Comb Filtering and Headphones

[mnarwold]

Ethan Winer has a good write-up about comb filtering here and blames it for the differences people hear between cables, power cables, etc. When it comes to speakers, I think  he makes a great point. Of course, that doesn't really effect headphones does it? I mean, I guess its possible that the sound changes a little bit every time you put your cans on since they will never be in the exact same place twice in a row.
 
Anyway, the article gave me a couple other ideas about HPs. First I thought, "yet another reason I'm glad I've gone the headphone route rather than the speaker route," but then I wondered if this somehow ends up being a practical disadvantage. Could part of (not all) the "unnaturalness" people talk about with cans be contributed to a lack of comb filtering? Real life doesn't have a flat frequency response. According to the article, we're used to comb filtering and each ear receiving a different frequency response.
 
I guess I don't have a specific question, just wanting to share my thoughts and see what you guys think.
 
P.S. Thanks Ethan for all your nice articles and sharing your experience with the world.

 

[Speedskater]

That 'comb filtering' paper does not apply to headphones.  It's about how the sound in a room will change when a listener moves his head a small amount.  So just the simple act of swapping anything makes it difficult to get back to the exact same listening position.

 

[mnarwold]

Quote:

That 'comb filtering' paper does not apply to headphones.  It's about how the sound in a room will change when a listener moves his head a small amount.  So just the simple act of swapping anything makes it difficult to get back to the exact same listening position.

Well yes and no. We don't have comb filtering in headphones, but it could still be an issue for cans since we are removing a natural phenomenon. So one could argue HPs are more accurate and therefore better, and another person could argue that they are less natural (no comb filtering) and therefore worse.

 

[xnor]

Quote:

Ethan Winer has a good write-up about comb filtering here and blames it for the differences people hear between cables, power cables, etc. When it comes to speakers, I think  he makes a great point. Of course, that doesn't really effect headphones does it? I mean, I guess its possible that the sound changes a little bit every time you put your cans on since they will never be in the exact same place twice in a row.

Yeah, placement changes the frequency response. Worn pads change the frequency response. I guess a lot of hair also reduce seal and therefore negatively effect low-end response. And so on..
 
Quote:

Anyway, the article gave me a couple other ideas about HPs. First I thought, "yet another reason I'm glad I've gone the headphone route rather than the speaker route," but then I wondered if this somehow ends up being a practical disadvantage. Could part of (not all) the "unnaturalness" people talk about with cans be contributed to a lack of comb filtering?

Unnaturalness in headphone playback stems from two major things: a) unnatural stereo separation and b) fixed positioning on your head.
a) If someone stands to the left of you and speaks to you your right ear will also receive sound pressure, just at a lower level and slightly time delayed. This is what our hearing expects. With a headphone, sound in one channel (usually a result of hard panning instruments to one side) will only reach the same ear, not the other. A possible remedy is crossfeed.
 
b) You unconsciously move your head slightly to precisely locate sound sources, but your headphone is fixed on your head so a sound source doesn't move with head movement. The Smyth Realiser uses a head tracker (sensors that track head movement) and processes the audio accordingly in realtime.
 
Of course there's a lot more to it, this is just a simplistic explanation. Research HRTF, sound localization if you want to know more.
 
I don't think a lack of comb filtering has a negative effect, because comb filtering itself is bad.
 
Quote:

Real life doesn't have a flat frequency response. According to the article, we're used to comb filtering and each ear receiving a different frequency response.

I've read the article but it was a long time ago. Does it really say that?
It says: "We don't usually notice these changes when moving around because each ear receives a different response, so what we perceive is more of an average." The important thing here is the averaging I guess, not that each ear receives a different comb-filtered response.
 
 
Back to frequency response: at low frequencies, headphones can provide a flat frequency response, but at higher frequencies it gets complicated. There's pinna, ear canal reflections and resonances. This is why we see very erratic high frequency responses in dummy head measurements. So I guess one could argue that at high frequencies there is comb filtering with headphones..

 

[EthanWiner]

Good thread, and thanks mnarwold.

Yeah, placement changes the frequency response. Worn pads change the frequency response. I guess a lot of hair also reduce seal and therefore negatively effect low-end response. And so on.

