SA4ME: The 1st International Workshop on Spatial Audio for Multisensory Experiences

Overview

Scene understanding provides geometry, user pose, and source locations, but this information is not yet systematically incorporated into spatial audio for eXtended Reality (XR). Differences in data formats, metadata, and spatial representations across sensing, scene understanding, and audio rendering further complicate integration and reuse. The Spatial Audio for Multisensory Experiences (SA4ME) workshop brings together researchers in audio, computer vision, mobile sensing and human-computer interaction to examine how scene and sensor data can inform spatial audio, and how sound can support XR telepresence and multi-user interaction. We welcome contributions on open challenges, emerging approaches, and shared evaluation scenarios for multisensory XR.

Key Topics

The workshop invites contributions and fosters discussion around the following interconnected areas:

Vision-to-Audio Translation for Dynamic Environments: Leveraging visual depth, scene representations such as neural radiance fields and 3D Gaussian splatting, and material cue to predict acoustic parameters and reverberation characteristics.
Audio as a Spatial Condition for Visual and Social Presence: Utilizing sound localization and speech semantics to drive avatar behavior, direct multi-user gaze, guide visual attention beyond the field of view, and enhance conversational presence.
Mobile Sensing and Egocentric Multimodal Understanding: Integrating on-device sensing, IMU-based motion tracking and spatial sensor data across mobile, wearable, and peripheral devices for multimodal sensor fusion, behavioral tracking, spatial understanding, and embodied interaction under constrained computational budgets.
Interactive Digital Twins and Multi-User Social XR: Spatial audio as an amplifier of co-presence in multi-user environments where remote spaces, not just avatars, are augmented into the local space, closing the presence gap left by occlusion bound visual rendering through omnidirectional, spatially surrounding sound which understands depth cues.
Dataset Interoperability: Defining unified data structures, metadata, and exchange formats that allow calibrated visual, acoustic, and mobile sensor recordings to be integrated and reused across modalities without dataset-specific conversion.

Important Dates

Submission deadline: December 20, 2026 (AoE)
Notification of acceptance: January 4, 2027 (AoE)
Camera-ready deadline: January 11, 2027 (AoE)

Keynote Speakers

Prof. Stefania Serafin

Stefania Serafin is an Associate Professor in the Department of Engineering Technology and Didactics at the Technical University of Denmark (DTU). Her research focuses on sonic interaction design, immersive audio, and multisensory experiences in virtual and augmented reality. She received the 2026 IEEE VGTC Virtual Reality Technical Achievement Award.

Prof. Tae-Hyun Oh

Tae-Hyun Oh is an Associate Professor in the School of Computing at KAIST. His research spans computer vision, machine learning, and computer graphics, focusing on multimodal perception, audio-visual understanding, and spatial and physics-aware AI for embodied agents.

Prof. Ting Dang

Ting Dang is a Senior Lecturer in the School of Computing and Information Systems at the University of Melbourne. Her research spans human-centered AI, speech and audio processing, and multimodal sensing, focusing on systems that understand human signals and support communication and accessibility.

Prof. Dooyoung Kim

Dooyoung Kim is a Lecturer in the Department of Computer Science and Information Technology at La Trobe University. His research combines XR, AI, and human-computer interaction, focusing on mixed reality telepresence, spatial memory capture and replay, and interfaces that connect people, AI agents, and physical environments.

Organizers