TOPIC:

Hi,

I am finally getting to play around with these wonderful design tools and I have hit my first major road bump.

I taken a few measurements of a loudspeaker I would like to use FIR Designer to correct its magnitude and phase response, but I am a bit confused about the best averaging mode to use for which I will appreciate your help. Here is the scenario;

I took 16 measurements (one speaker at the time) in a 4 x 4 grid around the my listening seat plus 1 reference measurement where my head would be located when listening.

The microphone location for the reference measurement is 1,3 meters away from the loudspeaker.
My intention was to average all 16 measurements by first time aligning all 16 impulse responses using the reference measurement as some kind of guide for impulse cross correlation, before importing into FIR Designer. I used the "Time Align all to Reference" feature for this.

I had therefore thought Power (min.phase + Ref. excess phase) mode would be a good option for this, but based on the help file description of the Power (with complex avg. excess phase) mode, it seems like it is the most appropriate for this job.

I am now confused about which one is the best for which scenario. Does Power (with complex avg. excess phase) mode execute impulse alignment first, before calculating the complex average? How does it calculate the excess phase without knowing the distance to the speaker being measured?

When is the Power (min.phase + Ref. excess phase) mode, the most appropriate mode to use?

Perhaps I am asking for a bit more detail on when to use each mode and how the Power (with complex avg. excess phase) works in practical terms.
Thanks.

Hi musan,

The two most common averaging modes are "Complex" and "Power (minimum phase)."

• Complex is as described, so if there is phase variation (even after time alignment) - particularly at higher frequencies - due to different different mic positions relative to all the drivers, then it can lead to some loss of energy in the magnitude of the average.
• Power (minimum phase) does a power average of all measurements - discarding the phase - then calculates the minimum phase response from the averaged spectrum. This is a fairly common averaging approach and using minimum-phase EQ to flatten this will also flatten the phase (except for any phase from crossovers).
• Power (min. phase + Ref. excess phase) is the same as Power (minimum phase) except the excess phase from the Reference measurement is added back in. It's a way to add back in non-minimum-phase phase from crossovers, so that you can - if you wish to - unwrap the crossover phase with a FIR filter (in addition to EQ'ing the loudspeaker). This isn't common and it's something we added in to see if anyone was interested in using it.
• Power (with complex avg. excess phase) is an unusual mode and came from a discussion with a pro loudspeaker manufacturer who wished to isolate and correct the crossover-based phase. In this mode the magnitude is a power average and the minimum-phase phase is calculated from this average magnitude. Then the minimum-phase phase is subtracted from the phase calculated from a full complex average. The remaining phase is just the excess phase and should contain mostly crossover phase.

Thinking over this some more, I realised there a useful averaging mode that's missing from the list. Specifically, power averaging for the magnitude and full complex averaging for the phase. This is a technique used by some measurement systems, particularly for time averaging. I've just implemented this mode and it will be released in Averager 1.1.0.

For your situation, I'd suggest trying the Complex mode, the Power (minimum phase) mode and the new Power (complex phase) mode when version 1.1.0 is released.

Kind regards,
Michael

HI Michael,

Just out of curiosity, how will the new averaging mode you will be implementing in the new version differ from the Power (with complex avg. excess phase) one?

I think I get it. The new mode is essentially the same as the Power (with complex avg. excess phase) except you are not subtracting the minimum phase from the phase derived from full complex averaging?

That's correct.