Figured this is a worthwhile article for folks to read through:
This corroborates a lot of the stuff we’ve been talking about in our content as well, that basically harmonic distortion isn’t something to worry about unless it’s awful.
10 Likes
I’m a little surprised . . . I know RTINGS.com as the premiere place for informed reviews of large televisions. Didn’t realize they did audio gear.
Interesting article. Can’t tell whether its result are peculiar to headphone audio, or whether they would generalize to speakers-in-a-room audio.
Two comments:
- I never heard of some of these headphones
- And the “n” (number of participants) is way too small to construct meaningful statistical claims from. Though given these data, a larger n probably wouldn change much.
1 Like
Very interesting paper and the primary finding as to audibility of distortion is not shocking.
I don’t think I understood the methods however when it came to the experiment looking for listeners to discriminate among headphones. Recordings were made binaurally in order to capture total output, but then what was the playback mechanism of those recordings versus the “clean” original?
1 Like
As someone that is just beginning to dabble in audiophile headphones and related tech I found this study to be very informative. In the end I will let my ears decide what headphones work best for me. Love that I have up to a year to decide if I purchase through headphones.com. This learning curve appears to be much steeper than I initially imagined.
3 Likes
Whilst I find the study interesting, can’t help but wonder if a ultra low distortion headphone such as the Verum-2 included in the mix would have bought different results.
1 Like
It seems to me that using something with ultra low distortion like the Verum 2 should have been the playback headphones instead of the LCD-X.
If memory serves, LCD-X has decently low distortion. At least on the units I’ve tested.
It does. They have it plotted in the article:
But, I mean
Y axis uses different units but I think the Verum 2 is still on top here. It probably doesn’t make any difference to the study results other than better feels.
Right but what I’m saying is that the other results are still fairly wide, so if it didn’t change a preference there, lowering the baseline wouldn’t make a difference. And if anyone is familiar with LCD-X distortion, it’s already quite low. To be clear, there is definitely some scrungle to do with this research, but the LCD-X as the example of low distortion probably isn’t an issue.
1 Like
You used that term “scrungle” multiple times in the latest The Noise Floor too. What do you mean? Surely not “to be cute. A cute object.”
Scrungle over here is shorthand for “there are some problems”. So when things don’t make sense or there’s something ‘scuffed’, imprecise going on.
I still wonder. The Verum 2, when connected to the Topping D900/A900, (subjectively), sounds more clear than any headphone I’ve come across. There was one subject who did hear differences. Who’s to say that there wouldn’t be more than that one with this setup?
We can be fairly confident that’s because of its in-situ FR being a better fit for you than other headphones and not due to any distortion products conferred by other equipment. That’s basically what this research indicates.
I understand what you are saying, but, FR is not the same as distortion. I think distortion is more audible than generally advertised.
I think the opposite. It’s typically far less audible than what it looks. Where that’s not the case is to do with higher order products that are farther away from the fundamental, because they’re outside the masking window. So 2nd order products… you can have like 10% and it still won’t be audible. But 5th order stuff… you’d want to worry about that at lower thresholds. Generally it’s far less common to come across higher order distortion problems in headphones, unless they’re excursion limited (or sometimes BA drivers have these problems), which is why we don’t typically need to worry about it.
This research is yet more indication in that direction.
Understand the relationships between even and odd order harmonics with regards to audibility. There does seem to be a correlation with users perceiving planar cans as sounding cleaner/more clear than the average dynamic driver. Hard to picture the perceived added clarity entirely to FR.
Two things:
- Planars aren’t unversally lower in distortion. Many of the large diaphargm or looser tensioned planars have relatively poor performance with respect to harmonic distortion.
- Planars are often brighter than dynamic driver headphones, and crucially behave more flat below 900hz, which could also contribute to being more strongly received for clarity.
The last thing I’ll say on this is that no two headphones exist where the FR in situ is the same, and it’s hard to actually match the FR of two headphones without in-ear mics. But you can simulate harmonic distortion with various tools, and it’s worth doing that to become more familiar with audibility thresholds for both test tones and music, particularly for 2nd and 3rd order distortion products.
1 Like
Fair points. The Verum 2 is a major exception to #1. It’s a very large diagram, with the best distortion measurements ever seen from ASR, which is pretty high praise from them.
Was not aware of #2. That makes sense. Good to know.
The other item that could be perceived as distortion is clipping of the audio signal. It’s a major issue with regular speakers, so it stands to reason clipping could be an issue with headphones. As I recall, dynamic transients where a instantons but very short burst of power is required is the cause.
What tools are you referring to regarding audibility?
I believe Klippel has a listening test, but… there are others. We put out an article simulating amplifier distortion where we created tones based around certain products, might be worth checking as well.
Actually, this might be a good tool for us to consider building, to see how well people can hear harmonic distortion of different types of equipment.
1 Like
Clipping is definitely a form of distortion that can occur in headphones as well as speakers. Transients aren’t really the cause per se. But an amp may begin to clip the signal if it doesn’t have sufficient voltage to drive the headphones to the volume you want. And this will be most noticable during the peaks in playback.
The headphones may also clip and distort if fed too much power for the drivers to handle. Most (good) headphones are designed to be driven pretty loud though. So this would be more likely to occur at listening levels which are getting uncomfortably loud for normal or extended listening, especially with music compressed for greater loudness.
If you’re listening to recordings that are quieter overall with a greater dynamic range, and have turned the volume up on your amp to hear the quieter passages better, then your headphones (or amp) could begin to clip/distort during some of the louder or peakier passages in the recording. So the type of content you listen to will come into play with some of this.
I think I’ve mentioned this before, but one of the reasons recordings are often boosted in loudness with DRC is to make them sound “better” on the inferior audio gear used by average consumers, which sometimes has a very limited dynamic range. Like the speakers on a computer, TV, or desktop. If you mostly listen on this type of gear, then you may prefer the additional loudness in the content, because you can listen at comfortable levels without overdriving and distorting your gear.
Clipping can also occur at the EQ stage if you do not employ appropriate pregain/preamping to prevent the audio signal from exceeding 0 dBFS, when using EQ filters with positive gain.
In every case, the clipping distortion is most likely to manifest during the (transient) peaks in the recording. Those peaks will be more frequent (but not as extreme) when listening to recordings with DRC and boosted loudness. They will be less frequent (but also more extreme) with full dynamic range recordings, where the average overall loudness is lower.