Fetching the paper…

On the Choice of Modeling Unit for Sequence-to-Sequence Speech Recognition · Around