Dataset · Sunbird
African multilingual speech and text corpus
Sunbird/salt
Parallel corpus of texts and audio recordings in English and six East African languages.
What's inside
Includes 25,000 sentences in English, Luganda, Lugbara, Acholi, Runyankole, Ateso, and Swahili (the description also mentions local topics like health and agriculture). Designed for machine translation, ASR, and TTS.
What you can do with it
- Training automatic speech recognition (ASR) models for African languages
- Creating text-to-speech (TTS) and machine translation systems
- Tasks
- Automatic speech recognition, Text → speech, Translation
- Languages
- Acoli, English, lgg, Ganda, Nyankole, Teso, ttj, Swahili and 1 more
- Size
- 10K–100K rows
- Downloads this month
- 892
- Likes
- 27
- Access
- on request on Hugging Face
Models trained on it
- Sunbird/asr-whisper-large-v2-saltAutomatic speech recognition 916
- mradermacher/lugbara-english-translator-GGUFTranslation 572
- Sunbird/asr-whisper-large-v3-saltAutomatic speech recognition 360
- Sunbird/translate-nllb-1.3b-saltText generation 187
- Sunbird/spark-tts-saltText → speech 147
- Sunbird/Sunflower-14BText generation 113
- ckuma/lugbara-multilingual-translatorTranslation 93
- Sunbird/translate-nllb-1.3b-salt-8bit 58
How to load it
from datasets import load_dataset
ds = load_dataset("Sunbird/salt")