Backwards compatibility in audio delivery can be somewhat contrary to the desire to preserve artistic intent. In order for backwards compatibility to exist, all of the sound must be available to not only the primary sound system, but also the secondary system. It is easy to say that a new SMPTE object-based format will be backwards compatible, but it requires careful thinking and strong guidelines to make this work in a manner that does not produce unintended results.
In the history of cinema, only a few formats have set out to actually accomplish backwards compatibility. Dolby Stereo, the matrixed format in which two channels of sound carry four channels of sound, could be played as a stereo track or a four channel track, without any loss of audio information. Not until recent years has this capability been again pursued. In Auro 11.1, also discussed in an earlier report, all of the sound targeted for an 11.1 system can also be heard as a 5.1 track. Dolby Atmos offers backup capability, which could be applied in a backwards compatible manner, although that is not Dolby’s intent. Atmos carries a full 5.1 or 7.1 mix in addition to the channels and objects that comprise the Atmos mix. The 5.1 or 7.1 mix is selected automatically by the Dolby processor should a problem occur.
To be backwards compatible in a manner that remains within the realm of artistic intent requires agreement as to which formats one is to be backwards compatible with. If sound artists are to have their say, then we can’t simply talk about being able to play the audio distribution on a system of arbitrary design. In this regard, there is agreement in the industry for the 5.1 and 7.1 formats, but above that plane of speakers, there is no agreement for speaker layouts. Both Atmos and Auro utilize different speaker layouts. This poses a problem in the general utility of a SMPTE OBA distribution, for if the format is to be successful, it must play on all defined speaker layouts.
How backwards compatibility is realized has not yet been a discussion in SMPTE or anywhere, for that matter. DTS claims an MDA mix can play on any speaker system, MDA has not been put to the test. No movies have been mixed in MDA, and it is the opinions of sound designers and sound mixers that matter more than demonstrations at trade shows. There is the example of Dolby Atmos that is completely contrary to the DTS position. Dolby surely is capable of designing a rendering engine that can fold down an Atmos mix to a 5.1 speaker system. But instead of taking that route, Dolby chose to include a discrete and separate 5.1 (or 7.1) mix in the Atmos distribution.
When choosing to carry a discrete and separate 5.1 or 7.1 mix alongside the Atmos mix, Dolby took the safe route. It was probably too difficult to get mixing engineers to agree to algorithmic mix downs. One would expect that mixers prefer to be in control of all aspects of their mixes, and not allow anything to chance. For example, a sound mixer may prefer that a sound pan over the audience’s heads in Atmos is better positioned in the rear of the room in a 7.1 mix – not the kind of mixing decision one can count on an algorithm for.
Of course, there are ways to guide an algorithmic mix in the rendering engine, such as the inclusion of mix coefficients for a particular speaker layout. But the inclusion of infinite ways of controlling infinite speaker layouts is obviously not practical.
A parallel discussion to that of backwards compatibility is the eventual elimination of the main audio track file in the DCP. If SMPTE OBA is successful, and audio rendering engines are eventually installed in all cinemas, the 16-channel main audio track file will become redundant. This may sound trite, but the digital cinema package, by definition, must include both an image track file and a main audio track file. If we are to one-day eliminate the main audio track file, it would be worth our time to plan that path today. Besides the long term impact, it will also have affect how we think about backwards compatibility.
The point made is that for all of the smoke blown about one object-based format versus another, core issues that will impact the future logistics of the industry, such as the eventual replacement of the 16-channel main audio track file and how backwards compatibility is accomplished, have not yet been seen on an agenda.