The HG 35xx gateways support VAD (Voice Activity Detection).

More detailed information on Voice Activity Detection (VAD)

Voice Activity Detection (VAD) serves to save bandwidth in the IP network during pauses in speech.

Based on the knowledge that only one partner usually speaks at a time, while the other partner listens, the bandwidth seized during a call is only really used in one direction (speaking station -> listening station). In the other direction (listening station -> speaking station), only ambient noise and “silence“ are transmitted. Voice activity detection makes use of this knowledge, interrupts the data transmission and drastically reduces the data rate as soon as silence is detected. As soon as voice activity is detected again, transmission for that direction is immediately re-established (with full bandwidth).

If NOTHING is transmitted on a route, the receiving end generally tends to identify the absence of background noise as line failure and therefore as undesirable. Therefore, depending on the codec used, highly compressed background noise is sometimes transmitted or noise is supplied.

The problem with VAD is the difficulty in distinguishing between ambient noise and silence in transmission. Ambient noise in the room should fall under the category of silence, but softly spoken words should be categorized as voice. The impairment (or in the worst case, truncation) of the beginning or end of voice activity in the transition area is unavoidable.

Linking is also a problem with VAD. Using VAD on multiple transmission paths in succession can increase impairment effects. The linking of transmission paths with VAD must therefore be avoided.

The resulting bandwidth reduction is in the range of 50% to 75%.

The VAD classmark is deactivated by default, but can be activated at any time for voice connections.