vaghawan/hausa-tts-inference-mar-23

Splits (for the dataset viewer): test_auto — language=auto samples from inference_output_auto/. One short Hausa line per row; audio + text columns. test_hausa — language=hausa samples from inference_output_hausa/. Short text, or long multi-sentence text when the filename contains _long. Model checkpoints: see vaghawan/hausa-hausa-lang-mar-23 (step500, step1000, …).