Yes you are reading correctly. I spent weeks a year or so ago finding the best locations for the speakers. Trust me moving the speakers or sofa inches make a huge difference. The biggest thing that gets you on moving the speakers / sofa is the Bass response. In my room which is not even close to being symmetrical it is very hard to find a sweet spot. I worked with Mike Major from GIK for a few days moving things around based on his experience and when I was working with Mitch Barnett to generate a filter I moved things around and unfortunately the spot I am in now measures the best and sounds the best. One thing I have not played around with is the speaker toe in and I need to try an optimize that by looking at ECT data. Any recommendation on how to use REW to adjust Toe in would be appreciated.
""Moving absorber panels or adding diffusion / scatter panels around, you could look at the ETC, spectral decay, and spectrograms of the left and right channels -- check that they're not too dissimilar -- ~200 - 500 Hz in the left has a bit more moderate strong reflective energy than the right side between 15 - 35 ms (hint: look a the wavelet spect)""
Question from the above statement, how do I get view of Spectral decay in REW that is easy to make comparisons when I move things around? Can you take a snagit and show me how I should have it set up in REW?
Here are a few more pics.
With regards to the toe-in, I'd probably just look at the on- and off-axis responses -- both at the center listening position and the side seats. Pick whatever looks the most linear overall. Although, listening tests should be the ultimate deciding factor, IMO...
There isn't much that I can say below 500 Hz in your
spectral decay views other than some uneveness or lumpiness here and there between the left and right which is common enough in any room (treated or not). But, more importantly, there are no obviously prominent, long-decaying room resonances to be found. Really, that's as far as I can tell from the spectral decay graphs of your single-point measurement at the MLP.
Now, if we want to observe some more finer differences, we could use other views already mentioned before, and maybe one more extra:
Wavelet spectrogram (1/6 res, 35 dB scale, normalized):
Envelope time curve (ETC):
Filtered IR (morlet/wavelet transform):
None of these particularly look egregious or god-awful "broken" to me... other than being a little warped or "imperfect" -- in audibility terms... maybe sounding just a tad less "defined" perhaps.
Any remaining discrete late arriving energy peaks/reflections still look rather low and much less in level than the initial peak; none appear to significantly resonate over time.