Master Complaint Thread About Headphone Measurements

[scolfaro]

Maybe just including another line in the graph, with quieter measurements of IEM's to prevent those pitfalls, would be everything needed?

 

[Mr. Haelscheir]

I don't agree with your assessment here, because if what we wanted was to single out response to "peak" levels, we would be looking into transient response (I don't know if there's an objective measurement for this, is there?).

The transient response can be deduced from the impulse response which can be obtained from the Fourier transform of the magnitude and phase response. Room EQ Wizard makes it pretty easy to retrieve that transient response or go the other way and calculate the magnitude and phase from a direct impulse response measurement.

Figure 1: Meze Elite hybrid pads with "V3.1 PEQ" left ear blocked canal magnitude and phase response.

Figure 2: Impulse and step response derived from the same.

What I don't know is whether dynamic compression or changes in FR at higher playback levels accurately translate; i.e. whether the same bass decrease seen in a sine sweep would be measured from directly measuring the equivalent loud Dirac impulse provided that the microphone, amp, and audio interface can handle it.

I would concur that taking a white spectrum distortion sweep at such loud playback levels would be quite unrealistic, whereby applying something like a pink spectrum sweep if that were possible, or like I had suggested in https://audiosciencereview.com/foru...phone-measurements.18451/page-61#post-1956895 (post #1,210), a pink spectrum multi-tone, might better simulate the realistic distribution of SPL levels for the frequency content within music.

 

[solderdude]

I don't agree with your assessment here, because if what we wanted was to single out response to "peak" levels, we would be looking into transient response (I don't know if there's an objective measurement for this, is there?).

They are actual measurements of the voltage on a driver and the SPL (excursion) of the transducer, assuming there is no compression, is real.
Music consists mostly of transients.
Low frequencies require larger excursions to reach the same SPL. For 20kHz to be reproduced at say 100dB SPL a much smaller excursion is needed.
To reproduce a square-wave of say 1kHz increasingly smaller amounts of higher frequencies are needed so when a headphone can reproduce 20kHz at say 110dB SPL and also can do 1kHz at 100dB SPL more 'speed' is not needed.

Here is the impulse response for MDR-7506

Below the step response for Sennheiser HD580 Precision.

One can also measure the a square-wave (40Hz and 440Hz) and needle (impulse) response

Below the Hifiman HE400SE:

Since what we're looking for are best "average" response representation, then it makes sense to use the most "average" of levels, which is also the level defaulted to by most mastering engineers to judge a master, and that is hovering around 85db.

Actually, averaging 85db is already quite loud, and louder than most people I worked with monitor for long periods, 85db is a level to raise up to as you work, then back down to "rest." We're talking about a level that's safe for 8 hours a day and may cause hearing loss (still average, so the fluctuations are implied, but nonetheless represented in this average). Talking about a constant sweep of frequencies for a response, a sweep of 85db should represent that average appropriately.

Yep, average 85dB is quite loud. 85dB peaks are not and that's what we listen to so one should measure at that peak level in order to ensure that lower levels are O.K. too.

That's exactly my point... I didn't measure the Reds, but I heard this over the course of a few days, and it's kind of reasonable to assume that that happened in a very similar manner even if the headphones depicted are different.

Given the distortion plots of the red even at 114dB there is no hint of distortion in the bass so there is no dynamic compression. The FR at 114dB thus is the same as at 70dB SPL.

Considering both that the test results Amir posted had nothing to do with what I heard, and the differing resistance values on the bass response of the iem's, I assume in your graph that the lines at 90 and 100db would've been closer to Amir's measurement, and since I rarely go much past the purple 80db line, I'm stuck with a "bloated" experience in real life, nothing like the loud tests...

When measuring at lower SPL you run into noise floors of the mic (test fixture) and surrounding sounds.
One can measure the FR at many different levels if the the distortion plots indicate there is compression (increased odd harm. distortion at higher SPL).
Not many people do this or publish that.

 

[scolfaro]

They are actual measurements of the voltage on a driver and the SPL (excursion) of the transducer, assuming there is no compression, is real.
Music consists mostly of transients.
Low frequencies require larger excursions to reach the same SPL. For 20kHz to be reproduced at say 100dB SPL a much smaller excursion is needed.
To reproduce a square-wave of say 1kHz increasingly smaller amounts of higher frequencies are needed so when a headphone can reproduce 20kHz at say 110dB SPL and also can do 1kHz at 100dB SPL more 'speed' is not needed.

