DavidVita/archive-bd66a225

Korean multi-speaker speech corpus 46,346 studio-recorded Korean utterances from 8 professional voice performers, 85h 45m in total, each paired with the exact sentence that was read. Seven of them work from the same five script families, so a large part of the corpus is the same sentence read by different voices — usable for speech synthesis, voice conversion and speaker-representation work. Performers are identified only by opaque codes. Nothing here names a person. A private… See the full description on the dataset page: https://huggingface.co/datasets/DavidVita/archive-bd66a225.