"Sloped neutral" target curves for headphones and IEMs?

[Matias]

Some days ago I was reading reviews of headphones and IEMs, and their frequency responses compared against a "neutral" target curve, and that got me thinking...

Contrary to a "preference" target curve, a "neutral" target curve, as far as I know, is defined by using a head (+torso?) simulator with mics where the ear drums are, and they measure a flat response speaker on axis in an anechoic room. The captured flat response gets a "pinna gain" in the high mids to treble area and voilá: this is the target curve for all headphones and IEMs to be compared to, at least using that specific head (+torso) simulator. Great.

A few things I don't get and maybe someone can clarify for me.

The first is:

1. It should not matter what type of transducer, be it a full sized headphone or IEM being measured, as long as it reaches the same "neutral" target curve captured by the mics, right?

But moving on, as we know, a speaker in an anechoic room is NOT representative of how it sounds in a typical room. Because of the speaker directivity (bass radiates all around the speaker, and as frequency increases, it gradually closes into on axis for mids and treble), in the main listening position, it measures as a "slope" response: Toole's book figure 12.4d, which he remarks as a "slope in the range of −0.4 to −0.5 dB/octave is a good descriptor".

2. Since an album is mixed and mastered using a speaker in a room as a reference, intended to be heard with speakers in a room as well, shouldn't a "neutral" target curve for headphones and IEMs use the same "sloped neutral" response as a target?

And that is not what I see: headphones or IEMs "neutral" target curves (why are they different? Question 1 above) have flat bass and mids until it reaches the pinna gain area.

3. Could this explain the "preference" target curves, which have increased bass, as an average of people trying to mimic the slope of a speaker in a room?

Adding all things, I suppose a true "sloped neutral" target curve should be defined by: a head (+torso?) simulator, in an anechoic room, with a flat speaker, playing a sloped response, and this reference be used by both headphones and IEMs.

And for fun.

 

[ADU]

You're not far off, Matias.

Some days ago I was reading reviews of headphones and IEMs, and their frequency responses compared against a "neutral" target curve, and that got me thinking. ..

Contrary to a "preference" target curve, a "neutral" target curve, as far as I know, is defined by using a head (+torso?) simulator with mics where the ear drums are, and they measure a flat response speaker on axis in an anechoic room. The captured flat response gets a "pinna gain" in the high mids to treble area and voilá, this is the target curve for all headphones and IEMs to be compared to, at least using that specific head (+torso) simulator. Great.

There are a variety of different ways of calculating the transfer function of in-ear measurement systems. And the free field approach described above is only one!

And as you have correctly surmised, it is not one of the best for headphones, because it lacks the room gain that would be normal when listening to music on speakers in a room in your home, or in a mastering suite designed to emulate or approximate that type of acoustic environment.

"Pinna gain" is also not quite correct, because a good bit of the gain in the upper mids and treble that occurs when doing in-ear measurements actually comes from the resonant characteristics of the ear canal (which is separate from the pinna), and the concha.

A few things I don't get and maybe someone can clarify for me.

The first is:

1. It should not matter what type of transducer, be it a full sized headphone or IEM being measured, as long as it reaches the same "neutral" target curve captured by the mics, right?

In theory, this is true. Both should ideally be measured from inside the ear, at the ear drum reference point (or DRP). And this in-ear measurement of the headphones or IEM should approximate the in-ear measurement of a pair of neutral loudspeakers in a typical semi-reflective room like in your home, rather than an anechoic or echo-free room.

The reason the targets for IEMs are sometimes different in the bass than over-ear headphones is probably due to inaccuracies and/or discrepancies in the modeling of the simulated ear canals or couplers that have mostly been in use until now.

These differences will hopefully start to disappear as more accurate and anthropomorphically correct ear canal models come into use though, as explained here...

But moving on, as we know, a speaker in an anechoic room is NOT representative of how it sounds in a typical room. Because of the speaker directivity (bass radiates all around the speaker, and as frequency increases, it gradually closes in into on axis for mids and treble), in the main listening position it measured as a "slope" response: Toole's book figure 12.4d, which he remarks as a "slope in the range of −0.4 to −0.5 dB/octave is a good descriptor".

I am also a huge Toole fan. And currently use -0.75 to -1.0 dB/octave as my general approximation for a neutral loudspeaker's in-room response, as explained here...

