Improving lightly supervised training for broadcast transcription

Yanhua Long, Mark J. F. Gales, Pierre Lanchantin, Xunying Liu, Matthew Stephen Seigel, Philip C. Woodland