The Cardiff Conversation Database (CCDb) is a collaborative project between researchers at Cardiff University, Brandenburg Technical University, and Korea University. It is the first high-quality, multi-modal, non-scripted database of natural conversation between two people. Since the participants’ roles in the conversation are not fixed, they repeatedly swap back and forth between speaker and listener. 

The database currently consists of 44 conversations between pairs of speakers, each lasting approximately five minutes. There are 16 speakers in total, 12 male and 4 female, between the ages of 5 and 56. Thus, some speakers were paired multiple times with different conversational partners. Many conversations have annotations for facial expressions, verbal and non-verbal utterances, and transcribed speech (see below).

The dataset is useful for perceptual experiments as well as computer vision and computer graphics tasks, including for usage in machine learning tasks. 

Data acquisition
The full database includes audio, 2D video, and 3D video. Two 3dMD dynamic scanners (one for each conversational partner) captured the 3D videos, two Basler A312fc firewire CCD cameras captured 2D color video at standard video frame rate, and a microphone placed in front of the participant, out of view of the camera, captured sound (at 44.1KHz).

To ensure all audio and video could be reliably synchronized, each speaker had a handheld buzzer and LED (light emitting diode) device, used to mark the beginning of each recording session. A single button controlled both devices and simultaneously activated the buzzer and LED. No equipment was altered between the recording sessions, except for the height of the chair to ensure the speaker's head was clearly visible by the cameras.

Database access:
The 2D video can be found at huggingface 

Core Publication:
 If you use the dataset, please cite: 

Aubrey, Andrew J., David Marshall, Paul L. Rosin, Jason Vendeventer, Douglas W. Cunningham, and Christian Wallraven. "Cardiff conversation database (ccdb): A database of natural dyadic conversations." In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp. 277-282. 2013.

Derivative Databases and Annotations:

The CCDb Head Gesture (CCDb-HG) contains annotations for each frame with 6 head gestures: Nod, Shake, Tilt, Turn, Up/Down, and Waggle.

The Interaction Behavior Database contains annotated segments regarding smiles and laughs as well as their intensities, expressed by interlocutors in conversational contexts. They also contain annotation segments of the interlocutors' roles (speaker, listener or none) during their conversations