Oratory1990 Canjam Presentation on Headphone Sound

[Curvature]

See the attached document.

The (I think) uncontroversial message is that heads and ears modify response and that the same headphone can present different frequency response to the listener. Not super different, but enough. Measurement and adjustment is complicated by where you place the microphone (ear canal entrance or close to the ear drum), the kind of headphone and the impedance it presents (e.g., closed vs open back) and consistency of positioning.

For those who use EQ presents, the positive message is that by using them you get closer to the average, but individualization and messing around with the presets until you get to something you like is necessary for a more optimal sound.

Very interesting set of slides.

 

[kemmler3D]

The interesting takeaway for me was that two headphones that measure identically on a fixture could sound different to the same person due to variances in acoustic impedance. (did I read that right?) It's something I haven't thought about before.

For me, I think the idea of a headphone target being enough to get us all the way to real hi-fi is dead. It's even more dead after reading this. The path forward is no individualized measurements and EQ. Targets are still useful to get us in the zone without too much distortion, but it seems that the "last mile" will never be traversed at the factory.

I think the greater emphasis lately (at least it seems so to me) on individual variation is good progress in the industry.

I am reminded of talking to an engineer that worked for a well-known IEM brand. They had started trying to characterize pinnae / individual responses, with the idea of grouping them and creating IEMs for each group, possibly... they apparently gave up after measuring 50. I guess individual responses are very resistant to being grouped.

Random thought: Lasers can be good for measuring vibration and so are sometimes also used as microphones. Measuring response at the eardrum with a mechanical transducer is tricky because sticking things into people's ears without injuring them is tricky, not to mention you have to position it really accurately at the same time. Is there any such thing as a laser microphone that can measure the motion of the eardrum directly? I feel like that would be definitive in this arena.

 

[Apples]

The fact he eqs to harman makes me question if he knows what good sound quality sounds like. And his desire to eq every headphone to it is just wild. Only benefit I got from it was a measurement database to analyze headphone performance.

Harman has thin anemic midbass, shouty 5khz presence, and a rolled off treble. Many people agree with this. Regardless of study it's not a target many companies actually use. More so a reference. But we already have diffuse field for that. As a preference target, harman was found to not even rate that high and required custom bass and treble adjustments via their own research.

 

[Curvature]

The interesting takeaway for me was that two headphones that measure identically on a fixture could sound different to the same person due to variances in acoustic impedance. (did I read that right?) It's something I haven't thought about before.

That's right. It's the primary reason why the B&K HATS was developed to model the high frequency impedance such that it matched closer to an average of people all the way to 20kHz.

For me, I think the idea of a headphone target being enough to get us all the way to real hi-fi is dead.

There will pretty much be three groups:

1. Regular humans. A headphone target with bass and treble shelves will do almost all the work. What's especially important is the usual stuff during the design stage, like minimizing resonances.
2. Technical audiophiles. They will individualize with ear canal entrance microphones, ear probes, or ear/head scans.
3. Untechnical audiophiles. They will do whatever they want, including creating random new targets.

Random thought: Lasers can be good for measuring vibration and so are sometimes also used as microphones. Measuring response at the eardrum with a mechanical transducer is tricky because sticking things into people's ears without injuring them is tricky, not to mention you have to position it really accurately at the same time. Is there any such thing as a laser microphone that can measure the motion of the eardrum directly? I feel like that would be definitive in this arena.

It's actually not dangerous at all if using medical-grade probes, which are extremely soft. The old-school audiology technique was "bump-and-pull"--insert the probe blind and wait for the patient to complain of a loud sound (like tapping a microphone), which means you hit the eardrum, and then pull back a smidge. The other technique, because most ear canals are bent, is to "rest it on the bend", where you insert the probe and guide it visually with an otoscope until you get to that point. However, not doing anything stupid and avoiding injuring yourself requires some learning. Stupid includes making probes out of small plastic piping which, you know, can puncture that ear drum no problem.

I had my ear canals scanned with a laser, but due to the geometry of my ear canals and the amount of hair and earwax in there meant the probe had to be inserted deep. They have a little meter which estimates total depth vs. insertion depth. The scan was accurate up to 15kHz. If you don't get within 2-5mm of the ear drum, the readings will not be accurate (for laser or microphone). If you're interested: https://www.audiosciencereview.com/...irtualization-users.62664/page-5#post-2431765

There is good work on the level of accuracy for ear entrance vs. deep insertion microphone placement, and you can get reasonable results without going deep. It is still kind of a gamble on the final quality, since without accurate measurements in the HF there will be errors in frontal localization and externalization, elevation and front/back confusion, although there is also evidence that users can adapt to new HRTFs up to a point with repeated exposure and training.

Edit: Typos.

