卒論で参考にする研究
Listen, Attend and Spell
Joint CTC-Attention based End-to-End Speech Recognition using Multi-task Learning
seq2seq