Auro-3D: how does it work?25 January 2011
The Auro-3D format promises with-height surround in 9.1 channels, embedded in a standard 5.1-channel PCM mix. It doesn’t use conventional psychoacoustic or lossless coding approaches, so how does it work? A recent US patent application, USPTO 20100027819, provides some clues. It has its roots in an earlier European Patent, EP1592008 from 2005, invented by Guido Van den Berghe, one of the engineers on the Auro-3D team. There are various embodiments of the invention, but essentially it seems to be based on a method of combining samples from two or more audio channels in such a way that they can be unmixed. It’s not the same as conventional matrix down-mixing, but incorporates a form of information reduction of the original channels (involving the formation of data subsets and equating of adjacent samples). Corresponding samples of the resulting channel data to be combined are added to create a third channel that can be stored or transmitted. By embedding appropriate seed samples of the original data in the mixed signal to use as a starting point for reconstruction, the channels can be unravelled upon replay by means of a mathematical algorithm. Sample interpolation can be used at the decoding side to restore an acceptable version of the original signals. (In the original patent, interpolation of either the odd or even samples was done on the encoding side, leaving the intervening samples intact.) This is not an entirely lossless process, however an option exists to store an error approximation in some of the least significant bits of a 24-bit sample word. This can be used to improve the quality of the reconstruction process. The quality of the reconstructed audio channels is almost certain to be dependent on the original sampling frequency, because some loss of audio bandwidth is inevitable as a result of the information reduction process. The Auro-3D format offers an additional layer of loudspeakers for height information, elevated 30 degrees above the main channels. According to the patent application, the with-height information can be embedded into the 5.1 channel mix in a perceptually compatible way. The height channels can be combined in attenuated form with their nearest primary mix channels, being subsequently extracted and amplified in order to restore them to their correct level. In a similar way an artistic stereo mix can be compatibly embedded into a 5.1-channel recording. Mixes incorporating these alternative embedded versions can be played back satisfactorily on devices without the appropriate decoder mainly, it would seem, because the embedded signals can be sufficiently attenuated to be below the threshold where they are likely to cause perceptual disturbance to the primary mix signals. If played back through a decoder the embedded channels can be extracted and rendered separately. One key motivation for the new encoding format is clearly the limited number of channels available on even the most advanced DVD and Blu-Ray disks, which makes it difficult to deliver the number of channels required for with-height mixes and artistic stereo versions alongside conventional 5.1. Auro 3D uses the available channels to deliver a version compatible with replay on a conventional system. An alternative technology, if one is mainly interested in maximising the channel capacity of delivery media, is Dolby TrueHD, which uses lossless encoding based on Meridian’s MLP to deliver up to 14 channels.