MFLA: Monotonic Finite Look-ahead Attention for Streaming Speech Recognition

Open in new window