At a video conference endpoint including a microphone array and a camera, different camera framings are established to frame different views of a talker based on different sets of pan, tilt, and focal length settings of the camera. Different video frames of the different views are captured using the different camera framings, respectively. A sound source direction of the talker relative to the microphone array in a fixed three-dimensional (3D) global coordinate system is determined for the different views based on sound from the talker detected by the microphone array. The sound source direction relative to the microphone array is converted to different sound source positions in planar coordinates relative to the different video frames based on the different sets of pan, tilt, and focal length settings, respectively. The different video frames, the sound, and the different sound source positions in planar coordinates are transmitted.