One of the lesser known concepts proposed for immersive sound is that of the “reference renderer.” At first blush, it sounds like a blueprint that would allow anyone to build an immersive sound rendering engine, such as Atmos. That would be an extremely unlikely outcome, however. A renderer is complex. The quality of a rendering engine is a competitive matter. Unless source code is shared, its unlikely that we’ll see off-brand knock-offs of Dolby Atmos that perform identically to the branded product. So what, then, is the purpose of a reference renderer?
The reference renderer has been associated with the concepts of assuring interoperability and preserving creative intent. Let’s look at those concepts. The Merriam-Webster dictionary defines interoperability as the
“ability of a system to work with or use the parts or equipment of another system.”
Interoperability describes functional operation. Signals can be defined, for example, such that one make of equipment can generate a signal, and another receive the signal and do useful things with it. One can take the output of a Blu-ray player, for example, and plug it into a display unit and observe picture. Functional operation does not imply quality. The display unit connected to the Blu-ray player may look terrible. But there would be no question that it’s interoperable with the player. Functional operation can be described in documents. The signal that allows the Blu-ray player to drive the display is documented.
Unlike interoperability, creative intent is not functional at all. It is a purely qualitative concept. One looks at the display connected to the Blu-ray player, and knows if creative intent has been preserved, or if this display unit belongs in the dumpster, or possibly good enough for some other purpose. You know it when you see it. It cannot be documented. We can measure lots of parameters to improve the chances of the display to be hailed as one that preserves creative intent, but ultimately, it is a subjective determination that says that creative intent has been preserved.
Digital cinema already has a parallel concept of reference device in the recommended practice of SMPTE RP431-2 Reference Projector. The Recommended Practice defines the performance factors to be measured, including contrast ratio, luminance, and color accuracy. For color accuracy, 12 colors are defined for measurement. That’s 12 colors out of a theoretical 68,719,476,736 colors possible. Even if only a quarter of those colors are used for the DCI P3 color gamut, that’s a lot more colors than 12. While the measurement of 12 colors does not guarantee that the projector will meet creative intent, it does say something about functional performance and interoperability.
From these examples, we can see that it is possible for specifications to identify the means to interoperability, and it is possible for specifications to define how to measure the parameters that lead to quality. But it is not possible for specifications to define quality. There is more to the goodness of a device than can be described in a specification.
When translating this understanding to the reference renderer in immersive sound, it follows that the role of the reference device is to insure certain parameters of functional performance for interoperability. But it is not possible to define a specification that dictates that if one follows certain steps, creative intent is guaranteed to be met.
Unlike a reference projector, which produces light, a reference renderer can only produce electrical signals. It does not produce sound. In some ways this is good, because we don’t have a standardized immersive speaker system to connect the reference renderer to. (There are no standards for how to place speakers in immersive sound systems.) With a reference renderer, certain signals are applied to the device, which then produce certain signals at the output.
To accomplish this, however, one needs to define the configuration of the virtual speaker system, and drive the renderer with signals intended for that system. This point alone is likely to cause a boatload of arguments about the configuration of the speaker layout employed. Will an Atmos speaker layout be used? An Auro speaker layout? Any ol’ speaker layout, as DTS-X might indicate?
As with the many colors possible from the reference projector, there are many points in space where a rendering engine can position a sound. In addition, there are many dimensions to that sound, such as its spread. One cannot define performance for all of the possible permutations of spatial position. As with the specified colors of the reference projector, it may be more agreeable to participants to settle on 12 flavors of sound rendered for a mythical speaker layout. This appears to be a reasonable path for a functional specification.
Where things could get dicey, however, is when and if a proponent attempts to define an input signal that was intended for a speaker layout other than the target layout of the specification. This is the point where creative intent would be crossed: an attempt to use the reference renderer specification to define how sound intended for one speaker layout is folded into another speaker layout. While some automation may be acceptable, the mix on a specific speaker system is ultimately the work of a creative artist, and not a machine.
The image analogy to the problem of rendering sound to alternative speaker layouts is that of mapping an image created in the DCI P3 color gamut to the Rec 709 color space used in home entertainment. Again, while some automation may be acceptable, the final color grade ultimately requires the work of an artist to preserve creative intent. In practice, an artist oversees the re-mapping of color from one gamut to another, providing the adjustments needed to preserve creative intent. Likewise, it takes an artist to insure that the creative intent of sound mixed for one speaker layout is preserved when played on a different speaker layout.
The reference renderer, therefore, must be a simple specification. As with the reference projector, it can only define a very limited number of cases for measurement. It can provide basic functional guidance to the manufacturer, and it can assist with interoperability. But it cannot guarantee the preservation of creative intent. That is the role of an artist, not a standards committee.