On Tuesday, February 23, 2010 1:36 AM, George Brock-Nannestad wrote:
> let us remember that although I have asked for it for years in
> archive circles, there is still no comparison of systems available.
> A comparison would be to subject the systems to the same raw files,
> to agree on what type of result is desired and to determine to which
> degree the various systems enables the results.
George, I couldn't agree more for a system of testing noise reduction
tools.
A standard test suite of raw files and a clear set of metrics to apply
to the results would be useful. Not an easy task to design such a test
suite and metrics. Some things I would be interested in measuring:
- % of known defects removed
- type of defects removed and retained
- changes to dynamics (ie. compression)
- changes to frequency response / spectral balance
- changes to harmonics
I find crackle removal to be the most egregious of NR tools, and
even when applied most sparingly it seems to rob a recording of
some life. All crackle removal tools seem to do this (at least to
my ear). I would like to more objectively measure this perception,
for example.
Custom tools may be required to perform the metrics, and I suspect
that this is where any strategy to measure the performance of
NR tools will hit a stumbling block.
However, if there was a way to perform even simple metrics using
the existing tools available in most DAWs, that would be a step
forward.
As a Cube-Tec AudioCube user, I would happy to run such a test suite
and metrics and share the results with users of other NR tools.
Eric Jacobs
The Audio Archive, Inc.
tel: 408.221.2128
fax: 408.549.9867
mailto:[log in to unmask]
http://www.TheAudioArchive.com
Disc and Tape Audio Transfer Services and Preservation Consulting
|