Is there a way to average a particular Decay Time and generate filters on that

[mswlogo]

I'm trying to come up with better and more repeatable process of generating filters for my Meridian processor.

For more info read this http://www.meridian-audio.com/w_pape...ection_scr.pdf

Meridian believes in attacking just slow decaying frequencies.

I'd like to take all my measurements and average just all the 160ms LF Decay graphs. Then generate Filters on that. Currently it only Averages and builds filters off T=0ms.

You don't want to get rid of large peaks (yummy energy in a hometheater) if they decay quickly.

[OvalNut]

You don't want to get rid of large peaks (yummy energy in a hometheater) if they decay quickly.

Sure you do. You want the frequency response to tracks your target closely, and you want quick decay across the whole range. That's how you get audibly accurate bass.


Tim

[DrWho]

If you want quick decay, then listen in an anechoic chamber. I think you'll find the results most displeasing!

[OvalNut]

Well, I'm certainly not the sharpest knife in the drawer, but I've done a respectable job with bass trapping in my room. I know that it is much better now than the ringing mess it was before, that's for sure. I've got everything above about 35hz very much in check, and I wouldn't have it any other way. The bass just jumps out of nowhere now it seems. It hits you, hard, then it's gone like it was never even there. It's really an incredible sensation, and greatly articulated for music.

That's how I like it, and others most certainly may disagree.


Tim

[brucek]

Meridian believes in attacking just slow decaying frequencies.

Yeah, and these will usually be caused by room modes. The theory is that if you apply an EQ filter with the same Q and opposite gain of a room mode, it will completely counteract the effect of the mode in both amplitude and time. I did a post on this here that you can read and see what you think. REW attmpts to match all room modes with its filter recommendations.

I think you're on the right track, but I also believe that the response should be smoothed as much as possible in all domains including amplitude, but that's personal choice. Certainly you don't want to reduce the decay until it sounds like an anechoic chamber, but if the decay is too long it also won't sound right.

brucek

[mswlogo]

brucek wrote:

Yeah, and these will usually be caused by room modes. The theory is that if you apply an EQ filter with the same Q and opposite gain of a room mode, it will completely counteract the effect of the mode in both amplitude and time. I did a post on this here that you can read and see what you think. REW attmpts to match all room modes with its filter recommendations.

I think you're on the right track, but I also believe that the response should be smoothed as much as possible in all domains including amplitude, but that's personal choice. Certainly you don't want to reduce the decay until it sounds like an anechoic chamber, but if the decay is too long it also won't sound right.

brucek

I think OvalNut has it right. It's ok not to be flat. It's ok to have some reverb too. I did a full LF EQ (total of 8 filters across 3 speakers) I set the Target Level fairly high (80db) so it wouldn't be too agressive with filters. It sucked the life out the whole system. No LF impact at all now. The filters were few but fairly large. I started to do a target of 85 but the filters are still pretty large.

I know there is a lot of debate on this. But you want some reverb but too much is boomy. Not every large peak will ring. I just want to go after the ones that ring. Goals might be a little different for Music versus HT. If a an explostion isn't "flat" in a movie does it really matter !!

[mswlogo]

Bruce can you give me some advice then.

I did did 3 measurements on 3 speakers across listening area.

I averaged the 3 measurements for each speaker. Set Target to 80 and the did Find Peaks, Optimize and Adjust.

See below of one speaker.

[brucek]

First I wanted to add a note about graphs.
Always use LOG and not LIN graphs (click the Freq Axis icon for that).
Use a horizontal scale of 15Hz-200Hz for the subwoofer responses.
Use a verical scale of 45dB-105dB.

can you give me some advice

Well, I think you first need to establish how your Meridian defines bandwidth and the entry paramaters in BW or Q etc.
This is important because it will determine which equalizer type you'll choose in REW. REW's filter recommendations and entry units are consistent with the mathematics that each equalizer type uses to define bandwidth. Strangely enough, each type seems to have their own idea on how it's defined.

Here are the formulas that are used:

---------------------

For the DSP1124P, the bandwidth parameter is entered as BW/60. It defines bandwidth as:

Bandwidth (Hz) = centre frequency*(BW/60)*sqrt(2)

So, the Q formula becomes:

Q = 60/[(BW/60)*sqrt(2)]

For the FBQ2496 the bandwidth control adjusts in 1/60 of an octave steps from 1/60 to 5/60 of an octave and is entered in decimal, and so the formula becomes:

Q = sqrt(2)/BW

SMS
The bandwidth is entered as Q and the filter bandwidth in Hz between the half gain points is given by:

Bandwidth = centre frequency/Q

TMREQ
The filter bandwidth in Hz between the half gain points is given by:

Bandwidth = centre frequency/Q

R-DES
The filter bandwidth in Hz between the half gain points is given by:

Bandwidth = 1.766*centre frequency/Q

Generic
The filter bandwidth in Hz between the half gain points is given by:

Bandwidth = centre frequency/Q

--------

Anyway, you can see the problem there that you need to be aware of. Play around with the different eq types in REW for the same filter and see the different types of bandwidths that are entered and try and match what your processor uses...

brucek

[DrWho]

Yikes, I would turn down the gain on the sub first! If you get a hole in the crossover frequency, then play around with the phase and speaker distance settings...

[mswlogo]

First there is no sub. They are 3 full range speakers (that was just one). Everything on the audio system is set flat. It's the placement and room doing most of that. But since they are full range speakers I have little choice where to place them.

I do normally use Log scale. I was comparing the effect of Filters in Meridian Software vs REW to see if the parameters I was using from REW is compatible. Using Generic EQ the Gain seems the same, and the Hz (Width) displayed in Meridian seems to match exactly. I was Matching both vertical and horiz scales to look at the shape of the filters. The shape and interaction of filters looked Identical to generic.

