Coming back to review this thread. Very interesting stuff here. A few thoughts follow. The questions below are largely rhetorical, but I welcome your thoughts and discussion.. and hope I am not derailing the thread with this somewhat oddball post.
I make live music recordings using a multichannel microphone array, which I reproduce both via mulitchannel speaker playback arrays and via mix-down to 2-channel stereo.
My thoughts here are threefold-
1) In manipulating a raw recording for either playback method, I typically start by balancing and EQing logical stereo-pair subsets of the multichannel microphone array output to best effect prior to routing or combining those pairs (which can include further balance and EQ modifications of the baseline pairs to best effect in combination). In the case of multichannel playback this means working various L/R microphone pairs to best effect using L/R speaker playback alone prior to introducing the center channel. In the case of a 2-channel stereo mix it means getting the L/R microphone pairs working reasonably well on their own prior to bringing in center microphone channel content with subsequent balancing to get everything working together. In either case, introduction of the center channel content can spur me to go back and further modify the L/R content in an iterative process, yet proper initial balance of each pair on its own it makes for a good starting point. It is that part of the process which got Pano's words above concerning achieving an even tonal balance across the phantom stage resonating with me.
The post-production process I describe above has made me aware of the interaction of phantom and hard center imaging manifest across three different yet related regimes within the overall process- first with regards to the microphone array geometry side of things, second within the mix-regime, and third in a multichannel playback regime which includes a center speaker. The first two of those apply to 2-channel reproduction. The second is the regime in which you all have been applying and discussing the methods explored in this thread.
2) What might be the effect of this kind of compensation of the L/R content in a 2-channel mix (2nd regime above) of multichannel microphone array content that includes a discrete center channel in addition to several L/R pairs (the 1st regime)? What might be the effect of it when reproduced discretely via a speaker array that includes a physical center channel, where both phantom and hard center imaging is occurring at the playback end of things(3rd regime)?
3) It strikes me that such compensation would ideally be made at the mastering (2nd regime) stage rather than upon reproduction. However this assumes the effect is robust enough to work universally across different playback system arrangements, rather than needing to be tuned to a specific playback system.
Thoughts?
I make live music recordings using a multichannel microphone array, which I reproduce both via mulitchannel speaker playback arrays and via mix-down to 2-channel stereo.
My thoughts here are threefold-
1) In manipulating a raw recording for either playback method, I typically start by balancing and EQing logical stereo-pair subsets of the multichannel microphone array output to best effect prior to routing or combining those pairs (which can include further balance and EQ modifications of the baseline pairs to best effect in combination). In the case of multichannel playback this means working various L/R microphone pairs to best effect using L/R speaker playback alone prior to introducing the center channel. In the case of a 2-channel stereo mix it means getting the L/R microphone pairs working reasonably well on their own prior to bringing in center microphone channel content with subsequent balancing to get everything working together. In either case, introduction of the center channel content can spur me to go back and further modify the L/R content in an iterative process, yet proper initial balance of each pair on its own it makes for a good starting point. It is that part of the process which got Pano's words above concerning achieving an even tonal balance across the phantom stage resonating with me.
The post-production process I describe above has made me aware of the interaction of phantom and hard center imaging manifest across three different yet related regimes within the overall process- first with regards to the microphone array geometry side of things, second within the mix-regime, and third in a multichannel playback regime which includes a center speaker. The first two of those apply to 2-channel reproduction. The second is the regime in which you all have been applying and discussing the methods explored in this thread.
2) What might be the effect of this kind of compensation of the L/R content in a 2-channel mix (2nd regime above) of multichannel microphone array content that includes a discrete center channel in addition to several L/R pairs (the 1st regime)? What might be the effect of it when reproduced discretely via a speaker array that includes a physical center channel, where both phantom and hard center imaging is occurring at the playback end of things(3rd regime)?
3) It strikes me that such compensation would ideally be made at the mastering (2nd regime) stage rather than upon reproduction. However this assumes the effect is robust enough to work universally across different playback system arrangements, rather than needing to be tuned to a specific playback system.
Thoughts?
