_continuous_1gau/
-------------------

Contains 1 gau/state acoustic models.

cd_continuous_2gau/
-------------------

Contains 2 gaus/state acoustic models.

cd_continuous_4gau/
-------------------

Contains 4 gaus/state acoustic models.

cd_continuous_8gau/
-------------------

Contains 8 gaus/state acoustic models.

ci_continuous/
--------------

Contains context-independent acoustic models.

etc/
----

Contains the various associated files, such as the mdef files,
dictionary, fillerdict and a stripped down version
of the variable settings used to train the models.