Do Cables/Components Make a Difference? Here's How to Test for Yourself

msg

TL;DR Version
Use Audacity, a laptop, and an RCA/3.5mm cable to record a baseline and difference files to demonstrate audible differences in component changes.

This is long, and I spent a good portion of the morning/afternoon pulling this together and testing it in rudimentary fashion. This is as far as I'm going to go today, given that it is time consuming, and will, for me, only satisfy a mere curiosity re: the cable debate, essentially proving what I've already learned through personal experience - everything matters. This is for those who say cables don't make any difference or enough of a difference, for those who say, "Proove it." And also for those satisfied systems owners who may be curious as to just what more they're getting out of system improvements. This is a method for recording audio from your system demonstrating just how much extra you're getting out of your system component improvements.

========================================
Do Cables/Components Make a Difference? Here's How to Test for Yourself
In a recent thread, "Do Power Cables Make A Difference", @mikeyb128 referenced an article he read detailing an interview with AudioQuest's power products designer, Garth Powell. Here's the article.

In this article, author Guatam Raja poses the question to Garth Powell, "How do you back up claims of improvement?" This is a clip from that article generally summarizing a method of providing that proof by way of an actual audio file that plays the additional audio information missing without a particular component, in this case, one of AudioQuests Niagara power conditioners. It's called a "difference file".

...Garth stresses that he is an engineer, and all his claims can be measured. But how? It turns out to be elegantly simple: a difference file. Play a track on a digital system with stock power cables from the wall, and digitally record the output. Then add audiophile power cables or a power conditioner, and record the same track. On a computer, align the recordings perfectly so they start and stop at exactly the same point. Flip the phase on one track and add them together. What you get is a difference file—if the two files are identical, the result would be total silence. If the power cable or conditioner is making a difference (hopefully an improvement), the difference track should be playing all the information that was masked in the stock playback.

I love this idea, and got to thinking how I could set this up for myself. It's actually not that difficult. Here's how one can accomplish this with not much more than the audio system already you're already listening to.

REQUIRED

• Audio System for Testing (playback stage)
• Computer/Laptop (with mic/line in for recording stage)
• Audacity (free recording software)

msg

Okay, so the idea here is that we will use a computer (or a laptop if the source is a computer) with the Audacity audio editing software installed to record the output signal from a preamp or processor. This allows us to record a baseline, then make changes to the system configuration as one desires, and create additional recordings for comparison against the control/baseline recording. These changes can be anything from power source, power transport, component changes, cable changes, different DACs, even the file format used for playback - say DSD vs .flac vs .mp3.

In the Audacity software, we employ the Difference File/Phase Inversion method in order to reveal audible changes due to these various system configuration adjustments. I'll come back to this later.

This is not a difficult setup for one to try personally with the free audio editor, Audacity, and a computer. Ideally, one would want to go with a preamp/processor that offers digital output for the greatest accuracy, however not a lot of preamps seem to offer digital audio out, if any, though most AVR processors seem to.

In this test setup with my own particular equipment on hand, as an example, I can test the path from the power at the wall, up through the output stage from the processor. However, I do not have a computer with digital audio in in the form of digital coax or optical, so I would have to switch to an analog RCA/3.5mm cable in order to record. Given that this would be a relative constant in each of the tests, for this purpose, I think it would be "close enough" to reveal audio quality changes of various components in the chain up through that point. I've personally used the RCA/3.5mm cable previously in order to record (and demonstrate for a seller a series of blemishes on) a rare vinyl record I purchased by connecting the RCA/3.5mm stereo cable from a phono preamp to a laptop, and recorded the audio in Audacity.

