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

One-shot talking head generation has no explicit head movement reference,thus it is difficult to generate talking heads with head motions. Some existingworks only edit the mouth area and generate still talking heads, leading tounreal talking head performance. Other works construct one-to-one mappingbetween audio signal and head motion sequences, introducing ambiguitycorrespondences into the mapping since people can behave differently in headmotions when speaking the same content. This unreasonable mapping form fails tomodel the diversity and produces either nearly static or even exaggerated headmotions, which are unnatural and strange. Therefore, the one-shot talking headgeneration task is actually a one-to-many ill-posed problem and people presentdiverse head motions when speaking. Based on the above observation, we proposeOSM-Net, a \textit{one-to-many} one-shot talking head generation network withnatural head motions. OSM-Net constructs a motion space that contains rich andvarious clip-level head motion features. Each basis of the space represents afeature of meaningful head motion in a clip rather than just a frame, thusproviding more coherent and natural motion changes in talking heads. Thedriving audio is mapped into the motion space, around which various motionfeatures can be sampled within a reasonable range to achieve the one-to-manymapping. Besides, the landmark constraint and time window feature input improvethe accurate expression feature extraction and video generation. Extensiveexperiments show that OSM-Net generates more natural realistic head motionsunder reasonable one-to-many mapping paradigm compared with other methods.

OSM-Net: One-to-Many One-shot Talking Head Generation with Spontaneous Head Motions