The dataset consists of 3992 clips of Kinyarwanda TTS corpus recorded in a studio using a voice actress, it was collected in the mbaza project
Audio: 3992 Single voice studio recordings by a voice actress
Text: CSV with audio name and corresponding written text
Text collected had to include Kinyarwanda syllabes, which is made by… See the full description on the dataset page:
https://huggingface.co/datasets/mbazaNLP/kinyarwanda-tts-dataset.