A Comparison of Modeling Units in Sequence-to-Sequence Speech Recognition with the Transformer on Mandarin Chinese

Preprint English OPEN
Zhou, Shiyu; Dong, Linhao; Xu, Shuang; Xu, Bo;
  • Subject: Computer Science - Computation and Language | Electrical Engineering and Systems Science - Audio and Speech Processing | Computer Science - Sound

The choice of modeling units is critical to automatic speech recognition (ASR) tasks. Conventional ASR systems typically choose context-dependent states (CD-states) or context-dependent phonemes (CD-phonemes) as their modeling units. However, it has been challenged by s... View more
Share - Bookmark