naturerecordists
[Top] [All Lists]

Re: Noise reduction

Subject: Re: Noise reduction
From: "Rob Danielson" danielson_audio
Date: Sun Mar 1, 2009 8:54 am ((PST))
Hi Bernie--
Interesting project. I bet it would be very expensive to explore the
"no-needle" scan methods.  Its been a couple of years since this
(pdf) report: http://tinyurl.com/b4ulam but I've heard a related
process is now quite effective with optical sound on paper. Makes
sense to suppress the mechanical weaknesses/patterns working from
digital scans and avoid recreating them with needle playback. Then
you can go after it with developments in audio signal noise
reduction! :-) Rob D.

  =3D =3D =3D

At 7:01 AM -0800 3/1/09, Bernie Krause wrote:
>In the old transitional days, when analog tape was still the
>preference, and adaptive-predictive-deconvolution systems first
>emerged (digital systems in the early 80s) you could really hear the
>artifacts, and the result was unimpressive when compared to the cost.
>And I agree, to some extent the artifacts are still present. But when
>trying to distinguish detail inherent in language from whatever
>source, human or non-human, and where subtleties are important in
>decryption - especially with extremes like old wax cylinders - this is
>just another tool. The example I showed was where the noise was gone.
>Different levels of noise reduction left more of the original context
>intact.
>
>For the linguistic anthropologists at the Lowie Museum in Berkeley,
>where the Ishi archive is stored, and where the focus is on content
>and language recovery rather than audio fidelity, this type of
>approach means a great deal. If more than very subtly applied, it
>would certainly destroy all of the spatial imaging of an otherwise
>well-recorded stereo or MS recording not to mention the inherent
>details of the biophony.
>
>Noise deconvolution has a way to go but it's coming along in pretty
>amazing ways. Eventually, it may even work for continuously variable
>noise like the Doppler shift and audio levels of a jet flying overhead
>ruining an otherwise perfect natural audio clip.
>
>Bernie
>
>On Feb 28, 2009, at 11:13 PM, Dan Dugan wrote:
>
>>  Bernie, you wrote,
>>
>>  > The audio clip contains a 30 second example of the original
>>  > wax cylinder recording followed by the same clip cleaned up.
>>  > Considering the quality of the original very few artifacts remain
>>  when
>>  > compared with other options.
>>
>>  But all the naturalness is gone too, replaced by a strange robot-like
>>  quality. I've done quite a bit of noise reduction work, and I often
>>  hate myself in the morning. I get tuned into the noise, and it seems
>>  great to get rid of it, but later I realize that the cost was too
>>  great.
>>
>>  -Dan Dugan
>>
>>
>
>Wild Sanctuary
>POB 536
>Glen Ellen, CA 95442
>707-996-6677
><http://www.wildsanctuary.com>http://www.wildsanctuary.com
><chirp%40wildsanctuary.com>
>Google Earth zooms:
><http://earth.wildsanctuary.com>http://earth.wildsanctuary.com
>SKYPE: biophony
>
>
>
>


--









<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU