Complete Communications Engineering

Comfort Noise Generation (CNG) is used in vocoders to provide a level of background noise with which people have become accustomed using older telephone services. The absence of signal energy during a conversation can be disconcerting to the person listening to the signal. Due to the long time use of the Plain Old Telephone Service (POTS), people are used to having a certain level of background noise and static. If this low level noise is not present when there is no speech energy, our experience with POTS leads us to believe that the call has been disconnected. Thus our conversations would be full of “Are you still there?”

In speech coders, there are several different reasons why a frame may not contain any information. Many vocoders have a discontinuous transmission (DTX) feature where frames are rarely coded when there is no speech present. Also, if there is little energy in the encoded signal, it may happen that because of the encoding process, there is almost no energy in the decoded signal. Additionally, if a packet is lost when there is little speech energy present, it can lead to the Packet Loss Concealment (PLC) feature creating a frame of almost no energy.

In these situations, Comfort Noise Generation  is used to create the missing noise. CNG detects whether there is signal energy present. If there is, all is good and nothing needs to be done to the decoded signal. On the other hand, if the signal energy falls below a certain threshold, the CNG is activated which generates low level background noise to fill this auditory void. This noise is added to the original signal, which, in the case of a lost packet, can help mask the lost packet. Any discontinuities from the lost packet blend into the random noise.