mike1127
Member of the Trade: Brilliant Zen Audio
- Joined
- Oct 16, 2005
- Posts
- 1,114
- Likes
- 25
I have a new idea for testing the audibility of premium cables.
I have a personal theory that a key factor of quality is a system's impulse response, and more specifically its ability to present musical meaning by clarifying patterns of transients. And it's possible that the most important property of a cable is its effect on transients.
So I had the idea to use a test signal rather than music. I want to use an impulse train.
Yes! Me! Using signals instead of music!
Okay, here's the idea. I create two versions of a test signal. One of them is an impulse train with the impulses occurring ever 50 ms. The second is a modification of this. It's modified by moving some of the impulses slightly in time. For example, it might become
<impulse> (47 ms delay) <impulse> (53 ms delay) <impulse> .... pattern repeated
That is, every other impulse comes 3 ms earlier.
What's neat about this test is that I can write software that creates a series of signals A or B, without telling me. Then I can listen and try to guess each one, and compare it to a file that the software saved with the right answers.
So far I've tried this experiment with a shift of 4 ms, and a shift of 3ms. I burn a CD and play it through my good CD player and cables. With 4ms I get almost every trial right. With 3ms I usually get about 7/8 right. What I want to do is find the smallest shift that is audible with the good cables. Then see if it's still audible with the cheap cables.
Initially I can do this test without any help---I'll know which cables are hooked up. Note that the identity of each test signal, A or B, is unknown to me while I am listening. So that part of it is blind. However, later I can do it "blind" both to the identity of the cable and the specific test signal.
If I tend to do much better identifying A/B with the good cables, that's good evidence. Now, analyzing this mathematically to determine the confidence level is going to be a little tricky. It's no longer a simple, standard function.
I have a personal theory that a key factor of quality is a system's impulse response, and more specifically its ability to present musical meaning by clarifying patterns of transients. And it's possible that the most important property of a cable is its effect on transients.
So I had the idea to use a test signal rather than music. I want to use an impulse train.
Yes! Me! Using signals instead of music!
Okay, here's the idea. I create two versions of a test signal. One of them is an impulse train with the impulses occurring ever 50 ms. The second is a modification of this. It's modified by moving some of the impulses slightly in time. For example, it might become
<impulse> (47 ms delay) <impulse> (53 ms delay) <impulse> .... pattern repeated
That is, every other impulse comes 3 ms earlier.
What's neat about this test is that I can write software that creates a series of signals A or B, without telling me. Then I can listen and try to guess each one, and compare it to a file that the software saved with the right answers.
So far I've tried this experiment with a shift of 4 ms, and a shift of 3ms. I burn a CD and play it through my good CD player and cables. With 4ms I get almost every trial right. With 3ms I usually get about 7/8 right. What I want to do is find the smallest shift that is audible with the good cables. Then see if it's still audible with the cheap cables.
Initially I can do this test without any help---I'll know which cables are hooked up. Note that the identity of each test signal, A or B, is unknown to me while I am listening. So that part of it is blind. However, later I can do it "blind" both to the identity of the cable and the specific test signal.
If I tend to do much better identifying A/B with the good cables, that's good evidence. Now, analyzing this mathematically to determine the confidence level is going to be a little tricky. It's no longer a simple, standard function.