LIFE_G1.m is the top MATLAB file for LIFE processing.
It takes 2 inputs, a wavfile (MS-WAV format) and a file to write the LIFE processed features.
It optionally returns the features in a Ndim*Nframe matrix, where Ndim is the feature dimension and
Nframe is the number of feature frames.


features_MFCC.m program is the top file for feature extraction. 
Its inputs are a wavfile and parameter to indicate the type of nonlinearity (log or power-law).
Feature extraction is currently being done for a speech signal with sampling frequency of 16 kHz.
The feature extraction program will later be extended for other sampling frequencies.
Current features are obtained for the conventional 25 ms window length and 10 ms frame shift.
Assign the variable "temppath" to a temporarty path, it's used to save an intermediate file.


features_icpf_VML_G1.m applied LIFE processing. It's first input arguement is feature matrix and
the second input argument specifies the inclusion of CPF pre-processing.

The program is currently configured to work with power-law nonlinearity and includes CPF pre-processing in the
LIFE processing.

Please refer to the following papers for additional technical details:

% CPF (Cepstral Post-Filtering)
@inproceedings{CPF08,
title= "ENVIRONMENT-INVARIANT COMPENSATION FOR REVERBERATION USING LINEAR POST-FILTERING FOR MINIMUM DISTORTION",
author= "Kshitiz Kumar and Richard M. Stern",
booktitle= "in Proc. IEEE ICASSP",
pages= "4121-4124",
year= 2008
}

% LIFE (Likelihood-based Inverse-Filtering)

@inproceedings{LIFE10,
title = "Maximum-Likelihood-based Cepstral Inverse Filtering for Blind Speech Dereverberation",
author =  "K. Kumar and R. M. Stern",
booktitle= "in Proc. IEEE ICASSP",
year = "2010"
}