Dataset of audiovisual unscripted triadic conversations

Subject: This dataset contains audiovisual recordings of unscripted triadic conversations in German for acoustic and linguistic research. Data collection: Subjects (1f/2m) produced free conversations on self-chosen topics. Video Recordings were done using three Canon “EOS 700D” cameras (Firmware version 1.1.4), synchronized using linear time code signals. Audio recordings were performed using three microphones “Neumann KM-184” close to each speaker, an RME Micstasy microphone preamplifier and A/D converter. Camera set up such that subjects face the camera. Face and shoulders visible, face at the center of the video. Video (.mp4): video has a framerate of 25 fps, bit rate of 640 kb/s, audio has a sampling rate of 48 kHz and bitrate of 129 kb/s, 2 channels Audio (.wav): sampling rate of 48 kHz, 24 bits per sample, 3 channels (one mic for each speaker). ch1: male speaker, red t-shirt ch2: male speaker, blue t-shirt ch3: female speaker Date of data collection: May 2023 Location: Carl-von-Ossietzky university of Oldenburg. building W30, Gesture Lab File naming convention: For each take, video and audio data are stored in separate files for each take. The file naming convention is as follows: v<take_num>.<extension> take_num: running integer index indicating the take number. extension: taking the value of .mp4 or .wav Data manipulations: To ensure synchronized video information across speakers and recording takes, the separate speaker videos were spatially merged into a single video file by arranging them side by side on the horizontal plane. Video editing was performed using DaVinci Resolve. Audio editing was performed in Ardour. Funding: This research was funded by a research grant from the German Research Foundation (Deutsche Forschungsgemeinschaft; SPP 2236 project 444761144, and Deutsche Forschungsgemeinschaft; project-ID 352015383 – SFB 1330-B1), and by the Cluster of Excellence “Hearing4all.connects”, funded by the German Research Foundation (Deutsche Forschungsgemeinschaft; under Germany's Excellence Strategy – EXC 2177/1 - Project ID 390895286).