archive/: script related to older experiments (eventually move into directories for each of them)
controller.py: current controller using the combined model
data/*: scripts specific to data manipulation and preprocessing (consider splitting into subdirectories for each data set)
evaluate*: model off-line evaluation
train*: model training
A directory to store manually pre-processed data.
bag_parsing: helpers to extract data from
data: helpers specific to the structure of data of the training user experiment
defaults: paths to default data and model directory. This is to avoid hard-coded paths and a command line argument should always be preferred. Ideally only use for resources that need to be committed to the repository.
features: helpers to extract speech and context features