English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper, we present the speaker diarization system for theMulti-channel Multi-party Meeting Transcription Challenge (M2MeT) from teamDKU_DukeECE. As the highly overlapped speech exists in the dataset, we employan x-vector-based target-speaker voice activity detection (TS-VAD) to find theoverlap between speakers. For the single-channel scenario, we separately traina model for each of the 8 channels and fuse the results. We also employ thecross-channel self-attention to further improve the performance, where thenon-linear spatial correlations between different channels are learned andfused. Experimental results on the evaluation set show that the single-channelTS-VAD reduces the DER by over 75% from 12.68\% to 3.14%. The multi-channelTS-VAD further reduces the DER by 28% and achieves a DER of 2.26%. Our finalsubmitted system achieves a DER of 2.98% on the AliMeeting test set, whichranks 1st in the M2MET challenge.

Cross-Channel Attention-Based Target Speaker Voice Activity Detection: Experimental Results for M2MeT Challenge