that would be excellent; any thoughts on how you'll sync their respective clocks?
By calibration. Initialization offset is constant, and dilation can be measured and corrected on the fly by counting the number of samples requested by each sound card.
This is good for studio applications being able to use heavily multichannel ASIO, but passing 8+ channels to any system device is still not possible on windows. This is why I'd handle the sound cards with separate direct connections.
so...decode to multichannel PCM in PC, then HDMI out to pre/amp (or AVR) HDMI in, then to speakers? Is that what we are talking about?
You can completely bypass an AVR, but old 7.1 systems without Atmos support can now also be used as 5.1.2.
I read that Google is trying to develop a open source multi channel codec
Many others are, but none will reach the popularity of Atmos, it will stay for a long time. It's so integrated to the industry that even Dolby is unable to change AC-3 to AC-4.
Right, I mean applying e.g., room correction either in the PC downstream of decoding, or in a box between PC and speakers.
That's also a goal of the project to apply EQ on the Atmos bitstream to be able to use some higher quality calibration than what's in AVRs.
There are 2 reasons to prefer Atmos in TrueHD:
1) TrueHD is lossless, the difference when you listen in High quality HT is really big.
2) Streaming audio is Dynamic compressed to follow ITU 1770/1771 or EBU128 specs. Good for TV and soundbar, not for High quality HT listening.
So, please, don't discard Atmos in TrueHD: it's the only Atmos for HT ultimate experience.
Luckily the DRC is in the metadata, and can be easily bypassed, just throw away the values. This is not possible on an AVR, so the difference will indeed be great there, but we blind tested the DRC bypass method in a high-end HT demo room, and it passed as transparent.