Estimated In-Room Response - trendline & slope

Hi everyone. I know the Estimated In-Room Response should be sloping downwards. But how much? Moreover, depending on how the trendline is drawn, slightly different conclusions could be reached about where the EIR is over the trendline (potential brightness) or under it (potential recess). I...

www.audiosciencereview.com

I think this is about the same range that others also use on this forum.

For sound power, I use a slightly steeper slope, more in the -1.0 to -1.5 dB/octave range. (Or about -1.25 dB/octave on average.)

2. Since an album is mixed and mastered using a speaker in a room as a reference, intended to be heard with speakers in a room as well, shouldn't a "neutral" target curve for headphones and IEMs use the same "sloped neutral" response as a target?

And that is not what I see: headphones or IEMs "neutral" target curves (why are they different? Question 1 above) have flat bass and mids until it reaches the pinna gain area.

3. Could this explain the "preference" target curves, which have increased bass, as an average of people trying to mimic the slope of a speaker in a room?

Hopefully some of the info I've posted above helps to clarify some of this.

The subjective preference testing that Harman has done has indeed suggested that most people prefer a more "sloped" response that approximates the gain you get in the lower frequencies when listening to neutral speakers in a semi-reflective room.

And it also showed some differences in preference between the IEMs and over-ear headphones in the bass, which will hopefully be mitigated or corrected with the newer anthropomorphic ear canal models.

Adding all things, I suppose a true "sloped neutral" target curve should be defined by: a head (+torso?) simulator, in an anechoic room, with a flat speaker, playing a sloped response, and this reference be used by both headphones and IEMs.

There are various different approaches which can be used to model this type of response. Most involve compromises of one kind or another.... unless you are measuring some actual speakers in an actual semi-reflective room, which is the best approach imo.

The direction of the sound source can be of particular importance in these kinds of models btw. Because you will get a very different in-ear measurement if the source is directly in front of your listening position, at a 0 degree angle, than you'll get if it's at a 30, or 45, or 90 degree angle. And in most home audio systems (and mastering suites), speakers are generally positioned at about a 30 degree angle to the listening position.

Since I don't currently have an easy way to do accurate in-ear measurements of actual loudspeakers, I use a slightly different model, which involves combining a neutral loudspeaker's diffuse sound power response with the in-ear measurement system's diffuse field response.

I'll also use the measurements of other headphones which have a reputation for a fairly neutral response as one of my in-ear targets though. And also my own (informal) subjective listening tests. Or some combination of all of the above. In my experience, the net result is usually pretty similar with all these different approaches.

 

[ADU]

If you want to use the in-room response curve of a loudspeaker (or a slope which approximates that) as your target for a pair of headphones, there are also some ways of doing that.

One method is to EQ your speakers to a flat response at the listening position in your room using a standard omni mic. And then to re-measure this flattened response in your own ears with an in-ear mic. You can then take that in-ear measurement of the flattened speakers, and combine it with the in-room curve or slope of your choice, to add the necessary room gain in the bass.

This is exactly the method Harman used for some of its early in-ear tests.

 

[ADU]

It's generally easier to measure, and tune an over-ear headphone to a neutral or accurate response using methods like the ones described above than an IEM btw. This is for some of the reasons already mentioned in the video above. But also because it's more difficult to do precise measurements of IEMs on your own ears with a mic. And because IEMs interact with less of your personal head and ear anatomy, or HRTF, than a loudspeaker or over-ear headphone does. Some of this can be augmented though by doing more of your own listening tests and tuning/EQ adjustments by ear with the IEM. Or using newer technologies that create a virtual model of your own personal HRTF.

Variations in anatomy don't matter as much with speakers and over-ear headphones because these kinds of transducers interact with more of your personal HRTF than an IEM. This is one reason why I generally prefer to use over-ear headphones for my own listening.

The seal on a headphone is also important though in achieving a reliable and predictable frequency response. And this applies to both the over and in-ear headphones. Jude talked a little about this in the video posted above. But if a headphone doesn't provide a reliable and fairly consistent seal between individuals, or between different listening sessions on the same individual, then its also harder to tune or EQ it with good precision. Listening tests and individual adjustments by ear also become more important in this situation.

The seal mostly effects the lower frequencies btw. This is why closed headphones with a good seal will often be able to extend much better into the lower sub-bass frequencies than headphones which are either open on the sides or back, or which seal poorly.

