MAY 1, 2026
Reality Labs Research Conversations for Hearing Augmentation Technology (RL-R CHAT) Dataset
The RL-R Conversations for Hearing Augmentation Technology (RL-R CHAT) dataset is an egocentric, multi-modal dataset created for tasks related to improved hearing, such as estimating listening effort, identifying sound sources of interest, and speech enhancement. Using the Project Aria platform, a large dataset was created of group conversations in quiet and noisy backgrounds involving familiar participants with and without hearing loss.
Overview
The Reality Labs Research Conversations for Hearing Augmentation Technology (RL-R CHAT) dataset, an egocentric, multimodal conversational dataset that is the largest of its kind available to the research community. The dataset was created for tasks related to improved hearing, such as estimating listening effort, identifying sound sources of interest, and speech enhancement. Using the Project Aria platform, we collected an egocentric dataset of group conversations in quiet and noisy environments, involving 800+ familiar participants in 300+ conversations of ~1-hour each. Group size and hearing loss were allowed to vary for additional complexity. Along with our publication (citation below), we release the RL-R CHAT dataset for public use. This site includes the egocentric data and relevant meta-data.
License
CC-BY-NC-ND
Citation
If you use this dataset, or the accompanying processed data, models, or code, please cite:
Miller, C., Murdock, C., Ananthabhotla, I., Ithapu, V. K., Proulx, M. J., Brimijoin, W. O., & Lunner, T. (2026). The RL-R Chat Dataset: Egocentric conversations among familiar interlocutors for multi-modal hearing augmentation technology. In ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 22687–22691). IEEE. https://doi.org/10.1109/ICASSP55912.2026.11460964
Our approach
Latest news
Foundational models
Meta © 2026