New dataset tests speech recognition on accented academic discussions

Original: PAREDA: A Multi-Accent Speech Dataset of Natural Language Processing Research Discussions

Source: arxiv.org ↗