cross-posted from: https://lemmy.ml/post/43120463
TLDR (I’m very long-winded): this YT video took measurements of three different audio cables, including a 200€ one, and found differences where there should be none. My physics knowledge (and general consensus among the scientific community) says the measurements of the different cables should be identical or near-identical — or I am, at least, under that impression. My own measurements, because the channel does provide the files, confirm that the expensive 200€ cable does measure differently from the others. But surely something else must be causing this? Please help me find out what that is, Lemmy!
Below, I go into a little more detail and context, and I go through what I have tried, etc.; I tried splitting everything up into chunks to make it easier to read, but I was never good at being succinct. Sorry about that 😬 :P
Context: What Am I Talking About?
Firstly, I should clarify that this isn’t relevant to most people, only really physics and audio-technology nerds. But I’m desperate. This question has been tormenting me for days...
I should provide some context. For some reason, YouTube gave me a rather odd video suggestion. I am very interested in audio technology, mixing, and mastering, but I’m not a snake-oil audiophile type. I guess you could call me a lover-of-audio, e.g., audiophile, but I don’t attach myself to that community. So you can imagine my confusion when this video was suggested to me on YouTube.
For anyone unwilling to click on an ambiguous YT link like that, here is a brief description: audiophile guy (who believes that cables affect sound) compares three cables, two of which are "cheap" (e.g., approx. 50€) and one of which is expensive (like 200€ — for a cable). His conclusion is ultimately (yes, I watched that far) that they don’t really affect sound, because they’re just interconnects (so RCA cables), and not speaker or power cables. But he actually records each cable and provides those music files — which means I can compare them as well...
By "records each cable," I mean that he used each cable to connect a CD-player to a pre-amp (for some reason) and then an analogue-to-digital convert (ADC) into his computer, where he can record the output; this way he gets just the supposed difference the cables make. I realize that the CD-player might have some inconsistencies, since CDs are a moving medium, but jitter compensation and stuff like that is very advanced these days, so this is getting pretty close to an accurate measurement of the cables, I think (correct me if I’m wrong!).
In the video, this guys claims that the measurements show a difference between the cables. This debate around cables is something I thought was quite contested, so I naturally decided to download the tracks (which he provides in the description) and compare them myself. Here’s what happened.
My Own Testing: They Don’t Null
So I downloaded his recordings, phase-aligned (which they weren’t already) everything and normalized to peak (which, again, was necessary, and creates the lowest delta compared to LUFS, etc.), and then null-tested. That means I inverted the phase of one, played two of the files at the same time, and thus got the delta between them.
The Mogami and Belden cables (the cheap ones) are nearly identical; there is a slight difference in the null-test, but it is statistically small enough, that I attribute it to measurement inconsistency (e.g., CD-player, DAC, or ADC performance, but also maybe natural conditions, etc.). I especially suspect — given that most differences occur at higher frequencies but aren’t really audible when listening — a jitter related issue, causing the timing to be just a little off, thus creating a delta in the higher frequencies. This could be due to the CD-player, but I have no idea.
...the Neotech (the expensive one), however, is significantly different.
Firstly, he provides the file for the Neotech cable with a significant phase delay compared to the others, of about 50 ms, and it is also about 0.02 db louder than the others. That is rather suspicious on his part. This alone makes it sound better, but when I correct this volume and phase difference — it still sounds better, and the null-test confirms that it is still quite different (we're talking differences up to -40 db here, which should definitely be audible).
It turns out, the volume difference actually changes throughout the song; meaning that in some places the difference is 0.01 db, and in others 0.03 db. But I can correct for that! I don’t know why it happens (someone smarter than me, please explain), but I can correct for it... And yet, the null test still shows a clear difference (especially when the singer makes s-sounds, i.e., sibilance).
Please Help Me
What is going on? My physics knowledge tells me this should be impossible. ~~I can only imagine that for some reason the Neotech is more conductive, or something like that, and therefore recreates the harsh and very fast dynamics of sibilance more accurately.~~ (edit: crossed out because this is a little too ridiculous) But the difference in conductivity should not be enough to cause that... I really am confused. Someone with more expertise please explain this!
For context, I cannot blind test ABX the Neotech. (edit: meaning, I cannot hear any difference whatsoever.) This is very subtle stuff here. But I can see a difference, the null test shows differences as high as -45 db when the singer does those s-sounds and everything has been normalized. So clearly something is happening (and again, this is phase aligned and normalized and everything). So what could it possibly be?
Edit: here, for clarity, screenshots of the null test at different moments in the tracks from the youtube vid (here nulling between the magomi and neotech).
Normal part of song, no sibilance or cymbals, but otherwise singing, piano, bass guitar, drums, etc.:
When the singer creates sibilance with an s-sound:
The null test was performed in the Reaper DAW with Voxengo SPAM and Reaper’s built-in phase align, normalization, and phase inversion.
Lemmy, please show me what I have missed! Show me the obvious error I or the video creator made. Sorry to make such a long post on c/asklemmy, but I don’t know where else to ask.
Thank you! I don’t know the YouTuber’s gear’s noise floor, so I don’t know how much of an influence that should be having, but I would presume quite low. But I’m not entirely sure if I understand what you mean. The null-test I did (I’ll attach a screenshot to this comment) mostly showed a delta in higher frequencies, specifically when the singer makes sibilant sounds. Does the graph look like what you would expect from system noise?
Null test while the singer is making a sibilant sound (via Voxengo SPAN and Reaper):
Null test while the singer is not making any sibilant sounds and the cymbals aren’t playing:
Thank you for taking me as good faith. I realize now that my post-title sounds a little click-baity, like I’m suggesting that there’s an authentic delta in the null test. I very much don’t think that’s the case; I’m just trying to figure out what a plausible reason might be, for why I’m measuring a delta in the first place.
I guess ei incumbit probatio... the burden of proof lies upon the YouTuber here, so I shouldn’t get too worked up about this :)
Your graph has a signal of -40dB. Are you working in high-energy physics? If it's only nano-tech or something of this magnitude, you don't have equipment that is this precise.
And all the "high-intensity" findings on your posting (at audible 0dB) are probably because the expensive cable didn't connect well to the socket. And are still nothing, and have no impact on your sound quality.
I'm not sure if you're reading the graph correctly, this is the delta between two of the digital files from the video's description. So a signal of -40 dBFS is quite audible, since it's all relative to 0 dBFS (full scale).
And it isn't the recording itself, it's just the difference between two of the recordings provided in the video's description. This is commonly known as a digital null-test, and let's you find the amount (and significance) of difference between two digitally encoded recordings, and in particular at which frequencies those differences lie.
You can try doing it yourself by downloading the audio from the YouTube video's description and then playing two of them at the same time in audacity, but with the phase inverterted for one of them. Just make sure the phase and volume are aligned. Then you can hear the difference between the recordings yourself!
The question is, where does this difference come from.
At -40, it doesn't make any difference what kind of dB you are talking about.
(Well, if you are looking for superstrings evidence or even quantum gravity, it does, but I guess it's not your case.)
https://en.wikipedia.org/wiki/DBFS
Relevant line:
That means the US definition you can't hear anything below -24 dBFS. It's the largest value there, it's -18 for the EU.