Reproduce “Talking Spell: A Wearable System Enabling Real-Time Anthropomorphic Voice Interaction with Everyday Objects” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Xuetong Wang, Ching Christie Pang, Pan Hui

Researchers at The Hong Kong University of Science and Technology developed Talking Spell, a wearable system that enables real-time anthropomorphic voice interaction with any everyday object. This system, combining advanced computer vision with large vision-language models, achieved a System Usability Scale score of 79.09 and demonstrated high object detection accuracy, fostering emotional connections through personalized dialogue.

Key insights

  • Anthropomorphism, when applied to familiar objects, substantially enhances emotional connection and user engagement, particularly in recreational and companionship scenarios.
  • A wearable, user-centric system allows for a flexible "radiative network" where the user communicates directly with multiple objects, shifting from device-centric AI.
  • Advanced large vision-language models can effectively generate rich, consistent anthropomorphic personas for objects, outperforming general-purpose LLMs in this specific task.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.