Which smoothing?

[dcashel]

Which smoothing in REW is recommended when looking at frequencies below 100Hz and above 100Hz?

 

[sm52]

There are no recommendations for this case. As you prefer.

 

[Wayne A. Pflughaupt]

I like 1/3-octave smoothing above 100 Hz because it generates a graph that looks like what you actually hear. Finer-resolution smoothing, especially beyond 1/6-octave, gets scary-looking graphs that compel people to go overboard with the equalizing, chasing every little ripple in response. Below 100 Hz no smoothing is best. You can utilize REW’s variable smoothing feature to get both in a single graph.

Regards,
Wayne

 

[Harrycr]

I like 1/3-octave smoothing above 100 Hz because it generates a graph that looks like what you actually hear. Finer-resolution smoothing, especially beyond 1/6-octave, gets scary-looking graphs that compel people to go overboard with the equalizing, chasing every little ripple in response. Below 100 Hz no smoothing is best. You can utilize REW’s variable smoothing feature to get both in a single graph.

Regards,
Wayne

What frequency range do you EQ?

Are you still using the Z9

 

[Wayne A. Pflughaupt]

I’ve typically EQ’d full range, but at this time have none on my current speakers (although they could us a bit in the 4-6K range). Center speaker is mainly EQ’d to address program material shortcomings.

Still using the Z9, but as an outboard multichannel amplifier. Currently using the Z7 as front end.

Regards,
Wayne

 

[teyhyrh4r]

I look at the envelope of the unmoothed graph. we don't hear like any smoothing since our ears filter dips out, while any smoothing treats dips and peaks the same; creating an average

 

[John Mulcahy]

while any smoothing treats dips and peaks the same; creating an average

No, smoothing is applied to squared magnitudes so peaks have a greater influence than dips.

 

[teyhyrh4r]

No, smoothing is applied to squared magnitudes so peaks have a greater influence than dips.

ok, but if any smoothing existed that shows what we hear (in-room at least, the smoothings were created for loudspeaker meassuring) it would be flat for a neutral response.
in fact if I eq a var smoothed graph of my LP with a rather unfamous target found in a book of Dr Tool

it aprouches a flat envelope on the unsmoothed graph

I think house curves are just adoptions to the distorsions the smoothed graphs produce, and we should use unsmoothed graphs

 

[sm52]

If you use the minimum smoothing 1/48, then such a graph is already readable and is as similar as possible to an unsmoothed one, including the phase graph. Greater smoothing, 1/24 gives better readability, but the phase graph is already far from the original. And an unsmoothed graph is very difficult to read.