The Web Audio spec is more a user manual than a spec, to be fair. We are implementing the API at the moment, but the spec is so unclear that it is impossible to implement it properly. We have a couple things done and working, though.
And about "We innovate early, often", keep in mind that we had a counter-proposal to the Web Audio API, that we implemented. The W3C has chosen Google's instead of ours, and I believe that the fact we are late to implement this particular spec is quite normal.
I happen to be working on my own implementation of Web Audio in Haxe(for easy portability into Flash or outside the browser). I'm still at an early stage and admittedly am heavily guided by the Webkit implementation - while working on parameter changes I had to stop and take notes on their code to see exactly what was going on. What parts of the spec do you find most difficult?
So why did the W3C choose Google's audio proposal? Was there some kind of political wrangling going on there or were there purely technical reasons? I've wondered about that for a while.
The Google proposal was more full featured than the Mozilla Audio Data API. It provided more functionality baked in vs having to implement it in JS. This was considered important for mobile devices.
Another advantage of the Google proposal was they had a spec and implementation from the start. The Mozilla proposal had a wiki page write up and an implementation and I don't think there was as much effort put into promoting it as Google did into theirs.
Because it was the better API? Game devs all over the net have praised the Web Audio API
The Mozilla proposal required audio processing in JavaScript which in some ideal world is a great idea but in the real world where you want audio running on Intel ATOM and mobile processors and still have time to run other code is the not such a good idea.
And about "We innovate early, often", keep in mind that we had a counter-proposal to the Web Audio API, that we implemented. The W3C has chosen Google's instead of ours, and I believe that the fact we are late to implement this particular spec is quite normal.