 

[Curvature]

The fact he eqs to harman makes me question if he knows what good sound quality sounds like. And his desire to eq every headphone to it is just wild. Only benefit I got from it was a measurement database to analyze headphone performance.

Harman has thin anemic midbass, shouty 5khz presence, and a rolled off treble. Many people agree with this. Regardless of study it's not a target many companies actually use. More so a reference. But we already have diffuse field for that. As a preference target, harman was found to not even rate that high and required custom bass and treble adjustments via their own research.

You ever send an email to someone whose reply made you think they didn't read it? This reminds me of that.

 

[Chagall]

My takeaway:

1. Buy headphones with low distortion and good EQ headroom - because EQ is basically mandatory.
2. Apply Oratory EQ preset - get closer to target.
3. Use frequency sweeps - correct obvious peaks/dips you actually hear.
4. Fine-tune with actual music - broad bass/mids/treble adjustments to taste.
5. Stop - you’re probably 95-98% there without actually measuring your in-situ response.

- Get a closed-back only if absolutely necessary (isolation).
- Don't judge headphones just by looking at one FR response.
- Don't flame people for liking xyz headphones - they just might sound fine on their head.
- You can't EQ comfort, so get the ones that are actually comfortable.

 

[Curvature]

Stop - you’re probably 95-98% there without actually measuring your in-situ response.

This advice specifically applies to stereo. If you are getting into binaurally-rendered spatial audio then individualization via measurement is more or less required. Along with headtracking.

 

[Redacted]

Many of us have been saying for ages that physical ear shapes make a difference to perceived FR, in a similar way to speakers in different rooms....
The only real way to know is to try IEMS or headphones for yourself. Quite a gamble if you have to have boutique gear.

 

[shuppatsu]

The fact he eqs to harman makes me question if he knows what good sound quality sounds like. And his desire to eq every headphone to it is just wild. Only benefit I got from it was a measurement database to analyze headphone performance.

Harman has thin anemic midbass, shouty 5khz presence, and a rolled off treble. Many people agree with this. Regardless of study it's not a target many companies actually use. More so a reference. But we already have diffuse field for that. As a preference target, harman was found to not even rate that high and required custom bass and treble adjustments via their own research.

Uh, he's kind of a legend in the field. And in open collaboration with other the leading lights in the field. Believe me, he's well aware of the latest meta when it comes to tuning.

 

[RandomEar]

My takeaway:

1. Buy headphones with low distortion and good EQ headroom - because EQ is basically mandatory.
2. Apply Oratory EQ preset - get closer to target.
3. Use frequency sweeps - correct obvious peaks/dips you actually hear.
4. Fine-tune with actual music - broad bass/mids/treble adjustments to taste.
5. Stop - you’re probably 95-98% there without actually measuring your in-situ response.

- Get a closed-back only if absolutely necessary (isolation).
- Don't judge headphones just by looking at one FR response.
- Don't flame people for liking xyz headphones - they just might sound fine on their head.
- You can't EQ comfort, so get the ones that are actually comfortable.

I would add that not everyone likes "Super Best Target XYZ" - be it Harman, or any other. So the second step should probably be "Test out a range of targets to see/hear which one best fits your preference or goal.". You could also try that with any headphone you already own, just to get a rough idea which targets are at the top of your list. So it could also be step one, but it doesn't have to be.

 

[Curvature]

Many if us have been saying for ages that physical ear shapes make a difference to perceived FR, in a similar way to speakers in different rooms....
The only real way to know is to try IEMS or headphones for yourself. Quite a gamble if you have to have boutique gear.

It's not new information, just well-presented information.

 

[Redacted]

It's not new information, just well-presented information.

Well, if it's sufficiently authoritative, it may save a few people money by getting them to try before they buy, rather than relying on the all-too-common fake reviews!

 

[Curvature]

I would add that not everyone likes "Super Best Target XYZ" - be it Harman, or any other. So the second step should probably be "Test out a range of targets to see/hear which one best fits your preference or goal.". You could also try that with any headphone you already own, just to get a rough idea which targets are at the top of your list. So it could also be step one, but it doesn't have to be.

On one hand, you have a blinded, subjectively tested and statistically analyzed headphone target; on the other, you have something essentially random.

Anyone who cares enough to try multiple targets should go for individualization instead.

 

[Chagall]

I would add that not everyone likes "Super Best Target XYZ" - be it Harman, or any other. So the second step should probably be "Test out a range of targets to see/hear which one best fits your preference or goal.". You could also try that with any headphone you already own, just to get a rough idea which targets are at the top of your list. So it could also be step one, but it doesn't have to be.

I think that would mostly happen naturally during the final music-tuning step. Once you start making broad adjustments to bass, mids and treble to taste, you are effectively defining your own preferred target rather than choosing one in advance. Sure, you might end up with a lot of filters at first, but you can always simplify the final EQ curve afterward.

