Help with EQ in REW

[Rivers]

Hi folks, I have a pair of Focal Alpha 65 Studio monitors that I recently purchased and am trying to set up a flatter response with Equalizer APO. I took a couple of measurements with my Behringer ECM8000 measuring microphone and am surprised at the extremely un-flat response and have a couple of newbie questions if you guys don't mind:

1. Is it the norm for studio monitors to have a frequency response like this?
   
   
2. What could be causing the big spike and dip in the 100-200 hz area?
   
   
3. Could room acoustics make a difference to the frequency response in this case? I have the monitors next to the wall but the speakers are front ported though. My small room is slightly treated and has about 5 sound absorption panels.
   
   
4. After applying EQ filters, there is still two large dips of almost 10db, is there any advice as to how I can potentially solve these 2 dips? I am using the Harmon curve as the house curve

Many Thanks.

 

[John Mulcahy]

The response is due to the room rather than the monitors. The big spike just over 100 Hz is most likely a modal resonance of the room. You have gone a bit overboard with the EQ, it is best applied below about 200 Hz or so to help deal with the more isolated modal resonances that occur at low frequencies. It is also best for the target to pass through the measurement rather than sit below it, otherwise the filters end up being used to adjust level.

 

[Rivers]

Thanks for your advice John.

Do you mean I should be doing the EQ filters from 100 Hz to 20 Khz? Was trying to get rid of the spike that occurs roughly around 120 Hz.

Also would this target below at 49db work better? With the predicted settings, there is still a big 10db dip at 120 Hz. Do you have any recommendations on how I could get rid of this with EQ?

 

[John Mulcahy]

No, I mean use EQ up to 200 Hz and not above 200 Hz. The target looks better, perhaps still a bit low. The low frequency rise in the target is a bit optimistic since the system doesn't seem to be producing enough at the low end to deliver it. Dips tend to be strongly influenced by location of speakers and listener, so try moving things around to see what helps. EQ doesn't often help with dips. They can be very position dependent, so applying EQ to a dip in one measurement position may create an unwanted boost a short distance away.

 

[Rivers]

Hmm I am quite new to this John but I running a pair of studio monitors without any subwoofer, does the low frequency rise play a part here?

Also, if I am doing EQ filters only from say 40 Hz to 200 Hz, does that mean I should not do EQ filters for 200 Hz to 20 Khz? Or are you suggesting that I split the EQ filters into parts, eg: 40 Hz - 200 Hz followed by 200 Hz to 2 Khz, 2 Khz to 10 Khz etc?

Thanks.

 

[John Mulcahy]

I'm suggesting you don't have any filters from 200 Hz to 20 kHz. At higher frequencies what you hear is the direct sound from the monitors but what you measure is the direct sound plus the room's contribution. If you EQ against the measurement you will generally make things worse.

For monitoring usually better for your bass response to be flat. If you want to tweak things to allow for wall proximity use the LF Shelving controls on the monitors.

 

[Rivers]

Thanks John. I am considering adding more sound treatment to my room. Would you say that if frequency response measurements above 200 Hz did not improve after that, it would probably be a better idea to upgrade the monitors?

 

[John Mulcahy]

There isn't much the monitors can do to influence the result, it is pretty much all about the room and placement within the room. A monitor with a cardioid response down through the mid range might show some differences, but not dramatic. Treatment is a good idea, and there are other options for EQ (FIR rather than parametric filtering) that can help above the bass range - Dirac, for example.

 

[Wayne A. Pflughaupt]

Hi Rivers,

A few things.

First, you are still going way overboard on the equalizing. You’re making the classic mistake of shooting for a ruler-flat line. You should instead only look to correct the worse peaks and troughs in response, because those are the ones you actually hear.

For instance above 1000 Hz there are eight filters, when only three are needed, at ~1300 Hz, ~1800 Hz, and ~ 10 kHz.

