Would people like to see 86dB distortion measurements in Headphone Reviews here on ASR, in order to bring it more inline with speaker reviews?

[Robbo99999]

My thinking is that it'd be interesting & useful to see headphone distortion measurements at a lower SPL level in addition to what is currently done here (94dB/104dB/114dB), so I'm thinking 86dB as a proposed extra SPL for testing - and this would directly mirror the 86dB SPL that speakers are tested at for their lowest test. I think it would be interesting in terms of making direct comparisons to speakers performance, but also some people listen at low levels so in that case the proposed lower SPL value of 86dB distortion test would make more sense for them. I'm aware that this is Amir's choice ultimately, but I thought it would be interesting & useful to see what proportion of the community would welcome distortion testing of headphones at a lower SPL too, and then of course it's down to Amir if he wants to include one more SPL to test at, I suppose it doesn't take that much more time to run it (for instance at least not when I measure my own headphones' distortion on my miniDSP EARS rig using REW software). Hopefully he's not too mad at me for conducting a poll that may ultimately be asking for "one more test"! (Perhaps, as an idea, the 86dB test would only be added when a particular headphone didn't show low distortion at 94dB, so perhaps it would only occasionally make an appearance.)

 

[Blumlein 88]

I think it a good idea. If your headphones are blocking out some ambient noise, then the lower level probably is enough.

 

[IAtaman]

I don't see the point of it. Most of the time what distortion measurements tells us are whether a headphone is "broken" or not, and how malleable they are for EQing. Multiple levels also gives us some info about compression effects I suppose. So if a pair of headphones are doing well at 94dB, 104dB and 114dB, they will also do well at 86dB - there will be no new insight of info gained for the additional time spent for its use case.

What would would be the point of comparing the distortion figures of headphones and speakers?

 

[Robbo99999]

I don't see the point of it. Most of the time what distortion measurements tells us are whether a headphone is "broken" or not, and how malleable they are for EQing. Multiple levels also gives us some info about compression effects I suppose. So if a pair of headphones are doing well at 94dB, 104dB and 114dB, they will also do well at 86dB - there will be no new insight of info gained for the additional time spent for its use case.

What would would be the point of comparing the distortion figures of headphones and speakers?

I agree that if a headphone tests at 94dB with low distortion then there's no point doing the 86dB distortion test, which is why I had put my bit in brackets at the end in the OP where I say that as an idea the 86dB test would only be run if the 94dB test didn't show low distortion.

Regarding 86dB and comparing to speakers, I think it's just interesting to see the performance difference between a given headphone and a given speaker at that level - it helps to put in perspective the performance of that headphone vs a speaker of your choice - a point of interest in comparison. But ultimately also the lower 86dB level could just be more useful for people that don't listen loud and want to know if the headphone will suit their needs from that perspective (if it hasn't already passed with flying colours at 94dB). EDIT: and specifically it's probably more relevant for the mids & treble portion as these areas are often recorded at lower levels in tracks, so it's almost like seeing real world mids & treble.

EDIT: and thanks for your input & vote, it's good to get everyone's perspective on it - by no means do people have to agree, as the objective is to find out what people think re testing distortion at lower levels.

 

[tomissoawesome]

Is there some segment of the headphone market that needs this differentiation? Anything beyond the very low end likely has some offerings that have low distortion at higher dB levels. In fact, I seem to recall seeing some pretty cheap headphones that perform well wrt distortion using the current levels.

The other reason to do it might be to “prebut” arguments about real-world listening levels. Personally I don’t find those arguments all that convincing because, absent extraordinary performance in other areas, I’d rather have the headphone that performs well at 94 dB. Even if that’s not how I listen. But I’m neither doing the measurements nor arguing about their absence in the threads, so what do I know?

 

[doalt]

I like the idea. And to avoid adding more work Amer can do 3 measurements at 86db, 100db, 114db.

94 is roughly in the middle of 86 and 100. Based on general trends of existing measurements I think it's safe to assume the distortion level is in the middle too.

 

[Robbo99999]

Is there some segment of the headphone market that needs this differentiation? Anything beyond the very low end likely has some offerings that have low distortion at higher dB levels. In fact, I seem to recall seeing some pretty cheap headphones that perform well wrt distortion using the current levels.

The other reason to do it might be to “prebut” arguments about real-world listening levels. Personally I don’t find those arguments all that convincing because, absent extraordinary performance in other areas, I’d rather have the headphone that performs well at 94 dB. Even if that’s not how I listen. But I’m neither doing the measurements nor arguing about their absence in the threads, so what do I know?

I'd rather have a headphone that performs well at 94dB too, just I'd find it interesting to have more fine-grained information at lower SPL levels which would be more applicable to the real world levels I listen at - particularly when it comes to the mids & treble. To be clear, I think for headphones that measure well at 94dB then for that headphone the 86dB test would be skipped.

 

[Robbo99999]

I like the idea. And to avoid adding more work Amer can do 3 measurements at 86db, 100db, 114db.

94 is roughly in the middle of 86 and 100. Based on general trends of existing measurements I think it's safe to assume the distortion level is in the middle too.

It's true that those 3 marry nicely in terms of spacing, but me personally I wouldn't want to see less information - I think the current 94dB/104dB/114dB is some good spacing, just I'd like to see one more lower level SPL tacked onto those if the headphone doesn't have low distortion at 94dB already. I'm not sure I'd want to see the measurements spaced out more by the 14dB that you mention. But you do agree with me on the 86dB test though, we'll see how the poll goes.

 

[Dazerdoreal]

I am not against 86db, but I would really like to keep 94db.

