Efficient implementation of dynamic EQ?

[synthpark]

Hi guys,

I regard the naive implementation where some analysis filter (here, lowpass) is used to shape the detector signal, which is used to control the gain of some shelf (here, lowshelf) in the usual way, leading to one filter calculation per sample, as way too much effort, especially if you want to make a system for a lot of channels. This is true when comparing the effort to a normal EQ which might require one BIQUAD filter recalculation every 1-10ms.

Is there any well-known efficient implementation of a dynamic EQ?

For example, for the positive part of the peak filter, one can construct the filter using two terms where the gain controls the amplitude of one term. In this case you don't have to calculate any filter per sample at all.

But for the negative part it seems impossible, also for other filter types like shelfs.
Am I wrong? In this regard, filter banks for multiband compressors seem much more CPU friendly, but not an expert opinion.

[Burillo]

I think there's a DSP section for these sorts of questions?

viewforum.php?f=33

[synthpark]

I think there's a DSP section for these sorts of questions?

viewforum.php?f=33

alright ...

[Ah_Dziz]

I use the Melda dynamic EQ and it's fine even with every band working and all the visualization going as well. Even a very well constructed fft based EQ barely makes any of the various cpus I have take notice. I'm not a big fan of sloppy coding but there's not much of an issue with poorly coded eqs out there dynamic or otherwise.