I suggest doing a search for an on-line article, “Minimal EQ and Hard Knee House Curve.” It mainly deals with equalizing subwoofer response, but the principal carries to the main speakers as well. Here’s another one that deals specifically with full-range EQ:

Spridle’s Experiment

But as John noted, it’s very possible that upper-range EQ won’t help. Trust your ears; if things don’t sound better with the EQ filtering, then eliminate it.

Also, is your graph combined response, or a single speaker? If the former, I suggest doing separate measurements. It’s possible that 118 Hz peak is being caused by only one of the speakers. Below ~2-300 Hz you can use separate filters for each speaker, if needed.

Last, due to differences in size, every room needs its own house curve. I would never use an “off the shelf” house curve. Look for another on-line article on the topic of “House Curve, how to do it.”

Regarding treatments for the upper frequencies, you’ll probably find that they help make the room more dead, but have minimal effect on frequency response. However, deadening the room more would obviously help remove room’s contribution and allow your upper frequency measurements to be a better reflection of what the speakers are generating.

Regards,
Wayne

 

[Rivers]

Thanks folks. The graph shown is a combined response from both speakers. I will measure individual responses and post updated later.

I will have a read up on the links provided, though I am highly contemplating purchasing the MiniDSP with the built in Dirac Live room compensation software, I guess having the FIR Equalisation throughout the frequency range might probably take care of things 'better'?

For the newly measured graph below, would the below corrections work below 200 Hz? I lowered to match range between 40 and 200 Hz though I increased the max boost for both individual and overall boost. Seeing that I do not use an amplifier since my 2 studio monitors are connected to my PC via an Audio interface, would this amount of boost be OK? I know they are generally frowned upon for amplifier applications and was wondering if this was OK in my case?

 

[Wayne A. Pflughaupt]

I’m not averse to a little boosting, but you have TWO filters boosting 18 dB. No way would I ever do that.

Again, the goal is not perfect response, but addressing only the worse peaks and troughs.

Frankly, everything between 200-600 Hz I’d just leave alone, unless you think that range sounds emphasized.

Everything below 200 Hz, I’d like to see individual measurements. It’s possible that you’re getting some cancellation between the two speakers (e.g. 91 and 159 Hz).

Regards,
Wayne

 

[Matthew J Poes]

I’m not averse to a little boosting, but you have TWO filters boosting 18 dB. No way would I ever do that.

Again, the goal is not perfect response, but addressing only the worse peaks and troughs.

Frankly, everything between 200-600 Hz I’d just leave alone, unless you think that range sounds emphasized.

Everything below 200 Hz, I’d like to see individual measurements. It’s possible that you’re getting some cancellation between the two speakers (e.g. 91 and 159 Hz).

Regards,
Wayne

I’d listen to Wayne’s advice here, it’s right on.

If you are getting cancelation that is actually it’s own problem, but good to know.

DIRAC is great, certainly worth experimenting. Before you buy the mini dsp you can try using the free software trial. You just need to use your computer as a source.

Keep in mind, DIRAC can only do so much. Acoustic treatment could be a worthwhile, but i wouldn’t look to the steady state necessarily to see what is going on. I’d actually look at time domain measurements more, including possibly the RT60 and more valuable, EDT. You don’t want to get hung up on these numbers as they are almost universally low in domestic spaces, but it’s a good starting point. IF EDT and RT60 are not flat from 200hz to 10khz that is a good sign you could use a little treatment and such. As John mentioned, your measurements don’t fully match what you hear and no amount of spatial averaging will change that. Our ears are very good at filtering out reflections at mid and high frequencies. A speakers anechoic response looks much better, usually, than the in room.

I think unfortunately forums have ruined people on what a speaker response should look like. We’ve been shown these disturbingly smooth/flat graphs and assume that is what an ideal speaker would look like in an ideal room. The reality is that is simply what modern DSP has allowed. That is what we get when we either correct for a single spot, or for DIRAC, average a set of spatial measurements which minimized average deviance to nill. However even DIRAC won’t be a flat response at different individual spatial locations. That is totally ok.

 

[Rivers]

Thanks guys. Here is what my individual speaker response looks like, response looks pretty similar. Any thoughts about them and if MiniDSP would still be a good option here?

