Does cables matter, a small self-test

Whether there is an audible difference between cables is an eternal debate among audiophiles. Here is a short self test for you.

First, start by downloading this file – the archive contains three FLAC files. While downloading, continue reading this post (It’s not really possible to TL;DR it, I’m afraid):

This archive contains three different recordings in FLAC format of the Nine Inch Nails’ song “Discipline” from the album “The Slip” (NiN is chosen because they are one of the few bands that release their stuff under a permissive license, in this case CC-BY-NC-SA

The first file is made by recording the output from a Focusrite audio interface, using an RCA cable that has been cut in half, and spliced together by hand (no soldering, just twisting the wire together), as shown in this image:

RCA cable cut in half, and spliced together by hand

The specific cable is a 1m RCA cable from Monster Cable, bought for about $5 in a bargain bin in the mid-90’s, and is connected from output to input using no-name RCA -> 6.3 mm (1/4 inch) TS-to-RCA adapters.

The second file is made with a Safecon AC36/5M cable from Morgan Instrument TS cable with 6.3 mm (1/4 inch) gold-plated TS connectors.

The third file is the original recording.

Each of the recorded files have randomly been assigned a name “Specimen A”, “Specimen B” and “Specimen C” and does not necessarily reflect the order in which they are mentioned in this document

The files were recorded in the following way:

First, the cut/spliced Monster cable was first connected to the “Line 3″ input on the Focusrite 8i6 interface, while the Morgan cable was connected to the “Line 4″ input, and the song was played back and recorded in to Reaper 4.32 on OS X. The left and right channels were imported in to separate tracks

Next, the cables were swapped, and a new recording was made, recording the Morgan cable to track 3, and the broken Monster to track 4.

Then, each of the “broken cable” recordings (Line 3 from the first recording, and line 4 from the second) were imported in to Audacity (version 2.0.2), assigned to left/right channel as needed, and aligned with the original file, so the beginning and end of the file matches.

Likewise, each of the Morgan cable recordings (Track 4 from the first, and track 3 from the second recording) were combined into a single file and aligned.

(Note that a tiny bit of audio has been trimmed from the end of the file, since this is a much easier way to align it.)

The Focusrite interface by default has -10 dBFS gain on the rear line inputs, and so the volume was normalized to match that of the original recording for each of the copies, so they can be directly compared against the original file, and the resulting file was then exported as a 16-bit 44.1 KHz WAV file to match the original.

Finally, the files were converted to FLAC.

That was the where my work ends, and yours begin – it’s time for you to do an ABX test. If you don’t have an ABX tool, here are some examples:

  • OS X: ABXer – Download
  • Windows: foo_abx component for foobar2000 – Download

Note that I don’t know of an ABX tool for Linux, so if you know one, please post it in the comments – alternatively, foobar2000 and the foo_abx component is known to work in WINE.

To perform the self test, do three test runs:

  1. Compare A vs B
  2. Compare A vs C
  3. Compare B vs C

For each of the tests, you need to run ten trials – post the test logs in this thread. If you don’t want to post the test logs, but want to know if you can hear the difference: The generally accepted criteria for a “positive” result is that out of 10 trials, you need to have identified the “X” correctly nine times.

For fairness sake, and to not introduce bias: I have not attempted to identify the files through a blind test myself, and will not divulge what I think.

If you run the test, please post your results below – full test logs are preferred, but if you don’t post the full log, please include the following info:

  • Number of trials for each of the three tests. Number of correct responses.

If you are absolutely certain you can hear a difference (This basically means that you need to be correct on 9 out of 10 responses) on any of the three tests, feel free to theorize about which recording is which

Leave a Reply