For the curious, my current recording array consists of 8 microphone channels, which can be thought of as consisting of the following logical pairs (it is a single array, exclusive of spot or close mics) :
● A wide-spaced omni pair
● A forward-facing center coincident Mid/Side pair
● A forward-facing near-spaced directional L/R pair (forming an L/C/R triplet with center Mid microphone)
● A rear-facing near-spaced directional Ls/Rs ambient/audience/room pair
In the images below the omnis are mounted in small spherical baffles at the end of the telescopic arms, the larger windscreens contain the directional microphones, the center of which contains two microphones (the Mid/Side pair). [edit- forward axis of the array in the photos below is toward the upper left)
I'm posting this simply to clarify where I'm coming from in the post above, not intending to derail into a discussion of recording, or this array in particular!
● A wide-spaced omni pair
● A forward-facing center coincident Mid/Side pair
● A forward-facing near-spaced directional L/R pair (forming an L/C/R triplet with center Mid microphone)
● A rear-facing near-spaced directional Ls/Rs ambient/audience/room pair
In the images below the omnis are mounted in small spherical baffles at the end of the telescopic arms, the larger windscreens contain the directional microphones, the center of which contains two microphones (the Mid/Side pair). [edit- forward axis of the array in the photos below is toward the upper left)
I'm posting this simply to clarify where I'm coming from in the post above, not intending to derail into a discussion of recording, or this array in particular!
Attachments
Last edited:
It seems the recording side is also important. I am more into the Binaural Recording camp, but I also understand the need for other methods. Below is a listening session we played around with.
Augene - Hear the real thing!: Listening in an Anechoic Chamber
Recording methods will effect how one might process the files. Building a binaural recording head is one of my projects in progress, currently chose a head from a sex doll, but also trying to find a larger sized head whereas the current size is some 50cm circumference. Would like to find on one with 60 cm size.
Augene - Hear the real thing!: Listening in an Anechoic Chamber
Recording methods will effect how one might process the files. Building a binaural recording head is one of my projects in progress, currently chose a head from a sex doll, but also trying to find a larger sized head whereas the current size is some 50cm circumference. Would like to find on one with 60 cm size.
My advice in that case- Get the ears right, or go with a simplified model that intentionally simplifies the highly complex pina-interaction to avoid mis-match (such as miniature omnis placed at your own temples, simple wedge-shaped ears on a dummy head, or at the extreme, an earless head-sized spherical baffle).
If going with ears, best to either pull a mold from your own, or simply use your own head for making the recording. Easy to place miniature omni mics at the opening of the ear canals. More challenging to go full binarual and use soft flexible tubes to sample adjacent to the ear-drum ala David Griesinger, which requires EQ compensation for the ear-canal resonance, but is capable of true binarual reproduction of the experience from an actual seating location suitable for acoustic analysis.
^
That works when combined with headphone EQ correction matching to one's personal HRTF response for a single center speaker source (again see Griesinger).. which brings us back full circle to achieving truly-correct tonality of all angular positions across the playback stage (for headphone reproduction at least).
I leverage binarual cues in mixing to 2-channel stereo, but do so via interactions of the microphone array channels mentioned above.
[edit- looking forward to giving your link above a listen when I get a chance, thanks!]
If going with ears, best to either pull a mold from your own, or simply use your own head for making the recording. Easy to place miniature omni mics at the opening of the ear canals. More challenging to go full binarual and use soft flexible tubes to sample adjacent to the ear-drum ala David Griesinger, which requires EQ compensation for the ear-canal resonance, but is capable of true binarual reproduction of the experience from an actual seating location suitable for acoustic analysis.
^
That works when combined with headphone EQ correction matching to one's personal HRTF response for a single center speaker source (again see Griesinger).. which brings us back full circle to achieving truly-correct tonality of all angular positions across the playback stage (for headphone reproduction at least).
I leverage binarual cues in mixing to 2-channel stereo, but do so via interactions of the microphone array channels mentioned above.
[edit- looking forward to giving your link above a listen when I get a chance, thanks!]
Last edited:
I actually had done recording from my own head with modified earbuds. The special reproduction was so convincing when playing back through earbuds, that I hope to see how different head size and ears will effect the result. But I also noticed some shortcoming during that recording.
In the reference blog post, the Chesky binaural recording was very convincing as well, the guy really is whispering in my ear! This is what seemed promising to use an artificial head.
In the reference blog post, the Chesky binaural recording was very convincing as well, the guy really is whispering in my ear! This is what seemed promising to use an artificial head.
Be aware that your ears are yours alone. Super convincing recordings made with them that work great for you may not translate well to others, which may or may not be important to you.
