Depth of soundstage - controlled directivity or in-wall?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The last sentence you wrote i find interesting. Do you have some more information about that to share (reading, experience)?
Off topic, a bit, but are asking about visual stereopsis (which is always veridical, at least in principle) being over-ruled by other cues (which aren't always veridical).

Almost the entire world has seen that demonstrated in the old favourite of science museums: the distorted room.*

YouTube

While there is no stereopsis looking at the illustration on your computer screen, can you recall seeing it at a museum?

The distorted room demonstration illustrates the fudging of visual cues just as you do with sound cues for getting Carnegie Hall into your music room at home.

B.
*as a grad student, I was the keeper of the original Ames demonstrations... which included special eye glasses that changed stereopsis for purposes of demonstration
 
Last edited:
I m a big fan of Maurits Cornelis Escher's work.
That talks to me and the analogy does make sense to me (at least).

No disrespect, but never met anybody who isn't a fan!

You are right to mention Escher's staircases (and a bit differently, the tessellation's). They also illustrate putting Carnegie Hall in your music room in the sense that they clearly denote things that can't be true.

Sticking with the analogy to a soundscape, the engravings sort of look one way, but not quite... not quite..... high fidelity.

For sure, you can't use an Escher print as a blueprint to make a staircase. Nor can you use a physics textbook to make a soundscape.

B.
 

ra7

Member
Joined 2009
Paid Member
The eyes do deceive, but I've found it's not the whole story. In your new house you really do have depth and you can hear it. The visuals help, but the reality of actual depth is there and you hear it. Your fortunate to have that space. Are you getting various levels of depth?

Hey Pano! My point was that in both locations, having no front wall reflections (through the use of corner loading), the imaging is about the same, but it is just more believable in the new place because there is no wall there anymore. In the garage trick you experienced, new early reflections from a front wall were added and that will definitely cause the depth to reduce.

Yes, I do get layers of images. For example, the horns and flutes are located in the back relative to the violins in an orchestra. Even the timpani has a location and it is in the back (when well recorded, of course).
 
I'm posting to this interesting older thread with some insights into the perception of depth as it relates to the recording side of the equation. I've followed a number of discussions here for years yet don't post much as I'm more deeply involved with aspects of recording.

My experience is as a live music recordist and my in depth discussions with other recordists about the capture side of things are analogous to the discussions here at DIYAUDIO about the the reproduction side of things. My particular interest over the past 10-15 years has been developing unusual techniques designed to better capture depth and other spatial auditory cues that produce a more convincing auditory illusion of the live performance in the listener with reproduction via 2-channel and multi-channel playback.

Some background-
Historically, most amateur live music 'taper" recordists record for 2-channel stereo playback using simple 2-channel microphone techniques. These recordings range from terrible to excellent. Most are not especially good but serve as valuable documentation of live performances for aficionados. Yet when done well, some of these recordings are capable of providing a shockingly convincing "you are there" perspective, which is heavily reliant on how well the recording setup is able to translate the experience "as heard by a listener present in the performance space". For some tapers this is the most reasonable and appropriate goal, yet it represents a listening experience dissimilar from almost all commercially released music. With exceptions, for the most part there is frequently significant depth in such recordings, yet little "close perspective" other than audience reaction and audience noises which happened to be in close physical proximity to the recording position. In a way this is the opposite of the presentation of most commercial music which tends to place the listener in close intimate proximity with the performers. It may be relatively accurate to a blind listening experience at the performance, yet is generally considered undesirable. Other tapers, even those who strongly value "you are there" type fidelity with the "as heard live" experience are more strongly influenced by a lifetime of listening to commercial stereo releases - primarily the increased perception of articulation, clarity, sense of close proximity to the performers and a very clearly delineated stereo image, among other aspects. This may not be as much of an accurate analog to the live listening experience, yet is pleasing and tends to work better on lesser playback systems and for the way in which most folks consume music.

