The HG 35xx gateways support voice compression following standard G.711A encoding at 64Kbps. Compression according to G.729A and G.729AB is supported as well. Also supported codecs are G.711U, G.729 and G.729B.

Of course the actual bandwidth requirements per connection depend on the added IP overhead and various other settings (see Section 2.5.4, “Required Bandwidth per Connection”).

More detailed information on voice compression

Voice compression uses complex algorithms to reduce the bandwidth of the voice signal, for example G.729, to an eighth of the bandwidth required by the original PCM signal. Impairment of voice quality is only minimal in this case.

Voice compression can be problematic when it is applied a number of times in succession. In other words, the original PCM signal is compressed using a specific method, transmitted and then expanded back to a standard PCM signal, then the whole procedure is repeated on another section. In the worst case scenario, with another procedure. Applying voice compression a number of times, particularly using different compression methods, severely impairs voice clarity and in the worst case scenario it can result in unintelligibility.

The linking of compressed transmission segments - particularly using different methods - must therefore be avoided wherever possible.

Given that all digital speech memory systems for announcements, voicemail etc. also compress speech, this type of equipment must be monitored carefully when used with compressed transmission paths.

Within the OpenScape 4000 system, connections to announcement/music on hold devices as well as to conference units are configured without compression in order to guarantee undistorted music or good voice clarity for conferences.

The factors specified in association with voice compression, such as compression to an eighth of the originally required (uncompressed) bandwidth, for example, always refer to the actual voice data. If this data is transferred using IP, the IP packaging, the “protocol overhead“, must also be taken into account. The voice data to be transferred is reduced by the specified factor by means of compression, but the overhead remains the same. The bandwidth required in the IP network is therefore never reduced by the specified compression factor, but instead by a lot less. Always take the values specified in Chapter 3, “Load Calculation” in the document “IP Distributed Architecture (IPDA)“ into account when dimensioning the bandwidth.