[78-L] Cambridge Process

[Chris Zwarg]

At 17:50 16.12.2008, you wrote:
>This raises an interesting point. Collectors who have spent a great  
>deal of time listening to acoustic records end up with trained ears  
>that benefit from the fact that except for echo, size of the horn and  
>distance from the horn, the sound doesn't vary very much.

Do you think so? Sorry but you are obviously not listening (or you are *listening acoustically* in which case the uniform coloring is due more to the playback horn).

It's true that all acoustic recordings have a distinct peak somewhere in the lower midrange and a response petering out towards both ends (top and bottom frequencies), not continuously however but in form of "ripples" much like a comb filter. Both the actual frequency of the peak, the width and depth of the ripples, and the steepness of the falling-off of "extreme" frequencies (we're talking 100 to 4,000 Hz here) varies wildly however, often even between adjacent matrices cut during the same session, as the system - much like a wind instrument - was very susceptible to changes in room temperature and humidity, not to mention numerous little "tweaks" employed by the engineers, ranging from the use of differently-shaped horns to carefully varying the diaphragm tension.

The difference in average level at both ends of the useful response (i.e. frequencies that are cleary audible above the noise) against the main resonance peak is usually 20 dB or more, so finding the peak(s) and toning them down (resp. bringing up the weakened top and bottom) is just as important for proper reproduction as the EQ curve is for electrics. The great difficulty is that you have to find the maxima and minima by spectral analysis for every record separately.

Chris Zwarg