A 70-hour subset of Luhya speech data collected by Digital Divide Data in Kenya. The dataset includes recorded sentences from native speakers and is intended to support research and development in Automatic Speech Recognition for low-resource African languages.