 

[Matthew J Poes]

Thanks guys. Here is what my individual speaker response looks like, response looks pretty similar. Any thoughts about them and if MiniDSP would still be a good option here?

Try the free software demo. That is the only way to know if you prefer the effect. Just do me a favor, judge with your ears. The measurements will look massively better, but it’s what you hear that matters. I know that’s obvious, but it’s easy to get lost in the measurements and lose site of the sound.

 

[Rivers]

Thanks Matthew. I guess therein lies the issue of what 'accuracy' is when we use our ears to judge since everyone has a different opinion to what their ears like. One of the reasons why I am trying to set up the monitors to have a 'flat' response (my preference is now to go via a MiniDSP DDRC-24 that is Dirac Live enabled) is so that I can hook my guitars up to them and EQ them using a modeler.

By the way, I greatly appreciate both John, Wayne (John Wayne!) and your thoughts on this subject, they have been extremely insightful to me as a newbie.

 

[Matthew J Poes]

Thanks Matthew. I guess therein lies the issue of what 'accuracy' is when we use our ears to judge since everyone has a different opinion to what their ears like. One of the reasons why I am trying to set up the monitors to have a 'flat' response (my preference is now to go via a MiniDSP DDRC-24 that is Dirac Live enabled) is so that I can hook my guitars up to them and EQ them using a modeler.

By the way, I greatly appreciate both John, Wayne (John Wayne!) and your thoughts on this subject, they have been extremely insightful to me as a newbie.

This is true but what I’m describing and what John mentioned is a different issue yet. So yes, the definition of accurate is hard to objectively measure. No speaker is accurate, just an approximation. All make compromises and the ones we tolerate are subjective.

That is a different issue from how a measurement of a speaker in a room compares with what we hear. There is an objective difference. When sound radiates out of a speaker it travels out from the speaker. It expands as it comes out in a spherical shape. Kind of like an Icecream come. As soon as it travels out, it starts to touch the surfaces around it. The speaker enclosure, walls, ceiling, floor. It then will diffract, reflect, or be absorbed as it does this. Usually some combination of the three. The sound recombines with the direct signal as it travels toward the listener or mic. The speed of sound is fixed in air but the length of a period is frequency dependent. What this means is that for a fixed time window in an FFT the measurement contains more and more of these room interactions as the frequencies get lower. At mid and high frequencies the measurement contains much more room interaction in the FFT than our ears hear. This is because our brain is very good at filtering out early reflections. We could filter them out of the FFT, but it would compromise the LF resolution. At low frequencies the FFT is an acceptable approximation of what we hear. Not perfect but not bad. Especially if we take multiple spatially distinct measurements. However at mid and high frequencies the measurement isn’t a very good approxinatuon or what we hear and it really isn’t easy to approximate the two.

The problem is exacerberated by the fact that what we hear over longer periods of time does include room reflections. Our brain filters early reflections, but we recognize more delayed reflections as reverberation. This information is technically in the measurements, but not in the way we hear it. Those reflections should ideally have a similar frequency response to the direct sound, at best with a roll-off at high frequencies (same general shape and smoothness). This is more natural. A measurement in a room doesn’t tell us much about the response of the reflections. That information is compromised by the many influencing factors. This is part of why we can’t eq the mids and highs. We can’t know what of the variation in reflection response is a result of the speaker vs room. Ideally we just want a speaker with a flat anechoic response and a falling angular response (reflected by a flat elevated DI). This naturally leads to a response with somewhat elevated bass, which we perceive as more natural, and better behavior in room.

DIRAC is different from anything we can do manually. By taking many measurements over a relatively large space, you can examine differences in the response over both different space and over different time. This lets you begin to understand what aspect of the sound is direct and what is the room. Eq can then be applied to ensure a flat average spatial response that corrects the right things in the right way. This can’t be replicated with normal IIR PEQ filters.

Hence why we are saying, don’t eq above about 200hz. What you are Eqing is as much an artifact of the measurement approach as it is reflective of what you hear. As such, you are introducing errors that can actually make things worse.

