All About Headphones…



I just finished several days of educating myself about headphones. I’ll try to summarize what I learned, and show how it can be used along with or without Crescendo.

First off, there is an excellent resource for all headphone information to be found at There is a wealth of data to be had, created by a fellow who has been making precise headphone measurements for nearly 20 years. But you have to dig really deeply to find all the information.

From all the discussions presented on various threads surrounding the InnerFidelity data, it seems the consensus among audio professionals is that a flat room response from a well designed speaker system and carefully crafted room is NOT what sounds best. Rather the best profile seems to exhibit a gentle spectral rolloff of 10 dB from 20 Hz to 20 kHz, or about 1 dB / octave.

Secondly, flat headphones are most definitely NOT the desired goal. To measure headphones, they first measure the sound intensity at the eardrum in response to a perfect room with perfect loudspeakers, as mentioned in the previous paragraph. The result is anything but flat due to the resonant cavity of the ear canal.

So audio experts are closing in on a target profile for perfect sounding headphones. Good headphones would be expected to produce the same spectral profile at the eardrum as is heard in the room with perfect speakers. The primary player seems to be Harman International, of Harman-Kardan fame. The fellow at InnerFidelity has his own thoughts, and slightly modifies that target profile.

So I took the Harman target profile, modified it according to the suggestions at InnerFidelity, and digitized that curve so I could use it to derive suitable equalization for each headphone model.

I did the Sennheiser HD650 headphones yesterday, and today a new pair of Status SM-CB1 headphones just arrived. So I’ll present both of these.

The Sennheiser HD650 has been highly regarded as a neutral reference headphone. I took the data from the InnerFidelity website and digitized it for this analysis.

The top curve shows the modified Harman target profile in red. The light green is the measured performance of the HD650 headphones without any additional equalization. The dark green shows the expected response after suitable equalization.

The bottom graph shows the theoretically needed equalization, along with points indicating octave band frequencies at 32, 64, 128, 256, 512 Hz, and then 1, 2, 4, 8, and 16 kHz — just as you might find on an octave-band graphic equalizer.

However, many of the adjustments are too difficult to achieve with a graphic equalizer. So I used the 4-band parametric EQ built into my MOTU Ultralight interface. And I fiddled with the filters until I got as close as I could to the target curve. Those parametric EQ curves are shown in light blue. And when you combine those EQ curves with the original headphone response, you get that bold green curve in the top graph.

Agreement isn’t perfect, but I don’t think that matters. What matters is that we are a whole lot closer to target with EQ than without.

For comparison, here are the graphs for the Status SM-CB1 headphones:

Both sets of graphs show a notch in the neighborhood of 6-7 kHz. That seems to be intentional, and not an accidental artifact. Apparently, excessive response near 6-7 kHz incorrectly colors the sound at lower frequencies, based on listener surveys. So headphone manufacturers try to plant a notch in that region.

After adjustments with the Ultralight EQ I have to state that the headphones both do sound pretty good. And they sound a whole lot different than without EQ applied.

If you are using a Crescendo, the EQ is applied after the Crescendo corrections, in an effort to make the headphones as correct as possible before hitting them with Crescendo corrections.

We aren’t trying to color the headphone response in any way with this EQ. Rather we are trying to make them present the same spectral profile at the eardrum as a perfect speaker system would present.

If you want to color the headphones, you can apply additional EQ ahead of Crescendo. Remember, we don’t hear EQ the same way that people without hearing impairment hear it. So coloring EQ must be applied ahead of Crescendo so that we can hear what everyone else hears.

  • DM


Author: dbmcclain

Astrophysicist, spook, musician, Lisp aficionado, deaf guy