Note that I was calibrated at 84 so everything there is 9db high than what you are used to looking at.

I will Cal to SPL 75 and use Log scale in the future.

[mswlogo]

Ok, I did it by the book this time. Still way to much filtering.

Cal to 75 SPL. 3 Mic Positions across the couch.

If I implemented these filters I'd have NO bass.

I include a couple shots of entering filters in Meridian software.

[Wayne A. Pflughaupt]

mswlog,

REW is recommending filters on the assumption that you’re using a subwoofer. With a sub, after flattening out peaks with an equalizer, you can easily turn it back up to compensate for the loss in output.

Naturally, in your situation that’s not possible - so yes, it looks like REW’s recommended equalization is going to leave you with much less bass. I think what I’d do is simply smooth out that 50 Hz peak, maybe along the 80 Hz line or a little higher. I’d start with a single broad filter centered between 30-60 Hz and adjust the bandwidth as you cut, and possibly the frequency center as well, so as not to make the area between 60-100 Hz any lower than it is.

Actually it makes no sense to do averaging in your situation. With three full-range speakers, it’s going to require an equalizer for each speaker, since each is a discrete channel. So it makes more sense to equalize each speaker independently, according to the response it’s generating for its location.

Regards,
Wayne

[mswlogo]

Ah! thank you Wayne for the explaination. That's sort of what I've been doing. I created Filters for 92, 90, 88, 86 and I can dial through them while listening to music to see what's the best trade off.

I attached a measured graph with EQ in place. Even though it looks like I should still have a lot of bass that's not what it sounds like. (If I had sub it sound like I shut it off). The prediction from REW was off quite a bit to, but it was in the correc direction. But the impact is huge.

I think I can boost the "Virtual Sub" with the Meridian. It does everything virtually in digital form then distributes the Bass to the speakers that can handle it. But I don't know if it does that (a boost Sub) before or after EQ. Would it matter?

Wait if it's the room that's the problem wouldn't I be back where I started if I boost the sub?

[JohnM]

You can just load a house curve to move the target up at the low end so REW offers less cut down there. Ignore the filter recommendations between 100 and 200Hz for the moment, just apply the largest filters at 48 and 43 and see how that is.

If you want to see what the response looks like after 160ms or so and work out some filters based on that, just use the IR Windows dialogue to move the window ref time to 160ms and reapply the windows. The difficulty with setting up filters for the later part of the response is knowing what the correct reference level is, so you'll have to tweak the target level by hand until it looks right before asking REW to work out some filters.

[mswlogo]

Totally awesome thanks.

I attached the differences you get in generating filters at two different references.

This is what Meridian does, they guess at the IR Window Reference (they call it Target Decay) to begin analysis at and you can override it. But Meridian doesn't let you adjust the target reference level. But now I can with REW.

In the example below I picked similar aggressiveness of the Target Reference Level for the corresponding responses (about 8db down from the peak). You can see the filters are at very different frequencies.

This will focus attack on LF reverb (boominess) rather than attack non flat (high-impact) good boosted bass your system may have. It focuses on the room rather than the driver.

[brucek]

I didn't realize the Meridian system was so sophisticated.

I take it you don't like the results of its Auto system ?

But now I can with REW

For sure. And you can also add a house curve file to the target reference and then hit Apply Windows to get a new result that might be better than the flat reference level.
Once you have your measurement mdat file stored you can then play around with REW off line and create all sorts of what-ifs.

brucek

[mswlogo]

brucek wrote:

I didn't realize the Meridian system was so sophisticated.

I take it you don't like the results of its Auto system ?


For sure. And you can also add a house curve file to the target reference and then hit Apply Windows to get a new result that might be better than the flat reference level.
Once you have your measurement mdat file stored you can then play around with REW off line and create all sorts of what-ifs.

brucek

It's auto system isn't too bad (the theory behind it is very good). But now that I understand better what it's missing is it doesn't allow you to override the target level at that target decay you are allowed to set. So you basically keep pushing in the target decay in (towards 0) until a reasonable set of filters come in. It might not be at the ideal target though (or the best for all speakers). With REW it looks like I can control both. The target decay also has to be the same for all speakers but you can work around that with a lot of cutting pasting.

[mswlogo]

How come if I look at the LF Decay plot and look at say the slice for 200ms and if I set the IR Reference Window to 200ms I get something different. Shouldn't they be similar?

Basically how do I want to set my IR window so that Low Freq Filter generation focuses on reverb and ingnores direct and direct reflect energy (ignore early stuff).

I thought I just set the IR Reference where I thought it starts but it looks a little more complicated that based on the differences I see at that time in the LF DEcay plot.

Picure is worth a 1000 words I guess

First one is LF Decay Slice at 210
Second is IR Reference set at 210.

[JohnM]

The window settings will make a bit more sense if you look at them on the Impulse graph page. On the LF Decay page the slice time is the start of the window, the Window setting then controls the width. There is a narrow window to the left of the slice start. On the main response you have window settings that extend 125ms to the left of the reference time (the Left Window) and 500ms to the Right, so overall the window width is much more than for the LF Decay. That extra width corresponds to higher frequency resolution (hence the 1.6Hz figure on the Impulse Response Windows dialog compared to 3.3Hz on the LF Decay page). Iif you widen the LF Decay window or reduce the main response windows you will see much closer results.

[mswlogo]

That's what I suspected. I just wanted to make sure I understood. Thanks.

[mswlogo]

I wrote up the procedure I used to tune me Meridian Processor with REW here.

http://www.meridianunplugged.com/wik...ewMrcProcedure

If you have comments or corrections I'd appreciate it.