Musings re: Harman's headphone/IEM preference adjustments and how they relate to speakers

I would agree that the directivity of the speaker is a much more important cue than that of any instruments. My main point is basically that since your idea is derived from the DI of a speaker, an “idealized” version of that might translate better because of the different possibilities of how, and frequencies at which speaker directivity can change. It might be more effective at “closing the loop”. It’s not a substitute for preference though, and from what I’ve seen, the 10dB tilt doesn’t exactly excel in preference tests. Your idea might be better.

I’m going to finally have time to play this weekend. In Peace EQ, we should be using High Shelf filter (Q as slope), correct?

I see that Listener’s squiglink also has the Average All option to compute an average from multiple FR curves. Seems to work pretty well too. Great to see that.

Still no stock 5128 DF though. That would be another useful addition imo, because I still like to use the stock DF with a slope as a coarse FR reference for sorting the best headphone responses.

What do you mean by “stock” DF? Is the DF there somehow modified?

The stock DF is HBK’s original 1/3-octave diffuse field measurement of their 5128 rig. Which is different than the 5128 DF curve that Headphones.com uses.

I believe the Headphones.com DF curve was derived from free-field measurements. And it is higher in resolution than HBK’s DF measurement. But also has less “air” in the higher frequencies above about 10 kHz. And is probably also a tad more forward in the mids.

The stock curve has limited resolution though. It’s the equivalent of a 31-band curve, which is not an issue at the lower frequencies. It is too lacking in detail though for precise compensation/equalization at the higher frequencies (particularly above 15 or 16 kHz, where there really should be more of a dip between the ear canal resonances). So it’s only useful as a coarse reference in the treble, up to about 15 or 16 kHz.

For more precision, especially in the treble/HF, I suggest using an average of several good headphones that are close to your preferred response or target.

1 Like

My takeaway from Dr. Olive’s remarks is that the Harman headphone curve reflects a preference for the in-ear response of neutral speakers that are flat on axis, but also well-extended into the sub-bass frequencies or accompanied with a sub-woofer. This seems to be reflected in the design of many recent (and not so recent) Harman products based on their research and target curve, including the AKG K371 and Mark Levinson No. 5909, among others.

A couple posts on ASR that might help to shed a bit more light on this…

This does not mean everyone will prefer/like this type of response though.

I personally seem to prefer a response that peaks in the bass somewhere around the resonant frequency of a kick drum, as opposed to one which continues to rise much below that. With a good seal, the F0 (sub)bass resonance of the AKG K371 is too low in frequency and too prominent for my taste, for example. And sounds over-emphasized.

I’ve tried this method with a few headphones I own. I started with my Focal Clear and Sennheiser HD800S. Using your 5128 Squiglink I EQ’d to DF using AutoEQ from 20-10k with 4 filters and Q < 3. Then I added 1 filter HSQ -3db 0.4Q. Then I loaded these presets into EQMac and A/B compared with my existing favorite preset. My Clear preset is based on your EQ preset in the EQ request topic. I don’t think I made any changes. The HD800S preset is based on the bass filters from the EQ request topic and my own adjustment for the 6k peak.

The results were good. I mean, I can say for sure that this technique doesn’t yields a bad result to my ears. Was it better than what I compared it to? Hard to say. I listened to a variety of music from my Equipment Evaluation playlist but never really determined a clear winner. It may boil down to personal preference. My existing presets were a bit warmer, and I prefer a warmer sound within reason. I ended up increasing the HSQ to -6db. The new presets present a bit more clear, but the sense of spaciousness and imaging did not seem to be affected.

Later I used the same technique to make a preset for my Hifiman Mini Shangri-La. I then made another preset using DF with a -10db tilt and no HSQ. Comparing the two, I definitely preferred the first preset - the one without tilt. I’m going to need to A/B it against the preset I developed some time soon. Problem is I much prefer listening to music over comparing EQ presets.

2 Likes