Comment by trashfindhunter | Hacker News Reader

Comment by trashfindhunter | Hacker News Reader

Parent Full thread

trashfindhunter

I've played with audio software for a long time and I recently experimented with Active Noise Cancellation. There are a few things to keep in mind:

* A "live" ANC process has no control over the environment from which it receives the audio signals that it acts adversarially against.

* When transmitting audio waves from one medium to another, there will be latency. Perhaps not much, but it will be there.

If you accept these two positions, then consider this:

* What happens when a sound wave that is being combatted (via phase inversion) suddenly stops, or inverts it's own phase? That's right, ANC could potentially double the amplitude of the frequency being combatted.

* I imagine that ANC technology takes advantage of latency to ensure that they don't damage people's hearing, but the nature of ANC requires low latency in general, otherwise you can't be sure that you are combatting the correct frequency (at which point you risk doubling the amplitude due to abrupt changes) - if someone more familiar with the actual algorithms could chime and correct me I will happily stand corrected :)

So first off, doubling the amplitude is a 6dB increase in SPL. So not that bad actually.

Second off, instantaneous sound is only a health issue when it's really loud, like a gunshot (130-140dB SPL) near the point at which the ear drum ruptures. That means that you need to be in an environment where the background noise is dangerously loud to begin with, and because of the way sound is made - this might be unlikely. Which is interesting, because early ANC did have these problems - when it was being used initially for military applications (helicopter/tank pilots iirc).

Lastly the important thing to remember is that ANC is usually part of a dual pronged approach to ear protection. Latency is a problem when you need to cancel high frequencies (where you get past about a quarter wavelength and interference can become constructive), but ANC excels at low frequencies (below about 500-1kHz it can be remarkable even). This is great because passive reduction strategies (sealing off the ear, thick padding, good fit/headband adjustment) are much more effective at high frequencies.

So TL;DR it was a problem, been fixed, and where it might happen is pretty rare for a consumer.

Also noise rarely spontaneously inverts phase at a particular frequency. That'd be weird.

Yeah, you're right about ear damage coinciding with exposure time - short bursts would have to be very loud to cause real damage.

And yes, it would be weird if a frequency range spontaneously inverted - the only scenario I can imagine that happening in is some jerk doing it on purpose.

The reason I became interested in ANC was because every night I would hear a terrible frequency being emitted from the air conditioner units above me (top floor apartment building), and during my experiments I quickly realized how hopeless it would be to effectively combat them due to the varying intensity of the sound throughout my apartment, the dynamic interactions of the sound with itself within my echoey wood floor studio, and my location at any one time. All valid points though, thanks for chiming in. I learned more about ANC :)

Edit: Btw my goal was ANC via speakers, not headphones. Headphones would be much easier since they only have a single, summed audio source.

Then again... won't that summed audio source be at the mercy of my loud speakers?

If you mean by that that it couldn't be louder than what the speaker could produce: No, because you're adding to the original noise entering your ear from the outside.

Will the physical sound waves that I produce with my loud speakers cause Active Noise Control enabled headphones to "engage" with me? Could I adversarially engage with them at that point?

I think I understand what you mean now.

While the noise cancellation is active it will attempt to neutralize (destructively interfere with) sounds from the outside, including those generated by your speaker. You could indeed adversarially engage through something like a spontaneous phase shift (so the interference will become constructive, making the resulting signal louder) or generating a frequency the ANC can't compensate.

Reply on news.ycombinator.com