Week 23
Spent this week outlining the LTASS method in MATLAB and took a look at the data preparation scripts that are under the VoxForge Kaldi example. I thought it was going to be straight-forward, but it appears there are some conflicts with the directory names of the CHILDES folder and VoxForge folders. More clearly, the bash script is looking for specific types of folder the names.
LTASS
The Long-Term Average Speech Spectrum (LTASS) is a representation of where sounds are regarding pitch and volume, superimposed over the result of a hearing test. Hence, we want to use the LTASS method to observe which parts of spectrum fall within their “hearing range” of the two corpora. Maybe they are really similar, if they are not, the results are still illuminating.
Kaldi
Working on changing the way some of the VoxForge scripts work to generate results from the CHILDES data. The CHILDES data might need to be reformatted, but at the moment we are trying to use a pearl regexp to accounts for both the folder and files of the CHILDEs data.
Best,
EO