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

Cued Speech (CS) is a communication system for deaf people or hearingimpaired people, in which a speaker uses it to aid a lipreader in phoneticlevel by clarifying potentially ambiguous mouth movements with hand shape andpositions. Feature extraction of multi-modal CS is a key step in CSrecognition. Recent supervised deep learning based methods suffer from noisy CSdata annotations especially for hand shape modality. In this work, we firstpropose a self-supervised contrastive learning method to learn the featurerepresentation of image without using labels. Secondly, a small amount ofmanually annotated CS data are used to fine-tune the first module. Thirdly, wepresent a module, which combines Bi-LSTM and self-attention networks to furtherlearn sequential features with temporal and contextual information. Besides, toenlarge the volume and the diversity of the current limited CS datasets, webuild a new British English dataset containing 5 native CS speakers. Evaluationresults on both French and British English datasets show that our modelachieves over 90% accuracy in hand shape recognition. Significant improvementsof 8.75% (for French) and 10.09% (for British English) are achieved in CSphoneme recognition correctness compared with the state-of-the-art.

An Attention Self-supervised Contrastive Learning based Three-stage Model for Hand Shape Feature Representation in Cued Speech