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

How secure automatic speaker verification (ASV) technology is? Moreconcretely, given a specific target speaker, how likely is it to find anotherperson who gets falsely accepted as that target? This question may be addressedempirically by studying naturally confusable pairs of speakers within a largeenough corpus. To this end, one might expect to find at least some speakerpairs that are indistinguishable from each other in terms of ASV. To a certainextent, such aim is mirrored in the standardized ASV evaluation benchmarks.However, the number of speakers in such evaluation benchmarks represents only asmall fraction of all possible human voices, making it challenging toextrapolate performance beyond a given corpus. Furthermore, the impostors usedin performance evaluation are usually selected randomly. A potentially moremeaningful definition of an impostor - at least in the context ofsecurity-driven ASV applications - would be closest (most confusable) otherspeaker to a given target.  We put forward a novel performance assessment framework to address both theinadequacy of the random-impostor evaluation model and the size limitation ofevaluation corpora by addressing ASV security against closest impostors onarbitrarily large datasets. The framework allows one to make a prediction ofthe safety of given ASV technology, in its current state, for arbitrarily largespeaker database size consisting of virtual (sampled) speakers. As aproof-of-concept, we analyze the performance of two state-of-the-art ASVsystems, based on i-vector and x-vector speaker embeddings (as implemented inthe popular Kaldi toolkit), on the recent VoxCeleb 1 & 2 corpora. We found thatneither the i-vector or x-vector system is immune to increased false alarm rateat increased impostor database size.

Voice Biometrics Security: Extrapolating False Alarm Rate via Hierarchical Bayesian Modeling of Speaker Verification Scores