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

Knowledge distillation (KD) is used to enhance automatic speaker verificationperformance by ensuring consistency between large teacher networks andlightweight student networks at the embedding level or label level. However,the conventional label-level KD overlooks the significant knowledge fromnon-target speakers, particularly their classification probabilities, which canbe crucial for automatic speaker verification. In this paper, we firstdemonstrate that leveraging a larger number of training non-target speakersimproves the performance of automatic speaker verification models. Inspired bythis finding about the importance of non-target speakers' knowledge, wemodified the conventional label-level KD by disentangling and emphasizing theclassification probabilities of non-target speakers during knowledgedistillation. The proposed method is applied to three different student modelarchitectures and achieves an average of 13.67% improvement in EER on theVoxCeleb dataset compared to embedding-level and conventional label-level KDmethods.

Emphasized Non-Target Speaker Knowledge in Knowledge Distillation for Automatic Speaker Verification