info.txt

this folder contains scripts for getting features from wav files and labelling them. They are meant to be used with transcripts made using kaldi. 

Files:

-cepstrum; functions for creating the MFCCs, option to stop at filterbanks can be set in main

-data_functions; functions for listing the input files, creating the phoneme dictionary and checking if the 
input files are correct.

-feature_table.txt; maps CGN phonemes to articulatory features.

-filters; functions to create and apply filterbanks.

-label_func; functions to handle the transcripts and label the data.

-main; this is to place your data path and other parameters like the number of filters and framewidth.
Run this script to create MFCCs or filterbanks for your data.

-melfreq; functions to convert Hz to mel and vice versa.

-preproc; contains the main preprocessing steps like fourier transform, padding, preemphasis and hamming windowing. 

-process_data; this is called by the main function and applies all necessary steps for each datafile and saves the data in a h5 file.

folders:

-AWD; same files as this folders but words with CGN existing AWD transcript files.  

-DNN; contains scripts for training a DNN.

-kaldi; scripts for creating new forced alignment phonetic transcripts for CGN using kaldi.

Acknowledgements:

feature_table.txt was created by Odette Scharenborg