[VNandor]

As the title suggests, I want to measure my ELC so I can do a better job at equalizing my headphones. But for this I need to have a good speaker system to start with and a good methodology to take the measurements.
 
The thing is, I don't have a multi-thousand dollar Hi-Fi system to start with nor the knowledge to conduct such experiment but I hope I can tweak the speakers to be good for this particular task and to get some help with it. I have access to a pair of Orion HS501s and a pair of Jamo 708s (they are floorstanders and I don't have any room treatments) and a pair of nearfield speakers taken from a cheap Sony mini-hifi system.
 
The most obvious thing I can do is to get a device which I can use to calibrate the speakers to flat. Problem is, I don't have the slightest idea of how such a device should look or if it even exists. I imagine something like this could do the trick (with a calibrator if needed) or am I totally wrong?
 
An other thing I'm concerned about is the distortion of the speakers. I mean, Let's say I'm measuring 90dB SPL while I feed a 100Hz sinus to the speaker but in reality the speaker makes a 80dB SPL 100Hz sinus and the rest is just distortion/harmonics? How much could that skew the results? Is there a way to compensate against it?
 
And then, there are the room acoustics of course with standing waves, resonances, who knows what. What should I pay attention to? Is it something to worry about as well?
 
And probably the most important thing is how I'm trying to measure and EQ. I'm planning to go from ~20Hz to ~20kHz (well, maybe only like 50Hz to 15kHz due to the quality of the speakers) note by note (major seconds), and putting down how much I have to boost or reduce the signal and make an EQ correction curve out of it. the 20Hz-20KHz note by note measuring would take about 60 steps if I'm not mistaken so it wouldn't take too much time and would be reasonably precise. Any thoughts on that?
 
After I made the correction curve, I would sit down where the microphone was and start playing  sine notes (at my normal listening levels) and apply EQ to make them sound equally loud. However I'm not quite sure how I could remain consistent with my judgement though. I imagine it would be easier to go up by octaves instead of seconds because the loudness differences would be more obvious. And then maybe doing thirds between octaves to make it a bit more accurate but I doubt I could remain consistent with too similar pitches.
 
So with that method I would get a pretty good ELC depending on how good the devices I used were and how attentive I was when I made the loudness-matching.
 
I would later use that ELC curve to add it into my headphone playback-chain and do the same loudness-matching thing and then remove the ELC curve.
 
To sum it up my questions are: What can I do to make the measurement as accurate as I can? Any obvious flaws I haven't thought of? Am I deluded to think I can end up with reasonably good results? Anything else to add?

 

[RRod]

  An other thing I'm concerned about is the distortion of the speakers. I mean, Let's say I'm measuring 90dB SPL while I feed a 100Hz sinus to the speaker but in reality the speaker makes a 80dB SPL 100Hz sinus and the rest is just distortion/harmonics? How much could that skew the results? Is there a way to compensate against it?

 
Another way to put this is:
How much distortion do I need to add to a 80dBSPL 100Hz sine wave to get back to a 90dBSPL signal?
 
I think you'll find it's a lot more than you think.

 

[VNandor]

   
Another way to put this is:
How much distortion do I need to add to a 80dBSPL 100Hz sine way to get back to a 90dBSPL signal?
 
I think you'll find it's a lot more than you think.

I pulled those numbers out right out of my butt. 10dB difference would be insanely much, I just used that number to make my point. However I could easily notice an 1-2dB local dip in a sine sweep and what if the harmonics can add that 1-2dB? I don't know how much I should worry about all these stuff that's why I ask.

 

[RRod]

 
I pulled those numbers out right out of my butt. 10dB difference would be insanely much, I just used that number to make my point. However I could easily notice an 1-2dB local dip in a sine sweep and what if the harmonics can add that 1-2dB? I don't know how much I should worry about all these stuff that's why I ask.

 
For the THD to increase the dBSPL by 1, you still need insane amounts. As a concrete example, say you had a 100Hz sine wave at 80dBSPL and you had only 2nd harmonic (200Hz) distortion that managed to get the sum signal up to 81dBSPL (rms). Then that distortion product would need to be at about ½ the amplitude of the fundamental, so 50% distortion.
 
