Fast speaker adaptation using triple diagonal and shared block diagonal transform matrices.

Guo-Hong Ding, Bo Xu, Juha Iso-Sipil, Yang Cao

Browse the full ICASSP paper archive.