Beginner at REW looking for advice

[Matt76]

Hi!

I just started using REW with the goal to flatten the in-room bass response of my Revel Studio2 loudspeakers. I'm using a MiniDSP 2x4 and a calibrated UMIK-1. Test signal is sent to the CD-input of my amp from the headphone jack on my laptop. Is that ok?

I've been able to get the bass response better than ±2dB, 19-125Hz, with 1/12 smoothing, but I had to use some pretty strong boost filters. How strong/sharp filters can I use without causing excessive distortion? Here's the result so far with DSP (with the speakers moved half-way toward the main wall compared to what you see in this photo).

Thanks,

Matt

P.S. I have not applied any filters above 125Hz as I'm worried I might be measuring and equalizing the wrong thing. I figured at <125Hz the sound in-room is alot more "omnidirectional" in it's nature?

 

[John Mulcahy]

Keeping the EQ below a couple of hundred Hz is a good idea. The overall gain of the filter set is important, using filters with a lot of boost isn't a problem if they are countering the effect of filters with a lot of cut outside the region the cut is required. Best keeping the overall boost limit quite low, or zero. You can check any effect on distortion using the Distortion graph.

 

[Matt76]

Thanks. I tune the filters manually as I feel I get alot more precise control over the result that way. It seems the bass needs alot of boost to sound good. Even though I have now achieved a very flat bass response, of ±1.7dB, 20-125Hz, with 1/12 smoothing, the bass doesn't sound strong enough without an additional +5dB of bass boost on my amp.

So a few more questions;

(1) is it safe to combine the MiniDSP filters with the tone controls on my amp, or can that cause some sort of artifacts?

(2) Is it better to use a low shelf filter in the MiniDSP to boost the entire ~20-125Hz range, and then use additional filters to cut peaks for smoothest possible bass response?

(3) When perfecting the bass response for a single seat, is the optimal measuring points just outside each ear, while I'm actually sitting in the listening chair, and then use an average of those two measurements as a reference when creating the filters?

 

[John Mulcahy]

(1) is it safe to combine the MiniDSP filters with the tone controls on my amp, or can that cause some sort of artifacts?

Yes, that's fine

(2) Is it better to use a low shelf filter in the MiniDSP to boost the entire ~20-125Hz range, and then use additional filters to cut peaks for smoothest possible bass response?

Yes, or just turn up the sub

(3) When perfecting the bass response for a single seat, is the optimal measuring points just outside each ear, while I'm actually sitting in the listening chair, and then use an average of those two measurements as a reference when creating the filters?

Put the mic where the centre of your head would be.

 

[Matt76]

Thanks. Is that more reliable; a single measurement where the center of my head would be (orange line)? The blue line is an average of the left and rear ear position while I'm actually sitting in the listening chair. It seem to result in two peak at frequencies where I have added boost filters. So if the blue is a more accurate representation of what I'm actually hearing it would be easy for me to flatten them.

 

[John Mulcahy]

You should not be present while measuring. Your body changes the sound field in ways your ears and brain automatically take into account.

 

[Greg Dunn]

I would recommend, as John said, be well clear of the room while the sweep measurement is taking place - REW lets you insert a time delay so you can move away before it starts.

As an alternative, you might try the moving microphone measurement technique. It smooths the response around the listening position very nicely (using the RTA, pattern PN and continuous averaging) and in my experience gives a good compromise between ease of measurement and accuracy of the frequency response with a person in the listening position. Some nasty dips in the response seem to go away, and the result, for me, seems to track well with the perceived sound of the system. It doesn't let you characterize delays and impulse response, but it's one more tool to use while tuning your system.

 

[Matt76]

Greg, by "the moving microphone measurement technique" do you mean manually placing it at two or more locations, as I suggested, while I'm sitting in the listening chair?

 

[Breeman]

You should not be present while measuring. Your body changes the sound field in ways your ears and brain automatically take into account.

Hi John,

I apologise for hijacking this thread, however your reply here caught my interest. Do you perhaps have literature references I look into on the ability of the ear/brain to auto-correct and account for the presence of the human body in the perception of sound? Thanks.

 

[John Mulcahy]

Do you perhaps have literature references I look into on the ability of the ear/brain to auto-correct and account for the presence of the human body in the perception of sound?

There is an element of the self-evident about that, we are always present when we hear something so that is our only frame of reference and any object in the sound field alters it. For more on how the changes our presence introduces are used there is plenty in the literature on HRTF.

 

[Greg Dunn]

Greg, by "the moving microphone measurement technique" do you mean manually placing it at two or more locations, as I suggested, while I'm sitting in the listening chair?

It's a continuous motion of the mic during the Pink PN measurement, in a spherical/cylindrical pattern surrounding the listening position. This document appears to be the original codification of the method:

http://www.ohl.to/audio/downloads/MMM-moving-mic-measurement.pdf

If you try it as a test, you'll see that (with continuous averaging turned on in the RTA) the FR tends to converge on a fairly consistent graph which in my experience can be repeated pretty accurately with multiple measurements. Slow motion of the mic covering the listening position and a radius of a couple feet around it in all directions, over a period of about 30 seconds, is the goal. In other words, give it time to run a PN cycle before moving the mic significantly.

