In my last post, I suggested a definition for ‘coloration-independent characteristics’:
A coloration-independent characteristic is sonic characteristics of a component/system that is:
(1) VARIABLE, in the sense that multiple values of the characteristic are possible, and
(2) COLORATION-NEUTRAL, in the sense that, for at least a limited range of values, differences in the value of that variable have either (a) no effects or (b) identical effects on the concealment and corruption of information about the music.
Cbw wrote:
I'm not sure I agree with #2. We've already identified resolution as existing outside of neutrality/coloration, but it would not pass part b of this test, because low resolution would conceal information.
Cbw’s objection to my definition of ‘coloration-independent characteristics’ is implicitly an objection to my definition of ‘coloration’:
COLORATION: Additions or subtractions to the playback chain that conceal or corrupt information about the music.
The apparent problem identified by Cbw with my definition ‘coloration’ can be expressed in the following argument:
(1) Resolution loss is not a type of coloration.
(2) Resolution loss satisfies my definition of ‘coloration,’ since it is a subtraction in the playback chain that conceals information about the music.
(3) Therefore, my definition of ‘coloration’ is flawed, and by entailment, so is my definition of ‘coloration-independent characteristics.’
My reply to this is that premise (1) is partly true, partly false. That is to say, some resolution loss is a consequence coloration and some is not. What I would like to propose is that resolution loss can be thought of as falling into three types...
THREE TYPES OF RESOLUTION LOSS:
(1) CONCEALMENT of information about the music.
(2) CORRUPTION of information about the music.
(3) ELIMINATION of information about the music.
The loss of information through concealment or corruption is resolution loss BY COLORATION. The loss of information through elimination is RESOLUTION LOSS PROPER.
CONCEALMENT is the kind of loss that results, for example, from comb filtering, where some frequencies are exaggerated, others attenuated, by constructive and destructive interference. CORRUPTION is the kind of loss that results, for example, from intermodulation distortion, where spurious frequencies are added to an amplified signal. And ELIMINATION is the kind of loss that results, for example, from the informational compression of an MP3.
I chose the words “conceal or corrupt,” rather than “eliminate” for my definition of ‘coloration’ precisely because I was hoping to define ‘coloration’ in a way that does not entirely subsume the concept of ‘resolution’ under the concept of ‘neutrality,’ which would not reflect the usage of those concepts by audiophiles, or their usage on this thread. My definition of ‘coloration’ forces me to acknowledge that some resolution loss is a consequence of colorations, but it allows me to preserve a concept of ‘resolution’ that is INDEPENDENT OF the concepts of ‘coloration' and 'neutrality,' thereby addressing Cbw’s concern with the definitions of ‘coloration’ and ‘coloration-independent characteristic.’
Another motivation for differentiating CONCEALMENT from CORRUPTION from ELIMINATION is the idea that these three types of information loss have different (1) TYPICAL LOCALIZATIONS; and (2) TYPICAL DEGREES OF RECOVERABILITY:
TYPE OF LOSS.......LOCALIZATION..........RECOVERABILITY
(1) Concealment........Listening room................Easy
(2) Corruption...........Equipment.......................Difficult
(3) Elimination..........Format............................Impossible
Information loss through CONCEALMENT, as typically happens in the listening room, includes phenomena such as room modes, flutter echo, and comb filtering. The information concealed by each of these phenomena is contained upstream in the system. The information is being CONCEALED by the listening room (Or more accurately, by the physical relations among the listening room, the speakers, and the listener). Because of this, the lost information is relatively EASY TO RECOVER. It can often be achieved with modest room treatments (as in the case of flutter echo) or change of speaker position (as in the case of comb filtering).
Information loss through CORRUPTION, as typically happens in equipment, includes phenomena such as intermodulation distortion, crosstalk, and speaker cabinet resonance. Once again, the information concealed by these phenomena is contained upstream in the system. The information is being CORRUPTED by the equipment. Because of this, the lost information is MORE DIFFICULT TO RECOVER. It can sometimes be accomplished by modifying components.
Information loss through ELIMINATION, as typically happens in the format, includes phenomena such as the informational compression of an MP3. The lost information does not exist upstream in the system. Hence, it is UNRECOVERABLE. The only way to get this information into your system is to change components, and possibly formats.
It is important to point out that I am proposing these three types of information loss as IDEALIZED CATEGORIES, in the sense that the loss of information in any real-world system will merely RESEMBLE these idealizations. Having said that, I believe these categories are valuable to the audiophile, insofar as they help him conceptualize what is wrong, where it went wrong, and how hard it will be to fix it.
