Quote:
Originally Posted by davvy
I have to respectfully disagree with your post. Firstly, ABX testing, while not without it's drawbacks, is the best objective method of testing that we have. If there is an audible difference between two files, surely you should be able to pick it out while listening to them.
|
Really? Remember that placebo effects are in fact based on expectations. So, a placebo effect should only be implicated if a person hears differences that he expects to hear. (The perceptual system is complex enough that there are other possible variables here besides placebo effects). But placebo effects work both ways. If we expect to hear no differences between two formats, what does a placebo effect predict, and how does ABX address this? (ABX will not restore differences that a person is not perceiving, even if the differences are real, and the absence of perceived differences is a placebo effect). If you're going to invoke placebo effects in what people report hearing, you also need to deal with their effects on what people report
not hearing.
In interpreting data, it is critical to know which inferences you can draw from any experimental design, and what the limits of interpretation of data are. In ABX testing, if a positive difference is detected, you can state with reasonable confidence that the difference is real. But this does not imply the converse. If no difference is detected, that does NOT mean that a real difference wasn't there, or that any difference is so small that it won't matter in personal listening. It simply means that it was not detected on that day in that test situation. A difference might manifest with a different test sample, different switcher, different test instructions, etc. etc. etc. or a difference might take long term familiarity before it became prominent enough to be detected with assurance. There are experimental designs and statistical methods that will allow you to draw inferences about negative results, but these experiments tend to be large and expensive to run.
Getting back on topic, my work system is an iPod driving HE60/HEV70 (Analysis Plus Silver Oval iPod cable). This rig should have adequate resolution to be sensitive to format differences. I had originally loaded the iPod using Apple lossless. However, there was a problem. The iPod filled too fast (even at 60GB), and battery charge time was low. Further, the files were not compatible with Nomad Zen or iRiver HP-140, which I also have. So, I converted everything to 320 kbps mp3. Was there a difference? Actually, I have no idea. The rig sounds good enough to me for the way that I use it, so I'm not particularly interested in paying the price for a sonic improvement in terms of space, charge time, and compatibility. Were I using it at home, for more critical listening, I'd probably be paying more attention to some sort of comparative sonic standard, but since a "better" format didn't meet my real world needs, there was no real point.
I would have a very different perspective if this system were my primary rig and was used for critical listening. In that case, my reaction to differences between lossless and mp3 would probably mirror those of the original poster, if I heard what he did (and I'd take any convenience hits necessary to maximize sound quality). In the end, it really is all personal.