Re the ELC, I can't remember the name of the thread but I recall a vigorous discussion of this very topic in regards to EQ.

 

[VNandor]

According to wikipedia THD is a big issue at low frequencies (especially in bad systems) but you are right, not in the way I've described it. It will be a problem when I'm listening, not when I'm tuning.
Due to the steep roll-off my ears have combined with the rather big THD I will most likely hear the harmonics instead of the fundamental at low frequencies which is still a problem but not when I try to EQ the speakers to flat.
 
Yes, I've read both EQ tutorial threads but none of them go into details about how to get my ELC, they suggest to use a generic one. I found one more thread about it but it suggests to use my headphones and points out that the measurements will be skewed depending on what headphones/speakers I use which is what I want to avoid in order to get proper results.
 
To be honest I was hoping that there are people who equalized their speakers in the past, and know what to look for. Pro tips like position matters a lot or only X microphone is good enough for this, stuff like that.
If someone ever had their ELC measured and shared how he did it, it would be great as well. It's an audio forum after all, I'm sure there are people who tried those things before.

 

[castleofargh]

  According to wikipedia THD is a big issue at low frequencies (especially in bad systems) but you are right, not in the way I've described it. It will be a problem when I'm listening, not when I'm tuning.
Due to the steep roll-off my ears have combined with the rather big THD I will most likely hear the harmonics instead of the fundamental at low frequencies which is still a problem but not when I try to EQ the speakers to flat.
 
Yes, I've read both EQ tutorial threads but none of them go into details about how to get my ELC, they suggest to use a generic one. I found one more thread about it but it suggests to use my headphones and points out that the measurements will be skewed depending on what headphones/speakers I use which is what I want to avoid in order to get proper results.
 
To be honest I was hoping that there are people who equalized their speakers in the past, and know what to look for. Pro tips like position matters a lot or only X microphone is good enough for this, stuff like that.
If someone ever had their ELC measured and shared how he did it, it would be great as well. It's an audio forum after all, I'm sure there are people who tried those things before.

a word of warning, while I've played around with most of the stuff you want to do, I'm far from what I would consider informed on the matter. so all the following is more like my own experience, not a statement about what is and what isn't. that's why I didn't post before, thinking you'd get a few end level bosses to give it to you like it is, but that doesn't seem to be happening so here I am with my low fi opinions.
 

to EQ speakers in a room, IMO a calibrated USB microphone and something like room eq wizard will be the cheapest method you can start relying upon. and that will show distortions(to some levels depending on the mic, ambient noises and stuff). if you have more money to put into this, there are tools for such purpose including the DSP part applying the corrections. it's really a matter of how far you're willing to go. knowing that room treatment will soon show it's pointy foamy nose if you really try to do proper stuff to the end.
 
 
about how to get your own equal loudness graph, sweep or single tones(depending on your preference)+ EQ seems like the only logical way that costs nothing. you could also go see an audiologist, or buy the ethymotic stuff to check your hearing(seems to be a MC5 and a little DAC). knowing that all those stuff will help you get only a limited range of frequencies as the audiologists are not really preoccupied by 18khz missing to anybody. last time I had my ears checked the test went to 8khz

.
anyway both ends of the audible range are hard to get right by ear because a +/-2db can become harder to notice than +/-0.5db at 2khz. also you have to be careful not to boost those extremes too much just so you can notice them and ruin your ears in the process. the end result is that you play with quiet(to your ears) hard to notice frequencies at both ends making the job of creating your ELC even harder. personally I rarely try to EQ my stuff past 10-12khz by ear, but perhaps others have more success/confidence.
 
 
about distortions, IMO just forget about it. yes speakers will have significant distortions in the low end, even more so if you start trying to pull the FR up with EQ, but it's not like you have a solution to that. so I would just let it go and focus on what I can influence/treat.  also with headphones, you'll lack tactile bass, so for your brain, dead flat speakers in a room might not be the frequency response you will want on headphones. I would instead suggest to approach the low end frequencies on headphones intuitively by relying on your own taste. then maybe you could try to measure that, or compare it to some ELC like Joe Bloggs explains in his tutos to copy/past a signature onto other headphones. 

