Examples are disclosed for responding to voice commands within an existing telephony call between two or more end user communication devices mediated by an IP based communications server. The server is mediating a telephony call among the two or more end user communication devices where each end user communication device has its own communication link to the server. The server may detect a trigger event from one of the end user communication devices. Upon detecting the trigger event, the server may receive voice input over the communication link from the end user communication device that generated the trigger event. The received voice input may be parsed into one or more voice commands. The server may then cause the one or more voice commands to be executed. In other embodiments, much of the intelligence and processing may be carried out on the end user communication device directly.