@bobbooo, our measurement you're referencing above was taken on the GRAS KEMAR (the manikin fixture next to me in the video thumbnail). As I show in the video, KEMAR has some features (cheekbones, jawline, nape of the neck, etc.) that can affect seal (versus a flatter interface like the GRAS 45CA). This is one of the reasons we have made the 45CA our primary fixture (especially for over-ears and on-ears), as it's easier to get a more consistent fit and seal on it. (Again, I do discuss this a bit in the video.) In terms of measurement result (specifically bass output), which is more human-like? I think the answer is somewhere in between the KEMAR and the 45CA.
Long story short, we likely did have a good fit on the KEMAR when we ran that measurement. We monitor the measurements (scope, RMS level, and frequency response) in real time as we position the headphone. Here's a screen shot of a typical positioning screen:
Fig.1
Like many do when positioning the headphones for measurement, we use a 30 Hz square wave and monitor the waveform in scope view. We can also concurrently observe the RMS level (just below the waveform above). So we
are monitoring the bass level while positioning, via scope and RMS level meter. Audio Precision's APx gives us another tool to use when positioning, though -- FFT spectrum view.
As shown in Fig.1 (above), what we've come up with here that we've seen nobody else use is the use of the harmonics of the square wave in FFT spectrum view to give an approximation of the frequency response while positioning, in real time. (I've discussed this technique with many engineers now, the feedback for which has been positive.) Perhaps it is this part of the video that suggests to you we're only looking at treble frequencies.
Typically, the FFT spectrum view window during positioning would look more like this (Fig.2 below):
Fig.2
The difference between the Fig.1 and Fig.2 screenshots is that the FFT spectrum view in the Fig.2 is set to the standard settings, with the length being 32,000 points (with a 48 kHz SR bandwidth) and no averaging. The other key difference is that in Fig.1 the X-axis on the FFT spectrum view was reduced from 20Hz-to-20kHz to 500Hz-to-20kHz to focus on mids and treble (since bass level monitoring is largely covered by the scope view and RMS level).
What led to this particular use of the FFT spectrum view in Fig.1 was thinking about what a square wave is. Here's a definition from
ProSoundWeb:
"A square wave consists of a fundamental sine wave (of the same frequency as the square wave) and odd harmonics of the fundamental. The amplitude of the harmonics is equal to 1/N where N is the harmonic (1, 3, 5, 7…)."
I figured I could use those harmonic spurs to approximate the frequency response in real time to help with positioning -- if we just looked at the
tops of those spurs. Even though the amplitude of the harmonics falls as you go higher, there's still enough energy to work with through the audioband -- and all I want is an approximation of the frequency response while positioning. So even the harmonics way up at the top of the audioband from a 30 Hz square wave provide enough usable energy for this purpose.
To do this, we simply
reduce the length of the FFT -- reduce its resolution -- so that there's not enough resolution to suss out each spur (especially in the higher frequencies). I used to reduce the length to 4000 points (with 48 kHz SR bandwidth) and then switch to noise (like in the video). However, since that video was posted, I've now reduced the resolution further, to only 2,000 points. (I'm also averaging, which stabilizes the real-time spectrum view.) Typically, I'm looking at a range from 500 Hz to 20 kHz while positioning. With this simple change, I no longer need to switch to noise. We can keep it on the 30 Hz square wave until we're ready to set the level for the sweep and run the sweep.
Long story short, we do use Audio Precision APx to monitor quantity of bass as well as treble response, both in real time, while positioning the headphones.
Yes, we do multiple seatings and average those seatings (which the Audio Precision APx software makes quite easy to do). Also, in the past couple of years, we've also made sure to note (for every measurement run we make) the way the headband was positioned and sized (if there are headband detents we note how many clicks out of each side we went with, etc.), which eartips were used, and any other noteworthy things about the positioning and/or environment.
As for multiple seatings, the only exceptions to this are when we're just doing a quick check of something, and also some in-ears we'll do fewer seatings than we'd do with over-ears. In particular, some IEMs that are more anatomically shaped fit into our anthropometric pinnae essentially only one way. If we do two or three re-seats and get identical results, we'll stop there. There's not a lot of latitude to move them at that point, the insertion depth and position defined by the anthropometric pinnae.