 

[Here2Learn]

Good info from all who posted above. I'm not a headphone guy (I prefer speakers), so I'd never looked in to why the speaker preference curve differs so notably from the one used for headphones, but the above posts explain it quite succinctly. Thanks.

 

[AudioKC]

Some days ago I was reading reviews of headphones and IEMs, and their frequency responses compared against a "neutral" target curve, and that got me thinking...

Contrary to a "preference" target curve, a "neutral" target curve, as far as I know, is defined by using a head (+torso?) simulator with mics where the ear drums are, and they measure a flat response speaker on axis in an anechoic room. The captured flat response gets a "pinna gain" in the high mids to treble area and voilá: this is the target curve for all headphones and IEMs to be compared to, at least using that specific head (+torso) simulator. Great.

A few things I don't get and maybe someone can clarify for me.

The first is:

1. It should not matter what type of transducer, be it a full sized headphone or IEM being measured, as long as it reaches the same "neutral" target curve captured by the mics, right?

But moving on, as we know, a speaker in an anechoic room is NOT representative of how it sounds in a typical room. Because of the speaker directivity (bass radiates all around the speaker, and as frequency increases, it gradually closes into on axis for mids and treble), in the main listening position, it measures as a "slope" response: Toole's book figure 12.4d, which he remarks as a "slope in the range of −0.4 to −0.5 dB/octave is a good descriptor".

2. Since an album is mixed and mastered using a speaker in a room as a reference, intended to be heard with speakers in a room as well, shouldn't a "neutral" target curve for headphones and IEMs use the same "sloped neutral" response as a target?

And that is not what I see: headphones or IEMs "neutral" target curves (why are they different? Question 1 above) have flat bass and mids until it reaches the pinna gain area.

3. Could this explain the "preference" target curves, which have increased bass, as an average of people trying to mimic the slope of a speaker in a room?

Adding all things, I suppose a true "sloped neutral" target curve should be defined by: a head (+torso?) simulator, in an anechoic room, with a flat speaker, playing a sloped response, and this reference be used by both headphones and IEMs.

And for fun.

View attachment 224398

Older I become, more and more I enjoy different tunning of headphones outside of box without any changes. Open, Closed, Planar, Dynamic, IEM with multiple drives, planars iem, two dynamics eim etc…

It seems to me that that thing with target is some kind of “average”, that you can like now, but pick something different tomorrow.

Rotating my daily iem and headphones weekly…

Somewhat similar thing with my main stereo setup. One week I like dirac live room correction for full speaker range (harman curve with 4db bass boost), next week I pick filter with curtains from 20hz to 2kz… and enjoy treble as it is in my room, because it is imperfect in a specific enjoyable way. Is it better? Better than what?

It is all experience, and variations in experience is enjoyable for me 100%

I hope it makes sense.

 

[AudioKC]

Since an album is mixed and mastered using a speaker in a room as a reference

I believe many metal bands and DnB producers mixing for headphones now. And you can hear it, on 4k stereo with flat response it sounds meh, in 20$ headphones it sounds awesome… go figure

 

[markanini]

I believe many metal bands and DnB producers mixing for headphones now. And you can hear it, on 4k stereo with flat response it sounds meh, in 20$ headphones it sounds awesome… go figure

It's more common to check whole mixes on headphone today. Speakers are still the main tool because work done headphones tends to translate poorly across different systems.

 

[Matias]

It's more common to check whole mixes on headphone today. Speakers are still the main tool because work done headphones tends to translate poorly across different systems.

Exactly what Steven Wilson says, among other things.

 

[Matias]

Back to this topic, now @crinacle has his new toy B&K 5128/4620, but more interesting, it has diffuse fields with tilts, like the -0.8 dB tilt that I used below as a reference and plotted 2 popular IEMs to compare.

In-Ear Fidelity – IEM Graph Comparison Tool

Easily compare industry-standard frequency response measurements of hundreds and hundreds of IEMs.

crinacle.com

 

[Matias]

@crinacle and I were on the same page apparently. This video is exactly what I described on the first post!

 

[Matias]

This squig has both headphones and IEMs measured with 5128 against various targets including DF with slopes. Bravo!

Listener's 5128 Database

Satiate your sordid squiggle yearning!

listener800.github.io