From January 1, 2021 to July 31, 2021
During this secondment at Elyadata, we aimed to develop speech recognition system (ASR) for low resource languages. Like many languages in the world, arabic dialects don't have enough (annotated) speech data available to train Natural Language Processing systems such as Automatic Speech Recognizer.
During this seven months secondment at ElyaData, we planned to develop an Automatic Speech Recognition (ASR) system for Arabic Dialects. In order to succeed we will use Elyadata language resources and build on the solid existing work carried out on Modern Standard Arabic, a closely related high-resource language.