Here are some examples of paths and components that one can test for with this method:

• Power Source > Power Cable > Preamp
• Power Source > Power Cable > Source/Transport
• Source/Transport > Digital Interconnect > Preamp
• Preamp > Analog Interconnects > Amplifier << NOT TESTABLE WITH THIS METHOD
• Power Source > Power Cable > Amplifier << NOT TESTABLE WITH THIS METHOD
• Amplifier > Speaker Cables << NOT TESTABLE WITH THIS METHOD

In the above, the only thing we would be unable to test in this setup would be preamp/amp ICs, amplifier, speaker cables, and speakers. The reason being, obviously, is that we have to record the output. To record the output from the amplifier/speakers, for relative consistency, this would require a studio and additional recording equipment, and even then, the sound quality could vary, since it's all sound after the speaker.

However, capturing the audio output at the preamp just before it would go into an amplifier would still let us test component performance up through that point, thus proving for ourselves whether a particular component change makes an audible difference. Granted, as has been noted on the aforementioned Cables Difference thread, one wouldn't know for certain which component provides the missing audio. To this I shrug and say, it's a limitation of an essentially free test method, it's close enough, and at this point simply fall back on one's ears for the decision.

We could test various power cables, digital interconnects, analog interconnects, and even changes to the power source itself going as far back as the electrical panel, and to include power distribution/managers and power conditioning components.

msg

I have a relatively modest system in the home office, but here's an example of something I could try and a list of each of the components I could test for changes were I so inclined to spend the time doing so:

• Computer > Separate USB Card (for isolation) > USB Cable > DAC > Analog ICs > B&K AVR 507 > Analog RCA/3.5mm > Recording Laptop
• Source Power > B&K AVR 507

I could also test for:

• Receptacle > PS Audio AC-3 > PS Audio UPC-200 > PS Audio AC-3 > {component}

GENERAL PROCESS FOR CAPTURE

1. Play a test file on the system, and Capture in Audacity. Save Project.
2. Change a component for difference testing, play test file, Capture again. Save Project.
3. Edit the files if necessary, cropping out white space so that the start times are the same for each (It's easy to clip/crop with Audacity)
4. Open the second file, select all, Effect > Invert. Select All, Copy.
5. In a new Audacity session, Open the Control/Baseline File, click white space in Audacity, Right-click, Paste.
6. Play.
7. You will hear mostly silence, except for where there are differences in the audio file.

I just tested this with a single track. I opened a .flac, selected a section for testing, and saved the clip.
Duplicated the clip, and inverted it.
Opened both files and played them. Complete silence.
Were there differences in these files, I would have heard that difference audio.

msg

As a quick test of this Difference File method, I took a clip of a favorite track, Spiritulized's Sway. I clipped a section of a 320k .mp3 file, then exported a copy of this clip down to 96k to demonstrate the obvious difference between a 320k .mp3 and a 96k .mp3.

I then exported a copy of this difference file to .mp3 and am attaching for review. This, again, is a really obvious test of difference in quality, but check it out, and you'll get a sense of what one might expect from changes in system components.

msg

I'll finish this with a quoted excerpt of something else I read in Guatam Raja's article/interview regarding Garth's response to those who trash his life's work. Skip to the end if you're lazy. He essentially says, and I love this part, "Get off your ****$es and do some work yourselves."

It's not hard, but it takes some time and effort. To those who have doubts and question component gains, this is your method for testing for yourselves. With this method, you can prove for yourselves, definitively, one way or the other, whether what gear and cable enthusiasts have been saying all along is true or not - "It all makes a difference, and that difference is audible."

Now you've got no excuse. Here's your measurement method, a simple one. Account for your positions, do a bit of legwork. Put up or shut up.

Garth has been using this difference method for years—from his Furman days—to check all of his designs. He mentioned it when I asked him about the people who get upset when you talk about audiophile power cables, and insist its either placebo or outright rapaciousness turned retail.

When the critic is a non-engineer, Garth simply and peaceably says he can't do anything about previous situations where a person has tried something and it hasn't worked. Nor can he take responsibility for any other company's "fanciful marketing campaign with science that may or may not subscribe to basic stuff like Ohm's law". He has less patience for people he calls the "Double-E's" (electrical or electronic engineers) who get on the internet and talk about Garth's life's work being snake oil and hokum. And life's work it is—Garth is 56 and has been doing this since he was 15. "I want them to get off their- and you can quote me on this... I want them to get off their lazy rear ends and do some work. Because this stuff is so easy to measure, it's pitiful that they're not willing to do some work."

mikeyb128

Nice work man! That’s great you actually tried this.

