[prescient]

I was curious as to whether anyone has used digital room correction (DRC) software (e.g. Dirac, Audiolens, Acourate, REW) to develop custom curves for IEMs? Given a calibrated microphone and a coupler for the IEMs it seems to me that it would be a fairly straightforward process (at least in comparison to headphones).
 
For those unfamiliar with DRC the systems create correction filters that allow the fitting of house curves to speakers as well as helping to fix peaks/valleys in speaker response. As a for example, the below image are the measurements from my own desktop monitors. The light blue lines indicate the average measured response prior to calibration. The orange lines indicate the target curve the filters will fit the speaker response to and the green lines indicate what the average measured response should be after calibration. 
 

 
The nice thing about these systems are that you can load a number of house curves (4) for use during different scenarios. In my home theater I have curves for action movies, typical TV movie/watching and a night time curve that lowers the levels of the subwoofer so I don't drive my SO crazy. This would be equally useful for headphones IMO.

 

[prescient]

A quick update on this. I decided to play around with developing some headphone target curves for my Westone W40s. The problem with these headphones is that they are both rolled off on the bottom end and rolled off on the high end when measured against a headphone target curve.
 
This can be seen in the headphone response curve measured by Tyll at Innerfidelity.
http://www.innerfidelity.com/images/WestoneW40.pdf
 
 

To try and address these issues I took some measurements using an Dayton iMM-6 and 2.5cm of 7mm PVC tubing acting as a coupler. You can see the results in the graph below and they look remarkably similar to Tyll's uncompensated measurements (grey measurements in the plot above) given my $25 outlay.
 

 
Given that I'm fairly confident my measurements are reasonable I added a compensation curve based on the Harman target curve. This wasn't as precise as I would have liked as i don't have access to the actual target curve, but I was able to eyeball a few values from Tyll's posts and fit a curve using splines to the rest of the target (below).
 


I figured that all I had to do now was to add the target curve values to the microphone calibration file. After doing that the results should measure similar to Tyll's compensated curve. The curve has the same general shape, but isn't an exact match. I suspect it is a result of slightly different measurements and a slightly different measurement curve.
 
 

 
So at this point if the headphones measured perfectly against the Harman target curve we would see a flat line in the above graph. Obviously that isn't the case so finally I used Room Eq Wizard to create convolution filters that I can load into Jriver media center. The measured results below look surprisingly good.
 

 
I haven't had enough time to evaluate the results, but I've found the process an interesting learning experience.

 

[andrewwardell]

Well done that man!

I thought this was a very sensible idea. Makes me think about getting Viper4Android running on my phone.

 

[prescient]

After messing about with the Jriver WDM driver for a bit I couldn't get it to stop having pops and clicks with my Oppo HA-1. I ended up using Equalizer APO which is a powerful equalization software (https://sourceforge.net/projects/equalizerapo/). The interface is minimal, but easy to use. If you open up the configurator.exe it allows you to add components to the configuration in a modular manner. 
 

 
To add a convolution impulse response file all you need to do is add the component (Advanced Filters >> Convolution with Impulse Response). Additionally, you can supplement this with Graphical EQ and Parametric EQ. So assuming you have an impulse response file you like for your headphones that is your base setup you can further tweak with the graphical EQ. 
 
The one thing I've noticed to watch out for is to make sure you avoid clipping. This can be done by adding a preamplification control. You can then check to see if any clipping will occur either via the analysis panel on the bottom that has your target curve or by running benchmark.exe that installs as part of the program. Overall this is a pretty cool piece of software.
 
Additionally, now that I've had some time to evaluate the EQ one my headphones I'm shocked how much better they sound post EQ. Switching between EQ and no EQ everything sounds muddy without EQ. I suspect this can be attributed to the large amount of roll-off after ~1K which can be seen in Tyll's graph. I've been listening to quite a bit of piano lately and the W40 sound profile doesn't work with that kind of music. However, the compensated curve works pretty well with some of the treble brought back in line.