The practice is quite dissimilar to modern professional live recording practices which include or rely primarily on close spot-mics and direct feeds from each instrument. It is more analogous to, yet also differs significantly from historical "purist stereo" recording techniques where the microphone array was (is) able to be carefully placed and re-positioned as necessary in search of the optimal recording location which achieves the desired balance between pickup of direct-arriving sound and indirect-arriving reverberant-sound. That relationship is arguably the most important factor in making a good audience-perspective live performance recording, and is the aspect tied most closely to choice of recording position in the venue. It is also the primary factor affecting the overall perception of distance of the listener from the primary sources of interest, not to be confused with the perception of depth among sources and depth behind the sources within the playback soundstage. Also affected by recording position are aspects such as stereo image (which can be adjusted to some degree by the design of the microphone configuration - see Michael Williams' "Stereo Zoom" paper linked earlier in the thread), timbre (which can be adjusted to some degree via the choice of microphones and EQ), and other things, but when using just a single stereo pair of microphones, the perception of how close one is to the performers is determined for the most part by how far the microphone pair was from them when the recording was made. Playback systems affect this somewhat, but the recording is far more influential. This form of distance perception often translates to an even greater extent over headphones.

The primary factor over which the recordist has some control, unfortunately often quite limited, is the placement of the microphone pair (or multi-microphone array) within the performance venue. However, the microphones are almost always considerably farther away than they would be for a commercial recording, even when it can be arranged for them to be placed on stage. More often they can only be placed out in the audience where they end up being orders of magnitude farther away. This produces a "natural but distant sounding" recording - natural in the sense of reflecting something of the actual distance from the source. Much of my focus has been on developing ways of making this reverberant portion of the recording and the audience reaction (both direct and diffuse) more more natural sounding. Surround reproduction achieves this best, but the same techniques are applicable to 2-ch reproduction as well. In such recordings, the reverberant sound dominates and much of the efforts of the recordist focus on capturing as much of and making the most of the lower level direct-arrival sound as possible. Alas, despite popular journalistic and marketing rhetoric, there is no way of defying the laws of physics.

Because of that, if one is capable of recording more than two channels, and if the performance is PA amplified, the most attractive option for an additional channel or two is typically recording a mix directly out of the soundboard mixer, assuming one has permission to do so. Mixed with the audience perspective microphone pair this increases the direct/reverberent ratio ratio - increasing the proportion of dry "direct-arrival" sound to that of the reverberant sound, improving perception of "clarity", "presence", and "proximity". Often this signal is the same that feeds the PA reinforcement speakers (in which case things which have significant on-stage volume may be missing or underrepresented). Often it is mono, or nearly so. If it is stereo it will contain stereo elements, yet rarely any significant stereo panning as that doesn't translate over PA reproduction for a large audience. Sometimes it may be a dedicated submix. In any case, this is a dry signal (perhaps containing some artificial 'verb on some elements) consisting of a mix of the close placed microphones on stage and direct instrument feeds. It essentially represents only direct-arriving sound, or rather a sampling of close direct sound which acts as a good surrogate for the direct-arriving sound lacking in a distant microphone pair placement. Later, the recordist can blend the two in search of an appropriate subjective balance between the dry, clear, upfront and flat sounding soundboard feed and the reverberant, distant, spatially more complex signal from the primarily stereo microphone pair. Playing with that blend affects a change in clarity and detail as well as the perceived distance to the source. The soundboard feed is close sounding, flat and headphone like. The audience location microphone pair is distant sounding and dominated by room reverberance, audience reaction and ambience. An important thing one discovers upon listening to a lot of these types of recordings is that the subjective determination of what is the most "correct" mix of these aspects varies greatly from recordist to recordist! Some like it up close and dry, some like it more distant and "there" sounding. Another thing one discovers is its not as simple as just sweetening the signal from a pair of audience located by mixing in some direct soundboard feed.. although that's a good start.

