Segment-level Training of ANNs Based on Acoustic Confidence Measures for Hybrid HMM/ANN Speech Recognition.

S. Pavankumar Dubagunta, Mathew Magimai-Doss

Browse the full ICASSP paper archive.