Training of error-corrective model for ASR without using audio data.

Gakuto Kurata, Nobuyasu Itoh, Masafumi Nishimura

Browse the full ICASSP paper archive.