Head-Fi.org › Forums › Equipment Forums › Cables, Power, Tweaks, Speakers, Accessories (DBT-Free Forum) › Question: Cheapest Cable That Has Made a Significant Difference (in your experience)
New Posts  All Forums:Forum Nav:

Question: Cheapest Cable That Has Made a Significant Difference (in your experience) - Page 4

post #46 of 48
11/24/09 at 7:59pm
Quote:
Originally Posted by nick_charles View Post
Actually it is trivially easy, though tedious but inexpensive to conduct DBTs on Analog cables. You need the following

A source with analog outputs
Two or more cables
A USB---PC recording device (I use Edirol and Behringer devices) with analog inputs
A PC
Audio recording software, including spectral analysis function (Audacity is free)
Optional (A spreadsheet program)

The protocol is easy. Connect the source to the recording device via Cable A, record several **identical samples** , trim and align to the same length, 10 samples is sufficient for each cable.

Repeat with Cable B.

You can then do direct specral comparisons, you can average differences using a spreadsheet and chart FR differences down to 100ths of a db.

For the DBT you just select two typical samples one from A and one from B and then run them through a music player with an ABX plug-in such as FooBar. Maybe someone should do this with a broad selection of different cables and make the samples available publicly....oh wait

http://www.head-fi.org/forums/f21/my...rprise-405217/



I tested silver, silver plated copper, solid copper and stranded copper cables ranging from 77c to $139 the measured differences between said cables were frankly tiny, and to date nobody has been able to tell any two cables apart in a blind test, my conclusion I now use bog standard cheap cables and have sold all my boutique cables.
Tediously easy for you because you alone decided what protocol was best for testing. In a sense you've made a whole set of assumptions about how we hear, why we hear, and what we hear and proceeded from there. But across the forums it's clear everyone has a different idea of how to go about testing. Then you've got to deal w/ issues like psycoacoustics, cognition, and other questions like should the listening material be familiar or new, what is the correct listening time, etc. That's just tip. If you're going to approach it as rigorously as other scientific research, tedious yes, but easy no. Just coming to a consensus on the setup would probably take forever.

edit -- and reading the cable enterprise thread it looks like your testing process was anything but easy! So far I've read about
lack of time, funds, and resources (can't swap, so digitized instead, raising money for different equipment?)...
Reply
post #47 of 48
11/24/09 at 11:36pm
Quote:
Originally Posted by wushuliu View Post
Tediously easy for you because you alone decided what protocol was best for testing. In a sense you've made a whole set of assumptions about how we hear, why we hear, and what we hear and proceeded from there.
It is not really necessary to speculate on internal workings we can treat the human as a black box, we have inputs and outputs, I am not normally a Skinnerian but it is a reasonable approach here, you either detect a difference or you do not.


Quote:
But across the forums it's clear everyone has a different idea of how to go about testing. Then you've got to deal w/ issues like psycoacoustics, cognition, and other questions like should the listening material be familiar or new, what is the correct listening time, etc.
Not all opinions are equally valid, sorry, but that is just how it is...there is at present no actual strong empirical evidence that familiarity helps at all for forced choice discrimination tests, through training does help, as it happens once you have done a few dozen ABXs you are getting substantail training anyway.


Quote:
That's just tip. If you're going to approach it as rigorously as other scientific research, tedious yes, but easy no. Just coming to a consensus on the setup would probably take forever.
No need, you can run such experiments in several different setups, you treat each as a set of different data points, it only becomes an issue of interest if any of the tests have different results, if all the results from blind tests in different settings end up the same...


Quote:
edit -- and reading the cable enterprise thread it looks like your testing process was anything but easy! So far I've read about
lack of time, funds, and resources (can't swap, so digitized instead, raising money for different equipment?)...
Compared to what Sean Olive does it was pretty simple, it was a bit fractured with me trying to finish up a Doctorate at the same time, but in the end I had a pretty slick protocol.

As for the digitizing, my protocol seemed able to reliably uncover actual measurable and statistically significant differences in direction of cable, which was unexpected. Roughly speaking the system was accurate to within about 0.002db, if you aggregated the samples, up to about 20K, beyond 20K it was somewhat less reliable. My question was thus, if the system could discriminate down to about 0.002db, using repeated measures then the average inter-cable differences would appear to be consistently at or below that level. Then I ask you if the inter-cable differences are at or below 0.002db across the conventional audible spectrum how likely is it that they are humanly audible ?
Reply
post #48 of 48
11/25/09 at 1:59am
Quote:
Originally Posted by nick_charles View Post
It is not really necessary to speculate on internal workings we can treat the human as a black box, we have inputs and outputs, I am not normally a Skinnerian but it is a reasonable approach here, you either detect a difference or you do not.




Not all opinions are equally valid, sorry, but that is just how it is...there is at present no actual strong empirical evidence that familiarity helps at all for forced choice discrimination tests, through training does help, as it happens once you have done a few dozen ABXs you are getting substantail training anyway.




No need, you can run such experiments in several different setups, you treat each as a set of different data points, it only becomes an issue of interest if any of the tests have different results, if all the results from blind tests in different settings end up the same...




Compared to what Sean Olive does it was pretty simple, it was a bit fractured with me trying to finish up a Doctorate at the same time, but in the end I had a pretty slick protocol.

As for the digitizing, my protocol seemed able to reliably uncover actual measurable and statistically significant differences in direction of cable, which was unexpected. Roughly speaking the system was accurate to within about 0.002db, if you aggregated the samples, up to about 20K, beyond 20K it was somewhat less reliable. My question was thus, if the system could discriminate down to about 0.002db, using repeated measures then the average inter-cable differences would appear to be consistently at or below that level. Then I ask you if the inter-cable differences are at or below 0.002db across the conventional audible spectrum how likely is it that they are humanly audible ?
I'll save my questions for the enterpruise thread.
Reply
New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Equipment Forums › Cables, Power, Tweaks, Speakers, Accessories (DBT-Free Forum) › Question: Cheapest Cable That Has Made a Significant Difference (in your experience)