I think Somwhere on Tidals website they use this method to show the difference between MP3 and Flac.

I read in the other thread someone quoted “it does not indicate FOR SURE if the difference is truly audible to a listener”

The evidence is right here in the test, because you are actually hearing the lost information with those things stuck to the side of your head...your Ears! So I would think If you can hear the lost information then it must be audible.

K_M

mikeyb128 wrote: »

Nice work man! That’s great you actually tried this.

I think Somwhere on Tidals website they use this method to show the difference between MP3 and Flac.

I read in the other thread someone quoted “it does not indicate FOR SURE if the difference is truly audible to a listener”

The evidence is right here in the test, because you are actually hearing the lost information with those things stuck to the side of your head...your Ears! So I would think If you can hear the lost information then it must be audible.

You mention something I said, and yes a difference signal is an excellent way to detect "differences", I agreed completely, and is not arguable for sure.
My comment referenced, even if there is a difference, that is shown with this method, it simply does not "Guarantee" it will always be audible when heard in the original context of a song or music. Masking effect of other sounds, could make the difference nearly inaudible or completely inaudible.

But still a great idea, that I applaud!

msg

mikeyb128 wrote: »

Nice work man! That’s great you actually tried this.

I think Somwhere on Tidals website they use this method to show the difference between MP3 and Flac.

I read in the other thread someone quoted “it does not indicate FOR SURE if the difference is truly audible to a listener”

The evidence is right here in the test, because you are actually hearing the lost information with those things stuck to the side of your head...your Ears! So I would think If you can hear the lost information then it must be audible.

Indeed - that 320/96k example attached above is proof that this the DFM (difference file method) works. Anyone wishing to test gear changes can use this to hear differences.

K_M

msg wrote: »

As a quick test of this Difference File method, I took a clip of a favorite track, Spiritulized's Sway. I clipped a section of a 320k .mp3 file, then exported a copy of this clip down to 96k to demonstrate the obvious difference between a 320k .mp3 and a 96k .mp3.

I then exported a copy of this difference file to .mp3 and am attaching for review. This, again, is a really obvious test of difference in quality, but check it out, and you'll get a sense of what one might expect from changes in system components.

Could you do the same song in 320Kbps and 16/44.1?

msg

No, but you can.

K_M

msg wrote: »

No, but you can.

Okay but possibly with another song. I do not have access to that song in the format I want.
I realize quite time consuming, but will give it a try maybe with something I do own....cool

msg

Sure, you can use anything you like. It's flexible. You create your own baseline, and can test any number of elements in the playback path, up through the preamp output, even through the RCA/3.5mm cable if you wanted to experiment with those.

Clipdat

Nice job, Scott!

https://www.youtube.com/watch?v=wlMwc1c0HRQ

Hermitism

If you didn't waste your time posting Simpson videos, you'd have time to write a review on your new Solidsteel rack!

Joey_V

Simpson videos are more important than solidsteel reviews

Clipdat

It's not setup yet, so how can I review it? Patience guys. Remember I'm married so by default I don't have tons of free time.

Hermitism wrote: »

If you didn't waste your time posting Simpson videos, you'd have time to write a review on your new Solidsteel rack!

Hermitism

Make her put it together!

Hermitism

At least if she did it, it would be level and square

marvda1

give her a choice of you putting together your rack or hanky panky in the bedroom and see how fast that rack goes up

Clipdat

She was getting fed up with me just talking about it the other night. I was going on and on and then I remarked "Wow, it's super hot in this room!" and she replied "That's because you're blowing hot air about your god damn rack!"

