Home/Ongoing research and projects
WARA Media & Language develops, tests and demonstrates AI for media and language together with its academic and industrial partners. The work is organised in four research focus areas and carried out in ongoing projects – from shared testbeds and benchmarks to PhD projects. Choose an area or a project below to read more.
Ongoing projects
Testbed for evaluating speech, gesture and conversational components – now in five tracks.
Avatar Project
Benchmarking and leaderboard for gesture-generation models, with KTH and Electronic Arts.
GENEA Leaderboard
3D Gaussian Splatting for real-time novel-view synthesis and scene rendering.
3DGS for Scene Representation & Rendering
More ongoing projects
Evaluation of Generative Models for Gaming
Reliable, explainable evaluation metrics for generative models in the gaming industry.
Language Models for Software Security
Using language models to understand source code, bytecode and other software artifacts.
Communicative Behaviour
Modelling nonverbal facial and body cues that carry emotional content in interpersonal communication.
Efficient Modularisation & Adaptation of LLMs
Making resource-hungry large language models modular, adaptable and efficient.