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 work, we present an analysis of a DNN-based autoencoder for speechenhancement, dereverberation and denoising. The target application is a robustspeaker verification (SV) system. We start our approach by carefully designinga data augmentation process to cover wide range of acoustic conditions andobtain rich training data for various components of our SV system. We augmentseveral well-known databases used in SV with artificially noised andreverberated data and we use them to train a denoising autoencoder (mappingnoisy and reverberated speech to its clean version) as well as an x-vectorextractor which is currently considered as state-of-the-art in SV. Later, weuse the autoencoder as a preprocessing step for text-independent SV system. Wecompare results achieved with autoencoder enhancement, multi-condition PLDAtraining and their simultaneous use. We present a detailed analysis withvarious conditions of NIST SRE 2010, 2016, PRISM and with re-transmitted data.We conclude that the proposed preprocessing can significantly improve bothi-vector and x-vector baselines and that this technique can be used to build arobust SV system for various target domains.

Analysis of DNN Speech Signal Enhancement for Robust Speaker Recognition