How much does frequency response actually matter?

Discussion in 'General Audio Discussion' started by Lyander, Sep 20, 2024.

  1. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    12,562
    Trophy Points:
    113
    Location:
    Philippines, The
    I'm working through the thread and I'd say this is pretty concise

    upload_2024-9-22_4-4-33.png

    I'd be curious to know whether all the time domain elements that the like of burst decay et al DO manifest differently at the eardrum level, what sort of minutiae are changed in frequency domain. I really am willing to accept that it IS all FR if there are good, definitive rebuttals, but "it's absolutely FR at the ear drum that's responsible" is a bit curt.

    upload_2024-9-22_4-11-5.png

    This one offers a bit more context and I do see how shaping overtones of instruments and voices via EQ at the mixing stage can change the character of a thing, I've been mucking about with that more recently, but I still don't necessarily see how this would translate to one transducer being able to resolve texture more evidently than another given the same signal input and presuming both are well-amped out of the same source.

    upload_2024-9-22_4-19-44.png

    But later on they do disclaim that FR isn't everything, simply that it encompasses more than we give it credit for.
     
    Last edited: Sep 21, 2024
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Oh, don't get me wrong. Frequency response gets a lot of credit here! I mean, it's the f'ing FIRST measurement I do on headphones!

    Frequency response will affect a lot of stuff, cloudiness, veil, edge, definition, stage depth, overtones. Don't get me wrong here.

    However non-linear distortion, decay, and transient response all help to paint a fuller picture in ways that many outside of SBAF do not realize.

    We simply have more weapons here and thus a better understanding of them.
     
    • Like Like x 3
    • Agreed, ditto, +1 Agreed, ditto, +1 x 2
    • List
  3. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    12,562
    Trophy Points:
    113
    Location:
    Philippines, The
    I get where Resolve is coming from saying that there's a difference measuring things *at the eardrum level of the listener* as opposed to on any standardised coupler, whether the latter be a flat plate coupler, a free-air coupler (@purr1n non sequitur could we get free air measurements of the R70x? Forgot I meant to ask too), a cheapo MiniDSP EARS rig, or a 5-digit HATS. None of those perfectly replicate the full human hearing system which include neurochemical pathways and all THAT fun stuff, never mind hoping to come within sniffing distance of an individual's personal hearing system.

    But I do wonder whether these factors somehow DO manifest later on in terms of the signal that actually strikes our tympani. I don't see how that'd be the case in terms of how it'd function as a matter of FR and not time-domain. I had a random half-awake thought yesterday that basically amounted to: "wait, if sound waves bounce and it's sound waves that strike our tympanic membranes, what happens to any excess energy that gets pushed back, does it muddy the incoming signal any?". Likely a dead-end ponderance there, but yeah this is all just weird and curious stuff.
     
  4. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    This "hits the eardrum thing" with respect to FR is a total non sequitur.

    I can say this because I am a big believer in the use of EQ and yet I haven't been able to EQ the R70X to "slam" or have dynamics like the Atrium Open ("hits like a freight train") or Audeze LCD2R2". I also haven't been able to EQ the Mini C to resolve fine detail like the JAR600. One can experiment with FR all day and never be able to turn one headphones into another.

    May as well make the argument: the frequency response as measured inside my arsehole.

    I mean rather than talk theoreticals that don't even make sense, give me EQ profiles. Give me an EQ profile that will make the R70X slam like the Atrium Open.

    I've already demonstrated above specific examples of non-linear distortion: a spectrum of a steady state signal, and attack and decay of a sine burst.

    These non-linear distortions are severe, especially with the attack and decay of the burst. I highly doubt that any transfer function (to the eardrum or inside the rectum) is going to mystically converge the FR to make any one headphone have the same technicalities as another. And let's not ignore the elephant in the room: non-linear distortion cannot be corrected. It's specific garbage in, garbage out, doesn't matter where is it measured, doesn't matter the transfer function.

     
    • Like Like x 4
    • Agreed, ditto, +1 Agreed, ditto, +1 x 1
    • List
    Last edited: Sep 22, 2024
  5. Wilewarer

    Wilewarer Almost "Made"

    Joined:
    Sep 22, 2021
    Likes Received:
    394
    Trophy Points:
    63
    Location:
    Chicago
    FR at the eardrum is really verging on unfalsifiable statement territory. I mean, is there any headphone measurement rig out there that's ever been able to take a measurement across the entire audible range that would accurately reflect the soundwaves hitting a specific person's eardrum? Is it even possible to make one and verify that this is what it does?
     
    • Agreed, ditto, +1 Agreed, ditto, +1 x 4
    • List
  6. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I believe some German scientists have actually been able to shove mics into people's eardrums. There was a paper on this that I think @shotgunshane or @james444 may have posted.

    It's super interesting because it shows good variability, yet definitely a curve, of the "EQ" or transfer function for the entire ear appartus.