Hermitism wrote: »

Make her put it together!

gp4jesus

K_M wrote: »

You mention something I said, and yes a difference signal is an excellent way to detect "differences", I agreed completely, and is not arguable for sure.
My comment referenced, even if there is a difference, that is shown with this method, it simply does not "Guarantee" it will always be audible when heard in the original context of a song or music. Masking effect of other sounds, could make the difference nearly inaudible or completely inaudible.

The blind testing we* did on two Rotel amps I mentioned in the other thread. I set up the test w/just 4 variables - two sets of ears & the two amps. Listening to bass, low passed, 4th order @ 80 hz. We* could switch between amps driving identical speakers side by side in less than 5 seconds. The listener wouldn’t know which amp was doin the duty. Initially A, B. A, B. Later maybe A A B B A B B A... well you get the idea.
*My 17 year old son & I

Judged on musicality & thump, we* both consistently picked what became the preferred the amp for the task at hand. Granted the differences were small but...

Ears!

kharp1

I hate to be that guy, but, all that sounds good, but, what happens when you change out the equipment that you're recording the files with? What's the base line for that equipment? That would theoretically work if everyone was using that exact same setup. What happens when you go with a different computer? WIll it introduce different noise? What about the raw power that was used? IF we're talking about getting a base line for quantifiable results, what stock power cable are we using? What if one company has a better grade of "stock" power cable? If we are traveling towards finding quantifiable values, what power company are we going with? Is the power cleaned before reaching the test equipment? Way too many variables put a "label" on it saying it's the defacto answer to quantification of such an organic subject. If it makes a person feel better to think they can have a sstandard by wich all audio can be diluted in to test numbers, than so be it. I, however, believe it can't be done.

Ye, there can be some info gained from this, but, it has to be taken with a grain of salt...like most measurments in audio. Seriously, if we can't even agree on how to test wattage on amplification and hard info can't be reached with all the tests we have now, what makes people think they can answer this?

I like the human spirit to continue to search for the elusive answers, but, I know good sound when I hear it, not read it in a test report.

msg

Hey Kerry, not "that guy" at all. These are all valid questions and points. The bottom line is that you define your own baseline or control, this is relative only to your system and environment.

Any changes to the Recording Stage could very possibly skew any comparison with previously recorded files. In fact, I would assume that it would, and with any changes to the Recording Stage, a new baseline must be created, and previously recorded test files cannot be compared against this new baseline.

The Baseline recording you create is relative only to your system. It cannot be used against another's test files created from another system. It should be repeated, as noted originally by @FestYboy in the other thread - we cannot be sure, at least with the testing method/tools I've been playing with, which of the audio files the Difference information is coming from. The assumption is that it would be the "better" equipment, but that is a flawed assumption if we're being objective, and we want to be. That's the whole point of this process.

I have some more information to post about this. As much as I hate to say, I may have come across a potential flaw with my test "apparatus". I will post up more later, but waiting to hear from some empiricists.

I have to say, this is mostly just for fun with the tools I've been working with here, specifically Audacity, and even with the potential flaw I've discovered, the tests do still reveal difference information. How much difference this flaw will make and how to interpret the audio in the resulting Difference File are the questions I'm working on.

I believe the test method is valid, and still think it better than looking at a bunch of numbers, drawing conclusions. Perhaps Difference Filing can be used in conjunction with test measurements and specs to lend context to the improvements offered by component designs.

I think this process requires accuracy and consistency in the process and editing software, with strict attention to maintaining consistency in the Recording Stage. As you've noticed, something as simple as power fluctuation or quality change - which most of us have experience depending on grid power consumption/time of day - in any of the Stages could skew the results.

Anyway, the "fall back" here is always what? Our ears, right?

gp4jesus

@kharp1: yup. Further, those two amps spec’d the same except one made 10 WPC into 8 ohms more than the other*. Oddly we prefered the “other*.” Wish we had time to compare mids & highs. Regardless another case of specs & measurements don’t tell the whole story - how a component “sounds.”