This dataset contains short Luganda speech clips segmented from longer recordings, with corresponding transcriptions in Luganda.
It is intended for Text-to-Speech (TTS) research and prototyping.
wavs/: 250 WAV clips (22050 Hz recommended for TTS pipelines)
metadata.csv: Root-level metadata with two columns per line: path|text
Example: wavs/0001.wav|Gyebale ko ssebo…
bobi_interview_clean.wav: Source interview audio (cleaned)