Originally Posted by sdurani
Some coded soundtracks (as opposed to PCM soundtracks) use a core+extension structure. I'll use 7.1 TrueHD as an example. The core of a TrueHD soundtrack is a lossless 2-channel mix, either downmixed from the multi-channel original or a dedicated stereo mix. In any case, it contains all the information in the 7.1 soundtrack, just in fewer channels.
The first extension packet contains the data that, when combined with the core, will give you a lossless 5.1 soundtrack. Again, it contains all the information in the original 7.1 mix, but as a 5.1 version. The second extension packet contains the information that, when combined with the core and first extension packet, yields a lossless copy of the original 7.1 soundtrack. And so on, with more extension packets and more channels.
As a short-hand, you can say that the first extension packet contains 3.1 channels. But what it really contains is the specific data needed to convert the 2-channel core into a 5.1 track. As you can guess, some of the data in the stereo mix will have to be tossed or re-arranged to other channels as part of this conversion to 5.1.
Likewise, Roger used a short-hand to say that the second extension packet contains the last 2 channels of the 7.1 mix. What it really contains is the specific data needed to convert 2 surround channels back to the original 4. It's just easier to refer to these packets in terms of channels: 2-channel core, 3.1-channel first extension packet, etc. But it's really extension data, not actual channels.
This structure makes backwards compatibility easier. If your TV has 2 speakers, then the player only has to decode the core (no need to decode all 7.1 channels and downmix on-the-fly to 2.0). You'll still hear the entire soundtrack. If you have a 5.1 set-up, then your receiver only has to decode the core and first extension packet (no need to decode all 7.1 channels and downmix on-the-fly to 5.1). You'll still hear the entire soundtrack.