For audio the most basic forms of data compression /reduction involve reducing the number of bits and sampling rate of the audio. Lowering the sample rate reduces the high frequency content of the original while reducing the bits per sample lowers the fidelity.

Adaptive Differential Pulse Code Modulation

A few years ago a data compression algorithm (mathematical process) for audio was introduced called ADPCM (Adaptive Differential Pulse Code Modulation) This used a series of predetermined mathematical tables as a reference so that positional information about the waveform could be stored as smaller numbers than the 16 bit equivalent. By trying to predict the position of the next sampling point based on the previous few sample points and comparing that to the actual position, ADPCM allowed consecutive data to be stored in a positional manner, rather than as absolute data. These methods reduce the amount of data stored from 176 KB for stereo 16 bit 44 kHz CD quality audio to 2.25KBs for mono ADPCM 11 kHz 8 bit audio. [25]

While quite successful as a compression method, ADPCM did suffer from quite noticeable distortion, whistles and whine known as artefacts. Data compression ratios of between 4:1 and 32:1, in real terms, are available using ADPCM but degrade the original waveform quite significantly. This was until recently the most commonly used audio compression algorithm within the computer industry.

Back: Bandwidth problems | Contents | Next: MPEG audio compression