Deeper aspects of depth and dimension-
The availability of portable, battery-powered, reasonably affordable multichannel recording devices over the past 15 years or so has made the "hedged bet" approach described above more common, as well as opening up alternate or additional options and possibilities. As with the commercial shift toward multi-track recording 50 years ago, those other possibilities might include recording discrete instrument and vocal feeds from the soundboard rather than just the stereo or mono feed to the PA speakers or an alternate predetermined submix, but this is not especially common (partly because it is rarely allowed, partly because it is a lot of work, during and after recording). More common is introducing additional microphone channels at the recording position. The simplest, least problematic version of that is recording multiple stereo pairs so that one can later select the best sounding pair, discarding the others. This is especially useful in comparing 2-channel stereo microphone techniques. One hears all kinds of differences between them - it is the best way to learn how strong an influence stereo microphone technique has over what is heard later. More advanced techniques combine three, four, or more microphone channels. This can be very useful, yet requires careful array design consideration so as not to cause more problems than it solves. This is where things rapidly get complex as additional channels are introduced because all of the microphones interact with each other to a greater or lesser extent. Most tapers will approach this by attempting to combine "known good" stereo pairs arrangements, each of which work well on their own, or add additional microphones to a "known good" stereo pair arrangement that they prefer. This approach rarely works optimally. One does better to start anew when designing a multi-microphone stereo array rather than attempting to build upon an existing stereo pair arrangement that works well on its own.

Exploring more complex microphone arrays, informed by discussion with location recording engineers, research of existing literature and experimentation in the field is my focus. I initially gravitated to surround arrays supporting surround playback. Perhaps not surprisingly, when done correctly, this is the most powerful technique I know of for translating the depth, dimension, immersion, and spatiality of a "you are there" teleportation time-machine-like playback experience which emulates being in the original performance space at the performance. Done well, one can easily "hear through the playback room" as if it was not present. No question this is the way to go if that experience is the goal. This is not "stuff flying around" type surround, it is recreating sufficient cues to enable a robust auditory illusion in a listener exercising a "willing suspension of disbelief". A listener can get up and move around, even turn sideways or backwards and the illusion remains robust, as it does in real life.

What I did not expect was the degree to which designing for optimal surround recording arrays also applied to improvement of the 2-channel playback experience as well. There are a few important differences where they diverge, but for the most part, I find a really well optimized multichannel recording array intended for surround also works for improved 2-channel stereo output. I don't know of other amateur live music recordists recording for surround, but a number have adopted these techniques for stereo output, specifically mentioning "increased dimension, depth, and immersion" in the playback illusion. On cannot get up and move around in the same way, but within the constraints of stereo playback, depth and dimension is greatly increased.

I realize that much of this is about providing additional useful degrees of control (which can be manipulated after recording) in situations where the recordist otherwise has little. Given the ability to place the most appropriate pair of microphones in the most appropriate position, and confirm and adjust as needed by repeated listening during setup, prior to recording, the advantages of what I describe become less important. But almost no recordings outside a few audiophile labels are made that way anymore. Its too risky and too costly.

There are a number of microphone array design aspects I consider important, but I believe a most fundamental one is appropriately recording the sound arriving from all horizontal directions across multiple channel pairs (or triplets) with sufficient differentiation such that it all can be optimally balanced in a resulting stereo output, rather than relying on the well-behaved pickup pattern of a stereo pair for translating much more than just the forward "stereo window" at which it is directed and for which is has been optimized. Recording sound arriving from all horizontal directions provides far more control over direct/reverberant ratio and other aspects across all directions, which allows for better emulation of a natural 3-dimensional human hearing experience even when constrained to using only two playback channels. I often think in terms of the overall polar sensitivity response of the entire array, and subsections of it, in addition to that of the individual microphones making it up.

