In audio, if two people. A & B, do a listening comparison and A hears a difference while B hears none, then we know two things: there is a difference, and A is the more sensitive listener.
Of course B knows his own experience better than anyone else does, but A knows the reality better than B.
A scientific approach puts all sorts of caveats on the process by which A and B do their listening, but it does not affect the fundamental idea. And while I see nothing wrong with a scientific approachit would be odd if I did, teaching at Caltech as I doI do see something wrong with rejecting out of hand, prior to a formal test, the considered judgment of those who work in a field.
Do we ask the wine expert whether his rating of each and every wine was made under double-blind conditions? No, we accept his judgment, at least provisionally, once we know that he does indeed have a reasonable claim to being called an expert.
So when two audio professionals offer their judgments on digital processing, one saying that it degrades the signal he feeds into it and the other saying it does not, why not accept the idea that they are both reporting their perceptions honestly? Of course, if they are doing so, then as I pointed out above, the one who does hear a difference is likely to be closer to the truth. I expect that this is the sticking point for many proponents of digital audio, and the giveaway is the personal attacks one reads on those who hear problems with this new technology.
Moving on to the question of listening tests themselves, a point that is got wrong all the time is what you learn from a test in which no one can hear a difference. Such a result does not mean that there is no difference! It means that either there is no difference or your system has too little resolution to reveal the difference, or all your listeners are B's. Assuming that you have a good number of listeners, it's unlikely that they're all B's; so a truly scientific approach then turns to improving the system, in case its limitations are limiting the resolution of the test.
This is a major hidden point in the debate over digital audio. In our listening test at Caltech, we had a quite high-resolution microphone feed as our source; and I can attest that Sheffield Lab's system, which Doug Sax uses as a reference, has resolution undreamed of even by most professionals in audio. One can say high-resolution ‘til one is blue in the face, but it is a vacuous concept for most people, and understandably so; for where would they have heard that combination of power, delicacy and beauty?
Well, in live music, of course, the very thing we're trying to reproduce; and that's why a live microphone feed is the appropriate source for listening tests rather than degraded material stored on tape, disc or Compact Disc. But when Sax talks about high resolution, it's as though he were thinking of 8x10 view-camera photographs while some of those who disagree with him were thinking of snaps from a Brownie; or as though he were thinking of 35 mm film and they were thinking of ordinary television.
The burden of these burdensome remarks, then, is first, that if you do a careful test and no one hears differences, maybe the system has too little resolving power; and second, if certain people do hear differences, they may well be right!
|These sensitive people are a gift to the rest of us. If they use their sensitivity in creating with food, we call them master chefs; while those who consume the food with discrimination we call gourmets. If they create in paints, we call them artists, while those who appreciate the artists' work sensitively are called connoisseurs. But turn to audio and they are disparagingly called golden-ears or tweaks, and letters to the editor [1997: and Usenet postings!] are written impugning their intellectual honesty. Can you imagine Brillat-Savarin being called a food tweak?|
James Boyk Home
James Boyk's articles