ESPnet just released 1.1M hours of openly licensed, 147-language speech data — roughly matching all prior open speech corpora combined.
ESPnet just released 1.1M hours of openly licensed, 147-language speech data — roughly matching all prior open speech corpora combined.