The Hummingbird Clock: date videos by background mains hum

They really evoke the feeling of electricity. It reminded me of the soundtrack to the Chernobyl miniseries. It’s… powerful and eerie and calmly, effortlessly, sluggishly intimidating.

Feeding signal into e.g. a guitar pedal is easy enough, but mutating an entire orchestra in real time is a different beast. If anyone knows of similar works I'd be grateful for links.

Is there a way to watch the video without logging in? I tried multiple browsers and still hit the login wall.

I came here to check that someone had linked to that video, but in my mind it feels a lot longer ago than Dec 2021. I feel like that video should be 3 or 4 years old.

It's doubly weird because that normally happens the other way around, where I think something happened about a year ago, and it was actually 3 or 4 years ago. Or I think something happened 5 years ago, and it was actually 10-12.

    // hum.js
    
    // toggle html audio on/off
    // with cookie to maintain state
    // and use html5 audio synthesis
    // ios html5 audio must be triggered
    // call init_hum() after DOM loaded
    
    // 0. create audio context
    // 1. create oscillator, define
    // 2. connect oscillator to gain
    // 3. create gain
    // 4. connect gain to output
    // 5. start oscillator (on click)
    
    var audio;
    var audio_context;
    var control;
    var vco, vca;
    var hum_delta = 1;
    var hum_base = 200;
    var hum_min = 100;
    var hum_max = 400;
    
    /* init */
    
    function init_hum() {
        audio = get_cookie("audio");
        console.log("audio = " + audio);
        audio_context = get_audio_context();
        console.log("audio_context = " + audio_context);
        control = get_control();
        if (audio_context != "false") {
            if (audio != "off") {
                set_hum();
                hum_on();
            }
        } else {
            set_cookie("audio", "off");
        }
    }
    
    function get_audio_context() {
        var which_audio_context = window.AudioContext || // default
                                  window.webkitAudioContext || // safari
                                  false;
        this_audio_context = new which_audio_context;
        return this_audio_context;
    }
    
    function get_control() {
        var this_control = document.getElementById("control");
        this_control.addEventListener("click", hum_on_off);
        return this_control;
    }
    
    function set_hum() {
        vco = audio_context.createOscillator();
        vco.type = 'sine';
        vco.frequency.value = hum_base;
        vca = audio_context.createGain();
        vca.gain.value = 4.0;
        vco.connect(vca);
        vca.connect(audio_context.destination);            
    }
    
    /* on off */
    
    function hum_on() {
        set_hum();
        vco.start(0); 
        control.innerHTML="×";
        set_cookie("audio", "on");
        audio = get_cookie("audio");
        console.log("audio = " + audio);
    }
    
    function hum_off() {
        vco.stop(0); 
        control.innerHTML="+";
        set_cookie("audio", "off");
        audio = get_cookie("audio");
        console.log("audio = " + audio);
        cleanup();
    }
    
    function hum_on_off() {
        if (audio == "off")
            hum_on();
        else
            hum_off();
    }
    
    function cleanup() {
        vco.disconnect(0); 
    }
    
    /* cookies */
    
    function set_cookie(cname, cvalue) {
        document.cookie = cname + "=" + cvalue;
    }
    
    function get_cookie(cname) {
        var name = cname + "=";
        var ca = document.cookie.split(';');
        for(var i = 0; i <ca.length; i++) {
            var c = ca[i];
            while (c.charAt(0)==' ')
                c = c.substring(1);
            if (c.indexOf(name) == 0)
                return c.substring(name.length,c.length);
        }
        return "";
    }
    
    function check_cookie(cname) {
        if (getCookie(cname) != "")
            return true;
        else
            return false;
    }

Wife and neighbors are not happy.

Silk road was closed in 2013

There's a lot of information held within the time series data of an audio signal, and something like an FFT can tap into it. For a 60Hz signal the Nyquist frequency is 120Hz, which is well below any audio interface. It seems like accuracy of any processing would be limited by a crystal oscillator somewhere more than anything, and perhaps any unintentional signal aliasing if there's no low pass filter before the audio interface's ADC.

It has to be, that's the definition of a power grid. The relative phase can have minute variations and you get massive power flows as a result, which result in the frequency across the whole grid being the same.

I believe the whole UK is a single grid. The US is divided into three: The eastern interconnect, the western interconnect, and Texas. (Insert meme here.) There are a few DC-links that allow power to flow between them, but due to the expense of rectifying and inverting all that power (necessary for the frequency and phase to vary), they are rare. It's much cheaper to connect directly at AC, even though this requires that everyone on the system run at the same frequency.

Amusingly, this also means that if you could (theoretically arrange enough long shafts and bearings and neglect the curvature of the earth and) orient all the generators on the grid in a line, a single shaft could run through all of them, and the shaft would not twist -- they all spin in precise synchrony since they are, effectively, a single machine.

