Hearing and the future of loudspeaker design

markus76 · 2013-11-04 12:12 pm

David Wareing said:
Some are near coincident/coaxial, some are not.

?? Are you recording at the ear drum? Furthermore do you compensate the transfer function of headphone to ear drum in playback?

markus76 · 2013-11-04 12:18 pm

David Wareing said:
That sort of depends on how severe the recorded HRTFs are. If they are diffuse field equalized and the recording is of a diffuse field, then there is not much of a problem, for example.. If the dummy head is 'flat' in all directions then that would be even better, if it ever worked at all..

What do you mean with "how severe the recorded HRTFs are". They are what they are and what they need to be. Our torso, head and pinna is a spatial encoder. Each angle of incidence represents a new and unique transfer function which alters the direct sound. Removing those transfer functions effectively creates a stereo recording, not a binaural recording.

David Wareing · 2013-11-04 12:26 pm

markus76 said:
?? Are you recording at the ear drum?

No. Closer to a 'blocked meatus' approach.
If you look at a spectrogram of a recording of broadband noise, such as wind, you should learn nothing, and it should sound ok on stereo speakers.

Furthermore do you compensate the transfer function of headphone to ear drum in playback?

No, just one (your own) ear canal resonance involved.

David Wareing · 2013-11-04 12:45 pm

markus76 said:
What do you mean with "how severe the recorded HRTFs are". They are what they are and what they need to be. Our torso, head and pinna is a spatial encoder.

But nobody has really proved which bit does what, beyond the ancient 'Duplex Theory', although Duda identified a 'chest bounce' effect, he told me about.

A dummy head's HRTF is whatever it is, and not due to your head and torso, or what you say it 'should' be.

Each angle of incidence represents a new and unique transfer function which alters the direct sound. Removing those transfer functions effectively creates a stereo recording, not a binaural recording.

That is pretty much the received wisdom, but nobody has made a decent binaural recording using anechoically measured HRTFs, as far as I know.

markus76 · 2013-11-04 1:18 pm

David Wareing said:
No, just one (your own) ear canal resonance involved.

If the ear canal isn't involved in the recording then it indeed doesn't need to be compensated for. Nevertheless there are resonances from headphone to ear canal entrance. Furthermore and probably even more important is to compensate for the frequency response abberations of the headphone itself.

markus76 · 2013-11-04 1:24 pm

David Wareing said:
That is pretty much the received wisdom, but nobody has made a decent binaural recording using anechoically measured HRTFs, as far as I know.

This would involve creating HRTFs for a multitude of angles. The recording would need to have as many channels as HRTF angles involved. How many HRTFs are needed? Nobody really knows for sure.

David Wareing · 2013-11-04 1:38 pm

markus76 said:
If the ear canal isn't involved in the recording then it indeed doesn't need to be compensated for. Nevertheless there are resonances from headphone to ear canal entrance. Furthermore and probably even more important is to compensate for the frequency response abberations of the headphone itself.

It works well enough for me and my headphones without. There are data available for the KU 100, and headphones, if you are interested. Audio Group Download Server How should one measure headphones? What to look for? Learning Center - Build a Headphone Graph | HeadRoom Audio

markus76 · 2013-11-04 1:49 pm

David Wareing said:
How should one measure headphones? What to look for?

Simply measure the transfer function headphone to mic. Then inverse that filter and apply it via convolution to the playback signal.

David Wareing · 2013-11-04 2:04 pm

markus76 said:
Simply measure the transfer function headphone to mic. Then inverse that filter and apply it via convolution to the playback signal.

They normally measure with a head and torso device, but that makes it difficult to interpret. A mic free-field is no good - no bass for one thing. I have used a mic set flush in a flat plate placed against the headphone pad, but I am not sure of what use that is in general. I used it to measure changes whilst modifing (adding damping). To be meaningful, you probably need to use a probe in your own ear and compare the headphone response to the results using a speaker at some free-field position or in a reverberation chamber. Outside my means..

markus76 · 2013-11-04 2:25 pm

^
Mics in (dummy head) ears, put headphones on, measure.

David Wareing · 2013-11-04 2:27 pm

markus76 said:
^
Mics in (dummy head) ears, put headphones on, measure.

err no.. you have to drill into your own head dummy! See edited version above.

markus76 · 2013-11-04 2:44 pm

Why would you need "to drill into your own head dummy"? If you want to exclude the transfer function of headphone to ear canal entrance then simply measure it and create a inverse filter. By the way, that's exactly what the Smyth Realiser does.

David Wareing · 2013-11-04 2:50 pm

markus76 said:
Why would you need "to drill into your own head dummy"? If you want to exclude the transfer function of headphone to ear canal entrance then simply measure it and create a inverse filter. By the way, that's exactly what the Smyth Realiser does.

I thought it used _your_ head..

markus76 · 2013-11-04 3:00 pm

Yes it does and exactly what needs to be done. If we want to compensate for the additional effects of headphone playback then we need to measure them, build a filter and convolve it with the playback signal.

David Wareing · 2013-11-04 3:21 pm

markus76 said:
Yes it does and exactly what needs to be done. If we want to compensate for the additional effects of headphone playback then we need to measure them, build a filter and convolve it with the playback signal.

And you can use this for pre-recorded binaural recording? And my file still sounds like stereo? Perhaps you can afford to get some HD 219 for those camping trips..

markus76 · 2013-11-04 3:26 pm

^
No, I didn't do this. I was talking about an idealized scenario. At http://www.audiogroup.web.fh-koeln.de/ku100hrir.html there are a couple of compensation filters for headphones that could be used though.

I was listening to your files over a Stax SR-202.

David Wareing · 2013-11-04 3:37 pm

markus76 said:
^
No, I didn't do this. I was talking about an idealized scenario. At Audio Group Download Server there are a couple of compensation filters for headphones that could be used though.

I was listening to your files over a Stax SR-202.

I am honoured, but you could have spent less!

markus76 · 2013-11-04 4:17 pm

I didn't buy it because of you

David Wareing · 2013-11-04 4:32 pm

markus76 said:
I didn't buy it because of you

I just got a report from a friend who heard everything in the right places, unprompted. He has the HD219s (prompted).

jan.didden · 2013-11-04 4:37 pm

markus76 said:
Why would you need "to drill into your own head dummy"? If you want to exclude the transfer function of headphone to ear canal entrance then simply measure it and create a inverse filter. By the way, that's exactly what the Smyth Realiser does.

Markus do you have a Realiser? I recnetly got one and am discovering the possibilities.

Jan

Hearing and the future of loudspeaker design

Banned

Banned

Member

Member

Banned

Banned

Member

Banned

Member

Banned

Member

Banned

Member

Banned

Member

Banned

Member

Banned

Member

AX tech editor