Having spent 5 mins debugging it... it looks like the audio and video buffers are occasionally not exactly the same lengths, and it causes a glitch in the audio whenever the audio buffer is slightly longer.

It doesn't seem to happen when not CPU capped though, which I haven't quite figured out...

I don't know if it's what they mean but having your fans go absolutely nuts in every meeting can't be great for sound quality.

The video encoding/decoding could have leveraged hardware accelerated codecs and freed up resources for audio encoding/decoding

Audio encoding is not of particularly high complexity. Even the highest complexity modes of the best codec right now (Opus) are like 60-100x real time on commodity hardware, and not too far off on phones.