Richard Evans - 2016-04-01

The old method

In the previous method, all the side path audio went through 2 filters.
A fixeg high pass filter, to limit the frequency range of the noise reduction system.
Then a sliding high pass filter system, used to vary the amount of noise reduction.
This is how things worked in the analog circuit the program was based on.

Filtering will make some change to the phase of the audio, but I had assumed that
this potential problem had already been considered, so I didn't give it much thaught.

However, after creating some graphs, to see how well the program worked, it was
becomming clear the phase changes were an issue.

The new method

The idea was to try and pass the side path audio through just one filter, instead of
two. Which would mean less of a phase change.

Firsly I created a lower limit, to the frequency of the sliding filter, so that made
it do some of the job of the fixed filter. This is not quite complete, as the sliding
filter is a bit more complicated. The audio actually passes through a variable
attenuator, and then the sliding filter effectivly processes audio that does not get
through the antenuator. The solution is quite simple, basically pass the audio that
does get through the antenuator through a fixed high pass filter.

The combination of these two changes, mean that the first fixed high pass filter is
not needed. The audio effectivy passes through either the sliding filter, or the
fixed filter, but not both. So the filtering of the audio is similar, but with less
of a phase change, as the audio effective goes through only one filter.