These demo files are generated using the attached program.
 
 	Chanwoo Kim Apr. 13, 2009



1) please listen to input files first

 * Input_Inteference_6dB_300msRev.wav
 * Input_IsotropicNoise_15dB.wav


2) Then, please listen to the output files.
 * Output_Inteference_6dB_300msRev.wav
 * Output_IsotropicNoise_15dB.wav

You can generate those output files using the attached Matlab file in PC.

PDCW('output1.wav', 'Input_Inteference_6dB_300msRev.wav')
PDCW('output2.wav', 'Input_IsotropicNoise_15dB.wav')


+++++++++++++++++++++++++++++++++Caution+++++++++++++++++++++++++++++++++++++++

* Microphone distance


We used this package in getting the all speech recognition experiment shown
in this paper (Interspeech 2009 paper). We assumed the microphone distance is 4 cm.

* Sampling rate should be 16 kHz.

* If the sampling rate or the microphone distance is different, then dITD_TH
(We are using 0.2 in our experiment)
in the code should be changed. (Basically, it is proportionate to the
microphone distance and the sampling rate, but we didn't check the preformance of
the system other than 4 cm and 16 kHz case, and longer than 4 cm distance will result
in lots of space-aliasing so it is not recommended.)

* This program can run both in PD and PDCW mode. If you want to run in PD mode,
please change  bPDCW = 1; to  bPDCW = 0; in the PDCW.m code.


If you have any problems, please contact Chanwoo Kim (chanwook@cs.cmu.edu) or
Prof. Richard M. Stern (rms@cs.cmu.edu)
