The FR Curve of Headphones

I am not so similar with FR curve of headphones, and I saw the some of them, for example, the curve of hd800, it's so different compared with speaker drivers, curve of speaker (3-way or 4-way) are well, overall flat to 20k (of course the speaker bass part has some roll off, and the sr009 bass flats to 20Hz, wow), but on the headphones, the curve looks unlike the speakers', why are they so different?

 

http://www.innerfidelity.com/content/headphone-measurements-explained-frequency-response-part-one

In a nutshell: human hearing response isn't flat due to external physiology (ear shape, body, etc). A flat source like a speaker won't be flat by the time it reaches the eardrum. However, headphone are worn on or in the ears, so they bypass these effects for the most part. They need to recreate the effects artificially in order to be nominally correct in their response.

 

Super complicated and huge topic, but some (oversimplified) factors:

• Speakers can more easily be tuned to a personal target curve. Speaker drivers all have known FR curves, sensitivities, impedances, etc. ... This means everyone can buy off-the-shelf drivers and design a crossover network around them and end up with a very linear frequency response.
• The physics of speaker enclosures are well-known. You can use programs to calculate enclosure size, port tuning, baffle size etc., ... to make the speaker have a more neutral frequency response.
• Most headphone drivers are made by the manufacturer themselves. While this can be a huge advantage, especially with headphones, there are many factors to consider when making a new driver. Of course these issues aren't limited to headphones, but since virtually every headphone is a single driver, no crossover design, there are always trade-offs when aiming for a neutral response (Not that multi-driver speakers aren't a big box of trade-offs. Everything is a trade-off). A headphone that measures close to a flat line from 100Hz-1kHz may be relatively more simple, but making a headphone perceptibly neutral in the upper mids and treble is very hard.
• Speaker target responses have long been defined (curve similar to B&K target at listening position or a flat line in an anechoic chamber on tweeter axis at 1m), whereas headphone target responses are still changing. There's also not an industry-wide standard on how to even measure headphones, which makes the whole topic of neutral response in headphone even more complicated. I believe headphones are often tuned on dummy heads, but manufacturers also use other measurements. For example the HD800 4kHz dip which shows up on the flat plate couplers nearly completely disappears on a dummy head or a real head. Different headphones will differ in their interactions with different measurement couplers or dummy heads (and even real heads).
• While it may be harder to make a headphone perceptably neutral, I'd also argue that it is easier for a headphone to have a flat response up to around 1kHz. But even then there is still research if a flat line really sounds neutral or how the target response should be.
• The measured and published speaker frequency response curves are often much better than the real world results, because of how the room interacts with the speakers:
  • Especially the bass is nearly completely shaped by the room and not the speaker. Only a very huge room (with super high ceilings) can have a bass that doesn't show huge 5-15db room modes. Dipole radiation in the bass can help in this regard. Many people use multiple subs, EQ or even DSP solutions to get a flatter bass response in their rooms. The room interaction may get less with higher frequencies, but especially the response in the lower mids is still often very much dominated by the room response and not the speaker.
  • Related to the above: There are also trade-offs in crossover design. Many people seem to forget the vertical response curves, but the floor is often the first reflection surface! More specifically you will always get dips and peaks in the vertical response around the crossover region if the drivers aren't located in the same spot (which is basically never the case). This is less of an issue with 2-way designs where the drivers are closer together and further from the floor.
    • EDIT: Of course you can minimize the vertical dispersion unevenness with a D'Appolito configuration, but a real D'Appolito is very hard to do.
  • Speakers are also a reflective surface. Similar problems arise with headphone enclosures. Most of the time this isn't a huge issue.
  • There's also not a standard what a "neutral room" sounds like and there never will be. How much damping do we need? How big should a listening room be? Should we try to absorb the first reflections nearly completely? How much diffusion? Most of this comes down to personal taste.
• There's also the big topic of CSDs, impulse response, etc. While this is not directly related to the FR, there are also some companies that make speakers with less steep crossover slopes to get better impulse or step responses (less phase shifts), often sacrificing neutral FR curves and lack of resonances. This is also one of the reasons why many people are fans of the single-driver speakers, even if they have problems with lack of resonances, smooth frequency response, distortion, bass extension, ...
  • On the other hand the small drivers of headphones most of the time have far less resonances in the audioband than most wide-band speakers. You can't just have an 8" driver that doesn't have resonances in the upper mids and treble, while still having good enough extension to cover the treble region. I personally think this is the #1 advantage of headphones: Single driver while still having low resonance, low distortion, good extension to both ends of the spectrum, high sensitivity, ...
• The whole idea of neutral response is mainly based on speaker targets. Speaker listening has always been the reference for judging headphone tonal balance.

In the end, many speakers may measure neutral when measured in an anechoic chamber on the tweeter axis at 1m, but very few manage to sound neutral in most rooms. I'd even argue that a well-designed headphone (e.g. HD6X0) can sound more neutral than many speaker setups.

 

Dang. Thanks for the great info!

 

Speakers are measured usually in an anechoic chamber, with a microphone directly in front of the speaker (no room reflections measured; explains why FR charts look more on the flat side). Headphones are measured by inserting microphones into simulated ear caveties (not everyone's ears are shaped the same; explains why FR charts look NOT flat).

There is no standard measuring system that all manufacturers of both speakers and headphones have to adhere to.

Because of the above, all kinds of factors come into play and it's not a one-size-fits-all outcome.

 

Also, this topic (headphone neutrality) has been discussed at length on both Changstar and SBAF. Here's the thread on here:

• What is Neutral Frequency Response

Here's another thread on SBAF. I collected a bunch of links, especially old threads from Changstar and linked them in a post:

• Frequency response vs actual sound

Not to be mean, I really don't think we need another thread about this. It would be nice if we could somehow merge the threads. I also feel the thread on SBAF is a little lackluster compared to the information on Changstar. The most important things are there, but there are many interesting things that aren't on here.