AFMG Network Forum

Langston Holland · Member

Hello again. :)

I would like to be able to record an IR from a loudspeaker and multiply that at a later date with a loopback measurement through a processor to adjust the magnitude and phase response of the loudspeaker.

At this point I'm able to use the "Additional Processing" feature of EASERA to automatically perform a Cyclic Move backward to the Abs. Max of the processor IR. But I'm having to manually multiply that measurement by a copy of the IR of the actual loudspeaker made previously. (The copy of the loudspeaker IR also has been cyclically moved to its Abs. Max)

I want to have the multiply process done for me automatically after each measurement through the loudspeaker processor so that I can run continuous measurements and see the result of my processor filter adjustments in something approaching real time.

Is this goal possible to achieve with EASERA? Maybe I'm approaching this in the wrong way and there is a better way to perform post processing EQ?

Bruce · Member

Langston,

I've not tried this yet, so this is just a "waving of the hands" explanation.

You will want to create a System reference for your loudspeaker. Once you have created the new System, then select the measurement (Cyc Moved) file for the Reference file for that System in the External HW dialog. Finally, check the System checkbox under Compensate when you do the measurement.

Langston Holland · Member

Hi Bruce:

Thanks so much for the idea - it just about worked, but I need a little help on two points:

1. The magnitude and phase curves are inverted when the loudspeaker measurement is used as a system reference. This is reasonable given its intended use. Nevertheless I need a method to have these curves appear in their normal orientation during measurements. Charlie hinted in a SAC post that I could divide the IR by an equal length dirac to achieve this affect. If this is the best way to flip the curves, how do I create a dirac for 44.1kHz/3 seconds for instance?

2. Making single shot measurements correctly include the loudspeaker system reference in the measurement. Yet I've noticed that choosing continuous measurements from the View & Calc module clears the system reference checkmark and processes each measurement without the reference. Is there a way to include the system reference when making continuous measurements?

Thanks again!

Charlie · Member

Langston,

To create a delta Dirac function in EASERA of the measurement length desired start with a measurement of anything (it doesn't matter what). Just make sure it's the right length (time or number of samples) and sample rate.

In the time domain place the markers at the begining & end of the time record.

Go to Edit, Set To and 1 Unity. This will set the entire time record to 1 (Pascals not dB).

Set the left marker at 21u (for 20.83 microseconds). This is one sample assuming that you are using 48 kHz sample frequency.

Go to Edit, Set To and 0 Unity. This will set the entire time record except for the first sample to 0 (Pascals not dB).

You now have a delta Dirac function. Save it for future use.

I don't know why the Compensate System checkbox gets unchecked during continuous measurements. I tried this here & the same thing happened. This also happend with the Compensate Microphone at Input checkbox. I suspect this is an undocumented feature. Laughing

One caution when comparing & porocessing DSP measurements with a loudspeaker measurement. To EASERA, or any other measurement system, the saved loudspeaker measurement will seem to be a single source (radiator). In reality the loudspeaker may have more than one source. This can lead to errors when EQing in the crossover region between sources.

In the extreme case take a simple two way loudspeaker with a severe notch at crossover. By applying DSP correction to this measured curve, the notch at crossover can be made perfectly flat. If this same DSP correction is applied to the input of the loudspeaker, as will be done in practice, the notch will still be there. This is becasue the notch is caused by the magnitude & phase relationship between the two drivers. Applying the same EQ curve to both drivers will not change this relationship between them.

Hope this helps.

Langston Holland · Member

Could not have been clearer Charlie - thank you! You ought to teach this stuff - wait a minute... You do. :)

The additional perspective on EQ'ing at crossover is something I've noticed at times in the past. It happens often with passive crossovers, but not active where I was able to adjust the dalay of one driver so that its phase overlapped the other with the same slope. I never knew why EQ would work as expected sometimes and not others. Now it's so obvious I'm embarrassed. :)