Audiovisual recordings of acted casual conversations between four speakers in German

This dataset was used in a study by Hendrikse et al. (2018). It contains 7 audiovisual recordings of acted casual conversations between 4 actors (two males and two females), as well as 2-3 questions about the content of each of the conversations asked by one of the speakers. The conversations last between 1min 24s to 1min 39s and the topics are food, holidays/travelling, weather, work, plans, movies and anecdotes. Two of the actors are native speakers and two are non-native speakers with a C1 level of German. The actors memorized the script and acted as in a casual conversation. The audiovisual material was recorded in a circular space with uniform purple background using two cameras (Canon EOS 700D) and one microphone for each speaker (Neumann KM 184). 
Additionally the visual stimuli used in the aforementioned study with virtual characters is attached: a screencast of  a simulation of each conversation with virtual characters (Makehuman) with gaze to target speaker and lip-syncing (Llorach et al. 2016). The 3D scene is rendered with Blender3D with a camera with a 120º field-of-view.

References:

M.M.E. Hendrikse, G. Llorach, G. Grimm, V. Hohmann. Influence of visual cues on head and eye movements during listening tasks in multi-talker audiovisual environments with animated characters, Accepted in Special Issue on Realism in Robust Speech and Language Processing of Speech Communication, 2018.

Llorach G, Evans A, Blat J, Grimm G, Hohmann V. Web-based live speech-driven lip-sync. In Games and Virtual Worlds for Serious Applications (VS-Games), 2016 8th International Conference on 2016 Sep 7 (pp. 1-4). IEEE.