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

The primary characteristic of robust speaker representations is that they areinvariant to factors of variability not related to speaker identity.Disentanglement of speaker representations is one of the techniques used toimprove robustness of speaker representations to both intrinsic factors thatare acquired during speech production (e.g., emotion, lexical content) andextrinsic factors that are acquired during signal capture (e.g., channel,noise). Disentanglement in neural speaker representations can be achievedeither in a supervised fashion with annotations of the nuisance factors(factors not related to speaker identity) or in an unsupervised fashion withoutlabels of the factors to be removed. In either case it is important tounderstand the extent to which the various factors of variability are entangledin the representations. In this work, we examine speaker representations withand without unsupervised disentanglement for the amount of information theycapture related to a suite of factors. Using classification experiments weprovide empirical evidence that disentanglement reduces the information withrespect to nuisance factors from speaker representations, while retainingspeaker information. This is further validated by speaker verificationexperiments on the VOiCES corpus in several challenging acoustic conditions. Wealso show improved robustness in speaker verification tasks using dataaugmentation during training of disentangled speaker embeddings. Finally, basedon our findings, we provide insights into the factors that can be effectivelyseparated using the unsupervised disentanglement technique and discusspotential future directions.

An empirical analysis of information encoded in disentangled neural speaker representations