Organ pipes never speak the same way twice.  They even change their tone slightly and continually while they speak.  For this reason randomised active generation of speech, tone, and decay is required for completely realistic pipe sound.  Systems based on the continuous replay of sampled sound can therefore never be as successful; even the incorporation of digital sound processing into such a system, to give the appearance of some randomisation, ultimately fails to be as realistic.  The entire duration of the speech of an electronic `pipe' should be generated actively, without having the initial and decay transients provided from samples; these two stages of a pipe's speech are just as active as the sustained stage.

    Analogue organs that have a very large number of tone generators (or generators of `partials') do not have to be poor.  Their dependence upon circuit components that vary slightly with temperature and loading give them some degree of continuous variation of all the `partials' constituting a tone.  There is no doubt that some analogue electronic organs can be extremely effective, as has been proved by our revoicing and re-regulation of  the large organ in Bedale Parish Church, North Yorkshire.  The very successful `presence' results from its distributed speaker system, but its tonal success derives from the continuous slight variation of the pitches of its partials.

    Active computer generation of tone uses the same principal, and the latest technology is able to generate this variation in far greater abundance across the entire set of stops and their `pipes'.  It is also able to do this for the initial speech and final decay of each pipe, which an analogue system cannot do without a quite impractically large amount of extra analogue circuitry.

    Pipes also produce completely random noise when they speak, so it is important that the tone generation system should do so also.  The very best computer active generation system will do this as well.