Fix it where it fails: Pronunciation learning by mining error corrections from speech logs.

Zhenzhen Kou, Daisy Stanton, Fuchun Peng, Franoise Beaufays, Trevor Strohman

Browse the full ICASSP paper archive.