On Fri, 2006-12-08 at 09:36 -0600, Grant Youngman wrote:
> > Likewise, the phrase "building a filter around the signal"
> > which has been used in several posts in various forms does
> > not mean that there is an actual pass band being established.
> > I've seen that language used elsewhere as a simplified
> > explanation of DSP, referencing the more familiar analog of a
> > filter, instead of trying to explain the math, etc, of true
> > DSP noise reduction.
>
> In the v1 Orion it is most definitely the case. It's clearly observable on
> a spectrum analyzer, and isn't just the "analog thinking" of a newbie. This
> is as much "true DSP noise reduction" as any other DSP-based process
> targeting the task.
>
>
>
> Grant/NQ5T
>
While the effect on the output spectrum is that of a bandpass filter, I
don't think NR works that way. I think it works on correlation of the
pass band with a time delayed copy of the pass band. When the time delay
is a integral number of cycles of the tone signal, that tone adds up
while the noise doesn't because the noise isn't coherent. That can also
work for multiple tones, it gets a lot more difficult to create multiple
tracking filters for those multiple tones, yet NR should work for those.
At least every noise reduction book I've read in the library depends on
autocorrelation. The differences is how they choose the time delay.
Its a major problem in "noise" reduction when the roofing filter has a
narrow bandwidth. That tends to make the noise more correlated and so
makes NR less effective. And some things that would be considered noise
to listeners, such as computer hash, may not be random at all but be
well correlated and so the autocorrelation process won't eliminate it
but may enhance it. That can also apply to power line noise and possibly
lightning static from isolated events.
--
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
|