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

Voice data generated on instant messaging or social media applicationscontains unique user voiceprints that may be abused by malicious adversariesfor identity inference or identity theft. Existing voice anonymizationtechniques, e.g., signal processing and voice conversion/synthesis, suffer fromdegradation of perceptual quality. In this paper, we develop a voiceanonymization system, named V-Cloak, which attains real-time voiceanonymization while preserving the intelligibility, naturalness and timbre ofthe audio. Our designed anonymizer features a one-shot generative model thatmodulates the features of the original audio at different frequency levels. Wetrain the anonymizer with a carefully-designed loss function. Apart from theanonymity loss, we further incorporate the intelligibility loss and thepsychoacoustics-based naturalness loss. The anonymizer can realize untargetedand targeted anonymization to achieve the anonymity goals of unidentifiabilityand unlinkability.  We have conducted extensive experiments on four datasets, i.e., LibriSpeech(English), AISHELL (Chinese), CommonVoice (French) and CommonVoice (Italian),five Automatic Speaker Verification (ASV) systems (including two DNN-based, twostatistical and one commercial ASV), and eleven Automatic Speech Recognition(ASR) systems (for different languages). Experiment results confirm thatV-Cloak outperforms five baselines in terms of anonymity performance. We alsodemonstrate that V-Cloak trained only on the VoxCeleb1 dataset againstECAPA-TDNN ASV and DeepSpeech2 ASR has transferable anonymity against otherASVs and cross-language intelligibility for other ASRs. Furthermore, we verifythe robustness of V-Cloak against various de-noising techniques and adaptiveattacks. Hopefully, V-Cloak may provide a cloak for us in a prism world.

V-Cloak: Intelligibility-, Naturalness- & Timbre-Preserving Real-Time Voice Anonymization