DatoricAI/audio-video-conversation-4000h

Audio-Video Conversational Dataset 4,000 hours of synchronized speech and video of natural conversations: face movement, mouth motion, gestures, turn-taking, emotion, laughter, and interruptions across 20+ languages. This repository is a specification and preview listing. The production dataset is rights-cleared and delivered directly to buyers. Request access to see the full schema and get real samples. Overview The Audio-Video Conversational Dataset is a 4… See the full description on the dataset page: https://huggingface.co/datasets/DatoricAI/audio-video-conversation-4000h.