Skip to main content

I just got to play with a demo of the Massenburg Hi-Res Parametric EQ. It sounds pretty nice to me, but I would like to know more about it. Is it IIR or FIR? Linear phase or minimum phase? Nothing I can find online mentions these details, but maybe someone here has some inside info.


bouldersound Sun, 09/18/2011 - 21:32

Buying it isn't really up to me. The person who runs a studio I work in is enthusiastic about it and may find a way to fit it into the budget. I don't object on technical or artistic grounds because it sounds pretty darned nice, but it's a big expense for a little studio like ours and there are other things we could use.

I'll try to explain, but I'll probably just reveal the limits of my knowledge.

Minimum phase filters are the common form that all(?) analog and most digital filters follow. They induce phase shift that varies with frequency (as response varies with frequency). That can nudge parts of the signal slightly out of time alignment* with others. This doesn't really concern me that much but I wouldn't be surprised if it can smear transients. It could be more of a problem if some sort of parallel eq arrangement were used, but I don't think that's common. The upsides are that the digital versions induce little delay and take less processing power.

Linear phase filters don't throw anything within the processed signal out of time alignment. But they can be "ringy" (including "pre-ring"), require much more processing power, can only be done digitally (I think) and delay the processed signal quite a bit.

*Note that I use "time alignment" rather than "phase". That's because phase is always related to a given frequency, and here we are comparing one frequency to another.

Hopefully someone with more solid understanding will come along and correct as needed.

Boswell Mon, 09/19/2011 - 09:50

All filter responses are complex vectors (have real and imaginary parts), and these can be respresented in many different co-ordinate systems. For example, it's common to show an amplitude response and a phase response against frequency in Cartesian co-ordinates, or it could be an amplitude response against angle for every relevant frequency, as you would do for a microphone.

A linear phase filter has a specified frequency response but the phase response against frequency is a straight line, and that means it represents a pure time delay. A minimum phase filter is one that is not a pure time delay (i.e. its phase response deviates from a straight line), but that the deviation from the pure time delay is a minumum as measured by some criterion.

Non-recursive (FIR) digital filters are inherently linear phase. The "pre-ring" seen with this type of filter is not a predictive response ahead of an input signal, but is a reponse that occurs in time after the input signal but ahead of the specified pure time delay for that filter. It is anti-symmetrical with the filter response after the time delay.

There is no particular reason to prefer one over the other unless the aim is to approximate an analog filter as closely as possible. Since analog filters cannot be pure time delays (although a Bessell characteristic comes close over 80% of the passband), recursive digital filters (IIR) are often used in these cases. The "pre-ring" of a FIR digital filter is never seen in a well-designed analog filter, so FIR digital filters start with a disadvantage when comparing IIR/FIR digital with analog filtering techniques. Whether artifacts such as these are audible is another (entirely different) question.