Fetching the paper…

State-of-the-Art Speech Recognition Using Multi-Stream Self-Attention With Dilated 1D Convolutions · Around