With Web Audio, you can't ask for mic permissions without listening indefinitely, using up battery. You can't process audio offline with the exact same API, you have to send a 0 gain signal to the speakers and do it slowly, real-time. Want complex-valued FFTs for adaptive frequency domain filters? Sorry, all you get is a toy magnitude spectrum with shitty exponential smoothing. This isn't envelope pushing in the slightest, and I don't understand why people are impressed by this. Except of course that web developers are easily amused.
In all this HTML5 malarkey, WebGL is the only API that managed to avoid "not invented here" syndrome, and even that is years behind what is considered normal on mobile, with essential extensions still waiting for approval.
Are we supposed to be impressed by the audio capabilities of 15-20 years ago simply because now you get to do them in single-threaded JavaScript, with annoying black box async APIs, shitty number types and no serious high resolution timing control? Go build a web audio API based audio player with real-time effects that can play skip-free for hours on a busy mobile device. Then I'll be impressed.