模型:

speechbrain/soundchoice-g2p