Explore projects

Multimodal Convolutional Neural Networks for low-latency speech inference (C++), enabling assistive voice software in research and simulated environments. Scripts to curate & pre-process data, embedding image/audio at scale for model training and evaluation

Tools facilitating the manipulation and loading of data from different sources and different domains.

Ce projet consiste en la conception d'un clavier virtuel basé sur les panneaux de phonèmes d'Alexandre Do (inspiré des traveaux de Caleb Gattogno), croisé avec la liste des graphies possibles pour chaque phonème (Fidel, C. Gattegno).