 

[VNandor]

Thanks for the input!
 
 

 
to EQ speakers in a room, IMO a calibrated USB microphone and something like room eq wizard will be the cheapest method you can start relying upon. and that will show distortions(to some levels depending on the mic, ambient noises and stuff). if you have more money to put into this, there are tools for such purpose including the DSP part applying the corrections. it's really a matter of how far you're willing to go. knowing that room treatment will soon show it's pointy foamy nose if you really try to do proper stuff to the end.
 

On the hardware side, I could get that sound level meter I linked with a calibrator for about ~$75 a day and according to the standards it complies, it seems to be the strongest/most accurate part of my plan. A sound level meter can be used for this, right?
On the software side I would use either foobar with EasyQ, or FL Studio with its parametric EQ plugin. I don't need much more than these, do I?
 
Room treatment and more expensive speakers are out of question right now. I'll start thinking of it when I'm buying a house but that won't happen in the foreseeable future.

 

   
about how to get your own equal loudness graph, sweep or single tones(depending on your preference)+ EQ seems like the only logical way that costs nothing. you could also go see an audiologist, or buy the ethymotic stuff to check your hearing(seems to be a MC5 and a little DAC). knowing that all those stuff will help you get only a limited range of frequencies as the audiologists are not really preoccupied by 18khz missing to anybody. last time I had my ears checked the test went to 8khz

.
anyway both ends of the audible range are hard to get right by ear because a +/-2db can become harder to notice than +/-0.5db at 2khz. also you have to be careful not to boost those extremes too much just so you can notice them and ruin your ears in the process. the end result is that you play with quiet(to your ears) hard to notice frequencies at both ends making the job of creating your ELC even harder. personally I rarely try to EQ my stuff past 10-12khz by ear, but perhaps others have more success/confidence.

Good, I hoped I got that one right. What's the point of going to an audiologist? That would take away all the fun! Plus they would probably be less concerned about accurate results than me and since the health care isn't doing so well here the equipment they would use might not be up to date.
Fortunately my hearing is quite good in that regard, I won't have to boost the high frequencies just to hear them but to hear them as loud as an 1kHz tone. I'm not planning to add more than 20dB boost when the signal was originally at ~85 dB.
 
 

  about distortions, IMO just forget about it. yes speakers will have significant distortions in the low end, even more so if you start trying to pull the FR up with EQ, but it's not like you have a solution to that. so I would just let it go and focus on what I can influence/treat.  also with headphones, you'll lack tactile bass, so for your brain, dead flat speakers in a room might not be the frequency response you will want on headphones. I would instead suggest to approach the low end frequencies on headphones intuitively by relying on your own taste. then maybe you could try to measure that, or compare it to some ELC like Joe Bloggs explains in his tutos to copy/past a signature onto other headphones. 

Well yeah, I hoped there is a workaround for it.
Problem is, when I tried to EQ my headphones I could not make them to sound balanced overall. I did a pretty good job at tuning the local peaks and dips but despite the overall smooth sinesweep, they still don't sound balanced. I tried to tune the smoothed response further by listening to music and tweaking the EQ but to no avail. What sounded "right" Monday sucked on Tuesday, what worked for one music didn't work for an other.
 
Thanks for your thoughts again!

 

[gregorio]

  An other thing I'm concerned about is the distortion of the speakers. I mean, Let's say I'm measuring 90dB SPL while I feed a 100Hz sinus to the speaker but in reality the speaker makes a 80dB SPL 100Hz sinus and the rest is just distortion/harmonics? How much could that skew the results? Is there a way to compensate against it?
 
And then, there are the room acoustics of course with standing waves, resonances, who knows what. What should I pay attention to? Is it something to worry about as well?

 
Unless you're trying to get a lot more dBSPL out of your speakers than they're capable of, speaker distortion is not really much of a problem or rather, it should be largely irrelevant compared to room acoustics. There's likely to be at least one or two frequencies where you get more than 10dB of boost and another one or two (or more) where you get more than 10dB of cut, plus numerous frequencies where there is 2-10dB of cut and boost. There's also the issue that consumer speakers generally start rolling off the bass somewhere around 70Hz and then roll off steeply somewhere around 50Hz.
 
