A multilingual dataset of synchronized conversational speech and video, annotated with transcripts, speaker turns, gestures, emotion, overlap, and interruption events.
Datoric’s Audio-Video Conversational Dataset pairs natural multi-speaker conversation recordings with synchronized video. Its documented annotations include transcript segments, speaker turns and timestamps, diarization, face and mouth visibility, gestures, gaze, emotion or tone, laughter, interruptions, overlapping speech, and speaker roles. Datoric describes coverage across more than 20 languages, with the selected language and recording mix confirmed for a delivery. Available review materials include schema and sample metadata; production media and annotations are provided in a buyer review package. Listed delivery formats include MP4, WAV where tracks are separated, CSV, JSON, and transcript files.
Be the first Cosign
Add it to your Products I use list. You can also share why you use it.
