View Single Post
  #54 (permalink)  
Old 1st August, 2004, 09:55 PM
Aedan Aedan is offline
Chief Systems Administrator
Join Date: September 2001
Location: Europe
Posts: 13,075

Dolby Surround is a two channel matrix encoded sound. The rear channel is encoded by adding an inphase signal to the right channel, and a 180 degree out of phase signal to the left channel. As this is simply a two channel system, no compression is required to transmit it across an SPDIF link. Re-encoding Dolby Surround as AC3 leads to a loss in quality, as the phaseing between channels is highly critical. Lossy compression techniques do not attempt to keep phase coherence very well, which leads to poor performance once the Dolby Surround signal is decoded. Phase coherence is critical in Dolby Surround! Instead, taking the source material and re-encoding it into AC3 leads to a situation where better channel seperation is in place, and no signal steering logic is required to 'anchor' signals to a channel. So, in a toss up between AC3 encoding 4 channels, and Dolby Surround encoding 4 channels, AC3 will win each time around.

The AC3 specification clearly defines both the packet structure and the compression used. The compression technique used is based around psychoacoustic analysis, and works by throwing away parts of the sound that the encoder believes humans will not be able to hear. That's exactly the same technique used by MP3, OGG, ATRAC, DCC, MPEG2 and a whole host of others. That's nothing new, as psychoacoustics has been an expanding field for a good number of years now.

AC3 is definately not a 'container format'. That is, an AC3 stream can only handle AC3 encoding. It cannot handle DTS, MPEG2 or any other varient. Re-encoding DTS, MPEG2 or other compressed source material leads to a loss in sound quality, as the different psychoaccoustic engines regard different things are 'important'. This loss is noticable on high end systems.

Dolby released the details on their encoding system, as they want people to build encoders. Dolby receive royalties for every AC3 encoder built, and for every AC3 decoder built. When your revenue generation stream is built around royalties, you work hard to ensure that everyone knows how your system works.

DTS is 'Digital Theater System' and is the preferred format for use in cinema. Like AC3, it is also a lossy compression system, and transmits the compressed audio. Therefore, DTS cannot be AC3, any more than AC3 can be DTS.

MPEG2 is another commonly used multichannel standard. For digital terrestial, digital satellite and digital cable broadcast, MPEG2 audio is a far more common format than AC3 is.

Encoding a monophonic or stereo track into AC3 would be technically possible, but totally pointless. PCM audio across SPDIF can already handle two channels of uncompressed audio. Encoding those into AC3 merely results in compression and loss of material, that could otherwise have been sent across the link as is, as it fits within the bandwidth budget of the link.
Any views, thoughts and opinions are entirely my own. They don't necessarily represent those of my employer (BlackBerry).
Reply With Quote