Depth perception cues (and more) are most definitely captured or not by the recording technique. They can also be a product of post-production mixing techniques at times. And may also be a byproduct of certain playback systems. Just know there is far more defacto standardization of playback systems than there is of the recording techniques that may be feeding them! This is something I feel is rarely acknowledged. Of course commercial production is far more standardized, for better or worse.

In any case the entire endeavor is and always has been more about teasing the most convincing illusion than actual fidelity to the source experience (source meaning the actual performance, not the recording).
 
If you do the recording, when you hear the playback, you can picture the entire set-up perfectly and the sound is perfect. I say this without even calling on the ventriloquist effect.

Advocacy of using carefully located a pair of stereo mikes - AKA a purist approach - is no different in principle from any other kind of recording studio cooking. There is no recipe or guiding principle, you are just fishing for a location that you think cooks the sound about right.

Talk of surround systems seems to in the air these days, despite the errors of earlier efforts. Maybe better domestic surround systems will show up soon, in addition to movie theatres.

What we need is a knowledge of depth cues. Simply glomping on the ITD and ILD is only a start and not much good when a dozen other cues are in conflict with the synthetic pan-pot stereo. Those cues can't be removed from the mix like soap on a rinse cycle. They stay and corrupt the perception.

For example, you can't take a recording of a trumpet made 20 feet in the air and expect to foster the perception you are hearing it from Row H. OK, maybe you can cook that a bit, but you can't cook nothin' unless you understand the cues.

B.
 
Last edited:
@Gitbucket, no need for an apology in my humble opinion. I enjoyed the read!
Luckily I don't share the grim views of B. above. I'm having way too much fun with Stereo + some helper queues that can even make pan potted stereo a joy to listen to.
Teasing it into a convincing illusion has been the goal, playing with the variables that can make it happen.
 

Attachments

  • US3980829-2.jpg
    US3980829-2.jpg
    175.2 KB · Views: 101
If you do the recording, when you hear the playback, you can picture the entire set-up perfectly and the sound is perfect. I say this without even calling on the ventriloquist effect.

The ability to "close the loop" containing the design of both the recording and playback systems via comparison to what was heard live, applied iteratively while making adjustments over time is a useful working method for empirical improvement, fascinating, and a lot of fun. In a way, it's sort of a limited amateur emulation of the historic BBC era approach where a single organization was in control of the design of both the recording and the reproduction systems and engineers could essentially step from the live performance hall directly into the control room and to make a listening comparison with as little loss of acoustic memory as possible.

One thing that becomes challenging is identifying what aspects are "of the recording system" verses which are "of the playback system". There are multiple variables at play within the feedback loop.

I love heading home and listening the same night after making a recording while the live experience remains vivid and fresh for this reason. Doing that I notice different insights than listening the following day when listening back fresh, and listening many month later when the memory of the event remains but the acoustic memory is long gone. Each of these listening experiences are invaluable and serve to collectively inform in different ways.

You nail it though. Even if I know exactly what gear another recordist is using, how they set it up, am familiar with the room in which the recording was made, and know where the recording system was positioned, etc, the experience of listening to their recording of an event I was not present at is quite different than listening to something I was present for. Again, both experiences are valuable as is the contrasts between them. Having been there makes the illusion far richer, which is enjoyable and informative, yet at the same time it is useful to know how recordings translate to others that did not experience it live.

When able to do so, at some point after setting up my recording rig in the performance venue, before or after the performance, I like do a Alan Blumlein style "walking and talking" walk-around test, moving in a full circle around the array and actuating a dog signaling clicker while calling out the numbers of the clock. Sometimes I'll do so at a couple different distances from the array. It makes a great in-situ test-signal. The goal can be high directional imaging fidelity for source positions all the away around, but loosening that constraint for achieving the best playback portrayal of the particular orientation of primary sources in front (on stage) and mid-distant, audience surrounding at all distances including nearby, identifiable room boundaries and features (hearing the sound bouncing of the back wall of the venue, etc), and reproduction of diffuse reverberance is often the more important and desired goal. Again, all very interesting in how these things differ, and further confirmation that what it is all about is making the most appropriate choices for teasing the desired illusion. Accuracy to the desired illusion dominates over accuracy to any one specific measure.