 

[Robbo99999]

I am not against 86db, but I would really like to keep 94db.

Agreed from my point of view!

 

[amirm]

The history of these specific SPL measurements is that my headphone gear is calibrated at 94 dBSPL. So I thought it would be most accurate to measure at that level (both for SPL and frequency response). This is loud enough to overcome room noise but not too loud that most devices have trouble with it. I was using the Topping headphone amp which conveniently has three gain levels. I calibrate to lowest level being 94 dBSPL using its volume control. Then flip the switch to next gain which is 10 dB higher (nominally) and then again which is 20 dB higher. This is very fast and efficient since I just append the measurements to existing set. Doing anything else requires hand calibration to that level which can be error prone. And more work.

 

[usern]

my headphone gear is calibrated at 94 dBSPL

At 1 kHz?

 

[RayDunzl]

This is loud enough to overcome room noise

To illustrate:

A problem with taking distortion measuremnts at lower levels is that the environmental noise becomes higher than the distortion products, and the software misinterprets the noise level as dsitortion.

I have fairly low distortion speakers, and the distortion of the speakers doesn't rise out of the ambient/measurment noise floor until 90+dB SPL or thereabouts.

Three sweeps, around 50, 70, and 90dB SPL

The distortion figures for the 50 and 70dB sweeps are really just misinterpretation of noise.

Only the 90dB sweep has distortion products that begin to rise out of the noise floor.

Viewing distortion as a percent, the loudest trace has the lowest "distortion".

The hump at 48Hz is a room problem, cancellation of the fundamental occurs here.

 

[Robbo99999]

The history of these specific SPL measurements is that my headphone gear is calibrated at 94 dBSPL. So I thought it would be most accurate to measure at that level (both for SPL and frequency response). This is loud enough to overcome room noise but not too loud that most devices have trouble with it. I was using the Topping headphone amp which conveniently has three gain levels. I calibrate to lowest level being 94 dBSPL using its volume control. Then flip the switch to next gain which is 10 dB higher (nominally) and then again which is 20 dB higher. This is very fast and efficient since I just append the measurements to existing set. Doing anything else requires hand calibration to that level which can be error prone. And more work.

Yes, I remember you describing this process when this topic has come up before. As an idea though, if 94dB shows anything other than low distortion then that would be the only time it would be worth running 86dB - because if 94dB is already low distortion then so will 86dB - this means that it'll probably be a relatively rare occurrence that 86dB would be run in your suite if were to ever adopt it. So that would probably put the extra work into perspective.

 

[Robbo99999]

To illustrate:

A problem with taking distortion measuremnts at lower levels is that the environmental noise becomes higher than the distortion products, and the software misinterprets the noise level as dsitortion.

I have fairly low distortion speakers, and the distortion of the speakers doesn't rise out of the ambient/measurment noise floor until 90+dB SPL or thereabouts.

Three sweeps, around 50, 70, and 90dB SPL

View attachment 378308

The distortion figures for the 50 and 70dB sweeps are really just misinterpretation of noise.

Only the 90dB sweep has distortion products that begin to rise out of the noise floor.


View attachment 378307

Viewing distortion as a percent, the loudest trace has the lowest "distortion".

View attachment 378306

The hump at 48Hz is a room problem, cancellation of the fundamental occurs here.

I don't think that would mean we'd end up with false distortion measurements at 86dB because Amir would know based on comparison with the tests at the higher 94dB/104dB/114dB levels, so I don't think there'd be scope for misinterpretation. Whether or not the noise levels in Amir's room rule out any useful 86dB measurments then that would be for Amir to find out during testing. I did have an idea in another thread that Amir might be able to reduce AP fan noise by temporarily disabling fans by perhaps putting in a manual override switch for the fans that he'd flip off during the brief distortion sweep - that shouldn't overheat the AP in that short time. Or you could have a DIY soundproof box you put over the AP just during distortion testing.

 

[JSmith]

A problem with taking distortion measurements at lower levels is that the environmental noise becomes higher than the distortion products, and the software misinterprets the noise level as distortion.

A fair and good point to mention.


JSmith

 

[Robbo99999]

Hmm, this thread has had over 600 views, yet only 37 voters taking part. If we assume on average that everyone who's viewed the thread has viewed it twice then that would make 300 people would have viewed the thread and only 37 bothered to vote, which means I guess around 10 percent (rounded down) of people who viewed the thread decided to vote in it, maybe that's a reasonable assumption. Seems like maybe people aren't too interested in voting or the outcome, yet are still interested in viewing the thread somewhat. I'm just bumping this thread to increase viewership a little, but also to provide my quick analysis on number of voters vs number of thread views so far.

 

[Robbo99999]

@amirm this thread & poll has been open one week now, and we're not really getting much more viewership nor anymore votes, so this poll is probably close to conclusion. It seems that about half the people that voted would like to see 86dB distortion measurements in headphones whilst the the others don't care or 20% think it's a "bad idea". Not many people voted so it's not really exhibiting strong reactions in people. My overall take on it is that the community would generally welcome (tepidly) 86dB distortion measurements in headphones, but if you ever do decide to do 86dB measurements it's probably only worth doing in headphones that are already measuring not very well or showing issues at even 94dB - essentially this would cut down on your workload if you do adopt 86dB as an option.

EDIT: or maybe as an idea you could introduce "how low do you have to go" as a catch phrase for getting acceptable measuring distortion from headphones that distort particularly badly, so maybe you'd keep dialing down the SPL until you get acceptable distortion results....but then again that may exacerbate any background room noise measurement issues you have, so maybe the 86dB option is still the best, but figured I'd throw in this EDIT.