My advice, if you're going to have a go a getting your speakers/room right right, is to get a calibrated measurement mic (about $100) and some RTA software (REW is good and free!), rather than the sort of unit you posted. You might be disappointed at the sort of accuracy you can achieve with your speakers/room though, especially if EQ is your only treatment tool.
 

After I made the correction curve, I would sit down where the microphone was and start playing  sine notes (at my normal listening levels) and apply EQ to make them sound equally loud.

 
Are you talking about then creating a sort of inverse loudness contour? You're likely to struggle to achieve that, unless you really limit your bandwidth. I'm also not quite sure why you would want to?
 
G

 

[VNandor]

   
Unless you're trying to get a lot more dBSPL out of your speakers than they're capable of, speaker distortion is not really much of a problem or rather, it should be largely irrelevant compared to room acoustics. There's likely to be at least one or two frequencies where you get more than 10dB of boost and another one or two (or more) where you get more than 10dB of cut, plus numerous frequencies where there is 2-10dB of cut and boost. There's also the issue that consumer speakers generally start rolling off the bass somewhere around 70Hz and then roll off steeply somewhere around 50Hz.
 
My advice, if you're going to have a go a getting your speakers/room right right, is to get a calibrated measurement mic (about $100) and some RTA software (REW is good and free!), rather than the sort of unit you posted. You might be disappointed at the sort of accuracy you can achieve with your speakers/room though, especially if EQ is your only treatment tool.
 
 
Are you talking about then creating a sort of inverse loudness contour? You're likely to struggle to achieve that, unless you really limit your bandwidth. I'm also not quite sure why you would want to?
 
G

So, this is how I understand it: I don't have to get everything right on the system only the things that could affect how loud I would perceive the clear sine tones. I hope it mostly depends on the FR only. For example I assume I don't have to worry at all about phase distortion that the EQ would cause, nor the THD of the speakers (well, except at low frequencies).
I know the room will amplify certain frequencies but that shouldn't be a problem, the microphone will pick it up and I'm going to hear it as well. I mean, the speakers won't measure flat if I placed the microphone right in front of them, but it's going to be flat where I originally placed it and made the correction. If the room only messes with the FR I can correct it with EQ if the goal is to make it sound right in one given place, at least I hope so.
 
I'm not sure but the unit I posted probably cost many times more than the speakers I want to use so it may not be a good idea to get them if it's so pointlessly accurate compared to what I can get out of the speakers.
 
After I got my ELC I would apply this to my headphone playback chain and start playing sines tones/sweep. Assuming my headphones are not perfectly flat, despite adding my ELC it won't sound equally loud at all frequencies. So I EQ it until it actually sounds equally loud on my headphones, then remove the original ELC. With that I'm supposed to end up with a pretty good EQ curve for my headphones and as a bonus with the use of my ELC and this technique I could make any decent headphones to sound flat. At least that's the theory.

 

[gregorio]

  I mean, the speakers won't measure flat if I placed the microphone right in front of them, but it's going to be flat where I originally placed it and made the correction. If the room only messes with the FR I can correct it with EQ if the goal is to make it sound right in one given place, at least I hope so.

 
Unfortunately, no. The problem with room acoustics isn't so much directly about FR, it's more about time domain issues, which then affect FR. It's mostly about the reflection of frequencies and then the interaction of those reflections with the direct sound from the speakers. Therefore the use of EQ for correction will only be partially successful on some of the issues you're likely to encounter and completely ineffective on others. As a general rule of thumb in acoustic treatment, EQ as a corrective tool is only used for about the last 10% of treatment, absorption and diffusion of reflections being the primary methods employed. But even these methods are largely ineffective when it comes to the lower freqs. In relatively small rooms, acoustic treatment is always somewhat of a compromise and a flat response is not attainable. With EQ as your only available corrective tool, I think you're going to be surprised/disappointed with how close to "flat" you're going to be able to get. We're certainly not talking about fractions of a dB here, in commercial studios with significant budgets for (all the forms of) acoustic treatment, variation between frequencies of 6dB is considered good/very good.
 
