NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/rec/audio/19339

Internet Message Format | 1993-01-25 | 2.9 KB

Path: sparky!uunet!usc!sdd.hp.com!crash!cmkrnl!jeh
From: [e-mail]
Newsgroups: rec.audio
Subject: Re: Loud Music, Noise cancelation systems?
Message-ID: <[e-mail]>
Date: 24 Jan 93 20:22:20 PST
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Organization: Kernel Mode Systems, San Diego, CA
Lines: 46

In article <[e-mail]>, rmg53668@dcl-nxt53
(Ryan Martin Grant) writes:
> > [... re sound cancellation systems ...]
> They seem endless, yes.  Imagine an end to noise pollution as we know it.
> It would be beautiful.
>
> These dreams are very similar to my own, but I was wondering if any
> rec.audio wizards could bring up any hard facts for/against this technology
> ever going mainstream.

Sound cancellation can be made to work if the "cancelling" source is very close
to the "noise" source.  "Close" means a small fraction of the shortest
wavelength of interest.  An example is the active noise cancelling headsets
made for small aircraft pilots by Bose and one other manufacturer (Telex, I
believe):  They pick up ambient sound just outside the ear canal, and feed a
cancelling wave plus the headphone audio into the ear canal.  Here we are
fudging on the definition of a "source", but since everything you hear has to
come through the ear canal (ignoring bone conduction), this works.

If you want to cancel noise in "free space", say, in a city bus, things aren't
nearly so neat and tidy.  You can get an idea of how well it can work simply by
setting your stereo FM tuner to "mono" and reversing the connections to one of
your speakers.  Now you have two sound sources, identical except for phase, and
coplanar as well.  This should give optimal cancellation, yes?  Unfortunately
you'll find that the cancellation you can actually achieve isn't very
impressive, even if you put the speakers right next to each other.  (You will,
however, thoroughly confuse your sound location sense if you walk around in the
resulting sound field....)

Why doesn't it work better?  Well, if the sources are separated by more than a
half a wavelength (just five inches at 1 kHz, half an inch at 10 kHz), there
will always be locations that are 'n' wavelengths from one source but 'n'+0.5
wavelengths from the other... and in those locations the "cancelling" signal
will *reinforce* the "noise" signal.

Worse, these locations will be different for every component frequency of the
noise!

It would be possible with real-time analysis and DSP to delay the various
components of the cancelling signal to compensate for this at one particular
listening position.  But you can't do it for all positions at once!  You might
be able to produce quiet for yourself, but in the process you'll be making the
noise twice as loud for someone else.

    --- Jamie Hanrahan, Kernel Mode Systems, San Diego CA
Internet:  [e-mail], or [e-mail]  Uucp: uunet!cmkrnl!jeh