rarewolf
100+ Head-Fier
I get the impression reviewers smooth FR measurements is because the peaks and valleys are assumed to be “relatively” inaccurate, and for making the visual representation a bit easier to understand… but are the peaks and valleys really inaccurate measurements?
In my field of image analysis, if we suspect the value of a pixel is affected by the sensitivity of the measuring device, and that the sample measurement is affected by error, there is only one way to know for sure and without smoothing (averaging, blurring). That is, to integrate a number of measurements.
The time to measure the entire frequency range doesn’t seem that long. Why don’t reviewers integrate (add, average in) something on the order of 16 or 32 independent measurements? If the peaks & valleys are truly errors due to device sensitivity, you should end up with a more convincing representation of the speaker’s frequently response(?) Otherwise, what am I missing?
In my field of image analysis, if we suspect the value of a pixel is affected by the sensitivity of the measuring device, and that the sample measurement is affected by error, there is only one way to know for sure and without smoothing (averaging, blurring). That is, to integrate a number of measurements.
The time to measure the entire frequency range doesn’t seem that long. Why don’t reviewers integrate (add, average in) something on the order of 16 or 32 independent measurements? If the peaks & valleys are truly errors due to device sensitivity, you should end up with a more convincing representation of the speaker’s frequently response(?) Otherwise, what am I missing?