 

[shuppatsu]

I like where he's going with presenting this research. But while the slides have some good info, there's little depth and no references. Hopefully the talk will show up on YouTube at some point.

 

[Bow_Wazoo]

See the attached document.

The (I think) uncontroversial message is that heads and ears modify response and that the same headphone can present different frequency response to the listener. Not super different, but enough. Measurement and adjustment is complicated by where you place the microphone (ear canal entrance or close to the ear drum), the kind of headphone and the impedance it presents (e.g., closed vs open back) and consistency of positioning.

For those who use EQ presents, the positive message is that by using them you get closer to the average, but individualization and messing around with the presets until you get to something you like is necessary for a more optimal sound.

Very interesting set of slides.

That’s why it’s always been clear that an EQ preset is an individual starting point, not the end point.

 

[solderdude]

The fact he eqs to harman makes me question if he knows what good sound quality sounds like. And his desire to eq every headphone to it is just wild. Only benefit I got from it was a measurement database to analyze headphone performance.

Harman has thin anemic midbass, shouty 5khz presence, and a rolled off treble. Many people agree with this. Regardless of study it's not a target many companies actually use. More so a reference. But we already have diffuse field for that. As a preference target, harman was found to not even rate that high and required custom bass and treble adjustments via their own research.

You just don't fall in the 'majority of people' that seem to prefer the tonality of the Harman target.
Nothing wrong with that.
In your case it is caused by and explains your love/obsession of the DT 990 which has too way much (mid)bass and treble.
We call this preference.

 

[Blockader]

I've been trying to reach to people in these forums, and inform them that a headphone's frequency response may be different than what they hear.
Measurements are taken on a coupler built to approximate the average adult ear/ear canal. Useful for comparing two headphones on the same rig, but an average ear isn't yours, and IIRC they used only 32 people to average it.

A headphone isn't a fixed pressure source. It has an acoustic impedance of its own, and your ear is a load with its own. What reaches your eardrum is whatever survives the split between the two. Most headphones sit at the stiff end of that, meaning the diaphragm moves a roughly fixed amount of air and the pressure you get depends on what it's moving into. So the gap around your ear, how the pad sits on your jaw, whether you wear glasses, your canal length, all of it lands in the response.

Bass is a seal problem. Leaks act as a high pass, and leak size is down to your head, not the designer. Several dB of variation between people is normal. Above 2k it's anatomy. Your canal resonance and eardrum impedance sit downstream of the headphone entirely, so nothing the headphone does can reach them.

My solution, pick low acoustic impedance headphones that are compliant with your ear canals acoustic impedance instead of fighting with it. That way the response stays consistent from one person to the next, which is what makes a rig derived target worth aiming at in the first place. Then EQ them to a target you usually enjoy. Here's how to spot low acoustic impedance headphones:

• Open back. Back volume drops out of the source impedance entirely. Best single indicator and the one with real multi-subject data behind it: open-backs run ~1.5 dB standard deviation at 20 Hz across listeners, closed-backs 5–8 dB.(can link Sean Olive's presentation on this one)
• Doesn't change when you reseat it. Impossible to determine unless you have your own measurement rig.
• Measures the same on different rigs. 711 vs 5128. If the shape survives a change of load, it's load insensitive by definition. Most informative check available.
• Large pad cavity. Doubling the volume halves the source impedance. Strongest structural lever in the bass.
• Vented. A deliberate leak swamps the one your jaw and glasses make, so the bass corner is set by the designer. Bass only, does nothing above a few hundred Hz.
• Bass that rolls off early. Counterintuitive. Flat to 20 Hz on a rig means it needs a perfect seal to get there. A rolloff means the corner is intentional. (May not apply to all headpones)
• Large drivers. Referred impedance scales with 1/area², so this is a strong lever, but a big driver in a shallow cup eats the pad cavity and works against the item above.
• Low resonance frequency(where impedance curve peaks), low impedance peak height. Indicates a compliant, well-damped system. Only readable on dynamics: planars usually have flat electrical impedance with no visible peak, so the sign is missing on the class that's usually lowest impedance anyway.
• Angled or recessed driver, visible mesh in the front cavity. Broader illumination of the pinna, lower-Q modes. Weak signals but they point the right way.

 

[Curvature]

You just don't fall in the 'majority of people' that seem to prefer the tonality of the Harman target.
Nothing wrong with that.
In your case it is caused by and explains your love/obsession of the DT 990 which has too way much (mid)bass and treble.
We call this preference.

Preference can only be established under blind circumstances. Liking something is different. And the Harman target is not a fixed curve.

 

[Curvature]

visible mesh in the front cavity

Do you have data on the benefits of this?