Linguistics and HLT for Countries with Minimal ICT Infrastructure

Abstract

This paper presents a visionary Human Language Technologies (HLT) model aimed at enabling the Basotho, who are computer illiterate, and thus unable to access internet but have access to either a landline handset or a cellular phone, to browse and access electronic government information service in their language, Sesotho. This requires various speech-based HLT components such as pronunciation modelling, speech recognition, text-to-speech (TTS) and language identification. For purposes of this presentation, we concentrate on speech recognition and TTS. The paper also provides a preliminary experimental setup necessary for the proposed model. The proposed model is specifically intended to empower the illiterate and the underprivileged members of the society.