HuComTech Multimodal Corpus and Database
The HuComTech multimodal corpus consists of about 50 hours of video and audio recordings of 111 formal dialogues (simulated job interviews) and 111 informal but guided dialogues. The language of the recordings is Hungarian. The participants were university students aged 19-27, female 54 and male 67. The corpus was annotated for video (facial expressions, instances of eyebrows, gaze, headshift. handshape, touchmotion and posture) and audio (emotions, discourse, prosody and textual transcriptions). Its unique features in a wider comparison include its special attention to pragmatics focusing on a comparative study of the unimodal vs. multimodal features of communication (as compared to multimodality alone) as well as the study of the syntax and prosody of spoken language with respect to the above wide range of multimodal characteristics. The data can be queried in ELAN and in our web-based SQL database.
People who looked at this resource also viewed the following: