Gaming voice products figured out the delay problem long ago (eg if you scream "flank" your team needs to respond ASAP). Video conference products are woefully behind.
I think gaming voice chat usually has a keyboard trigger. They solve echo cancellation by assuming microphones are usually off! Modern video conference software assumes users aren't using headphones, so they need sophisticated echo cancellation. Now, whether the delay is a side effect of noise cancellation, or one of the mechanisms is not clear to me.
Gaming voice chat (discord/teamspeak/other, not in-game) solved it way better than just push-to-talk. Voice activity detection, adjustable voice gate thresholds (open and close), echo cancellation, background noise removal, individual person volume settings, adjustable application attenuation.
From my perspective, Zoom's one and only benefit is how easy it is to share a link and get into a video chat. Everything else about it is painful to use.