Also, on a deeper fundamental level, its all about creating a convincing auditory illusion in the mind of a listener with a willing suspension of disbelief.. even when that listener is you listening to a recording made with your own head. There is an underlying tension between true accuracy and producing the most engaging illusion.
Also, on a deeper fundamental level, its all about creating a convincing auditory illusion in the mind of a listener with a willing suspension of disbelief.. even when that listener is you listening to a recording made with your own head. There is an underlying tension between true accuracy and producing the most engaging illusion.
I agree with the creating an illusion bit, I'm not sure I understand exactly what you mean about a willing suspension of disbelief?
There have been different instances of experiment. Listening to “whisper into the left ear” track in the Anechoic room was the most promising experience that a good binaural recording can be produced for most people. I have not yet tried it with proper earbuds. Since most listening environment will be far from Anechoic, it seems more reasonable to aim for earbud listening experience. Actually, the artificial head would aim to understand what kind of design requirements for earbuds would be necessary.
I think it's worth mentioning that the contribution of the listeners' pinnae to the recorded sound was demonstrated in an experiment that was done perhaps 35 years ago with, I believe, Sennheiser mics of the stethoscope variety where the diaphragm was very near the eardrum, but didn't block the sound. Multiple mic sets were distributed to a group of sound engineers who listened to an acoustic performance by some sort of band. Each engineer's mic output was recorded to a stereo pair of tracks, and when headphone playback was performed, everyone could easily identify the recording made through their pinnae, moreover hear how differently the same event was rendered with other pinnae being the only variable.
This led me to think that the best solution for binaural recording is using a head shaped and sized structure but with mics located at the ear position, without pinnae—just flush with the surface. Then it seems that the most rational playback scheme would be circumaural headphones, in a pinch (because the angle of incidence is wrong) and preferably headphones which can swing out, like the AKG K-1000, the Raal Requisite and Mysphere 3.1s.
This way, the listener's pinnae are fully engaged.
This led me to think that the best solution for binaural recording is using a head shaped and sized structure but with mics located at the ear position, without pinnae—just flush with the surface. Then it seems that the most rational playback scheme would be circumaural headphones, in a pinch (because the angle of incidence is wrong) and preferably headphones which can swing out, like the AKG K-1000, the Raal Requisite and Mysphere 3.1s.
This way, the listener's pinnae are fully engaged.
Last edited:
What is required from earbuds or other headphones for correct binaural reproduction? Primarily matching the frequency response through the headphones to that of a frequency-flat loudspeaker placed in directly in front of the listener. This compensates for both the headphone response and the listener's HRTF, and the correction for each listener/headphone combination will be unique.
See the latest on this on Greisinger's website, currently at the top of the home page (latest update March 2, 2021) http://www.davidgriesinger.com/
Because it is essentially correcting the phantom center spectrum (as heard through the headphones) to match that of an actual center source, this strikes me as having at least a tangential relationship to the thread topic.
See the latest on this on Greisinger's website, currently at the top of the home page (latest update March 2, 2021) http://www.davidgriesinger.com/
Because it is essentially correcting the phantom center spectrum (as heard through the headphones) to match that of an actual center source, this strikes me as having at least a tangential relationship to the thread topic.
We as listeners of reproduced music are always aware on some level that we are listening to an illusion via the reproduction. It's great when the reproduction sounds "like your are there" (or they are here), but no sane listener actually believes she is there. The presence of sufficient cues which satisfy certain psychoacoustic listening thresholds allows us to relax our criticality sufficiently as long as we are willing to "go along with it" and accept the trick. In that way listener is and must be a willing participant. There will always be things a critical listener can point to that are incorrectly reproduced. We have the ability to switch back and forth between critical assessment and immersion in our enjoyment of the illusion. Well engineered recordings and well engineered playback systems make that mindset switch easier, be it at the level of intentional awareness on the part of the listener or not.
Flat relative phase response free from abrupt changes or jumps are the key to a stable and precise stereo image. Channel symmetry is also critical. The closer both channels are matched in phase, frequency response and overall amplitude, the more accurate and stable the stereo image. Higher end drivers are usually matched in pairs for a good reason, unless the manufacturer has extremely strict manufacturing tolerances.
Obviously speaker placement is very critical as is the listening room's affect.
Obviously speaker placement is very critical as is the listening room's affect.
