about average function

[jlohl]

hi all,
I'd like to know more about how the REW average function is calculated.
Especially when some of the averaged signals are non-minimal phase, how is the calculation done ?
Thanks for any info.

[JohnM]

It is the arithmetic average of the spl values of the selected traces.

[jlohl]

thanks a lot for the info

you've done a great job with REW !
have you ever considered the feature of extracting the minimum phase part of an IR ?

[JohnM]

Minimum phase extraction is on the features list to put in at some point, but there doesn't seem to be much demand for it. What would you plan to use it for?

[jlohl]

Minimum phase extraction is on the features list What would you plan to use it for?

Generally, if I use a hardware parametric EQ or any software IIR, I prefer only to correct the min phase response.
To do this, I measure at different places around listener. Then I extract min phase responses (together with time windowing) of all the measured REW curves with DRC and reimport it to REW.
At the end, I compare :
- all untreated REW measurements
- same measurements but with min phase extraction
So I can check what accidents are common to all curves and I only correct those ones.
It's some work but can give goood results.

[JohnM]

Can't say I follow what you are looking for. The frequency (magnitude) response of the minimum phase impulse is identical to that of the mixed phase impulse. What differences are you seeing in the responses you have exported, processed and re-imported?

[brucek]

Generally, if I use a hardware parametric EQ or any software IIR, I prefer only to correct the min phase response

My understanding is that REW only considers peaks that are modal resonances, and simply won't qualify peaks in the Find Peaks routine that aren't minimum phase......

brucek

[jlohl]

Can't say I follow what you are looking for. The frequency (magnitude) response of the minimum phase impulse is identical to that of the mixed phase impulse. What differences are you seeing in the responses you have exported, processed and re-imported?

sorry, I've not been clear enough :
I use the DRC processing mainly to do a frequency dependent windowing with quite short windows (equivalent to 3 to 5 periods) to keep most direct sound and avoid later reflections
And in the same process, I prefer to do a min phase extraction because when I do a correction with another min phase device, then when I re-measure after correction, I'm quite sure that it fits with the predicted curve.

My understanding is that REW only considers peaks that are modal resonances, and simply won't qualify peaks in the Find Peaks routine that aren't minimum phase......

that's very interesting, JohnM, can you confirm ? and can we know which peaks aren't min phase without using the Find Peaks function ?

thanks everybody

[JohnM]

brucek wrote:

My understanding is that REW only considers peaks that are modal resonances, and simply won't qualify peaks in the Find Peaks routine that aren't minimum phase......

Peaks are characterised to identify resonances but that doesn't include minimum phase dependencies - as I said, the magnitude response is the same in any case. The next version of REW, all being well, will accurately identify the poles of the modal resonances, however, and allow more precise correction that can target only those resonances. It may also incorporate some Common Acoustical Pole modelling to better isolate features that affect several measurement locations. All quite a ways off though.

[brucek]

I use the DRC processing mainly to do a frequency dependent windowing with quite short windows (equivalent to 3 to 5 periods) to keep most direct sound and avoid later reflections

Interesting, but wouldn't the very short gate time limit the lowest measurable frequency. For example, wouldn't a gate time of 5msec limit you to about 200Hz (f=1/gate period). That's above modal resonance possibility.

5msec would be a boundary ~0.86 meters away (d=(time*speed)/2). That distance could be the distance the mic is situated from the floor at the listening position and so a reasonable reflection free period. I don't see how you can avoid measuring reflections, or am I way out to lunch?

brucek

[jlohl]

Interesting, but wouldn't the very short gate time limit the lowest measurable frequency

no, this time is depending on frequency, that's why I mentionned "equivalent gate time" in periods : for example, if it is defined at 3 periods then it will be 30ms at 100Hz, 60ms at 50hz, 3ms at 1000Hz, aso...
So you don't really do full modal correction (would be more depending on listener place) but you get good first reflections correction (that is less listener place dependant)

The next version of REW, all being well,...

so please don't loose time reading this forum, write some code lines instead....

[DrWho]

I know it's a bit outside the scope of 'room' eq wizard, but minimum phase extraction would be extremely useful for measurement in performance halls (live sound reinforcement).

In larger rooms/venues, you can't just EQ the modal resonances or really implement any EQ since the magnitude changes very dramatically with the window used and the position of the mic (which is due to the number of reflections happening at very different times). However, minimum phase aberrations show up consistently...

The notion extrapolates back to the home though, just at the higher frequencies where the room is still effectively "large". I've got an interesting paper here somewhere that I can post when I find it that talks more about it...