    It also explains why IEMs, which bypass pinna and concha, are more likely to sound different to different individuals than headphones, and headphones to people.

    There is gonna be an inverse transfer function to the brain per person according to their physical structures (I believe humans mostly hear the same at the brain level), but bypassing ear appartus to varying extents means building in a transfer function into the IEM, and this transfer function may or may not gel with a specific individual.
     
  7. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    12,562
    Trophy Points:
    113
    Location:
    Philippines, The
    See that's why I think the narrative getting shifted to "it's all FR" when the original position by Resolve and Listener was always "it's all FR at the eardrum level" is a bit gross of the headphone community at large because it's a disservice to genuine research.

    Even then though, just speaking from a non-physics degree perspective EVEN IF there were massive shifts in FR after going through all our physical hearing hardware (which as @Wilewarer pointed out seems rather difficult to prove), that's just true of FR at ONE POINT IN TIME. Sound is dynamic, and similar to how you can get frequency masking effects playing two stimulus tones together (never mind that music often has WAY more than two things going on, no?) the nature of the incoming signal can often change wildly in just the span of a few milliseconds-- if anything, this just reinforces how important time domain elements ought be, no? That's also not to mention how you feel (and I agree,though mind I stick to much lower-tier gear) part of why planar headphones tend to have overly smooth and plasticky treble timbre is because of how those massive diaphragms are more of a beast to move and might actually self-dampen a lot of those fine textures and such.

    But yes, research funds and time. Bah.

    This seems like it'd be easy to disprove because of mechanical limitations of the A-T driver, but just to make things more interesting is there an EQ profile that can make the Atrium Open sound as pillowy and bouncy as most A-T headphones? I feel THAT would be a more fun challenge and theoretically feasible for the folks who say it's all DSP.
     
  8. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'm all ears. Someone send me a EQ profile that will make the Atrium Open sound like the R70X, as fast, as open, as dextrous, and with it's transients up high.

    Just do it.

    If there's one thing I have to say good about ASR, it's that they do it. HF tends to pontificate.
     
  9. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    What would be interesting: advance the burst measurements to the n-th degree.

    Take 10-cycle bursts at 100 frequency points, say from 50 to 5000Hz.

    The sine amplitude by the fifth cycle will likely reflect the FR as is typically measured.

    The sine amplitude from the first few sines will be very very interesting. As would the decay.

    Seriously I should f'ing do this because no one else will.
     
    • Epic Epic x 6
    • Like Like x 2
    • Agreed, ditto, +1 Agreed, ditto, +1 x 1
    • List
  10. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    9,370
    Trophy Points:
    113
    Location:
    Vancouver BC
    This post, I think.

    Also the link in this one. Fig. 5 (variations in "real ear unaided gain") is the interesting result for this discussion.
     
  11. rfernand

    rfernand Almost "Made"

    BWC Contributor
    Joined:
    Oct 10, 2022
    Likes Received:
    867
    Trophy Points:
    93
    Location:
    Kirkland, WA
    I’m just a caveman. Your world frightens and confuses me.

    An FR plot tells my eyes what the headphone voice is, so to speak. The contour plots and distortion stuff and decay tell me a lot about how the voice is projected.

    *porkers* who measure things at >5 volts and >100db and produce a single number do not have anything interesting insights for my simple mind to comprehend. Sorry, I’m not that evolved.

    “Fantastic FR” can sound like ass if everything else is ringing and fuzzy and stuff. I think that’s the edge sbaf discussions have, tbh. More (and more useful) dimensions to consider and describe. For instance, I really like the plots of the usable region @atomicbob produces because that’s where my listening happens most of the time, while the decay stuff that @purr1n puts out (very good for telling Grados apart in terms of attack and “naturality”, for me). Harmonics and shit above the hearing frequency are also interesting (that’s why the Yggdrasil MIL irritated so much folks it seems!)

    Back to my Eric Clapton Sunday binge.
     
    Last edited: Sep 22, 2024
  12. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,728
    Trophy Points:
    113
    Location:
    near Munich, Germany
    A couple thoughts:

    Yuck! You don't want destructive interference like that in the upper midrange. That will definitely sound weird.

    REW sort of calculates this from the sweep. Step response is key here. You can also plot the FR and only include the first 5 cycles. With headphones it won't be too interesting, but with speakers it's very interesting.
    This is because headphones have basically zero excess phase when measured on a coupler that doesn't introduce nulls.

    This also means that your regular old PEQ peaking filters can fix the burst response.

    One important bit here is that the recordings don't have a microphone FR filter, while the step response can take the microphone FR into account, something that should always be done. If the mic is peaky in the top octave like most electret measurement mics are, then the ideal response has some overshoot and ringing.

    Basically I think FR magnitude and excess phase are the most important determiners to how something will sound along with distortion behavior.

