Originally Posted by EB1000
Not sure if this question was asked before, but I'm curious to know. Is the post-calibration ("after") spectrum plot an assessment/prediction or actual validated results? I
Just to add to Kal's response, this is Chris Kyriakakis' explanation of the "predicted" after response:
"The "after" is as real as it gets. It is the result of convolution of the clustered "before" responses with the filter that MultEQ calculated. Linear systems theory tells us that this will be the clustered "after" response. As you said, in order to obtain this same graph through measurement you would have to re-measure with the filters running and the mic in the exact same (to the cm) locations. You can see why this is impractical."
So it's less of a "prediction" than it is a mathematical calculation which will be true IF you measure in the EXACT same spots, use the EXACT same weighting algorithm, etc.
This reinforces why it's extremely difficult to reproduce those graphs with a tool like REW since it's nearly impossible to replicate the precise measurement points and of course REW doesn't have Audyssey's proprietary weighting/clustering algorithm. If the mic moves a few inches all of the little peaks and dips will shift slightly from the input data that Audyssey used. And of course you're also using a different microphone in each case!
So I would suggest that if you are using REW, take a few measurements in the general area of the "bubble" you measured during Audyssey, average the responses, and if the general shape of the average looks pretty close to the Audyssey target curve, then it's working as intended.
Although this isn't a multi-position average, this is a measurement I took a few months ago of all 7 "ear level" speakers with full range Audyssey and the Reference curve, it's 1/3 smoothed to focus on the shape but you can see they adhere pretty darn well to the target (and FYI, the MRC dip is indeed left on for the 4 surround speakers but turned off on the LCR, which is why there's a deviation around 2kHz):