Self-and-Mixed Attention Decoder with Deep Acoustic Structure for Transformer-based LVCSR

Re-assign community