Anyone use REW to detemrine EQ for mains?

[robbnj]

I'll open with: Yes, I know that EQ'ing mains is a suboptimal way to improve things, but this is just an office system that I am tinkering with, and want to play/learn/enjoy.
I'll also add that I did a solid search for this topic with multiple keywords, and didn't find any threads on it. If it's been hashed over a hundred times, pls. let me know, and I'll go to those threads!
Some of you may recall my previous work with these speakers from another thread.

So, I want to do some EQ'ing on my main speakers with a DSP. They sound good, but not quite right. The "problem" is that I can't just arbitrarily move some EQ sliders until I get what I want. I am manually entering the parameters into a programmer, dumping it on the DSP, rebooting, ete, etc, before I can listen to the changes.
It's very tedious, and a quick A/B comparison is marvelous fun (going to try convolution this weekend, when I can be noisy in the house).

The second hitch is that I know the speakers sound pretty good, yet don't sound quite "right", but I'm not able to say "It sounds flat in the range of 1,500 Hz" or "It's running hot at 3.4KHz".

So, I'm hoping to use REW to identify what might be problem areas, and get a more educated running start on builing filters.
I tried making a few EQ curves, but now REW seems to ignore the range above 1,500Hz, even though I select full-size speaker, 60Hz cutoff, and the range of 20Hz-15Kz . It even does it when I change the range to 1KHz-15KHz. It has worked in the past for me, but I must have mucked soemthing up. I even see it build filters above 1.5KHz, then erase them.

There has to be something basic that I am totally missing, but I can't see what I can't see, I guess.
Can anyone see my error/s?
And, can anyone who is looking at the graph tell me what I might be perceiving as "not right" with the sound?

 

[Wayne A. Pflughaupt]

Well, the name is “Room” EQ Wizard. 1kHz is beyond the range where “room“ affects (i.e. modes and such – John, correct me if I’m wrong), so that might be why you’re not having any luck up there.

For starters, your upper-range graphs probably should be single-speaker, not both.

But, based on your graph a couple of things jump out.

One is the broad depression between ~1.5 – 5 kHz. A one-octave filter at about 3 kHz or a bit lower, boosted a few dB could help that. NOTE, when applying filters, pay attention to the “outer parameters” of the area you want to EQ. If there is a rise at 1.5 kHz and 5 kHz, then the filter is too broad.

Second, the spike in response at 5 kHz. If these are desktop speakers, it might be better to reduce this instead. This will take a fairly tight filter maybe 1/4-octave cut a few dB. Again, watch the edge frequencies – if they drop, the filter is too broad.

Keep in mind, the goal is an improvement in sound quality, not a pretty graph.

BTW, waterfalls are only useful up to 200 Hz or so – not for upper frequencies.

Regards,
Wayne

 

[John Mulcahy]

REW seems to ignore the range above 1,500Hz

You don't show the target, but REW won't attempt EQ above the last frequency where the response drops below the target. The latest builds have an option to apply shelf filters, which might allow more EQ to be applied, but the main issue is probably that a sensible target would have a room curve component to reflect the downward trend in the response at higher frequencies.

 

[ddude003]

You might look at the B&H house curve or a Harmon house curve...
Here is the B&H... I use the B&H centered around 75dB instead of 80dB as shown below... You can google for the Harmon curve...

 

[robbnj]

Well, the name is “Room” EQ Wizard. 1kHz is beyond the range where “room“ affects (i.e. modes and such – John, correct me if I’m wrong), so that might be why you’re not having any luck up there.

For starters, your upper-range graphs probably should be single-speaker, not both.

But, based on your graph a couple of things jump out.

One is the broad depression between ~1.5 – 5 kHz. A one-octave filter at about 3 kHz or a bit lower, boosted a few dB could help that. NOTE, when applying filters, pay attention to the “outer parameters” of the area you want to EQ. If there is a rise at 1.5 kHz and 5 kHz, then the filter is too broad.

Second, the spike in response at 5 kHz. If these are desktop speakers, it might be better to reduce this instead. This will take a fairly tight filter maybe 1/4-octave cut a few dB. Again, watch the edge frequencies – if they drop, the filter is too broad.

Keep in mind, the goal is an improvement in sound quality, not a pretty graph.

BTW, waterfalls are only useful up to 200 Hz or so – not for upper frequencies.

Regards,
Wayne

I hadn't even considered the software name vs what I was trying to do, lol!

I will do some test adjustments in the areas you think may be causing what I'm hearing. I'm not looking top create a good graph, per se, but I know that measurements/graph can halp explain what I might be hearing (but cannot describe with accuracy).

I am working on each speaker indendently, b/c they are in a non-symmetrical environment. The waterfall was just what showed when I let the program build the filters. I'm not even CLOSE to worrying abou that with this system. Just looking to sweeten it up a little.

Thank you, Wayne!

 

[robbnj]

You don't show the target, but REW won't attempt EQ above the last frequency where the response drops below the target. The latest builds have an option to apply shelf filters, which might allow more EQ to be applied, but the main issue is probably that a sensible target would have a room curve component to reflect the downward trend in the response at higher frequencies.

I try to put my targtet in the "middle" of the existing response to balance out how much boosting or cuting gets dine, but now I understand how that could cause the no-EQ situation I am experiencing. I'll experiment with moving that around, and see if I get any filters above 1.5KHz that way.

I'll also try again with a Harman curve. I had included one at one point, but thought maybe that was causing the "issue", so I eliminated that variable.

Thank you, John. Always appreciated.

 

[robbnj]

You might look at the B&H house curve or a Harmon house curve...
Here is the B&H... I use the B&H centered around 75dB instead of 80dB as shown below... You can google for the Harmon curve...
View attachment 87023

Thank you! My test curve looks something like this (just grabbed one off the web). I haven't really delved too deeply in a "good" house curve yet.
I WILL be, though!
63 4.9
80 3.7
100 2.5
125 1.3
160 0.6
200 0.2
250 0
1000 -1
32000 -6

 

[ddude003]

That looks like a Harmon style type curve... And that may get you in the ball park... At this point its all about what sounds good/best to you...