LMU–BR Study Finds AI Speech Models Struggle With Bavarian Dialects

Researchers attribute the performance gap to scarce dialect training data, urging larger corpora to enable future subtitling.