Yep.

 

[mnarwold]

Quote:

 
Unnaturalness in headphone playback stems from two major things: a) unnatural stereo separation and b) fixed positioning on your head.
a) If someone stands to the left of you and speaks to you your right ear will also receive sound pressure, just at a lower level and slightly time delayed. This is what our hearing expects. With a headphone, sound in one channel (usually a result of hard panning instruments to one side) will only reach the same ear, not the other. A possible remedy is crossfeed.

 
I use a crossfeed when listening on the computer.

 

[Speedskater]

'Ethan' isn't that more HRTF or maybe Head Phone Related Transfer Function than room 'comb filtering' ?

 

[xnor]

What do you mean with "that"? But anyway, headphones more or less disable the natural Head-Related Transfer Functions (HRTFs) depending on the type of head/earphone.

 

[Speedskater]

The "that" is about this part of your post:
 
Quote:

Yeah, placement changes the frequency response. Worn pads change the frequency response. I guess a lot of hair also reduce seal and therefore negatively effect low-end response. And so on.

 
Which I believe is more  a HRTF or maybe Head Phone Related Transfer Function than 'comb filtering'
 
 
I'm not disagreeing with your post, I'm just looking for a title or function or definition or something.

 

[xnor]

I see. To clarify, there is no "phone" in HRTF. A HRTF describes how the ear receives sound from a point in space.
 
For example with a normal stereo speaker setup, the right speaker's center of the tweeter is located at 0° elevation (same height), +30° azimuth.
If you measured what the ear received from a perfect speaker (flat FR from 20 Hz - 20 kHz) in an anechoic chamber at that point in space, you'd measure completely different frequency responses for the left and right ear and the left ear would also receive the sound slightly delayed. That's a HRTF.
 
 
The changes in headphone FR are due to different seal, different distance from the eardrum (worn, flat pads), and placement.
Peter/Tyll posted: "Even the smallest variances in placements alter the resonance artifacts (peaks and valleys) in the higher frequencies."

 

[jaddie]

Quote:

The "that" is about this part of your post:
 
 
Which I believe is more  a HRTF or maybe Head Phone Related Transfer Function than 'comb filtering'

It's the lack of HRTF.  The sound coming from headphone drivers has no opportunity for HRTF.  Crossfeed attempts to add a bit artificially to get the image out of our heads, the results are often just a little better.
 
If I may, HRTF is the difference in frequency response and time delay of a sound arriving at the ear farthest from the source vs the ear closest to the source.  The reason fixed crossfeed doesn't work is that real HRTF varies with source elevation and angle within the entire 360 degree sphere.  So no single crossfeed correction would be right for the entire sound stage, even if limited to within a 90 degree angle on a plane at zero degrees elevation directly in front.  
 
To be fair, it's only a little better with stereo speakers in a room, because the actual sound we hear is localized to the speakers in the room, and speared between them by the rather confused imaging cues in stereo music, which are a mostly a jumble of intensity-panned stereo and simulated spaces.  It's a mess, really, can't ever be right, but some like the forward position of some from speakers rather than the in-head position from headphones.  You can add inverse-crossfeed to speakers and get the image out of the speakers and all over the room, but the sweet spot where it works is almost microscopic.  You'd need a head clamp to hear it repeatably.  
 
And all the stuff xnor just posted...+1.

 

[Speedskater]

I just went back and re-read page 38 of Toole's book and I don't think that my Head Phone pun was all that far off the mark.

 

[jaddie]

Quote:

I just went back and re-read page 38 of Toole's book and I don't think that my Head Phone pun was all that far off the mark.

I see where you're going, but the change in response from the placement of headphones on the head in somewhat different positions, or seal quality, is nothing like the response changes from different arrival angles in space.  Yes, they both change response because of source movement, but what changes is entirely different in cause, magnitude and area of the spectrum affected.

 

[EthanWiner]

'Ethan' isn't that more HRTF or maybe Head Phone Related Transfer Function than room 'comb filtering' ?

Sure, or just slight differences in alignment, or loss of bass from a poor seal.

--Ethan