On Fri, 2006-12-15 at 16:19 -0600, Grant Youngman wrote:
> > > While the effect on the output spectrum is that of a
> > bandpass filter, I
> > > don't think NR works that way.
>
> NR works in a variety of ways, including using LMS or some other adaption
> algorithm to adapt the coefficients of an IIR or FIR filter, in the classic
> Wiener filter noise reduction algorithm. And the spectral response of the
> filter does indeed consist of multiple relative peaks in the passband,
> corresponding one hopes with the primary spectral components of the signal.
>
> Do I know that's what's in the Orion? No. I suspect on one else on the
> sidelines does for sure either, and those that know aren't really talking.
> Is it a Linear Predictor, as someone has suggested? For what it's worth, a
> Linear Predictor can be used, again, in Wiener noise filtering, to provide
> an estimate of the signal spectrum. Or is it correlation (or auto), based
> on DFT spectral estimation, or something else. There are as many ways to do
> this as there are chapters in a book, and the more processing power
> available to do it the merrier.
>
> If one sees what appears to be a bandpass filter on a spectrum analyzer,
> what else is it, regardless of the underlying algorithm? If it walks like a
> duck ....
>
> But I fear we're all just talking at, instead of with, each other at this
> point. Perhaps we should settle on a common text reference? :-)
>
> Grant/NQ5T
>
Something the ear and spectrum analyzer may show is the character of the
attenuated noise. I'd expect an adaptive bandpass filter to have a
considerable depth to the skirts and the noise left to be somewhat
narrowed by those skirts but to still have the same characteristic sound
as the unfiltered noise. In an autocorrelation noise reduction filter
the remaining noise has a mostly flat spectrum and doesn't sound like
the unprocessed noise. My Timewave works that way. Many find its reduced
noise unfriendly to further processing between the ears.
--
73, Jerry, K0CQ,
All content copyright Dr. Gerald N. Johnson, electrical engineer
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|