WAV2LETTER++

The dataset used in this paper is not explicitly mentioned, but it is implied to be a speech recognition dataset.

BibTex: