Best Way to EQ for Multiple Locations

[Tukkis]

Its all fine and great to get one seat Eq'ed to perfection, but what about the rest.

Is there a preferred method using REQ and Behringer EQ to measure and eq for multiple seats? I know with EQ's in some amps (Audyssey) it can take multiple measurements. In REQ you can take multiple measurements. But how do you go about figuring out a global average eq/filter combination? Obviously comprimises must be made.

Thanks

[brucek]

Is there a preferred method using REQ and Behringer EQ to measure and eq for multiple seats?

Since REW is so fast and easy to take measurements, the iterative approach works the best. Decide the positions you want to measure and create filters that are the best compromise across all positions.

It seems to me that those who treat their rooms enjoy the least differences across multiple positions.

brucek

[Ayreonaut]

I measured in four locations and used a spreadsheet to generate a median response, and then EQed for that.

[Sonnie]

You take the measurements on the various REW tabs and then select "Measured Averaging" from the "View" menu... which averages the responses that you measured.

[Tukkis]

Measured Averaging, displays the average in the AUX channel. If this is the target response, How do you then go about designing filters to achieve this response? What starting point do you work off?

[Sonnie]

That would be a good question for JohnM.

[brucek]

Measured Averaging, displays the average in the AUX channel

My understanding is that this isn't really the intent of the averaging panel.
The help file says: "Average Impulse Responses should only be used for averaging multiple measurements from the same location, it is likely to produce invalid results if used on measurements made at different locations.".

Anyway, it's really simple to accomplish what you want. Take a central position reading and enter the filters for it. Then take readings in several other positions and tweak the filters to arrive at a compromise. Whether a response is a few dB here or there is of little matter. It's those large reasonant peaks that are a problem. Try and reduce those as evenly as possible across the listening area...

brucek

[Tukkis]

Yeah I guess I could take the readings from the different seats and compare using the Measured Responses tab. Then fine tune the filters to find a average. As you said, it's more the big peaks that you want to avoid to get things smoother at all seats. It's never going to be perfect but it's better to have a range of seats that are pretty smooth than one seat thats perfect and the rest way off.

[Ayreonaut]

I would definately avoid using a straight average. You don't want a large null in one position to cause you to over boost in the other positions to compensate. The same is true to a degree with peaks, you want to tame all peaks but you don't want to over cut because of a peak at one seat and create big dips at the other seats.

I tried eyeballing the combined responses, using the mean, and finally the median. The median represents well what the responses have in common, and avoids over emphasis of the extremes.

[JohnM]

Sorry for the delay in chipping in here, I'm once again away on business and not home until middle of next week.

The frequency response average can be used as an overall target (not the impulse reponse average, which as Bruce noted is not suitable for different positions). However, it is typically better to view the various responses and make a judgement about what to correct depending on whether a particular peak is common to all or many locations (so definitely want to correct it) or only appears in one of a small number of locations (so may go for a smaller correction to balance out the positive effect where there is a peak versus the negative where there isn't). In general peaks are more objectionable than dips, so even corrections that are not needed in all positions are often worth applying. What you should never do is try to apply a boost unless it is to counter a shallow dip in all locations (and hence is countered by a fairly broad filter). Any narrow boost filter will make the response worse in most if not all locations.

[Tukkis]

Thanks for the replies guys.

I think I'll take the measurements at the different locations and print them out. Then circle areas that are common problems for the areas and go from there.