Waxal NLP Datasets

Waxal NLP Datasets Overview This repository hosts a large multilingual speech corpus for 27 African languages, split into two task collections: Automatic Speech Recognition (ASR) — natural speech paired with human transcriptions — and Text-to-Speech (TTS) — single-speaker studio recordings paired with the scripted text that was read aloud. The dataset card and data itself indicate the underlying recordings were gathered through partnerships with Makerere… See the full description on the dataset page:

Source: Hugging Face Hub (Ephraimmm/WaxalNLPr). Metadata imported from the dataset’s Hub tags.