In other words, while possibly better than nothing, your method will most likely be only a rough approximation, of which I'm sure you're already aware but probably not aware of how "rough". The advantage of the measurement mic + software is 1. Cost and 2. Flexibility. For example providing waterfall plots and a host of other tools/settings, calibrations, built in test tones/signals, etc. The disadvantage is a bit of a learning curve.
 
G

 

[VNandor]

First of all, I simply just can't imagine how I wouldn't have a total control over the FR with a parametric EQ but since I have never ever tried such thing and know that room acoustics are complicated I will take your word on that.
With that said, my next idea might be very stupid but what happened if I didn't try to correct the whole spectrum all together at once? For example first, I would EQ the 100-200Hz area to a given SPL (lets say 85dB) and do the ELC curve from 100Hz to 200Hz. Then with that out of the way, I would move to the 200-400Hz area, EQ it to the same SPL (85 dB again) without caring the rest of the frequencies and do the same. Sure, I can't achieve a flat response from 20Hz to 20kHz but maybe I could to that in limited bandwidths? Through an octave or two? Would that theoretically improve the accuracy?

 

[gregorio]

  First of all, I simply just can't imagine how I wouldn't have a total control over the FR with a parametric EQ but since I have never ever tried such thing and know that room acoustics are complicated I will take your word on that.

 
Actually it's not difficult to imagine at all. You can't just EQ the output of your speakers because it's the reflections of the output of your speakers which cause most of the problems in acoustics and as you change the EQ out of your speakers you're obviously changing the EQ of your reflections by roughly the same amount. Let's say you measure your room at the listening position and there's a big dip at 400Hz. Most likely that's caused by a phase cancellation between the direct sound from your speakers and a reflection from a table, wall, ceiling, floor or window. So, you get out your PEQ, boost 400Hz by 10dB and measure again ... result? Same as last time, your 10dB of boosted EQ has had no effect because by boosting that freq out of your speakers you've also boosted the (out of phase) reflection by the same amount and still have the same amount of phase cancellation and dip at 400Hz that you did before! EQ won't help here, the obvious solution is to stop (absorb) the reflection at that freq so it can't cancel (or sum) with the direct sound from your speakers. In the average untreated room, you're going to get numerous phase cancellations and resonances, peaks and dips all over the place!
 

  With that said, my next idea might be very stupid but what happened if I didn't try to correct the whole spectrum all together at once?

 
I'm not sure how that would help the fundamental problem?
 
G

 

[RRod]

Reading this might be helpful.

 

[VNandor]

   
Actually it's not difficult to imagine at all. You can't just EQ the output of your speakers because it's the reflections of the output of your speakers which cause most of the problems in acoustics and as you change the EQ out of your speakers you're obviously changing the EQ of your reflections by roughly the same amount. Let's say you measure your room at the listening position and there's a big dip at 400Hz. Most likely that's caused by a phase cancellation between the direct sound from your speakers and a reflection from a table, wall, ceiling, floor or window. So, you get out your PEQ, boost 400Hz by 10dB and measure again ... result? Same as last time, your 10dB of boosted EQ has had no effect because by boosting that freq out of your speakers you've also boosted the (out of phase) reflection by the same amount and still have the same amount of phase cancellation and dip at 400Hz that you did before! EQ won't help here, the obvious solution is to stop (absorb) the reflection at that freq so it can't cancel (or sum) with the direct sound from your speakers. In the average untreated room, you're going to get numerous phase cancellations and resonances, peaks and dips all over the place!
 
 
I'm not sure how that would help the fundamental problem?
 
G

Thanks for the quick explanation, I kind of understand it now. I mentioned the octave by octave measuring because I thought the problem would be that if I boosted certain frequencies I would somehow create a dip or a peak somewhere else unintentionally.
What you describe however doesn't seem like an unsolvable problem with just EQ. I may not be able to boost certain frequencies but I could reduce the frequencies as I please? Because if that's the case, I could just turn up the volume to the point where I only have to use the EQ for reducing the frequncies. Right now, I don't know if I'm going to have enough headroom for the insane boosts I would need for the low frequncies but I might just leave those out because the harmonics would be more apparent at that region than the fundamentals.

 

[VNandor]

  Reading this might be helpful.

Uh, that's something I will have to read more than just one time.