Thanks very much for asking my perspective. Now hang on for a longer
answer essay than you were expecting…
Given that every person's hearing acuity, personal hearing loss, and subjective tastes vary, every audio product will technically have mixed reviews. The question is: "What is the mix of positive/negative/neutral feedback?" I am delighted that feedback from NightHawk listeners at events like the most recent CanJam in SoCal and other shows across the globe (USA, Canada, UK, EU, South Asia, Japan, etc.) has been overwhelmingly positive.
I have to estimate that NightHawk has now been heard by over two thousand pairs of ears, and whenever a listener is kind enough to share their thoughts (either directly or by proxy), I write it down. So when I look back at my list of critiques, I see only a small percentage of criticism. And this criticism is really all over the map…too much highs, not enough highs, too much mids, not enough mids, too much bass, not enough bass, and on and on. There's really no common ground or consensus among this small percentage of criticism. However, there is a great deal of consensus in the positive reactions, which make up roughly 95% or more of the total feedback so far. I most frequently hear that NightHawk sounds like it should cost 4-figures, is one of the most comfortable headphones out there, and gives an experience much more like loudspeakers than headphones. These three sentiments come up over and over.
We can't please everyone. Which brings me to an interesting topic: consensus. During the first CanJam SoCal panel discussion on headphone measurement with
Tyll, Jacob from
G.R.A.S., and myself, the final audience question concerned consensus of listener impressions from a show environment like CES or CanJam. I struggled to answer the question initially, but Jacob had a wonderful response, similar to my first sentence of this post (after the preface). Tyll shared a great analogy that concluded with the idea that even
individuals have difficulty attaining consensus between
their own impressions over multiple listening sessions when demoing in short windows with unfamiliar gear/environments and medium/high ambient noise. My answer eventually settled around weighing feedback; for instance, I have to provide a different weight to feedback received at a show vs feedback received by listeners who are able to spend more time in a known, quiet environment. Similarly, I have to provide different weight to feedback based upon the listener's breadth of headphone experience.
NightHawk ($599 US)
is a different approach from most 4-figure headphones. I have put a serious focus on reducing distortion and creating a balanced presentation that will not cause fatigue even over lengthy listening sessions (and this has as much to do with sonic performance as ergonomic design). When A/B'ing NightHawk against typical flagship headphones, some listeners may feel like something is missing. And they would be correct…distortion is what’s missing along with transient edge enhancement and excessive high-frequency emphasis. Leading edge/transient accentuation can indeed be exciting and enjoyable, but usually only briefly before aural exhaustion sets in. Headphones that do this can have a certain "curb appeal" by imparting a sense of false detail arising from heightened treble response accompanied by distortion peaks. NightHawk simply does not play this trick and instead uses a truly pistonic driver that solves the problem of high-frequency diaphragm breakup distortion that most headphones suffer from. In fact, NightHawk has lower measured distortion than any headphone I have measured. This, of course, is no accident.
I know this is a strong claim, but I can back it up. Allow me to provide an example after first discussing a little background on the topic of distortion and measurement.
There are many methods to measure nonlinear audio distortion by observing harmonics (THD), intermodulation (IMD), dynamics, phase, non-coherence between a test signal and measured response, etc.
A recent AES paper by Steve Temme, Sean Olive, et. al. evaluated the correlation between most of the aforementioned methods and actual audibility in headphones. They found that measured non-coherent distortion is most correlated to audibility, THD is next, and IMD/multitone methods have virtually no correlation. Dynamic and phase distortion were not evaluated. After reading this paper, I have begun crafting some methodologies and custom-code for testing non-coherent distortion since it's not a built-in capability of most measurement hardware/software. But at least we have a bit more supporting evidence that THD is a useful metric.
By the way, Tyll has a
great article detailing how typical THD measurements are made. However, at AQ we use what I think is a more accurate method, a combination of reverse
stweeps and
HarmonicTrak. We normally see THD measured at some SPL (90dBSPL or 100dBSPL). While this is useful, it does not tell the whole story. THD measured at a fixed SPL won't tell us what happens at a variety of volume levels. Music—with its quiet sounds, loud sounds, and everything in-between—is so much more complex than swept tones at a fixed output level. I find that observing THD at many volumes helps build a better picture of how linear a headphone really is…how much detail is preserved and how consistent a headphone's tonal character is during dynamic musical passages.
Let's look at NightHawk versus a highly popular flagship headphone that retails at more than double our price. Here are THD vs frequency graphs at 85, 90, 95, 100, and 105 dBSPL output (calibrated by averaging SPL from 160Hz–8kHz). We see a changing relationship between the low and mid distortion profile of the flagship headphone as output is increased/decreased. I interpret this as a constantly changing coloration depending upon overall output and varying dynamics within a piece of music. This is exactly the type of non-linear behavior that we at AudioQuest try to avoid. NightHawk not only has much lower distortion overall, but also exhibits very linear behavior as output is increased/decreased. This means that the character of NightHawk's sound reproduction remains unchanged regardless of listening level and variations of complex dynamics within music. Just how audible is this? I leave that up to listeners to determine for themselves.
We can also learn a fair amount from the high-frequency region of a THD plot. Sudden, high-amplitude peaks above 1kHz can indicate resonances within the driver structure caused by undamped volumes and can also indicate diaphragm break-up. In the flagship THD graph, we see both. The large peak that increases with increasing SPL output and centered at about 3.6kHz is likely caused by a resonating air volume within the driver that has not been damped. The secondary distortion peak at just under 10kHz is a common characteristic of Mylar diaphragm break-up. I did not design this driver, so these are my speculations based upon experience. Regardless of the cause, the distortion spikes in question contribute to a type of high-frequency emphasis that can initially give the impression of
detail, but in fact is
distortion which becomes very quickly fatiguing.
NightHawk does not exhibit these high-frequency distortion problems, as I have carefully damped the cavity resonances within the driver and implemented a rigid yet self-damping biocellulose diaphragm that does not break up until well beyond what
ITU-T standard
measurement gear is capable of measuring. [Aside: we use a G.R.A.S. Kemar and 43AG inside a custom-made isolation chamber built to laboratory specs for scanning electron microscopes.]
There is a precision to NightHawk’s high-frequency reproduction that results in accurate detail retrieval without listener fatique caused by increased distortion and over-hyped highs.
To further understand
low-frequency accuracy/resolution, I use a linearity function to compare an excitation signal with the actual measured output from a headphone (calibrated to 75–105dBSPL) while slowly increasing volume at several frequencies. The ideal results are completely straight diagonal lines for each frequency, perfectly overlapping. Erratic lines, such as in the flagship headphone graph shown below, mean this particular headphone cannot accurately control how loud or soft bass should be reproduced. Spreading of individual lines means that varying bass frequencies are not reproduced evenly. In the case of the flagship headphone, this means greatly rolled-off lows. Conversely, NightHawk has a much tighter line pattern and minimal wobbling, resulting in tight, even and precise low-mid definition.
These measurements and charts give us a window into understanding performance, but please understand that I do not claim them to be definitive. These are simply measurements that have proven to be highly useful to me as I develop headphones.
Fortunately, each of us is equipped with measurement gear already calibrated to our individual perception…our ears. I believe NightHawk represents a new level of fidelity that hopefully inspires reflection upon the price-to-performance ratio of today's high-end headphones.
I will be supervising the final pre-production cycle of NightHawk starting next week with mass-production looking like late-May (no promises, though!). I look forward to continued discussion here, but please forgive me if my replies are delayed over the next two months. Thanks for reading, listening, and being all-around cool cats.