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

Recent advances in Talking Head Generation (THG) have achieved impressive lipsynchronization and visual quality through diffusion models; yet existingmethods struggle to generate emotionally expressive portraits while preservingspeaker identity. We identify three critical limitations in current emotionaltalking head generation: insufficient utilization of audio's inherent emotionalcues, identity leakage in emotion representations, and isolated learning ofemotion correlations. To address these challenges, we propose a novel frameworkdubbed as DICE-Talk, following the idea of disentangling identity with emotion,and then cooperating emotions with similar characteristics. First, we develop adisentangled emotion embedder that jointly models audio-visual emotional cuesthrough cross-modal attention, representing emotions as identity-agnosticGaussian distributions. Second, we introduce a correlation-enhanced emotionconditioning module with learnable Emotion Banks that explicitly captureinter-emotion relationships through vector quantization and attention-basedfeature aggregation. Third, we design an emotion discrimination objective thatenforces affective consistency during the diffusion process throughlatent-space classification. Extensive experiments on MEAD and HDTF datasetsdemonstrate our method's superiority, outperforming state-of-the-art approachesin emotion accuracy while maintaining competitive lip-sync performance.Qualitative results and user studies further confirm our method's ability togenerate identity-preserving portraits with rich, correlated emotionalexpressions that naturally adapt to unseen identities.

Disentangle Identity, Cooperate Emotion: Correlation-Aware Emotional Talking Portrait Generation