Digital processing

From Wikipedia, the free encyclopedia

Digital processing is the process of altering digital data in any form.

The most common situations where digital processing is involved are computer graphics and digital audio processing. The two are related regarding both the problems encountered and the solutions provided by the respective industries.

[edit] Overview

The digital data model was one of the greatest mainstream inventions of the 1980s, among other qualities, because it enabled artists to arbitrarily copy and alter their data losslessly throughout their respective creative processes, as opposed to analogue processing which was inherently lossy. The digital data model, throughout its various incarnations, is based on the idea of digitizing analogue data, processing and replicating the digital data, and finally re-compiling analogue data based on the available digital data (as in printing a digital image or playing a song originally stored in digital format).

Digitization of an analogue signal implies the existence of a conventional digital format which can emulate the analogue format of the original source. The digital format must necessarily consist of discrete signal levels which represent pre-determined levels of signal within the original analogue source.

Processing any digital signal involves manipulation of the discrete values describing the data stream. Since the data stream is discretely separated in distinct values, practically any procedure which involves complex manipulation of the digital data results in rounding real values to the discrete values used in the respective digital format. The effect of any such processing is a reduction in signal entropy because distinct signal levels end up being rounded to the same discrete value, as to match the pre-determined discretization model.

The final effect of almost any kind of digital data processing is rougher discretization of the resulting values compared to the discretization of the input signal. Depending on the type and intensity of the processing, the roughness of the resulting signal can be noticeable to a human operator, in the sense that it can end up noticeably degrading the quality of the resulting material.

Since relatively many industries rely on heavy digital processing of sound and image, some workarounds had to be found to circumvent the problem described above. The solution is to capture the original signal (sound or image) with a much greater resolution than the target specification. Processing of the respective signal is also performed using considerably more discrete values than the average human can notice. That way, even processes which considerably damage the signal still keep it within reasonable levels of resolution, in such a way that the signal deterioration throughout processing goes unnoticed for the average consummer.