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

Smart devices serviced by large-scale AI models necessitates user datatransfer to the cloud for inference. For speech applications, this meanstransferring private user information, e.g., speaker identity. Our paperproposes a privacy-enhancing framework that targets speaker identityanonymization while preserving speech recognition accuracy for our downstreamtask~-~Automatic Speech Recognition (ASR). The proposed framework attachesflexible gradient reversal based speaker adversarial layers to target layerswithin an ASR model, where speaker adversarial training anonymizes acousticembeddings generated by the targeted layers to remove speaker identity. Wepropose on-device deployment by execution of initial layers of the ASR model,and transmitting anonymized embeddings to the cloud, where the rest of themodel is executed while preserving privacy. Experimental results show that ourmethod efficiently reduces speaker recognition relative accuracy by 33%, andimproves ASR performance by achieving 6.2% relative Word Error Rate (WER)reduction.

On-Device Speaker Anonymization of Acoustic Embeddings for ASR based onFlexible Location Gradient Reversal Layer