Dynamic classifier combination in hybrid speech recognition systems using utterance-level confidence values.

Katrin Kirchhoff, Jeff A. Bilmes

Browse the full ICASSP paper archive.