Normalized Power Averaging

Normalized power averaging attempts to sidestep the limitations of power averaging in magnitude response measurements by ensuring that all data traces going into the average are approximately equal in overall level before calculating the average. It works by calculating a single-figure decibel average of all frequency data points within a given frequency range for each trace being averaged and then adjusting the overall level of each trace so that their average level within that range is identical. 

Transfer function power averages are adjusted so that their average decibel magnitudes for the range of 225 Hz to 8.8 kHz are all 0 dB. When averaging transfer function data, we can assume in advance that we want an averaged magnitude response (as opposed to an average level) and we have a natural reference point (0 dB) to adjust the levels to and so no additional intervention is required. The caveat is that you are also assuming that the system under test has significant energy within the normalization frequency range and if that is not true, e.g., when measuring a subwoofer, you may not get the expected result. In that case, a dB average with coherence weighting may be a better choice. 

Normalized RTA power averages work similarly to transfer function power averages except that Smaart cannot assume in this case that you want a normalized average; there isn’t a natural reference point to adjust the levels to, and the normalization range is different. Normalized RTA power averages use 125 Hz to 4 kHz as the normalization range to better accommodate cinema systems and in this case, you must designate one of the traces or measurements being averaged as the reference level.

Next: Coherence Weighted Decibel Averaging (Transfer Function)