[QUOTE=audiovideoholic;47243241]Yea they are slotted as well with absorption behind all of them. It's just KYD's preferred method of treatments.
Sorry for the delay in seeing and responding to this. I think audiovideoholic’s comment is well intended, but left by itself may lead to a distorted view of our actual process. To clarify, my "preferred method" is, well, a method rather than a product. I'm not sure how much discussion acoustic modeling has had on the forum, so some of you may need to bear with me here...
We bring our 3D CAD model into the engineering workstation, including the room’s…
- Shape and dimensions,
- Internal architectural features (seating platforms, columns, soffits, baffle wall, stage, doors etc.),
- Seating elements,
- Finishes (carpet, wall and ceiling fabric panels etc.,),
- The makeup of its envelope (important because the mass and stiffness of the walls, ceiling etc. affect low-frequency behavior),
- The intended Interior Acoustic Treatment Plan (what absorbers, diffusers, reflectors and hybrids at what locations and aiming angles), and
- Virtual speakers at our intended locations and aiming angles. (It's critically important that these speakers have the same radiation patterns as the "real" ones; since <1% of all speaker manufacturers have usable data to offer, we generally have to ship the speakers to well-equipped third-party acoustic labs.)
We set up virtual speakers in the model to flood the room with sound rays (one million per speaker). Each one of these rays will likely have some fraction of its energy ricochet off hard surfaces, some absorbed, and some scattered/diffused. To envision these 3 basic mechanisms – reflection, absorption and diffusion – you might envision how, in the first case, a pool ball bounces off a side cushion, where the angle of incidence equals the angle of reflection (Snell’s Law). In the second case (absorption) the pool ball disappears into a corner pocket, i.e., its energy has been “removed from play.” In the last case (diffusion) the pool ball hits the side cushion and explodes into a little poof of sawdust sprayed in many directions.
A ray's lower-midrange energy might ricochet per Snell's law, the treble range might get absorbed, and the sensitive upper-midrange might be partially diffused and partially absorbed (the latter a common but unintended of not knowing how far the “acoustically transparent” fabric [say, Guilford FR-701] needs to be spaced away from the front of the diffuser).
The part that ricochets continues until it interacts with another surface, where it’s modified further, and so on. Each of the million rays will likely experience many wall and object hits before becoming perceptually inconsequential. Modeling even a single hit can be a complex affair: A single ray impinging on a diffuser can generate 20 smaller rays heading off in various directions depending on the diffuser’s specific properties, driven by its construction and the math behind it. The software tracks the 20 re-radiated rays and how each evolves spectrally, spatially and temporally.
At some point some of these rays pass through the space that will be filled with audience noggins so we position virtual microphones at those locations to “catch” the rays, allowing us to look at what’s going on at each head position – e.g. what frequencies are showing up, from which directions, at what levels, at what points in time relative to the direct sound. The data is organized and processed to show us basic energy ratios that are known to map well to how humans perceive sound. When we tune the acoustic treatment plan (or, say, the changing the speaker’s toe-in angle or swapping it out for a model with a rad-pat that’s friendlier to the basic playback geometry we’re dealing with), we take note of what happens to the energy ratios and the perceptual metrics (clarity, definition, speech intelligibility, envelopment, etc.) that go along with them.
For frequencies 4 times the Schroeder frequency and below (about 500Hz down to 5Hz in Rob’s case), we switch from ray- to wave-based techniques, i.e. finite element analysis (FEA). There are several octaves where the two different modeling strategies (ray vs. wave) overlap, and it takes some thinking and experience to make an accurate assessment of how to “merge” them to reflect what you’ll actually hear and measure once the room you’re designing is built, furnished, equipped and treated. Why spend so much time and money on the 125, 250 and 500Hz octave bands where the 2 very different techniques overlap? Because they’re the ones with the highest speech energy for adult speakers, male and female. (For reference, 250Hz is around the middle key on a piano keyboard.) If you can’t get the speech range clean and tight, you force the guy in the money seat to fumble around in the dark for the remote, hit the rewind button and replay the passage a couple of times trying to figure out what one of the five guys talking in the bar said he’s going to do to whom. The interruption typically takes everyone out of the movie.
The upshot is that my engineering team and I choose acoustic treatments based on the objective and subjective performance metrics I'm looking for at key listening positions. It's not about "liking" or "preferring" a brand or a model. We’re brand-agnostic. (In fact, in many cases, the modeling reveals that no company offers the treatments that will meet the requirements in a certain location. In such cases we may model, optimize, fabricate and test something that will.)
The intent of KYD’s modeling programs is not just to generate and compare reams of acoustical statistics on room layout possibilities A, B and C, but to use that computational power and the experience of my engineers to allow us to transcend the numbers and actually listen to hi-res binaural simulations of the total system – room, speakers, acoustic treatments and playback geometry – before we finalize and ship all those drawings.
Once that’s done, there are two Moments of Truth still to come: 1.) How compelling and moving the system is on opening night, when we’ve finished our final calibration and are presenting it to the homeowner; and 2.) How well our measurement curves etc. correspond to the curves we’d predicted months (or even, as in Rob’s case, years) before.