There is described method and apparatus for processing signals of a telephonic communication, where the signals represent sensitive and non-sensitive information. A described method comprises processing the signals to provide a first version of the signals that is to be recorded and a second version of the signals that is to be output as audio; monitoring at least the first version of the signals to detect, in said at least the first version of the signals, one or more instances of one or more predetermined characteristics that represent the sensitive information conveyed by the signals; and modifying said at least the first version of the signals by removing the identified predetermined characteristics from said at least the first version of the signals.