Speech-recognition for Latex generation project.

Second year project at Ecole Polytechnique. This repo contains:

• A Django web-server with a working interface, including document, user management and audio communication

• A speech recognition module using CMUSphinx, with a custom grammar
• A custom implementation of the Myers algorithm for parsing mathematical formulae

• ML wrapper for formula (tree) choice based on multiple heuristics
  • Combines global model and custom models per document and per user

Link to the final presentation (in French).

Link to the final report (in French).

Project by Garance Cordonnier, Loïs Faisant, François Hublet, Baptiste Mourrain, Haris Sahovic, Fabrice Serret and Yolène Vanhaesbroucke.