Home Theater Forum and Systems banner

1 - 6 of 6 Posts

·
Registered
Joined
·
910 Posts
Discussion Starter #1
Does the Auto EQ function setup filters differently for average room spl responses vs say a single room spl response? When I enter an average mlp response for main speakers into auto eq the suggested filters do not fully reflect what is required to correct response to get me close to target curve. Running a room sweep after applying filters based on an average supports this as the average room measurement at the mlp is not close to the expected target curve. I find I have more success with a single measurement set of filters as the corrected response more closely resembles the target curve.

Just curious to find out if my setup is wrong for average room measurements?
 

·
REW Author
Joined
·
6,711 Posts
The EQ process is very similar, there are slight differences because the averaged measurement does not produce an impulse response but it should make little difference to the result - as an example, try comparing the result of an individual measurement with the result from averaging a measurement with itself. The variation should be no more than repeated runs on the same measurement.

Is the average after applying EQ close to REW's prediction?
 

·
Registered
Joined
·
910 Posts
Discussion Starter #3
Thnx. I will take a look later tonight.
 

·
Registered
Joined
·
910 Posts
Discussion Starter #4
Hey John

I took an rescent single measurement and then averaged it. Ran it through Auto Eq the over laid the single vs average measurement. The single full range measurement yielded 1 filter at
6686hz. The average measurement version of the single measurement was very close at 6670 hz also only 1 recommended filter. Seems very close. I recently had a situation where a single measurement yielded 5 filters & the average yielded only 1 with the same house curve. Is there perhaps something in my setup that I should look at that could skew average measurements results?
 

·
REW Author
Joined
·
6,711 Posts
The EQ process for equalisers which don't allow smooth variation of filter parameters (i.e. some parameters only vary in steps or over a limited set of values) use a random perturbation optimiser which can give different results run to run - using the 'Generic' setting tends to be more consistent as it can use a more deterministic optimisation method. You could try generating filters with Generic, then changing the equaliser type to the XP2040 to have REW convert the filter settings and either try those, or use them as a starting point and use the 'Optimise Gains and Qs' or 'Optimsie Gains, Qs and Frequencies' controls to better match them to the equaliser.
 

·
Registered
Joined
·
910 Posts
Discussion Starter #6
Thnx John
I worked on it today and ran several measurements over 3 seats to arrive at an average response that I used to generate auto eq filters. Once loaded I reran the measurements and the resulting average closely matched my target curve. When I have some time over the weekend I will try what you suggested & substitute in the generic eq to see if that changes anything. Not sure why it didn't work for me earlier but I suspect I made some errors during the eq setup process. Your comments regarding smooth variation of filter parameters sounds to me like it either works or it doesn't. As I had some success today
leads me to believe that I made some errors during the eq setup process earlier this week. I will let you know the outcome. Appreciate your input.
 
1 - 6 of 6 Posts
Top