In my experience, what you see is a rough FR graph that smooths itself out over the time of the measurement; dips tend to fill in and peaks flatten a little - but not to the extent that they are obscured. The resulting graph responds well to tweaks via EQ, and again, seems to represent the audible characteristics of the system surprisingly well. It does take a little practice, but I found a pattern which is not hard to implement and is quick/repeatable.

I think the idea expounded in the document is that we don't hold our heads perfectly still during listening, and there is some subconscious integration of the sound field by the ear/brain. You can't make the FR exact at your listening position, but you can achieve a good balance over a small volume and to me at least, it makes for a better listening experience. I can only say try it and see if it works for you, if you're interested.

 

[Matt76]

Thanks for the link. Please note though that I'm only equalizing frequencies below 150Hz or so. Shouldn't that make the measurements much more reliable, as we cannot seperate the direct sound from reflected sound at lower frequencies? We just hear the combined total, just like a microphone, at such low frequencies?

 

[Greg Dunn]

It should - but (again, I can only report what I see here) it does seem to correlate fairly well with the balance of the system so if there are big peaks or dips they might affect the sound in some way. I have EQed the range below 100 Hz using this method and it - at the very least - does not worsen the sound; I won't claim it is better, but I like the result. It does seem to show balance issues around the crossover frequency which I have tweaked. Even if you're not EQing above 200 Hz or so, it's not a bad ear training device IMHO. Ultimately, use what you think gives you the most pleasing results - it's not like we're in a contest here.

 

[Matt76]

I've noticed that with some filter settings the woofers makes a rattling sound at the beginning of the sweep (at the lowest frequencies). Is this distortion I'm hearing?

 

[Matt76]

One more thing; I'm using the headphone output to do the test sweeps, and it's an onboard soundcard on an old laptop from 2006. Should I be worried about any of this causing significant fluctuations in the measurement results?

 

[John Mulcahy]

I've noticed that with some filter settings the woofers makes a rattling sound at the beginning of the sweep (at the lowest frequencies). Is this distortion I'm hearing?

Subwoofer drivers can be poorly controlled at very low frequencies (below 10 Hz or so) and flap about or run into excursion limits. If you are applying boost filters that could also cause the driver to run into its excursion limits. The distortion graph will show you what effect it is having.

I'm using the headphone output to do the test sweeps, and it's an onboard soundcard on an old laptop from 2006. Should I be worried about any of this causing significant fluctuations in the measurement results?

No, should be fine.

 

[Matt76]

I have no subwoofers yet and the Studio2 loudspeakers are likely not designed to handle the lowest frequencies very well. Might be best to use a high-pass filter that cuts of quite sharply at ~30Hz? The music I listen to don't have much energy below 31Hz anyway (the lowest note on 5- and 6-string bass guitars.

 

[AJ Soundfield]

I would cut peaks only. I would not attempt to boost to fill dips.

 

[Matt76]

After upgrading to the HD version, and spending some more time on the filters, I've managed a ±0.45 dB in-room, 27-129Hz, with 1/12 scale smoothing. 129Hz is the Schroeder Frequency of my room. I figured it's best not to EQ above it. This was a single measurement, with the microphone centered between where my ears would normally be. I also added a high-pass filter to take some strain off of the amp and woofers. The music I listen to have no fundamentals below 31Hz anyhow.

 

[Matt76]

If I instead measure at both ear locations and then average those two measurements, I get a ±0.85dB, 31-146Hz, with 1/12 scale smoothing. That's with an empty room.

If I average the same measurements but with me in the listening chair it is worsened significantly to ±1.8dB, 30-120Hz, with 1/12 scale smoothing.

An average of 5 different measurements gives me a ±1.15dB, 31-146Hz, with 1/12 scale smoothing.

Not sure what microphone positions gives the most accurate representation of what I'm actually hearing?

 

[Matt76]

I did some more measurements and tweaking. I measured and EQ:ed the left and right speaker seperately this time. The average of all four measurements (left speaker from left ear position, left speaker from right ear position etc) came down to about ±1dB, 30-145Hz, with 1/12 scale smoothing. Is this the best measurement technique when trying to smooth out the bass response for a single listening position, i.e. giving the most accurate representation of how my ears/brain would actually interpret the sound? I was not in the room during these sweeps.

 

[John Mulcahy]

Independent EQ of each speaker is a good idea at low frequencies as they will typically be quite different. It can be problematic if EQ is applied at higher frequencies as what is being EQ'd is often not the direct sound that is perceived in that region, so the result can be imaging issues.

Note that there can be resonances that appear in left and right speaker measurements individually that are not present when both play, e.g. for predominantly mono content, if the resonances are being driven in opposition by each speaker.

When the variation is down to a few dB it isn't worth spending time to get it lower, small changes in head position will usually cause variations greater than that anyway.

 

[Matt76]

I was able to find a better speaker placement / listening position. Here I was able to achieve the same response, 30-200Hz, of ±1dB but this time without (as far as I can tell) audible distortion. This because the worst dip to begin with (without DSP and under 200Hz), was only 5dB, so I didn't have to boost much to flatten the bass.

It seems the MiniDSP 2x4 HD also, rather successfully, removed ringing? Again, this is only using stereo speakers and in a so far (except for an area rug) untreated room. I'm new to this distortion / waterfall thing though, but hopefully I have interpreted the graphs (and my ears) correctly.