This repository contains the code for the model published in 'Murakami et al. (2015): "Seeing [u] aids vocal learning: babbling and imitation of vowels using a 3D vocal tract model, reinforcement learning, and reservoir computing." International Conference on Development and Learning and on Epigenetic Robotics 2015 (in press).'

The code is written in Python 2.7 and was tested in Ubuntu 14.04. Apart from standard Python libraries, it requires:

• numpy
• scipy
• matplotlib
• brian
• Oger
• mpi4py

You can run create_python_environment.sh in order to create an python environment and install into that environment all the dependencies. You need to have virtualenv installed.

Additionally, you'll need the VocalTractLab API for Linux, which you can download at vocaltractlab. After downloading it, extract the content into the VTL_API subfolder.

generatedata.py :

generates training samples for the auditory learning.

learndata.py :

does the auditory learning with the BRIAN hears and the echo state network (ESN).

rl_agent_mpi.py :

does the sensomotoric learning. While doing the learning it uses environment_fun.py as a resource.

environment_fun.py :

supplies the environment for the reinforcement learning. It transforms the motor signals into reward signals by using the trained ESN.

example.py provides a step-by-step tutorial of how to use the code.