Apple AirPods Max - New Over-Ear Headphones from Apple

After having tested the Apple Airpods Max on a number on both a recent Macbook and on an iPhone 11, I can conclusively say that the below frequency response behavior is accurate. The iOS ‘Headphone Accommodation’ settings can be used to boost the mids, but the ‘Adaptive EQ’ functionality built into the headphone is always on, and is meant to ensure a consistent frequency response regardless of coupling variation.

For reference this is the Airpods Max frequency response with BT and ANC on, on the GRAS 43AG-7:

Update - this headphones is very difficult to measure because it seems to have two different behavior types for the bass response depending on coupling and pressure.

Behavior 1 (moderate clamp):

Behavior 2 (severe clamp):

These results seem to be repeatable as well. At this point I’m confident that second of these two behaviors is accurate for how I hear it. It may be the case that this other behavior is just the result of adaptive EQ producing a weird result on the test rig, and that when it’s coupled to the side of the head, this just compensates for leakage (wearing glasses, or a poor fit).

Airpods Max using the 3.5mm wired adapter ($35):

We can expect the wired connection to sound about the same as when using BT. There’s some minor differences in the treble, notably a more prominent 9khz peak, and less crazy ‘air’ presence in the upper treble. I don’t expect this to sound all that different, and remember that upper treble information on 711 couplers isn’t as accurate - in part because of how significant positional variance is as well.

Channel matching:

Distortion characteristics:


Wired and wireless distortion looks pretty similar as well. Both generally not high enough for anything to be audible (maybe in the bass? but it’s fine). Interesting to note though the 3rd harmonic distortion behavior in the lower mids.

After testing this on modern Apple devices (iPhone 11 Max and recent MacBook), the frequency response behavior was identical to the way it behaves on other devices. This has been corroborated by Crinacle as well. What does change the tuning is Apple’s built-in presets on modern iOS devices. More testing to follow.

Update:
So I was finally able to measure the preset differences using in-ear mics and the methodology outlined in the long post below - just using in-ear mics instead of the GRAS rig. From this testing, it looks like the presets may be part of the adaptive EQ, and so when wearing the headphone and measuring with in-ear mics, we can get a sense of it. This is because the coupling position doesn’t change, and it’s also a realistic use situation.

These cannot be compared with measurements done on the GRAS rig unless we’re only evaluating pinna effects exclusively.

Here is the Vocal Range preset with the delta applied to the existing measurement - This is approximated data, so not 100% accurate, but it’s as good as it gets with this device. It’s also worth noting (as explained below) that there’s a possibility the vocal range preset increases ear gain by a more significant amount, but since I’m unable to measure at my own ear drum reference point, additional gain factors that could potentially influence the actual effects of this are absent:

There is a downside to reading these graphs though. Remember this image?

image

Really all we’re able to evaluate with the preset measurement is how the preset differences show up with respect to the effects of (3) Concha, and (4) Pinna flange. So there may be additional effects that don’t show up here, simply because it doesn’t include the ear canal and drum effects - which are more significant than (3) and (4).

By and large, at the very least we can expect a 2-3dB difference for the various presets at different points. Perceptively it also sounds like it’s at minimum a 2-3dB difference, potentially more. But the trend seems to be similar to what shows up in the graph.

I have no idea what’s going on with the 3rd harmonic distortion in the mids but again I don’t think that’ll be significant to the experience. Potentially this is the result of adaptive EQ, or whatever DSP is being used to correct for the acoustic limitations in general of such a device.

Some comments on comparisons with other ANC headphones:

ANC Attenuation: Bose 700 > APM > XM4 (XM4 and APM have worse high frequency attenuation than Bose 700) > PXC 550 ii
ANC Noise Floor: Bose 700 > APM = XM4 = PXC 550 ii
Tuning (default): Bose 700 = APM (depends on preference but they’re both ‘decent’) > PXC 550 ii >>>>>> XM4
Technical performance: PXC 550 ii (detail) > Bose 700 > APM = XM4
Feature integration and ease of use: XM4 >= APM (depending on ecosystem and what you’re doing) > Bose 700 > PXC 550 ii
Movie watching: APM >>>> the rest for compatible content, otherwise about equal.

Live stream with Crin and Tyler on the APM:

14 Likes