When a new generator is brought on-line, it has to be phase-matched to the grid. These days there are probably fancy PLCs to handle it, but the simplest instrument is called a synchroscope, which is a motor driven by the phase relationship between two AC waves. You can find these on YouTube, they're pretty neat. Basically if the isolated generator is running slower than the grid, the synchroscope needle spins one way, and if it's running faster, the needle spins the other way. You adjust the throttle on the generator until it's running _ever so slightly_ faster, like 60.00Hz on the grid and 60.10Hz on the isolated generator, (50.xx in the UK) which will make the 'scope revolve once per ten seconds. Then you wait until the needle is pointing straight up which indicates zero phase shift, and slam the switches to connect it to the grid, and that's that!

If you did it just right, there's a minimal shudder and the generator now begins sending power into the grid, its frequency constrained to 60.00Hz and the extra throttle showing up as positive power flow. If you did it wrong, there's a tremendous amount of torque exerted on the rotating machinery by the grid power, which may simply force it into compliance, or in more extreme cases, may shred millions of dollars of equipment in an instant. Bigger synch setups have lockouts that won't even allow the button to be pressed unless the phase is _pretty close_. The most colossal kind of cockup is when someone wired the synchroscope wrong, so the rotation still works (it has to, that's impossible to screw up) but in-phase would be represented by some needle angle other than straight up. Joining the grid 120 degrees out would be a career-limiting event for someone.

Chances are internally it uses a much higher sampling rate and digitally downsamples.

This was used in a court case in the UK where the defence claimed the police had tampered with a recording - editing together separate recordings into a single one to make their case. An academic 'expert' analysed the recording and corroborated the claimed recording times and the accused were found guilty.

In practice, most people who manufactured videos before very recently were simply unaware of this possibility. I learned about it many years ago but I still forgot about it and probably would not try to do anything about it if I was manufacturing a video.

https://www.maxon.net/en/red-giant/pluraleyes

If you filtered the audio to exclude all but the 50Hz/60Hz signal then it might also work based just on the power hum, but why make it's job more difficult?

In other words, the base oscillation of 50 or 60 Hz is not stable, but instead varies a little above and a little below at higher rates.

The magnitude is significantly lower than the base mains frequency. So possibly not audible above any other noise in the recording, but present nonetheless.

That's immaterial because of harmonics.

While the AC wave itself is nearly a perfect sinusoid, various devices introduce all sorts of nonlinear properties, and produce higher frequencies that are harmonics of the base powerline frequency, and thus vary right along with it. These are insidiously hard to remove. Your best bet would be to encode all the voices with a vocoder, ditch the background sounds entirely, and play the coded voices back. Which of course makes it obvious that the audio has been altered...

I suspect that similar artifacts will be visible in lights too, since the brightness of an incandescent or neon source varies with the AC wave. Some fluorescents too, though LEDs all have DC drivers so they'd make it more difficult.

Me and some friends tested them and tried transmitting to a neighbours house (connected to the same electrical post) and had no success at all. There was just too much noise, coming from other houses. However with digital technology I'm sure you can get a much better range.

I expect the incidental signals your usage generates are going to be hard to detect past your power meter and almost certainly un-resolvable past the distribution step down transformer.

Chrome has a list of special websites that are given that access automatically, though I doubt this website is on that list...

Possibly, if someone went through and fingerprinted all the inverters.

If it's a plain simple inverter with no grid-tie capability (i.e. one that might be installed in a car), then it might not have any distinctive characteristics, particularly under load. Just plain simple 60Hz all the time. Lightly loaded, some inverters have a power-save mode where they alter the waveform to that it produces less wasted power in the MOSFET stage, but I don't know if that would show up as a distinctive sound.

If it's a grid-tie-capable inverter, even if you're not using it in that mode, it probably has some distinctive characteristics. These units try to wibble their frequency around a little bit to see if anything "pushes back", as a way of ascertaining whether there's grid power present. They may also shift their frequency as a way of communicating with AC-tied solar panels on microinverters, analogous to how the grid's own frequency varies with load, but not so smooth.

All of those behaviors would likely produce pretty distinctive sounds, and they may vary per model, and certainly per manufacturer.

The UK is all one grid, Ireland (the island) is another, there's a Nordic grid, a Baltic grid, and one for the whole rest of continental Europe (plus Morocco and Tunisia). The US has two plus a separate one for Texas, Canada uses the same two plus a separate one for Quebec (the jokes just make themselves), and Japan weirdly is cut in half, with a 60Hz grid in the south and a 50Hz grid in the north.

https://en.m.wikipedia.org/wiki/Wide_area_synchronous_grid

Some of them have phase-locked loops to sync the output frequency with the mains, but probably won't stray as far from 50Hz (or 60Hz) as the mains themselves seem to do. If they do, unplugging the mains and using the battery is probably enough to get pure 50Hz back!