The usual practice is to start the generation of tone at a high energy level, so a strong signal is available at the receiver, for crisp tone detection. If the tone remains on for a significant period, the energy level is reduced, to minimise crosstalk. During the signalling transitions, only the tone is sent through the channel, and the media signal is suppressed. This means the signalling receiver has a very clean signal to work with, allowing for crisp detection of the signalling tone. However, when the signalling tone is on for extended periods, there may be supervisory information in the media signal, such as voice announcements. To allow these to pass through the system, the signalling tone is mixed with the media signal. It is the job of the signalling receiver to separate the signalling tone and the media. The necessary filtering may degrade the quality of the voice signal, but at least supervisory information may be heard.