    As for distortion: measuring at different levels to check how it handles medium to high SPL is key.
    This way I found interesting stuff like how some speakers which "come alive" at higher SPLs have relatively high-ish distortion at lower levels, yet the distortion doesn't change as much with different levels.
    Other speakers which still sound vivid at lower levels may have a very low baseline distortion that rises for higher levels. At 90dB those two speakers may be similar, but at 70dB the 2nd is much lower in distortion. This mostly happens for the odd harmonics. Another data point for BA IEM timbral weirdness.
     
  13. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,396
    Trophy Points:
    113
    The worst thing we can do is to reify the frequency response or any other kind of measurement. Any measurement is a simplification of the infinitely complex for the benefit of the beholder.

    We don't listen to the frequency/phase response, THD, or IMD. We listen to the physical phenomena of which a facet is illustrated (or not) in these measurements. Claiming that any of these measurements show everything of anything is grossly unscientific. They are incredibly useful but they are NOT the reality.
     
    • Agreed, ditto, +1 Agreed, ditto, +1 x 8
    • Like Like x 2
    • Epic Epic x 1
    • List
    Last edited: Sep 23, 2024
  14. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    It's what I've been saying for almost fourteen years now. The more varied measurements we have, the fuller picture we get. However, as we've seem, human beings tend toward simplification for difficult to understand topics. Is this S or A rated IEM? How much SINAD?

    It won't. PEQ will just correct linear distortion (frequency response). All PEQ will do is increase the amplitude of every wave in the burst cycle, and probably in a non-linear way. EQ won't fix non-linear frequency or non-linear temporal distortion. How is one gonna fix the decay? Even an IR filter can't fix that. Non-linear stuff cannot be fixed.

    I'm more than happy to experiment with different PEQ to see if I can perfect the attack envelope of the burst for the R70X Refine or get it to be like the LCD-R.

    R70X Refine
    50Hz burst
    upload_2024-9-23_9-45-55.png

    LCD-R
    50Hz burst
    upload_2024-9-23_9-44-12.png

    I haven't played with REW in a while? Can you show me? This would be really cool to see the burst attack envelopes for 20Hz, 50Hz, 100Hz, 200Hz, 400Hz, 800Hz, etc. in one get-go. The first five cycles is plenty because by then the driver has stabilized from its transient state to steady state.

    ARTA has a step response too, but that's just what it is, a step response. I have only found the step response useful for aligning tweeter / woofer and xover work, and even then, it's more academic. JAR600 step response below. Not sure what glean from it?

    JAR600
    Step response
    upload_2024-9-23_9-37-6.png
     
    Last edited: Sep 23, 2024
  15. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Another interesting thing to measure would be compression. Compression is a very real thing and can be heard on woofers driven hard over time. JBL used to provide specs for compression -X db @ Y watts for some of their drivers.
     
  16. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    oh, I've wanted to do this measurement for a while...

    upload_2024-9-23_11-46-34.png
    This is at the 60Hz cursor, not RMS for the entire signal, which would include noise and distortion.
     
    • Like Like x 2
    • Epic Epic x 1
    • List
    Last edited: Sep 23, 2024
  17. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    12,562
    Trophy Points:
    113
    Location:
    Philippines, The
    Correct me if I'm wrong, but basically there's more difficulty reproducing the 60Hz signal as designed when pushing SPLs higher? Never thought to measure that, damn. It'd basically look like a within-variance difference if looked at within the full scope of FR, but you're thinking it could be something indicative of difficulty achieving slam?
     
  18. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Added a more few more headphones.

    Typically what will happen is when a driver heats up is that there will be compression. It just won't get as loud. Folks working in sound reinforcement or as DJs may have experienced this with woofers. The R70X follows this expected behavior. Usually what happens is a softening / pillowing of the bass.

    What's interesting in some of the above cases is that things get slightly louder. For example the RS1X does get louder at the 60Hz component (again, this SPL doesn't include distortion), but Grado drivers are oddballs. The ESP950 shows the most consistent behavior. I'd maybe chalk up the slight rise to measurement variability. Caldera Closed is interesting. It's more or less linear, but suddenly goes up before going down. Let me redo Caldera.
     
  19. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    12,562
    Trophy Points:
    113
    Location:
    Philippines, The
    Yeah I'm a dumbass but with how headphone drivers are often *significantly* smaller than most speakers (besides the ones you get in phones anyway) I'm wondering how pronounced a difference there'd be here. Is this going to be something primarily evident in low bass or will it also affect higher-freq content?

    Would be curious to see the HD600 and Focal Elear/Elex/Clear/whatever for comparison here mainly. Wondering if this is a matter of tech (planar/dynamic/estat) or similar. Grados are basically open baffle so would theoretically be sensitive to minute changes.
     
  20. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    93,889
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I think I screwed up on one data point for Caldera. I also redid the RS1X. The Grado tends to drift a lot! Here are the latest results. FWIW, I was able to replicate R70X.

    upload_2024-9-23_12-11-31.png
     

Share This Page