Measurement in DRC

[hajame]

Hi, I am a newbie in DRC. I have recorded two sweep samples. One of them was recorded at higher speakers output volume. Two samples resulted in different impulse response. The more loudly recorded one has more high magnitude spikes before the main spike(the highest one if I am correct).

My questions:
1. Which sample should be more accurately describing the actual impulse response of my system?
2. Will the highest spike be the result of strong wall reflections, but not the actual main spike?
3. For the correction filter, is it a better for the main spikes of the left and right channels to align as closely as possible in time? Can this be achieved only if the distances from the microphone to both speakers are exactly the same?

Thanks~

 

[breez]

Hi

Strong spikes before main impulse in this case (logsweep) are nonlinearities. While it seems the method is robust and usable with nonlinear systems I'd probably use the measurement with lower SPL.

If you can plot the data you could check the noise floor of the samples and see if there are great differences.

 

[hajame]

Thanks breez~

I have used frequency plot of CoolEdit to analyze both samples and correction filters. Both of the filters make correct compensation of the respetive impluse responses. So I believe that DRC does not calculated the wrong filter. But the frequency response of both samples are slightly different. I think I would use the filter generated in low SPL for low SPL listening and that in high SPL for high SPL listening.