SautiLearn : improving online learning experience with accent translation
Date
2021-05
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The paper demonstrates the use case for SautiLearn, a tool that can help students to learn in localized language accents. The work focuses on converting audio speech from one accent to another using a sequence-to-sequence neural network model that uses speaker-independent linguistic features such as the phonetic posterior gram as input. SautiDB is a dataset collection platform that collects speech recordings of various Nigerian accents through the power of crowdsourcing. Together with SautiDB, the SautiDB-919 dataset contains 919 speech samples collected via crowdsourcing through the SautiDB web application platform. The dataset covers a wide range of ethnicities and Nigerian accent variants.
Description
item.page.type
Conference Paper
item.page.format
Keywords
AI4D, ARTIFICIAL INTELLIGENCE, AFRICAN LANGUAGES, DATABASES, LANGUAGES, LANGUAGE BARRIERS, TRANSLATION, LOCALIZATION, LINGUISTICS, MACHINE AIDED TRANSLATION, COMPUTER APPLICATIONS, SOUTH OF SAHARA, NIGERIA