Friday, January 5, 2007

VoIP Codecs

Coders/decoders (Codecs) are used by VoIP networks for converting analog voice signals into digital pulses and then reconvert the digital pulses into analog signals. In order to communicate, the codecs have to be compatible with each other. The algorithms that the codecs use for conversion of the data streams affect the quality of voice as well as the bandwidth consumption.

The solutions to algorithm usage are either proprietary or covered by international standards, where everyone has access to the algorithms. Pulse Code Modulation (PCM) was responsible for the development of the T-carrier systems that are used even today. PCM could yield a data rate of 64 Kbps. The signal was sampled in two ways, Mu-Law in the US and Japan and A-Law in Europe. Both these forms of sampling allowed for a high resolution as the discrete levels were apportioned logarithmically and not linearly.

Recommendation G.711 has been instituted by the ITU in 1988 and is the standardized form of the PCM encoding. PCM does not eliminate the redundancy in the signals, which can result in a high data output rate unsuitable for certain situations, especially when there is a bandwidth constraint. This is the reason why several speech algorithms have attempted to reduce the data rate. Reduction in data rate by half can double the call-carrying capacity of the given bandwidth. G.722.1, G.723.1, G.726, etc are codec standards that reduce the bandwidth requirements. Their data rates are 24/32, 5.3/6.3, 16/24/32/40 Kbps, respectively.

Apart from these open standards, there are proprietary algorithms that may or may not offer an advantage over the ITU-defined algorithms. However, they can tie a business to their implementation for the economic life-cycle of the VoIP system.

No comments: