Language
The catalogue lists 5 Swahili datasets. Most listings are audio and text data, typically used for speech recognition, conversational AI and speech synthesis (TTS). Where providers state a volume in hours, it ranges from 227 hours to 43.1k hours.
Regional variants listed: Swahili (macrolanguage).
Audio · Text
Clean, high-quality scripted recordings aligned with transcripts, built for training expressive and natural-sounding TTS systems.
Audio
A collection of conversational audio with speakers separated by channel, in swahili
Audio
A collection of live, conversational audio with speakers separated by channel, in swahili
Audio
A collection of live, conversational audio in swahili, in single-channel and mp3 format.
Audio · Text
High-quality transcribed Swahili speech audio dataset collected from native speakers.