Google: Jingle (created at Google), an extension of Jabber/XMPP for signaling, RTP for media, H.264 video, GIPS (acquired by Google) audio
Apple: SIP/STUN/TURN/BuzzwordBingo for signaling/NAT traversal, RTP for media, H.264 video, AAC audio
Is there a reason everyone can't agree on a standard?
a) Google: not-invented-here-syndrome
b) Apple: massive lock-in
c) Skype: see Apple
To be fair, I'm not aware of any standard that existed when Google first released Google Talk video/voice. It is pretty strongly standards-based, though, and uses appropriate extension points in standard protocols (Jingle on XMPP).
I spoke to Chris Sacca, who was part of the Google Talk team (if I'm not mistaken) at Startup School a couple of years ago, and he told me that the 'SIP gateways' would be coming online shortly. This must have been in 06/07'.
If memory serves well, the reasons why Google didn't go with SIP at the time had to do with federation and trust between different SIP domains. So it was mostly security and anti-SPAM related. They were meaning to do the right thing - i.e. being open - but found that SIP could not serve their needs.
RTP is a complete unusable disaster: it's incredibly overcomplicated and almost no system supports more than an arbitrary and incompatible subset of the spec, so almost all RTP systems end up being incompatible with everything else.
We will have to see how much they open things up. My guess would be that they allow others to make FaceTime clients, but keep control of the server side.
In theory, Apple could make an end-run around the telcos, since they will eventually have a huge directory of phone numbers and email addresses for all FaceTime users. At that point the carriers truly become a dumb pipe. Maybe that is Apple's plan, after all.
Google use Speex in Jingle now.
Also, Google speaks many audio codecs, including speex. ISAC (the codec from GIPS) is just the default.