Car environment - minimum phase

In a car, does matching the EQ of both the left and right side using IIR filters mean that I'm matching the phase of each side? Some people are telling me that the car's environment is minimum phase and that matching the left and right side will match its phase response. I have never heard of such a thing. I know they would match amplitude but would highly doubt they're matching the phase which is something an FIR filter would actually do. Any help would be appreciated.

Hi John,

Let me give you a brief context explanation. A new DSP is coming out called the Helix Ultra. I was a bit disappointed that it doesn't come with FIR filters which I found to be very useful since it retains phase information. Another well-known member goes on to tell me, you don't need FIR filters since, in a mostly minimum phase environment like a car, you could just match the magnitude of the frequency response to match its phase response. I clearly thought what he said is wrong. I'm now on a quest to find out if this is true.

Here's what was said on another forum.




I'm pretty sure what he is saying is incorrect but I just want to get a second opinion from someone who is more knowledgable than me. I'm not trying to win an argument but really just understand what he is saying exactly. It doesn't make any sense to me.

When using an IIR filter and you're applying different EQ curves to both side, how in the world would the IIR filter (which causes phase shift in varying amounts depending on the amounts or cuts or boosts) find its way back to the correct phase at the listening position? What he is saying doesn't make sense to me so maybe some of you could make sense of it for me. Mainly the stuff underlined above.

I'm in the process of testing FIR filters in a car and I found that when the phase information is preserved I was able to hear much more detail in the sound stage and the transients are much sharper. When I switch my DSP over to IIR filters using the same EQ curves, I lose all the realism and everything collapses to a semi-smeary image. Both IIR and FIR frequency response curve closely matches since I had my processor do the FIR to IIR conversion (it was an option they offered). So I'm pretty sure that the phase retention feature of an FIR filter helps a lot compared to the IIR counterpart.

Basically, John, what I'm asking is... Does matching the magnitude of the left and the right speakers also match the phase response when using an IIR filter? I think what you said is, NO it doesn't. The magnitude of an IIR filter is linked to its phase response so if you were to EQ each side differently you will in effect cause phase distortion that doesn't match which will most likely result in a smearing effect. Correct?

Let's say I have the example below:


Let's pretend that I apply the EQ curves below to the L and the R side:



Then let's pretend that now both the L and R side are now matching with 1dB tolerance. The question now is the phase response of the L and R side now matched? Simply by just matching its magnitude. My brain says NO. But that other forum member is saying that in a minimum phase system like a car, the phase response will match as long as you match the magnitude.

Thanks in advance John.

Thanks a lot John! I think I'm understanding this better now.

So the key part is that the phase will only match if the frequency region is minimum phase. But a car's environment is still no better than a room where there could be many non-minimum phase regions that may cause dips or peaks when both L/R play together.

Do you think FIR filtering will be advantageous within a car environment? I really like how sharp the transients are compared to IIR filtering.

John Mulcahy said:

In a minimum phase system magnitude and phase are linked. If a minimum phase system has flat magnitude it will have flat phase (at zero or 180 degrees, depending on whether it is inverting). Flattening the magnitude corrects the phase under the assumption of minimum phase. Similarly, if both channels of a minimum phase system have the same magnitude response they will also have the same phase response.

When a reflection occurs doesn't have much bearing on whether the system is minimum phase, it is the amplitudes of the reflections that matter. If a reflection (or a combination of reflections) equals or exceeds the direct sound the system will exhibit non-minimum phase behaviour. There is more on that in the help on minimum phase. You can get a good indication of how close to minimum phase a measurement is by looking at the excess group delay plot.

Click to expand...

This is crystal clear to me now. Thanks again for all your help.

Flak said:

Excellent and detailed answer as usual by John Mulcahy
let me quote it!

Click to expand...

Flak - I spy Dirac Research....

I hear good things about the Minidsp CDSP 8x12 with Dirac Live.

What are your opinions about FIR filtering in-car environments? I have a Bit One HD which is a bit more manual work but any tips you could provide with regards to best utilize FIR filtering is highly appreciated. I'm currently using a combination of FIR for higher freqs (anything above 300hz) and IIR for the lower freq range. My results have been pretty good in terms of staging and clarity compared using solely IIR filters. I know that Minidsp and Audison are one of the few mainstream manufacturers that are offering FIR filtering. It's somewhat new for car audio so not many people have a good understanding of how to best utilize it.

One feature that I've been trying to utilize is phase alignment using all-pass filters. I was told that this is helpful for aligning phase within an active speaker setup. I have available to me 1st and 2nd order all-pass filters @ the crossover frequency (with Q and 360 degrees phase adjustment). I'm not even sure if this is useful when using FIR filters with steep crossover slopes such as 48db/oct (used in order to cross tweeters and mids lower for better front emission). My setup is a 3-way active front with a subwoofer.