Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!ames!amdcad!amd!intelca!mipos3!omepd!randys
From: randys@mipon3.intel.com (Randy Steck)
Newsgroups: rec.aviation,sci.electronics,rec.audio
Subject: Re: Noise-cancelling microphone
Message-ID: <716@omepd>
Date: Fri, 22-May-87 19:45:15 EDT
Article-I.D.: omepd.716
Posted: Fri May 22 19:45:15 1987
Date-Received: Sat, 23-May-87 19:12:45 EDT
References: <1027@mips.UUCP> <689@cod.UUCP> <867@lll-lcc.aRpA> <2849@mit-hermes.AI.MIT.EDU>
Sender: news@omepd
Reply-To: randys@mipon3.UUCP (Randy Steck)
Organization: Intel Corp., Hillsboro
Lines: 41
Xref: mnetor rec.aviation:1663 sci.electronics:732 rec.audio:1709

In article <2849@mit-hermes.AI.MIT.EDU> jpexg@mit-hermes.AI.MIT.EDU (John Purbrick) writes:
>
>Regarding noise-cancelling speakers and mikes, there was an article in the
>British magazine "New Scientist" a while back about the use of "anti-noise"
>near large sources of acoustic noise like stationary gas turbine installations.
>The idea was to put a microphone some way from the device and drive a speaker
>adjacent to the device; an electronic system would try to null out the
>resultant signal as measured at the mike. As I recall the system was claimed
>to work fairly well, but I don't recall the db figures or how they solved
>the obvious problem of delay from the noise sources to the mike.

The algorithm of adaptive signal processing apparently used in this system
was invented by Widrow of Stanford.  I audited a class he taught on the
subject and he suggested that one of the students try to build something
similar to the above for a class project.  He wanted to call it a "Cone of
Silence" (I hope it works better than Maxwell Smart's!).

However, there seems to be a small problem with the application of such a
system.  It seems to be possible to cancel noise at the microphone if the
source of the noise is reasonably periodic.  The problem comes in when a
person moves away from the microphone since the hearer may no longer be in
the "trough" of the waveform, but at a "peak".  If the noise was large
enough to cancel in the first place, then the resultant additive component
would probably blow a person's ears out.  This is minimized by putting the
speaker near the device, but the area near the device would have many
"peak" regions.

And if the microphone moves (maybe being worn by someone?) the adaptive
filter has to adjust itself.  This may take some time depending on the
algorithm used, so the user better not move too fast.

And if that weren't bad enough, what about the use of more than one
microphone?  An additive peak will likely result somewhere.  It seems that
the system described in this mail would be limited to cancelling noise for
1 or 2 people in a stationary position for long periods of time.  

But, it's really interesting technology!  Makes for alot of fun in the lab,
too.

-- Randy Steck
   Intel Corp.       ...intelca!mipos3!omepd!randys