
The author presents the first publicly available multimodal dataset, REWIND, for speaking status segmentation from body movement in real-life mingling scenarios with high-quality audio recordings.


coremsg

REWIND-Dataset-Privacy-preserving-Speaking-Status-Segmentation-from-Multimodal-Body-Movement-Signals-in-the-Wild


REWIND Dataset: Privacy-preserving Speaking Status Segmentation from Multimodal Body Movement Signals in the Wild


title_rewrite


Recognizing speaking in humans using multimodal signals for privacy-preserving segmentation.


rewind-dataset-privacy-preserving-speaking-status-segmentation-from-multimodal-body-movement-signals-in-the-wild


Recognizing speaking in humans using machine learning models trained on video and wearable sensor data.
