Ant Group’s Bailing unit said it has open-sourced the Ming-3.0 speech series,
offering five models across three lines: speech synthesis, audio understanding
and dialogue, and speech recognition. Developers can pick models by task or
combine capabilities into their workflows.