bhigy/natural-speech

This repository contains a codebase to build automatic speech recognition (ASR) systems for iCub and run them within YARP. It also proposes new articulatory-based and unsupervised models for ASR.

Copyright 2017 Istituto Italiano di Tecnologia

Authors: Leonardo Badino, Bertrand Higy, Luca Pasa, Raffaele Tavarone, Alessio Mereta, Claudia Canevari

natural-speech

The current folder contains the following sub-folders:

• icubrec. Codebase to build automatic speech recognition (ASR) systems for iCub and run them within YARP.

• biorec. Contains Matlab toolboxes Articulatory Phone Recognition and for Unsupervised ASR systems submitted to the 1st Zero Resources Speech, Task1. The toolbooxes use the Parallel Computing Toolbox for fast deep neural network training with GPUs.

A brief description of each function/script is provided within the README file of each sub-folder.

Acknowledgements. This work was funded by the European Commision projects Poeticon++ (grant agreement No. 288382) and ECOMODE (grant agreement No. 644096).