Here is the impulse response for MDR-7506

Below the step response for Sennheiser HD580 Precision.

One can also measure the a square-wave (40Hz and 440Hz) and needle (impulse) response

Below the Hifiman HE400SE:

Yep, average 85dB is quite loud. 85dB peaks are not and that's what we listen to so one should measure at that peak level in order to ensure that lower levels are O.K. too.



Given the distortion plots of the red even at 114dB there is no hint of distortion in the bass so there is no dynamic compression. The FR at 114dB thus is the same as at 70dB SPL.

When measuring at lower SPL you run into noise floors of the mic (test fixture) and surrounding sounds.
One can measure the FR at many different levels if the the distortion plots indicate there is compression (increased odd harm. distortion at higher SPL).
Not many people do this or publish that.

Thank you!! That sums up a lot of what we were talking on the other thread as well, I suspect my new LCD-2 Classic to perform better than all the headphones pictured, and the HE400SE and 580 prints definitely seem like what I was listening in my head that made my life harder when judging higher frequency percussion and some other high frequency content with the Sundara, and definitely has to do with the plot you posted of the resonances and poor damping altogether. There's "speed" (or lack of) depicted for sure...

Now my point with the Reds, and I may or may not be correct, but it's a hunch, is that the crossover and varying resistance has an effect when more current comes to play, and that wouldn't necessarily be perceived as compression. Not just Fletcher-Munson while listening (although it would be an effect for us) because the measuring rig isn't a human ear. The higher you go in volume with a headphone that has different impedance across the spectrum, the less increase you will get where it's harder to drive. So the Reds might start with way more bass than its given balance point in loudness, Amir's test might have gone past (that's why he measured "less" bass), while I was residing underneath (and had overwhelming bass). The fact that they play with that with the dongle should be an indication that the unit doesn't behave consistently across different loudness levels. I assume, because I haven't measured, but...

 

[solderdude]

Due to its impedance it is highly likely the Red will sound incorrect from higher output impedance sources.
Most interfaces have a too high output resistance and some studio gear may also have that. The reason for that is usually that the headphone out of such devices is often an afterthought and just an op-amp with some 'safety' resistors in the output.
It is also the reason why the bass boost using an in-line resistor works.

The rest of the measurements do not suggest tonal changes caused by level differences.
Differences can come from equal loudness contours and could come from a too high output impedance from the source is is connected to. But this is not level dependent.

IMHO Harman bass is too much bass boost for critical listening. It is fine for listening to music while commuting or enjoying bass-light recordings or listening at low listening levels.
Just my personal opinion.

 

[scolfaro]

Due to its impedance it is highly likely the Red will sound incorrect from higher output impedance sources.
Most interfaces have a too high output resistance and some studio gear may also have that. The reason for that is usually that the headphone out of such devices is often an afterthought and just an op-amp with some 'safety' resistors in the output.
It is also the reason why the bass boost using an in-line resistor works.

The rest of the measurements do not suggest tonal changes caused by level differences.
Differences can come from equal loudness contours and could come from a too high output impedance from the source is is connected to. But this is not level dependent.

IMHO Harman bass is too much bass boost for critical listening. It is fine for listening to music while commuting or enjoying bass-light recordings or listening at low listening levels.
Just my personal opinion.

I used a Fireface UCX-2, which has an 1 Ohm output with 210mW power per channel. So I assume the sound was correct (didn't get so far as plugging the Reds into my Adi-2 FS because that's where I run the Audeze/Sundara).

Maybe it was just an equal loudness thing then, or maybe they have bigger unit variance than we assume...

As for Harman being too much, I am ok with that amount of bass on big over-ears since I have the feeling most modern music is produced to play on relatively bass-boosted systems. That provided the high-mids are nicely filled out too to compensate for that... Somehow nothing can be too scooped out otherwise there's only bass left.

So If I go through my references and have the feeling they sound "right", then I'm good with it. It only bothers me when we get to notoriously bassy tracks (like a few Run The Jewels tracks, vs Aquemini from Outkast which sounds "right" everywhere even if it's clipping like there's no tomorrow).

Btw the only tracks that sounded "correct" in the Reds to me were a few of the hardest mid-pushing pop tracks on my list... And I must say the past 10 days were of intense listening while producing because of the change up, the Audeze with sunken low-mids and boosted high-mids and low-lows made most of my playlists make sense, so I started trusting it quick, the 7Hz Zeros too! Just the Reds were far off...