%%
%% ======================================================================
%%
%% README file for GEDI and mr-GEDI tools
%%
%% Copyright (c) 2019-2020 Wakayama University and Nippon Telegraph and Telephone corporation (NTT).
%% All rights reserved.
%% By Katsuhiko Yamamoto, 15-03-2019
%% Mod by Irino, T., 29-08-2020
%% ======================================================================
%%

The tools and program included in this package are designed and distributed for the evaluation of the intelligibility of enhanced speech processed by nonlinear algorithms. 
This program can be used only for evaluation purpose. Please refer to the license file for details.

Below is a list of related papers:

	Katsuhiko Yamamoto, Toshio Irino, Shoko Araki, Keisuke Kinoshita, Tomohiro Nakatani (2020) "GEDI: Gammachirp Envelope Distortion Index for Predicting Intelligibility of Enhanced Speech," Speech Communication, 2020, DOI: 10.1016/j.specom.2020.06.001
	
	Yamamoto, K., Irino, T., Ohashi, N., Araki, S., Kinoshita, K., Nakatani, T. (2018) "Multi-resolution Gammachirp Envelope Distortion Index for Intelligibility Prediction of Noisy Speech." Proc. Interspeech 2018, 1863-1867, DOI: 10.21437/Interspeech.2018-1291
	
	Yamamoto, K., Irino, T., Matsui, T., Araki, S., Kinoshita, K., Nakatani, T. (2017) "Predicting Speech Intelligibility Using a Gammachirp Envelope Distortion Index Based on the Signal-to-Distortion Ratio." Proc. Interspeech 2017, 2949-2953, DOI: 10.21437/Interspeech.2017-170

Before starting, please read carefully the license file and the following,
   - The package includes a MATLAB code version of the GEDI/mr-GEDI algorithm (GEDI_OutdcGC.m/mr-GEDI_OutdcGC.m and GEDI.m/mr-GEDI.m) .
   - To run MATLAB files, you need the signal processing toolbox.
   - Note that Wakayama Univ. and NTT may stop releasing this code at any time without prior notice.

In this package you will find the following
   (1) The license file (License.txt)
   (2) A MATLAB version of the GEDI/mr-GEDI algorithms (GEDI.m, mrGEDI.m, GEDI_OutdcGC.m, and mrGEDI_OutdcGC.m) and the frame-based version (mrGEDIfb.m and mrGEDI_OutdcGCfb.m)
   (3) A sample script describing how to use  (test_GEDI.m, test_mrGEDI.m, and test_mrGEDIfb.m)
   (4) Please download the package of the dynamic compressive gammachirp filterbank (dcGC-FB) and put it under "package/" directory. "package/GCFBv211pack/"  is for original GEDI/mrGEDI. "package/GCFBv221pack/"  is for mrGEDIfb.
   (5) Sample data in wav_sample/sample_sp1.wav ~ wav_sample/sample_sp4.wav, and wav_sample/sample_sp_clean.wav


For any inquiries, please contact us at gedi-ml@hco.ntt.co.jp

This software and related documentation include the following files,
|-- package
|   `-- EMPTY: Please download "GCFBv211pack" and/or "GCFBv221pack" from AMLAB-Wakayama’s GitHub repository (URL: https://github.com/AMLAB-Wakayama/gammachirp-filterbank) and put them into this directory.
|-- wav_sample
|   |-- sample_sp_clean.wav
|   |-- sample_sp1.wav
|   |-- sample_sp2.wav
|   |-- sample_sp3.wav
|   `-- sample_sp4.wav
|-- README
|-- License.txt
|-- GEDI.m          (original GEDI)
|-- GEDI_OutdcGC.m
|-- test_GEDI.m
|-— mrGEDI.m       (original mrGEDI)
|-- mrGEDI_OutdcGC.m
|-— test_mrGEDI.m
|-— mrGEDIfb.m     (fast, frame-based mrGEDI)
|-- mrGEDIfb_OutdcGC.m
`-— test_mrGEDIfb.m


-	GCFBv211pack and GCFBv221pack are Matlab packages of the dynamic compressive gammachirp filterbank. 
-	test_GEDI.m, GEDI.m, and GEDI_OutdcGC.m are Matlab codes for the gammachirp envelope distortion index (GEDI). These calculate speech intelligibility of noisy/enhanced speech sounds. GCFBv211pack is necessary. Execute test_GEDI.m first to learn the codes.
-	test_mrGEDI.m, mrGEDI.m, and mrGEDI_OutdcGC.m are Matlab codes for the "multi-resolution" gammachirp envelope distortion index (mrGEDI). These calculate speech intelligibility of noisy/enhanced speech sounds. GCFBv211pack is necessary. Execute test_mrGEDI.m first to learn the codes.
-	test_mrGEDIfb.m, mrGEDIfb.m, and mrGEDIfb_OutdcGC.m are Matlab codes for frame-based mrGEDI. These calculate speech intelligibility of noisy/enhanced speech sounds sufficiently fast. GCFBv221pack is necessary. Execute test_mrGEDIfb.m first to learn the codes.
-	The files in wav_sample are example sound files of noisy speech a clean speech reference and noisy speeches.