JMS
100+ Head-Fier
- Joined
- Sep 16, 2002
- Posts
- 106
- Likes
- 19
There's been some recent discussion over in the Sound Science forum about new research on the measured and subjective response of headphones, including a revelatory paper (summarized here) on the subjective impressions of equalized frequency responses.
I've been doing a bit of measurements and experimentation with various frequency response curves myself, and wanted to share the results because they have resulted in significant improvements to my Sennheiser HD650 and HD700 headphones. The improvements are so drastic that I could not listen to these headphones anymore without equalization.
The measurements were taken using a MS-TFB-2 in-ear microphone on a dummy head (mine). I've compared the free response of this microphone against calibrated microphones, and it compares very well.
First, here's a graph of the natural and equalized response of the HD700, mounted on the dummy head. The equalized response was predicted using Room EQ Wizard's "Generic" EQ.
The unequalized response matches well with measurements from purrin, as well as InnerFidelity's post-HRTF curves. The equalization target was motivated by the cited paper's findings, and adjusted based on my own subjective impressions. Subjective listening was done using jRiver's parametric eq. The exact settings were:
38hz +4db Q=2
50hz +5db Q=1
2500hz +5db Q=1
5900hz -8db Q=5
Based on the measured response matching the other published curves, my subjective preference matching well with a flat equalized response, and my previous measurement of my own HRTF using this microphone, my conclusion is that measurements taken this way (at entrance of ear canal) matches well with perceived response. Further compensation with HRTF does not seem necessary.
Second, here's the same for the HD650:
The settings were:
38hz +4db Q=2
50hz +5db Q=1
2500hz +1db Q=1
So, the difference in settings is less of a boost at 2.5khz for the HD650, and no removal of the peak at 5.9khz. The equalized curves match well up to 10khz.
After equalization, both headphones have about the same, neutral spectral balance, comparable to my hi-end, room-corrected speaker system. They are both significantly improved by the equalization, the HD700 more so than the HD650. However, that's not to say they now sound the same The HD700 still sounds a significant margin better to me.
Both before and after equalization, the HD700's bass takes the HD650's to school, with more authority while remaining tight. From the curves, we can see the HD650 rolls off earlier. Despite the EQ, I suspect its drivers simply could not respond to the bass boost. In the treble, the HD650 still has a sort of haziness that makes violins sound smoother, less detailed than they should be. I feel that the HD700's after-equalization sound is top notch, perhaps besting the unequalized sound of some of the other top headphones I've heard, like the T1, Q701, PSB M4U2, LCD-3, and perhaps even my former favorite, the HD800, although I don't have those headphones handy for a re-comparison.
As for why there remains a subjective difference besides the bass, my only guess is the remaining slight differences in response, and differences above 10khz, whose measurement are known to be "BS" due to sensitivity to positional effects.
I'd welcome your thoughts and comments.
I've been doing a bit of measurements and experimentation with various frequency response curves myself, and wanted to share the results because they have resulted in significant improvements to my Sennheiser HD650 and HD700 headphones. The improvements are so drastic that I could not listen to these headphones anymore without equalization.
The measurements were taken using a MS-TFB-2 in-ear microphone on a dummy head (mine). I've compared the free response of this microphone against calibrated microphones, and it compares very well.
First, here's a graph of the natural and equalized response of the HD700, mounted on the dummy head. The equalized response was predicted using Room EQ Wizard's "Generic" EQ.
The unequalized response matches well with measurements from purrin, as well as InnerFidelity's post-HRTF curves. The equalization target was motivated by the cited paper's findings, and adjusted based on my own subjective impressions. Subjective listening was done using jRiver's parametric eq. The exact settings were:
38hz +4db Q=2
50hz +5db Q=1
2500hz +5db Q=1
5900hz -8db Q=5
Based on the measured response matching the other published curves, my subjective preference matching well with a flat equalized response, and my previous measurement of my own HRTF using this microphone, my conclusion is that measurements taken this way (at entrance of ear canal) matches well with perceived response. Further compensation with HRTF does not seem necessary.
Second, here's the same for the HD650:
The settings were:
38hz +4db Q=2
50hz +5db Q=1
2500hz +1db Q=1
So, the difference in settings is less of a boost at 2.5khz for the HD650, and no removal of the peak at 5.9khz. The equalized curves match well up to 10khz.
After equalization, both headphones have about the same, neutral spectral balance, comparable to my hi-end, room-corrected speaker system. They are both significantly improved by the equalization, the HD700 more so than the HD650. However, that's not to say they now sound the same The HD700 still sounds a significant margin better to me.
Both before and after equalization, the HD700's bass takes the HD650's to school, with more authority while remaining tight. From the curves, we can see the HD650 rolls off earlier. Despite the EQ, I suspect its drivers simply could not respond to the bass boost. In the treble, the HD650 still has a sort of haziness that makes violins sound smoother, less detailed than they should be. I feel that the HD700's after-equalization sound is top notch, perhaps besting the unequalized sound of some of the other top headphones I've heard, like the T1, Q701, PSB M4U2, LCD-3, and perhaps even my former favorite, the HD800, although I don't have those headphones handy for a re-comparison.
As for why there remains a subjective difference besides the bass, my only guess is the remaining slight differences in response, and differences above 10khz, whose measurement are known to be "BS" due to sensitivity to positional effects.
I'd welcome your thoughts and comments.