Abstract: A system and device that enables a security officer on site or an agent in a remote central station, to speak to, and optionally receive speech from, a potential intruder observed or otherwise detected in the field of view of a particular one of a multiplicity of surveillance cameras at the site. This is achieved by locating speakers and optionally microphones at each camera, and locating an audio unit at the camera head unit in communication with the speakers, by which the security officer or central station agent can select and speak through a particular speaker, thereby warding off an intruder who is near the particular camera. If the speech communication is two-way, the officer or agent can interrogate the potential intruder to determine if access to the site is authorized.