Repository for the speech prediction experiment
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
hrc_speech_prediction
hrc_speech_prediction_msgs
.gitignore
.travis.yml
LICENSE
README.md
dependencies.rosinstall

README.md

Action Prediction from Speech Build Status Codacy Badge

Repository organization

Scripts

  • archive/: script related to older experiments (eventually move into directories for each of them)
  • controller*: controllers
    • 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

Data

A directory to store manually pre-processed data.

Main modules

  • bag_parsing: helpers to extract data from rosbags
  • combined_mode
  • 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
  • models