Search

6 May 2004: Conversational Telephone Speech System, acoustic modelling[English/Chinese]/language modelling[Chinese]]. Broadcast News Systems, acoustic modelling and adaptation]. ... Kit Thambiratnam (ajkt2@eng.cam.ac.uk) [ Broadcast News segmentation and clustering ]

14 Jun 2004: task of identifying 'who spoke when' in broadcast news audio is particularly interesting and challenging. ... Tranter, L. Wang, and P. C. Woodland. Recent Advances in Broadcast News Transcription, [in Proc.

25 May 2004: MPE training with a dynamic MMI prior. • New baseline development system (bnacTDT4)– 10x framework performance. • ... A new model set, λk1, is then trained using. single pass retraining and standard Baum-Welch estimation.

13 Jul 2004: so new segmentations can be tested without the need for computationally expensive decoding runs.

27 May 2004: IMPROVING BROADCAST NEWS TRANSCRIPTION BY LIGHTLY SUPERVISEDDISCRIMINATIVE TRAINING. H.Y. Chan & P.C. ... broadcast news data (TDT4 corpus) that were carefully chosen byclosed-caption filtering [3] [7].

27 May 2004: Broadcast News – English (BN):– Training dataset: bnac (144hr)– Test dataset: dev03 (3hr) & eval03 (3hr). • ... SPAM for CTS;– HLDASPAM for Broadcast News. • Candidate system combination branch for BN and CTS;. •

22 Mar 2004: will be evaluated in more detail when new. RT-04 diarisation dev data becomes available. • ... Tools:– What are the schedules for providing/debugging the new scoring tools for diarisation/SA-.

27 May 2004: INTRODUCTION. The majority of state-of-the-art speech recognition systems aretrained on found data, for example broadcast news and telephoneconversations. ... This is consistent with the gains that were obtainedon the Broadcast News task [11].

29 Apr 2004: Reference: The LIMSI Broadcast News transcription systemJean-Luc Gauvain, Lori Lamel and Gilles AddaSpeech Communication, Volume 37, Issues 1-2, May 2002, Pages 89-108.

27 May 2004: The large rescoring lattices produced in the first stage are thenrescored with the adapted triphones and new lattices are gener-ated on output. ... Niesler, A.Tuerk, E.W.D. Whittaker and S.J. Young, “The 1997 HTKBroadcast News Transcription System”