Swivuriso: The South African Next Voices Multilingual Speech Dataset

TL;DR

Swivuriso is a comprehensive 3000-hour multilingual speech dataset for seven South African languages, aimed at advancing ASR technology and benchmarking in underrepresented languages and domains.

Contribution

The paper introduces Swivuriso, a large-scale, ethically collected speech dataset for South African languages, filling a critical gap in resources for multilingual ASR development.

Findings

Baseline ASR models trained on Swivuriso show promising performance.

Swivuriso outperforms existing datasets in domain coverage and language diversity.

The dataset enables improved benchmarking for South African language ASR systems.

Abstract

This paper introduces Swivuriso, a 3000-hour multilingual speech dataset developed as part of the African Next Voices project, to support the development and benchmarking of automatic speech recognition (ASR) technologies in seven South African languages. Covering agriculture, healthcare, and general domain topics, Swivuriso addresses significant gaps in existing ASR datasets. We describe the design principles, ethical considerations, and data collection procedures that guided the dataset creation. We present baseline results of training/finetuning ASR models with this data and compare to other ASR datasets for the langauges concerned.