SALT Dataset
RESEARCH CORPUS
The Sunbird African Language Technology (SALT) dataset is a multi-way parallel text and speech corpus of over 25,000 sentences across 6 languages: English, Acholi, Ateso, Luganda, Lugbara, and Runyankole, covering healthcare, agriculture, legal discourse, and civic life.
Key Features
- 1.7B+ tokens of multilingual text across 75 languages.
- Domains: healthcare, agriculture, legal, civic, and social discourse
- 20,000+ hours of speech across 52 languages for ASR and TTS.
