基于注意力(Attention)机制的端到端系统,又被称为LAS端到端构架。
[6] W. Chan, N. Jaitly, Q. Le, O. Vinyals. Listen, Attend and Spell: A Neural Network for Large Vocabulary Conversational Speech Recognition. ICASSP 2016.
来自 <https://mp.weixin.qq.com/s/c64XucML13OwI26_UE9xDQ>
为了更好地进行LAS模型的训练。可以使用以下技术:
- Schedule Sampling
- Label Smoothing
- Multi-head Attention