Q) Why do measurements from Golden Ears disagree with data from other websites?
A) The raw data of earphones and headphones are unique to each measuring equipment. And this sort of raw data is different from what we actually hear - which necessitates some calibration to shape the data into what is perceived. At GE, we do this by taking the Diffuse Sound Field and the Room EQ into account, but the method of calibration varies between websites.
Q) Which data is more accurate?
A) This is not a good question since what we mean by a 'flat' sound signature is not set in stone for the reasons that are discussed below. That is, because both the measured data and how to display them depend on setup, calibration and methodolgy, there isn't a single right answer.
Q) So you have random, arbitrary standards?
A) While there might not be a single right answer, some answers are more right than others. At GE, we measure and display according to the following standard:
1. Setup: A Bruel & Kjaer HATS (Head and Torso Simulator) Type 4128C is used - this equipment is generally considered to yield the most accurate measurements (built to the average torso and ears of people worldwide). It is also among the most expensive HAT Simulators.
2. Calibration: A correction must be made to take distortion due to the setup (such as shape of the ears) into account - in case of the Bruel & Kjaer Head and Torso Simulator Type 4128C, the equipment comes with a set of data to be used for calibration. At Golden Ears, the Diffuse Sound Field data provided by B&K is used in conjunction with the Room EQ characteristic (the distortion caused by the room when listening to speakers) from Etymotic Research to correct distortion.
3. Methodology: At Golden Ears, RTA(Real Time Analysis) is employed to check for fit (earphones and headphones) before we proceed with the actual measurements. If there are multiple earfoams, we measure and post data for each pair.
Q) Do you adjust your measurements, then?
A) You'll notice that earphones and headphones sound significantly different based on how they are worn - likewise, the measurements tend to change significantly. Users who visit GE are very surprised when they see the RTA data change so much before their eyes. There are several things that we could do about this:
1. Measure once and post
This is very inaccurate and unreliable.
2. Measure several times, and post all of them
While this method has the most relevance to real-life usage, the amount of data might be overwhelming for some.
3. Measure several times, and post the average
This would be the preferred method since it is fair and accurate, but unfortunately not too many softwares support averaging (ARTA used by GE also lacks this feature).
4. Using RTA to obtain optimal data
The Real Time Analyzer is able to process data as the signature (due to the fit) changes, and thus a single measurement with RTA is effectively several without. Furthermore, the RTA provides immediate feedback on how the fit impacts the sound, making it the best alternative.
At GE, the second and fourth methods are used.
Q) Is there a reason why GE does not publish raw data before corrections?
A) Raw data from each test product(earphones, headphones) varies depending on hardware (HATS or Artificial Ear) used to measure them. Thus, the raw data is useless even if it is published - this also means that you should not compare across websites or sources unless the data from both were calibrated using the same corrections. For instance, if two sets of data were calibrated using 'Diffuse Sound Field only', then it is possible to compare them to each other, though some difference might remain when comparing across sources.
Q) Shouldn't the acoustic impedance be calibrated differently for dynamic and balanced armature drivers?
A) Acoustic impedance is related to the structure and dimensions of the measuring equipment, and the relevant corrections are provided by the equipment manufacturer - also, the difference in the mode of operations (balanced armature or dynamic) does not necessarily mean that the calibration needs to be done differently.
In fact, if this were the case, Etymotic Research's ER-4Ps (which uses Balanced Armature) and MC5s (Dynamic) should have completely different trebles, which is not the case.
Q) I don't really know, but I've heard that GE's data is wrong sometimes.
A) As explained before, the discussion of audio measurements being 'wrong' and 'right' is in itself a show of ignorance - the unfortunate part is that the people who call us out usually have little understanding of how we measure, and frankly, it is difficult to understand where we are criticized. If you have any criticisms, please let us know in the Suggestions Forum so we may address it.
Q) Does it make me a GE Hater or Skeptic if I criticize or pose difficult questions?
A) Questions are always welcome - they are a necessary part of learning and furthermore the life and soul of science, which is what Golden Ears is about. Especially welcome are criticisms where the intent is to correct GE where it is wrong - all of us make mistakes, and corrections to these mistakes are essential to ensure that GE moves forward.
What we do not appreciate is the sort of dogmatic criticism that are hurled without sure knowledge or clearly defined points (for instance, regarding places where there is no clear cut line between right and wrong as in measurements).
Q) Are you an expert? What makes your data so reliable?
A) How would you define an 'expert'? What makes a website reliable?
If I proclaimed myself to be a 'master of measuring gifts', would that make me an expert? Or, would having a degree from a prestigious institution in the field of acoustics such as the ISVR make one an expert? As for the site itself, would it make GE more publicly credible if our net worth became larger than Microsoft? Would using more expensive equipment to measure establish credibility?
For some, the things listed above might be exactly what sets an expert and a trustworthy website apart from others; but I am sure that there are many more of us who believe that credibility is not established overnight merely by adhering to those things.
My belief is that if I continue to publish empirical, reproducible results with which more people could relate, and if my logic is convincing enough, I would be considered an expert, and that once more and more users find the similarities between the measured data in reviews and what they hear, Golden Ears would naturally gain credibility. :)