Audio Coding

What is coding and why is it required to transmit audio over ISDN?
Without data rate reduction, high-quality audio requires a transmission capacity of about 700kbps for each audio channel. Channels that can handle data rates that high are very expensive and hard to get. More affordable and accessible channels, such as the two 64kbps channels in each ISDN circuit, offer a rate of only about 9% of that of a compact disc. That means you must do some coding to get "12 gallons of water into a one-gallon container." (Note that some refer to coding as "compression" but are generally referring to the same process.)

• Facts about MPEG4 AAC
• White Paper:MPEG AAC
• Facts about MPEG Layer
2 & Layer 3

How can coding be accomplished?
One might think that lossless, redundancy-reducing methods (such as those used for computer hard-disk compression) would be ideal for audio. Unfortunately, there is not enough redundancy in the audio signal for the significant reduction required by ISDN.

Early coding schemes, such as Adaptive Delta Pulse Code Modulation (ADPCM), take advantage of the fact that it takes fewer bits to code the difference, or delta, between successive audio samples compared to using their individual values. Further efficiency is had by adaptively varying the difference comparitor according to the nature of the program material. However, the reduction power of ADPCM is insufficient for transmitting full bandwidth audio over ISDN, as evidenced by 7kHz codecs that use G.722.

To develop coding algorithms with sufficient power to achieve the desired reduction, the audio industry has turned to psychoacoustics. Using carefully researched psychoacoustic principles, coding processes have been designed to reflect the way in which human hearing interprets audio information.