Consistent Training and Decoding For End-to-end Speech Recognition Using Lattice-free MMI

Open in new window