Speaking of test-signals, I also find applause especially interesting and useful. It is similar enough in how it is generated performance to performance and venue to venue, is source-distributed throughout the space, typically surrounding the recording position at near, mid, and far distance, and consists of discrete impulses and their longer-term average into a more steady-state noise. Getting it to sound natural is especially useful when EQing things by ear, and makes for a particular confirmation of spatial performance.


Advocacy of using carefully located a pair of stereo mikes - AKA a purist approach - is no different in principle from any other kind of recording studio cooking. There is no recipe or guiding principle, you are just fishing for a location that you think cooks the sound about right.

Although similar techniques can be used, recording in this way is considerably different than studio recording, it also differs from most professional location recording. I frequently emphasize this difference on recording forums. Those differences are what drives the design of the recording arrays I use, and are what make them unusual. On a recording forum I frequent these particular techniques have been long dubbed OMT (Oddball Microphone Technique) because they are quite unusual in comparison to commonly used traditional stereo pair arrangements. They are closer akin yet still different from classical arrays such as Decca Tree and all its variants with extensions with outrigger pairs and ambiance channels.

There are certainly guiding principles and recipes. I feel too many rely on following standard recipes.. or apply the wrong recipe to the situation. The first thing I try to help other recordists with is simply selecting the most appropriate cookbook recipe. My personal pursuit has been identifying the appropriate guiding principles in pursuit of my own improved recipes.

As mentioned, positioning the recording array in a location where one achieves an appropriate ratio of direct/reverberant-arriving sound is one of the most fundamental principles. The appropriate ratio is primarily determined by recording position and great efforts are made in swaying it to some degree when the microphones are not able to be placed with total freedom. That this tends to be the rule rather than the exception is one of the primary things that makes this style of recording quite different than most others.

The Michael Williams charts and tables mentioned are another example of cookbook recipes (flexible, well considered ones) based upon guiding principles determined by empirical research. The extension of Stereo Zoom to more than two channels partly informs the design of my arrays. Williams refers to that as MMAD (Multi-Microphone Array Design). Taking some guiding points from that and combining them with other approaches - some designed from well understood acoustic principles (such as OCT, Opimum Cardioid Triangle), some more by ear, experience and tradition (Decca Tree, wide-spaced omnis) - is part technical skill, part art.. informed by science to various degrees.
 
wesayso said:
..no need for an apology in my humble opinion. I enjoyed the read!
Thanks, I've very much enjoyed following your posts over the years, including the build and subsequent development of your line-array towers and all the interesting thinking surrounding them. You are an inspiration and always a good read!

I've got a longtime back-burner project going which I may start a thread at DIYAUDIO about at some point- development of a portable playback system for these recordings which I can relatively quickly deploy outdoors in a backyard or wherever as a way to share the experience with others. Like the recording arrays, this is another unusual and interesting project which differs significantly from most else in significant ways. It will be based around a pop-up shade canopy frame providing speaker array geometry, prewired for attaching 8 or 9 satellite speakers + stereo low-frequency / subs. The listening area will be near-field under the canopy- no wall reflections, but no room reinforcement to assist in ambient immersion and LF. I have a 10' diameter hexagonal 6-leg canopy I intend to use initially, but may end up adapting parts from other folding canopies to build a 7, 8 or 9 leg polygon version. The initial satellites I intend to use all fit into a single large Pelican-case. Another will house the amplifiers and control unit (perhaps MiniDSP DDRC-88A, unless I can press a surround receiver into appropriate service) The LF reinforcement/subs will likely be bucket subs placed on the ground at the base of the Left and Rightmost legs. Fun stuff.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.