Let’s talk about headphone measurements and also the MiniDSP
EARS measurement system and why I use it. Can you trust the MiniDSP
measurements, are they accurate? Can you trust any measurement of a headphone?
Well, let’s talk about that.
Before we get into this let me preface this with a
disclaimer, I am an enthusiast, not an engineer and as such, this is going to
be at a very basic level. I’m sure there are many out there who know a lot more
about this than myself, equally I’m sure there are many others who know less.
I’m not going to try and talk up to the experts here, just break this down to a
basic enthusiast level.
Let’s start with the human hearing range which is generally accepted to be 20Hz – 20kHz. There are of course the few outliers whose hearing may exceed this at a very young age however for the most part 20kHz is the maximum frequency a human can hear. This decreases with age to varying degrees depending on an individuals own makeup or exposure to causes of hearing damage such as sustained loud noise or health concerns. For my part, I can’t hear much above 14kHz which is about right for my age, perhaps a little on the low side.
Most headphones are designed to replicate sounds within the
range of human hearing. However, how they actually perform between 20Hz to
20kHz can vary wildly. It is not enough to say that a headphone can reproduce
sounds between 20Hz and 20kHz, this does not tell us how well it can reproduce
any individual frequency.
To break this down simply, some headphones are able to
reproduce lower frequencies better than higher frequencies, we would call those
a “bassy” headphone. Some are able to reproduce higher frequencies better than
lower frequencies, we would call that a “bright” headphone. Some are able to produce
the low and high frequencies well but the mid frequencies less, we would call
this a “V-shaped headphone”. Some are able to reproduce frequencies throughout
the frequency spectrum more evenly, we would refer to these as a “neutral”
So why is this important, surely your ears are the ultimate arbiter of how good something sounds? Who cares how something measures if it sounds good/bad? The more creative types here will attest to the fact that when it comes to making music, throw away the rule book, it doesn’t matter if you’re using any creative tool the right way, the important point is does it sound good? And on a personal level, all judgement on a headphone is subjective. There is no right or wrong headphone there is just do you like it or not? And yes that includes the M50x.
However, when it comes to explaining how something sounds to
another person or comparing headphones against each other, subjectivity must
take a back seat. If I say headphone X is “bassy”, what does that mean? Does
that mean this headphone is bassy or does that mean I think it’s bassy? One
person’s light bass is another’s heavy bass. So how do we resolve that?
Typically this objectivity can be obtained using a control
or reference point. Such as a well known and popular headphone on which to
compare everything. This is why I use the HD650/HD6xx in all of my reviews as a
point of reference. If I say headphone X is bassy compared to the HD650 this is
a lot more objective and much more meaningful than the simple statement.
“Headphone X is bassy”.
A more scientific approach is to look at the numbers and
define an objective description of the headphone. And that is why we measure a
headphones frequency response. Once we have a graph we can easily see how a
headphone performs and compare them between each other.
So how do we do that? We can test how a headphone performs
by playing a frequency at a set volume and measure the sound pressure level of
that frequency as reproduced by the headphone. Every frequency will measure a
different sound pressure level depending on how well the headphone is able to
reproduce that frequency. We can do this measurement for every frequency
between 20Hz and 20Khz and plot the results on a graph. And that, in essence,
is how we measure a headphones frequency response.
Of course, we don’t run a test for every frequency,
generally, this test is done by playing a sine wave sweep that moves from 20Hz
to 20kHz and the measurement rig records the output at each frequency on the
For this task, we need to use a measurement rig. Full-scale
professional rigs built to a specific standard, such as the ones from GRAS, are
the industry recognised way to do this. However, for as long as there have been
audiophiles and computers, there have been homebrew methods for getting
A rather old school way to do this is by mounting a calibrated measurement microphone into a flat plate, in a box for example, mounting the headphones over the plate and taking measurements this way. Measurements can vary wildly from one flat plate to the next and are generally not always comparable against professional rigs. But they can be useful for rough measurements that can be compared against other measurements from the same rig.
However, the flat plate has many flaws. There is no
standardisation and also measurements on a flat plate are not reflective of how
that headphone would be heard on a human head. The presence of your ear in the
headphone alters the frequency response vs measuring the headphone on a flat
plate with no pinnae. If you look at measurements done on one person’s flat
plate, do you know how it was designed, built and operated? No. However,
building a flat plate is relatively cost-effective compared to spending tens of
thousands of dollars on a professional industry standard rig. And for that
reason, there are a lot of flat plate measurements out there for all kinds of
headphones. Because the DIY flat plate has allowed the masses to perform their
In steps MiniDSP. So a few years back, MiniDSP released a
small cost-effective headphone measurement rig for enthusiasts called the
E.A.R.S or Earphone Audio Response System. This system could be seen as the
natural evolution of the DIY flat plate for homebrew headphone measurements.
