Are headphones frequency response measurements irrelevant?

[DTNT]

I've been obsessed with headphones measurements for years, specifically with frequency response measurements.

But, I am yet to find measurements that match with what I actually hear.

More specifically, none of the measurements match with what I hear when I run the SineGen. They are accurate up to 1khz or 5khz at best(!), but pretty much all of them are very inaccurate after that.

I know there is a lot of equipment and there are different compensation curves, but no matter what EQ I apply to my headphones, I don't get the flat response when I run the signal afterwards.

Are there frequency response measurements that perfectly match what you hear, and what those are?

 

[DVDdoug]

Frequency response is the most important thing! But you're right. It's tricky and imperfect, especially in the higher frequencies where the short wavelengths bounce-around in the headphone and your ear. And in the low-end it depends on how they fit around the ear and how well they seal (with closed headphones).

We don't usually want flat response. There is a "Harmon Curve" that matches most listener's preferences.

This thread has a good video about headphone measurements. The ironic thing is, the video has poor audio...

The most interesting thing I learned from the video is that listeners describe headphones with more distortion as "detailed". (I HATE that useless kind of audiophile terminology anyway.)

 

[Cote Dazur]

But, I am yet to find measurements that match with what I actually hear.

Measuring transducers is not 100% precise, hence it is an indicator with a margin of error. Does not make it irrelevant. It will point us in the right direction even if we are on our own to cover the last few steps. As we get better at measuring we might eventually get transducer measurement as precise and relevant as electronics measurement.

 

[DTNT]

Frequency response is the most important thing! But you're right. It's tricky and imperfect, especially in the higher frequencies where the short wavelengths bounce-around in the headphone and your ear. And in the low-end it depends on how they fit around the ear and how well they seal (with closed headphones).

We don't usually want flat response. There is a "Harmon Curve" that matches most listener's preferences.

This thread has a good video about headphone measurements. The ironic thing is, the video has poor audio...

The most interesting thing I learned from the video is that listeners describe headphones with more distortion as "detailed". (I HATE that useless kind of audiophile terminology anyway.)

This is such a treasury of a presentation, thank you so much! It contains everything I was asking about.

 

[solderdude]

Are headphones frequency response measurements irrelevant?

They are indicative of frequency response/tonal when having a good seal and presented in an understandable way.

 

[Loomynarty]

I've been obsessed with headphones measurements for years, specifically with frequency response measurements.

But, I am yet to find measurements that match with what I actually hear.

More specifically, none of the measurements match with what I hear when I run the SineGen. They are accurate up to 1khz or 5khz at best(!), but pretty much all of them are very inaccurate after that.

I know there is a lot of equipment and there are different compensation curves, but no matter what EQ I apply to my headphones, I don't get the flat response when I run the signal afterwards.

Are there frequency response measurements that perfectly match what you hear, and what those are?

Over ear headphones interact with your outer and inner ear which causes a HRTF that is unique to your head and ears. Measurements of frequency response are made on rigs that attempt to mimic an average HRTF which is not similar to any one individual. There are so many factors that affect how you hear over ear headphones that you should really only be using their frequency response measurement for comparative purposes or for matching a target response.

Basically, if you want something to sound like how it measures, buy IEMs.

 

[oleg87]

If you apply an EQ to loudspeakers measured in some other person's room to your room, you probably won't get ideal results.

They are useful (to a point) for apples-to-apples comparisons on the same rig, but I think you're trying to extract more value out of them than they're capable of offering.

 

[DTNT]

I think the biggest takeaway from Dan Clark's seminar is our ear shape and impedance, and how we differently perceive peaks. Which makes it really difficult to make universal measurements.

However, I absolutely agree there is a value in FR measurements, especially for comparison with other models. No matter what compensation curve was used, you can still get the orientation if you have a decent model to compare it.