 

[fantasyer]

Hi folks, I have a pair of Focal Alpha 65 Studio monitors that I recently purchased and am trying to set up a flatter response with Equalizer APO. I took a couple of measurements with my Behringer ECM8000 measuring microphone and am surprised at the extremely un-flat response and have a couple of newbie questions if you guys don't mind:

1. Is it the norm for studio monitors to have a frequency response like this?
   
   
2. What could be causing the big spike and dip in the 100-200 hz area?
   
   
3. Could room acoustics make a difference to the frequency response in this case? I have the monitors next to the wall but the speakers are front ported though. My small room is slightly treated and has about 5 sound absorption panels.
   
   
4. After applying EQ filters, there is still two large dips of almost 10db, is there any advice as to how I can potentially solve these 2 dips? I am using the Harmon curve as the house curve

Many Thanks.

I think you need check the level first.

 

[FargateOne]

I’d listen to Wayne’s advice here, it’s right on.

(...)

I think unfortunately forums have ruined people on what a speaker response should look like. We’ve been shown these disturbingly smooth/flat graphs and assume that is what an ideal speaker would look like in an ideal room. The reality is that is simply what modern DSP has allowed. That is what we get when we either correct for a single spot, or for DIRAC, average a set of spatial measurements which minimized average deviance to nill. However even DIRAC won’t be a flat response at different individual spatial locations. That is totally ok.

I so agree. I think I fell in this trap and forget to trust my ears. Dirac is special but I let it go out of my system and now I tried to work with 2 to 5 filters for each channel and want to reducate my ears!

 

[Matthew J Poes]

I so agree. I think I fell in this trap and forget to trust my ears. Dirac is special but I let it go out of my system and now I tried to work with 2 to 5 filters for each channel and want to reducate my ears!

and I do think Dirac is special, I just think most people don't understand how it works and as such attempt to replicate it based solely on measurements. Basically, DIRAC can't be replicated by the average user, and IIR filters found in consumer products cannot do what it does, even with the right correction calculations.

2-5 bands sounds reasonable. I really do think that EQ should be focused mostly below around 100-150hz or so. That above this point, up to say 300-500hz you can get away with a little EQ. For example, there is a difference in the tonal balance between my mains and surrounds, it extends up to about 500hz, where there is a broad and pronounced bump in the midbass response. I have removed that via EQ. I think that is ok. It shows up in all measurements, it's of low Q, etc.

In any case, a flat response is not our goal. the only way to achieve a flat measured response in a room from any single measurement is via an anechoic chamber and a speaker with a perfectly flat response at all radiation angles. Our listening rooms should not be anechoic chambers, those reflections are critical to our perception of realism. Further, no speaker has such a good response.

It is amazing how our hearing becomes accustomed to certain sounds. I don't think that is so bad, I hear people talk about trying to flatten their response to make it more neutral and accurate. As we've already discussed, that is a pretty abstract comment. Accurate in what way? First, we know that a flat response doesn't actually mean a flat response, so we very well may have already made things less accurate. But, we also have no idea what the real event is supposed to sound like. There are no standards for studio sound, so we have no idea what the artist intended. Sure we may have heard the same recording performed live, but who says that is what the artist intended the album to sound like? I've attended some atrocious sounding concerts. When it comes to modern music, where the "real event" or "absolute sound" is some abstract concept we can never really "know," I say go with your ears. Whatever sounds right is probably as good as any answer. The exception to this is live acoustic music, especially symphonic performances. In this case, the recordings are done of the event itself, there is a real event and we can know it. I want my system to perform both of these well, but because of the "circle of confusion" with studio sound, a system that is accurate to a real live event recording is not necessarily more accurate of a studio recording.

I can certainly point to some concerts or musical experiences where the "bar was raised" for what the real event is supposed to sound like. In general, I find the most significant issue being dynamics, few systems are capabpe of accurately capturing the dynamic scale of a live event. They often fall just too far from reality. My current system is the closest I've ever been, but it still doesn't like like a real live symphony.