Featuring a USB soundcard and a pair of calibrated
microphones mounted in an artificial ear canal inside a rough approximation of
the human ear made of silicone. It has never been easier to perform your own
So does the MiniDSP EARS spell death to professional
equipment? No, of course not. But you never expected that for $200 right? It
certainly could be argued that the MiniDSP EARS should be the death knell for
the DIY flat plate. However, that doesn’t seem to be going away any time soon.
So is the MiniDSP EARS accurate? Well no, not by
professional standards. The EARS are able to provide an approximation of a
headphones frequency response to a level of accuracy not seen in the home
before, but it’s not a patch on professional standardised gear. There are a few
quirks with the EARS such as some often odd results seen around 4kHz-5kHz and
higher up in the frequency range things do get a bit more unpredictable. You
will often see peaks or dips a few dB higher or lower than they should be, or
peaks/dips that are slightly out of place, such as appearing at 8kHz instead of
9kHz. There is also the concern that the artificial pinnae on the EARS are not
anatomically correct. It is an approximation of the human ear, albeit far less
supple which can lead to odd measurements at times particularly when dealing
with headphones with small and shallow ear pads or on-ear headphones.
So if it’s not accurate then what’s the point? Because it is
one of many tools available to us to help us understand and describe how a
headphone performs. If a professionally measured headphone shows a treble
elevation between 7-9kHz, rising to a peak at 8.5kHz and the mini DSP EARS
shows an elevation between 7-9kHz with a peak at 7.5 kHz does that matter to
the average person in terms of describing the sound? If I am trying to describe
a headphone and say I think I can hear a peak somewhere around 8kHz and then
the graph shows that same peak between 7-9kHz, then it has done its job at
confirming that what is being heard is objective and not subjective.
The MiniDSP EARS are able to produce some excellent useful graphs that are especially useful for comparing measurements all made on the same rig. Measurements on the MiniDSP EARS are never going to be so out, so wrong as to be completely useless. But measurements performed on a MiniDSP EARS or flat plate are not meant to be taken literally and as accurate. They are just one of the tools we have to help us be objective when dealing with the subjective.
Ok Ok, so the Mini DSP EARS is a useful tool, that’s all
well and good. But in a world that exists professional standard equipment and
professionals who publish those measurements, why does anyone need a home
measurement rig? Can’t you use the graphs made by the professionals?
Well yes of course, and we do. I always check out the
professionally made measurements from Oratory1990, RTings, Inner Fidelity,
Head-Fi etc as well as my own MiniDSP measurements. I show my own measurements
because they are my own, I feel it would not be right of me to be grabbing the
graphs off of Rtings and showing those in my videos. I always encourage people
to check out other graphs if they want to see something more accurate. I show
my own graphs as a tool to help me describe the headphone, it’s a visual aid.
As I said, MiniDSP or Flat Plate measurements are not to be taken as accurate
but as an approximation and a visual representation, after all, a picture
speaks a thousand words.
Another useful job for the MiniDSP EARS for home use is for
modders. It is a superb tool for measuring changes and effects of modification
for those who like to tinker with their headphones.
Yes, it is true that ultimately how a headphone sounds is
more important than how it measures. However, it is also true that if a headphone
measures poorly, it will almost certainly sound poor too.
Lets us also not gloss over the fact that even professional level equipment operated by professionals is not infallible. Even professionals are prone to human error. We also don’t always know what sort of compensation curve is being used, Rtings, for example, use their own target curve, which seems to be close to the Harman target but I’m not sure about that. Some headphones are notoriously difficult to measure due to the size of their ear pads. Some models are prone to wild unit variations in sound signature as a result of sloppy quality control.
This is actually an area where the democratization of
headphone measurements can be useful versus a professional rig. Many
professional outlets get to measure only one unit of a particular headphone.
Unit variance might become more obvious when multiple units have been measured
across multiple similar headphone rigs. Back when I had some QC issues with the
K371 I was able to clearly demonstrate through measurements the huge difference
in driver tuning on a faulty unit. Something that would have been extremely
difficult to describe with words alone.
So in summary then homebrew headphone measurements and rigs such as the MiniDSP EARS are useful tools in helping us understand and describe a headphone’s performance and how it compares to others. They allow us to be more objective when discussing something that would otherwise be rather subjective. But they are not to be taken as 100% accurate. In fact, no headphone graph, whether it has been made on a DIY flat plate or a $20,000 professional rig should be taken as infallible. Regardless of where the measurement has come from, it should be just one tool in your arsenal in describing a headphone. But it should never be the only tool.