1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Headphones Frequency Response: Challenges & Solutions

Discussion in 'Sound Science' started by samvafaei, Jun 20, 2017.
1 2
  1. samvafaei
    - Headphones don't have crosstalk, so for them to create a proper Soundstage, they need to artificiality add some correlated crosstalk.

    - Headphones don't interact with human's HRTF like loudspeakers do. To tackle this, you either need very large and deep drivers like the HD 800 S (which only interacts with the pinna - check out PRTF on our website), or a software that simulates the whole HRTF of the individual (and not just the transfer function of the outer-ear area)

    - Headphones don't have room effects (reverb, group/phase delay). Now either a headphones has to do this naturally or artificially. For doing it naturally they would have to be infinitely open and loud like loudspeakers (something open-backs try to get close to) or they should simulate room effects using software.

    There's nothing a Smyth Realiser does that can't be done using an iPhone app and couple of in-ear mics. It's just a matter of practically and mass market appeal. It'll come sooner than later.

    No headphones that I've listened to have a speaker-like Soundstage. The AKG N90Q seems to one of the few that tries to get Soundstage right, but I haven't listened to them.

    Our reference for frequency response at the moment is Harman's critical listening room. You can read about it online (e.g. Sean Olive's blog). But the rest of what I said was theoretical. That is, the ideal headphone should sound like an ideal speaker in an ideal room, and from there, it's relatively easy to theorize what the difference between a pair of headphones and speakers are.
  2. bigshot
    Thanks for the clarification on soundstage. I've never experienced anything remotely resembling soundstage in cans myself. It's interesting to see that there are theories for improving that aspect now. It's like DSPs for multichannel speaker setups. I really think that DSPs are the future of home audio.

    Have you considered employing an equalizer to correct headphones to calibrate them to your reference response? It would be interesting to compare various models of headphones all with the same response. It might make it easier to discern differences in other areas like dynamics and timing. It would also be useful to publish EQ settings to make a set of headphones meet the ideal curve. People could apply it themselves and see what "reference" really sounds like. I know there was an iPhone app that did that.
    Last edited: Jun 27, 2017
  3. pinnahertz
  4. bigshot
    I was asking if they would publish the info so you could dial in any equalizer and know exactly how various models of cans compare- not have to buy a canned app.
    Last edited: Jun 29, 2017
  5. pinnahertz
    No, not published directly, but the app is free, the "tunings" are free for 30 minutes, then $2. Theoretically you could sweep the tuning during the free period and get all the data you need. Of course that does require at least some work.
  6. bigshot
    It would be great if it was published, because then you would be able to directly compare the sound of different headphones and their abilities to perform with EQ graphically. You also would be able to turn the sound of one brand and model of headphone into any other brand and model... at least when it comes to frequency response.
    Last edited: Jun 29, 2017
  7. pinnahertz
    Yes, that would be great. Audyssey Labs is a business, though, and they've invested in the measurement and calibration system, as well as the man-hours to calibrate many, many models of headphones. Their tunings are worth something, they're really very good. I don't know anyone who can duplicate the effort and get that result for $2. I don't blame them for not publishing. On the other hand, with very little investment in time and money you can extract their result.
  8. bigshot
    I was asking if Sam would consider doing this, not the company that makes the app.
  9. samvafaei
    We have thought about doing something like that, but providing a corrective EQ based on measurements done on other people will be an approximation at best. It only works without issues if the measurements were done on you, using your headphones.
  10. bigshot
    One or two dB isn't going to matter as long as it's consistent. It would just be a lot easier to look at a chart showing deviation from the ideal, rather than have to look at the Harman curve every time.

    But I went to design school where visual clarity is important.
  11. samvafaei
    Depending on the quality of the earpads, and glasses/hairstyle, there could be a lot more than a couple of dBs of variance:


    Correction softwares should come with their own microphones for calibration, otherwise there could be too much speculation.
  12. bigshot
    I'm not talking about correction software. That was a tangent. I'm talking about how you display a headphone's response curve graphically. The idea is that headphone graphs showing big bumps that conform to the Harman curve don't clearly show the difference between one set of headphones and another. If you make the baseline of the graph a flat line that represents the ideal target curve and the graph the particular cans' deviation from the ideal, then you can look at the plot and instantly see its sound signature in a graphically clear manner. You could EQ the inverse to bring the cans back to Harman, or easily see the differences between different models so you could make one model sound like another. Why do we have to see everything related to some abstract notion of flat when our goal is in practice is Harman? Does that make sense?
    Last edited: Jun 29, 2017
  13. samvafaei
    Our frequency response plots are already compensated, so our dotted target line is the flattened version of our compensation curve.
  14. ScareDe2
    Last edited: May 21, 2018
1 2

Share This Page