That choice double encodes for the frequency response portion of HRTF - once in the binaural dummy head during recording, a second time in the playback headphones - and never made sense for me. Granted I couldn't find much to choose between, for example, diffuse field contoured Etymotic and ruler-flat Audio-Technica IEMs on most binaural recordings, chalking it up to the differences in anatomy. The most convincing headphone reproduction for me remains the few Cereni Holophonic cuts.
There's alot to making a believable binaural recording. It's so much more involved than walking around with a cellphone and ear bud mics (pretending to be accurately binaural).
Even if you do record with a real expensive Neumann binaural dummy head setup, there's still lots of room for error. The recording has to be properly equalized, free from compression and noise.
The Jecklin disc with very good omni mics makes the most believable binaural recording IMO. There still is the issue with high end response lift, etc.
Art Halperin and Barry Diament have done some really good binaural work. "Simplicity" by Work of Art is a good example of a decent binaural recording, free from compression, limiting and processing.
Even if you do record with a real expensive Neumann binaural dummy head setup, there's still lots of room for error. The recording has to be properly equalized, free from compression and noise.
The Jecklin disc with very good omni mics makes the most believable binaural recording IMO. There still is the issue with high end response lift, etc.
Art Halperin and Barry Diament have done some really good binaural work. "Simplicity" by Work of Art is a good example of a decent binaural recording, free from compression, limiting and processing.
This is a similar thought as what I had in mind, but rather put the mics at a similar location where the earbuds exit would be rather than the current location in most dummy heads. Then measurement would be conducted with sound source with minimum HRTF because. This would tell what kind of response the earbuds should have. Then equalize earbud playback response in a way that it would measure flat. And compare listening experience.
Why earbuds? Because this is the most convenient and popular way of listening, almost eliminating other effects.
Last edited:
Where objectively accurate reproduction is required, the response of any earphone needs to be calibrated to the unique response of the user due to the close-coupled nature of headphone use.
Greisinger uses this personalized calibration technique in hall acoustics research as a way of accurately reproducing binaural recordings for specific individuals that were made at the eardrums of an alternate listener with sufficient accuracy that particular hall acoustics phenomena can be experienced and compared. The recommended headphone type for this level of binaural reproduction accuracy is IEM / in-ear types which completely bypass the pina, or supra-aural types which effectively negate most pina effects. Corrections are necessary at both ends of the chain for this to work. Such a recording, made at the ear-drum, requires correction of the ear canal resonance specific to the person making the recording. What about that fellow's pina response which gets encoded in the recording? I'm not sure. However I do know that with this approach DG is able to sufficiently reproduce differences in sound from, say, seat 31 row N, from seat 28 row G with sufficient objective accuracy across various listeners to be useful in researching relatively arcane hall acoustic phenomena. The relative accuracy required for that application is considerably greater than what is needed for providing convincing binaural cues which might get one to exclaim, "those scissor's sound real and right here next to my ear!"
The current Neumann head includes an equalization compensation (earlier versions did not).
A Jecklin disc recording is not true binaural but a near-spaced baffled omni technique. It can produce quite good general binaural cues which are not listener specific, as can a spherical baffle of approximate head diameter, with either an acoustically reflective surface and flush mounted omnis, or an acoustically dead surface and tangentially placed omnis. These approaches are the way to go IMHO for more universal recording as they tend to work well for non-specific, non-calibrated headphone listeners in addition to loudspeaker reproduction.
I've built and used a number of Jecklin disks of various sizes and configurations and corresponded with the man himself, along with exploring other multichannel baffled techniques. They can work very well and I generally recommend them over true binaural techniques for most applications.
If interested in pina-less dummy-head recording for cross-talk cancellation reproduction check out Robin Miller's Ambiophone- https://irp-cdn.multiscreensite.com...9dfa207e8f9/files/uploaded/THE AMBIOPHONE.pdf
The above paper sourced from here: Rec Engineers Corner
We should probably continue this binaural recording conversation in an new thread. Thanks for taking it up, its a fun discussion, however my intent in posting here today was to discuss the potential applicability of the stereo phantom center techniques explored in this thread to multichannel mixing generally, and multichannel reproduction more specifically.
My apologies for the unintended threadjack.
Thanks for the clarification. Perhaps it can be tricky for some to switch out of critical listening mode, for most though, I would suggest the brain's need to make sense of the information received from the ears makes it very easy to accept the illusion without any conscious effort at all, thankfully.