Efficient Monotonic Multihead Attention
This addresses the need for real-time translation systems, though it appears incremental with improvements in training and inference strategies.
The paper tackles the problem of simultaneous speech-to-text translation by introducing Efficient Monotonic Multihead Attention (EMMA), which achieves state-of-the-art performance on Spanish and English translation tasks.
We introduce the Efficient Monotonic Multihead Attention (EMMA), a state-of-the-art simultaneous translation model with numerically-stable and unbiased monotonic alignment estimation. In addition, we present improved training and inference strategies, including simultaneous fine-tuning from an offline translation model and reduction of monotonic alignment variance. The experimental results demonstrate that the proposed model attains state-of-the-art performance in simultaneous speech-to-text translation on the Spanish and English translation task.