TIL that it’s possible for earphones to be misconnected such that they’re playing the instrumental to a song without playing the...
znk:
TIL that it’s possible for earphones to be misconnected such that they’re playing the instrumental to a song without playing the lyrics?
This is super weird, but I assume it has to do with how the audio encoding works. I assume different tracks are layered over each other such that the singing is a different stream of bits from the instrumental? Or something? IDEK.
I learned this because I was listening to Kartel song after Kartel song, becoming more and more distressed by the idea that the worlboss had decided to only upload instrumentals from now on. It’s only on a whim that I thought to pull out my headphone jack.
I believe sometimes tracks are set to have lyrics in one ear and music in the other, so if only one side of the headphones is working you could get that effect.
I’m not confident this is what’s happening, but I have no other immediate guesses.
Nah, both ears were working. Even to the point where instrumental effects that used one ear or the other worked fine. However, even though the lyrics were supposed to play through both ears, I heard none of them.
Analog stereo audio cables have three separate wires: left channel, right channel and null. If any of these breaks entirely, the effects will be obvious (no left channel / no right channel / no sound whatsoever)
On the other hand, in one failure type, which I believe is a short circuit between the right and left channels specifically, what you end up with is hearing the subtraction of the two channels in both ears. Probably phase-inverted with respect to each other, but that’s usually not an audible distinction.
Most music has some degree of stereo mixing, which means that you will not be left musicless. Also, since the voltage remains, volume is not really affected (though I would assume fidelity is).
Vocals, on the other hand, are usually mixed dead center, which means that they will be zapped from the mix entirely.
…The fact that this is a thing that can even happen is so cool and so annoying.
Fun fact: this is a pretty reliable method to remove vocals from a stereo track. Try it in audacity!
@tropylium is right. Connecting the left and right wires1 yields the difference between the signals.
[Attention conservation note: Press J to skip some hobby audio engineer infodumping applications of linear algebra to stereo audio signals & going off on various tangents. Maybe I should do a separate post on how to use it for fun?]
It‘s called the S (side) channel. Instead of getting stereo by having separate left and right tracks, you can encode the same information in a mid and a side channel:
M = (L+R) / 2
S = (L-R) / 2 2In linear algebra terms, it’s a change of basis in 2D vector space, and can be easily reversed to get the original representation back:
L = M+S
R = M-SL/R is the natural scheme for stereo playback: L goes to the left poweramp + speaker, R to the right. Simple. That is the obvious thing for CDs, cassetes, tape reel … - the playback device doesn’t have to do much.
But M/S encoding comes in very handy for lossy data compression (called Joint Stereo in MP3, OGG etc.):
Good quality M with bad quality S yields overall good quality for the important stuff in the middle (most things get mixed to near the center anyway, and even the rest is all added in there, just quieter3. The main mono tracks for vocals, bass, kick drum and snare are usually dead center.). And then, even the few information you get out of a shitty S channel adds spatial stereo fanciness. Nice.It also ensures mono compatibility: What was in the middle before compression, stays there.4 If you highly compress L and R seperately, all the important center stuff gets spread out over both channels, but might be rounded off a bit differently by the compression, depending on what else is mixed in from each side, which could result in a ~wobbly effect. Not nice.
It’s also how vinyl LPs do stereo. On mono records, the groove zig-zags left & right, making the needle follow the waveform of the one channel signal: Electron microscope video of a phono stylus moving along the grooves on a vinyl record.
You can’t move up & down as much, because then the needle might bounce off the track. That means you can’t encode as much information that way. You wouldn’t want to just use the horizontal movement for one speaker and vertical for R - then the violins might sound crystal clear, but the cellos sound like muffled, noisy garbage. Asymmetry, yuck…
Instead, keep left-right for M, and use up-down for S, the difference between the L and R channels. (Equivalently, you can think of the needle moving diagonally upleft-downright / upright-downleft as the L / R channels. Change of basis!). That also means older phono players will just ignore the up-down motion and just use the left-right signal as usual.5Radio adds the S to the usual mono transmission on a different frequency so receivers can make it stereo optionally. (FM is a bit complicated…)
Finally, there’s a neat analogy to video: When broadcasters introduced color to analog TV, they kept the black&white luminance (brightness) signal where it was, and just added two chrominance channels on another frequency6. That way, people could keep using their old monochrome TVs. Yay for smart standards!
And just like with the spatial S audio information, you can also
(a) save bits/bandwith by compressing the hell out of the color information, b/c we are far better at distinguishing shape & brightness than color; and
(b) be sure that if you transmit black&white footage, you get less weirdly fluctuating colorations added in because the brightness isn’t split up over three separate, low quality RGB channels.
This can happen if you don’t plug in the TRS jack all the way (such that the contacts of the plug touch multiple contacts of the socket. Not sure about the maths, but it might also involve feeding one of the signals into the ground, against which the voltage is measured); or if the isolation surrounding the three wires inside is broken. ↩
Halving (= subtracting 6dB) right away ensures that you stay in the original volume range. If you don’t, you might get clipping b/c you go over -0dB (max. Volume allowed by the data encoding / poweramp transistor). Might not be important for S though, since it should be a lot quieter than M anyway. ↩
Only components that are completely out-of phase (signal ) will not be present at all in M ↩
Sometimes you already want to ensure that during recording, if you mostly care about the center signal, and the stereo-ness should be subtle and natural instead of wide and flashy (e.g. classical symphony?). Using a normal (though probably ridiculously expensive) unidirectional mic, angled up against a weird bidirectional figure-8-characteristic mic, M and S actually correspond to physical microphones, which rules out weird phase issues from the start. ~so pure~ ↩
Though apparently old mono styluses are larger and more rigid, so they can “dig” into the groove and destroy the stereo information, making your LPs mono even on stereo players… whoops. ↩
Since regular humans have three types of color receptors in their eyes, colorspace has 3 dimensions (which is a pathetically lossy projection down from the ~infinite-dimensional continuous spectrum, even only considering the visible wavelengths). So in addition to the luminance, you need 2 chroma components (e.g. U for blue and V for red, each normalized by subtracting the brightness. Then low U and V but high brightness means green.).
You can encode these two chroma channels in one frequency carrier, by using both phase and amplitude (which might also have different fidelity, so apparently you do another basis transformation such that the phase means hue and amplitude means saturation, landing you in ~HSV, the neatest, bestest color space… Abstraction! :D) That’s how most of the world (PAL & NTSC) does it. SECAM (Russia, France, …) only uses amplitude and alternates between sending U and V over the same frequency. ↩