What is (an) Antisignal?
The birth of active signal control
Signals and Noise (In The Abstract)
For now and for the remainder of this article, when I speak of a signal, I am generally speaking of a sine wave playing over a channel. I will be speaking of signals and so forth in abstract terms and this is the representation that I want you to see when you think of the term signal.
Here is a simple experiment. In the image that follows, the first channel at the top is playing a sine wave at a frequency of 440 Hz. The second channel beneath it is playing a sine wave with frequency 220 Hz. For purposes of presentation, I have included a third channel that is what you would get if you added the two first sine waves. That is, if you literally added the two channels together, mixed them down to produce one new channel with the sum of the first two signals, you get the third signal represented at the bottom.
There is something very intuitive about visualizing signals in this way, in the frequency-domain. The sine wave is a periodic function. It’s easy enough to understand, it helps one think about signals and systems in the abstract. When one is doing signal processing, one is usually staring at these kinds of signal representations at one time or another.
This next experiment is also simple, but different. Here, I have one channel with white (Gaussian) noise and a second channel with a pure sine wave at a frequency of 440 Hz. The third channel, once again, is what you would get if you added the two first channels, or another way to put it, if you mixed them down to one channel.
Whenever I hear someone speak of a signal-to-noise ratio, or of signals and noise in general, this is what I picture in my mind. I have a signal, the sine wave, and some noise (white Gaussian noise in this case). If I am listening to both a signal AND some white Gaussian noise — which sounds terrible, by the way — then I am hearing exactly what the visual representation at the bottom sounds like, in the third channel.
In this example, I can still make out the sine wave, it has a rather distinct sound. That’s not what we really care about when we speak of noise in a channel a lot of the time. What I mean is that, yes, in the extreme case, the signal I am looking for is absolutely hidden, inconspicuous, and I am trying to clean it up somehow to reveal the hidden signal. Sometimes I might not even know what I am looking for. I process it in a dozen different ways, trying to isolate something which I suspect exists, but haven’t seen or heard yet.
In the next example, I take two sine waves with the same frequency, I invert one of their phases, and guess what happens? They cancel each other out entirely. The resulting signal is perfect digital silence in this case (in real-life applications, this is almost never the case).
And finally, here it is folks, active noise control! I took the original 440 Hz sound wave, all 12 seconds of it. I added 12 seconds of white Gaussian noise. I copied the exact same 12 second white Gaussian noise sample, inverted the phase, and the result was my original signal — minus all the noise. One caveat: I needed to use the exact same noise signal to be able to invert the phase and create what is called the anti-noise, which did the cancellation. If I had a different 12 second sample of white Gaussian noise, and inverted that, I would have gotten a 12 second sample of complete crap, the sine wave plus twice the noise.
What is (an) Antisignal?
So far, I have spoken of sine waves, signals, channels, white Gaussian noise, phase inversion (anti-phase), noise cancellation, active noise control, and of course the always elusive and proverbial anti-noise. This is in fact where the term antisignal comes from, at least in the way that I have been trying to use it.
After hearing so many people speak of signals and noise, of the signal-to-noise ratio, I began to do some thinking. Let’s take Twitter for example. People often speak of signal and noise with regard to the Twittersphere. The idea is that one is looking for some signal, some information, and anything that corrupts this signal, or that gets in the way somehow, is called noise. In this case, it is information that could be said to be irrelevant, impertinent, or unwanted. But as I have shown, noise can be a signal too. In fact, that’s exactly what I did when I added the anti-noise: I fed a new signal into the system — the anti-noise signal — which cancelled out the noise, cleaning up the channel, and allowing me to reclaim my original signal (the sine wave).
In acoustic ecology, where noise control is part of the established tool-set, one can speak of noise as environmental noise, as a form of pollution which one may want to control. In the case of Twitter, I would argue that the unwanted signals are a form of information pollution that one may want to minimize or filter out.
One problem that I discovered is the following. What do I do when there’s too much signal? Coming back to the example of Twitter, what happens when I have filtered out all noisy signals, all unwanted tweets and Twitter users, cancelling out all that is irrelevant, and all I have left are a hundred or a thousand beautiful signals? I call them beautiful signals because to me that’s what they are: If these tweets are signals, I argue that the information they contain is aesthetically pleasing. That is to say, it’s not that they are more relevant or less noisy, but that they are to my taste. I argue that this is in many ways a matter of aesthetic judgements, of judgements of taste. That is why I call them Beautiful.
What do I do when there’s too much Beauty in my Twitter feed? How do I filter or cancel out Beauty? A similar problems occurs with one’s taste in music. Say I have spent many years, maybe several decades cultivating my own unique musical taste. Imagine now that I find myself sitting at my desk trying to decide what to listen to next. I have 100,000 songs or 1 million songs on my iPod or my computer, which one am I going to listen to next?
Ideally, I am the one that carefully chose every last one of these songs, these signals.They all were filtered in by the fact that they were well-suited to my personal aesthetic taste in music. I have therefore come face to face with a massive, impenetrable wall chock full of beautiful signals, 100,000 or 1 million of them. What on earth can I do?
I am postulating the existence of an Antisignal. This is the missing element which makes the full set: Signal, Noise, Antinoise, and Antisignal. (There would also be pure digital silence, of course, which is always an option and a form of Antisignal — just turn off the stereo, turn off your computer, get off the Twitter Train.) But the antisignal is more than that. It is the first element in what I am tentatively calling active signal control. Or, if you like the sound of this better, if active noise control is adding an antinoise signal to cancel out the noise and free up the signal, then with the introduction of the antisignal, what we really have is: Active Signal Liberation.
How does one Free The Signal? Stay tuned and you may find out.
