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/

Zendy Plus

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

Zendy Tools

Zendy Open

Target-speaker voice activity detection (TS-VAD) has recently shown promisingresults for speaker diarization on highly overlapped speech. However, theoriginal model requires a fixed (and known) number of speakers, which limitsits application to real conversations. In this paper, we extend TS-VAD tospeaker diarization with unknown numbers of speakers. This is achieved by twosteps: first, an initial diarization system is applied for speaker numberestimation, followed by TS-VAD network output masking according to thisestimate. We further investigate different diarization methods, includingclustering-based and region proposal networks, for estimating the initiali-vectors. Since these systems have complementary strengths, we propose afusion-based method to combine frame-level decisions from the systems for animproved initialization. We demonstrate through experiments on variants of theLibriCSS meeting corpus that our proposed approach can improve the DER by up to50\% relative across varying numbers of speakers. This improvement also resultsin better downstream ASR performance approaching that using oracle segments.

Target-speaker Voice Activity Detection with Improved I-Vector Estimation for Unknown Number of Speaker