precise minimum phase correction for headphones


There is a principal mistake in your description, that linear filters are best for final critical listening.

Mixed and linear correction profile are not adequate for headphones. 90 % of the headphone’s effects are minimum phase effects and cannot be compensated by linear phase filters. The rest are mixed phase effects.

A headphone acts a single sound source and therefore has a minimum phase behavior. Only reflection effects occurring in the highs includes more sound sources and cannot be compensated by minimum filters. However, they also cannot be compensated be linear filters.

As a result, linear filters produce the worst phase response, 

The available zero latency filter is the only minimum phase correction filter available, but unfortunately it is not as accurate as the linear phase filter.

I made intensive test with individually calibrated headphone over more than a year.
I got the best results with my one minimum phase filter derived from the sonar linear filter response outside sonaworks.

Second with the zero latency (minimum phase) filters inside sonarworks.

Whenever you enable linear filtering the sound stage gets more diffuse and the size of instruments tends to become bigger and less precise.

The tonal balance tends to get brighter.
Defining the needed amount of hall is much more difficult witth linear filtering than with minimum phase filtering.

Please make a minimum phase version derived from the linear filter available. 

Please correct also your documentations concerning linear phase filters.



Official comment

Hey Jörg,

I apologize if I wasn't clear enough before. 
We really don't attempt to correct for headphone phase response in itself.
Imagine that the filter title describes the signal characteristics, not the headphone behavior.
While the phase discrepancies indeed do affect the performance of the headphones, they don't really cause us issues measuring the frequency response and correcting for it.

Linear Phase filter does not correct for the phase response of your headphones but allows you to retain the phase integrity of the signal coming from the original source whether it is a DAW or playback software.

Running a minimum phase filter will by its nature affect the frequency response in ways that will more often than not inhibit achieving a flat target.

This article and videos included on the page explain in a very straight forward way how the FR would be affected using either of the filter types -,between%20low%20and%20high%20frequencies.


Hey Jörg,

Thank you for voicing your concern!
However, the Linear Phase would describe the filter characteristics, not the hardware performance after correction.
Reference 4 software does not address the phase response discrepancies of your headphones or speakers.
Using the Linear Phase filter would allow you to retain the phase characteristics of the original signal when passing through Reference 4 software, with some minor pre-ringing effects of course.


I agree, but this is not the point:

You can optimize the phase response of a headphone with a minimum correction filter. A linear phase correction filter however will never improve phase response.

You don`t need to measure phase. You can predict from physics that the majority of the headphone response is minimum phase.

As an example correcting bass roll off and low drive resonances can completely corrected by minimum phase filter, resulting in perfect timing of the final sound.

The recommendation for headphones should be, that the best phase response will be achieved using the minimum phase correction filter.
The only minimum phase filter, you have implemented in Sonarworks is the zero delay filter.

However there are some restriction on this filter, because it is optimized for lowest possible delay. So it does not always produce identical frequency response to the linear phase filter.

So the request is to provide a minimum phase correction filter with identically identical frequency, minimum processing load and medium delay.
Extensive listening test done with 6 individually calibrated phones confirm, that the linear phase is the worst sounding filter.


Hey Chris,

Yes, I am experienced in digital signal processing and the characteristics of different filter types.

It is the purpose you want the filter use for, that makes the difference:

If the purpose is changing the tonal balance as a sound designer during mastering you are right, that minimum phase filters will in many cases not achieve what you want. Here linear phase filters can help a lot. 

However, it is a completely different story if you want to compensate a filter, that was already applied to the signal. In the case of equalizing the response of a headphone the filter you want to compensate is the transfer function of the headphone.

The aim is to design a filter that result in a flat frequency response and best phase response when combined with transfer function (filter) of the headphone.

Showing the phase response of the filter alone in Sonarworks is misleading, so are your recommendations for the filter types.

I know that you do not measure phase response of the transfer function,
so we must try to model the transfer function of the headphone and try to determine which main parts build it.
If we can model the headphone as a single sound source, the transfer function will be minimum phase and can be fully compensated using minimum phase filter, including phase.
With rising frequencies reflections will occur. Reflection are additional sound sources and add excess phase to the transfer function.
You cannot compensate excess phase by minimum filter nor by linear filter.
Reflections must be considered as separate sound sources when the wavelength gets smaller than the dimensions of the headphone. You do already address reflections during your measuring process in a way, doing some averaging instead. So, we ignore that range in discussing phase behavior.

A linear phase compensating filter will not result in any phase improvements. A minimum phase filter however will result in perfect phase at least in bass and midrange without the need of measuring phase.

Dipl. Ing. Jörg Stumpp


Hi Jörg,

Thank you for your elaborate description. 

I have forwarded your request to our Audio team for review. 



Please sign in to leave a comment.