Abstract: Echo suppressor break-in circuitry must be able to distinguish between double-talking speech on one path of a transmission channel and echos on that one path due to speech on the other path in order that suppression can be removed when both parties are talking. The present disclosure relates to an improved method and arrangement for approximating the echo that may result on an echo return path from speech on another path, which approximation may then be compared with the actual signals on the return path to determine the presence of double-talking speech. In an illustrative embodiment, a first memory store arrangement follows increasing magnitude signals and, during decreasing magnitude signals, provides an output signal corresponding to a predetermined decay rate. A second memory store arrangement follows increasing signals, less a predetermined loss, and holds the magnitude of the last peak while the signal is decreasing.