Another long and exhausting day working on the EQ programming. I'm getting closer but it's been far more difficult than I imagined. Here is a trace from my EQ at the moment. There are test peaks +3dB at 225hz, -6dB at 250hz, +6dB at 900hz, -inf. at 7200hz, and -inf at 15000hz:
Here's the same arrangement on my Sony plugin:
Pretty similar. If anything, mine is a little more accurate in frequencies and a bit smoother - this is good. The biggest deficiency are the high frequencies, mine just won't fall off! My plan of using a bank of low-pass filters was flawed, it couldn't cope with zero values (they could never be multiplied up again). So In the end I have my old plan of lots of high and low pass filters. I didn't appreciate the problems that the Q factor can cause until they started to appear. Too little Q and the bands vanish with no effect. Too much and everything wobbles alarmingly at the corners, like that little dip I have at 2khz. The little bump at 6khz that the Sony trace has is probably caused by the same thing. Both are supposed to have -infinity as 7200hz remember - zero evidence of that on the Sony trace, and only that tiny bump on mine. If I increase the Q it works better but everything else gets more wobbly.
This not bad, but those low and high ends need tackling, so I plan on applying separate low-shelf and high-shelf filters there.
There are 4 filters per band, so a whopping 40 filters for this; two low pass, two high pass, a lot of processing, but I expected this to be the case. Whether all of these experiments will be worth it is uncertain. Such is all science.
