I think it's possible if the two devices (1) are close to one another (2) have the volume turned up (3) use a "regular" speaker + microphone (4) the background isn't very noisy

I doubt it is feasible via walkie talkies. I assume their range of transmissible sounds doesn't include much outside the range of human hearing, plus they transmit lots of static, interference, and environmental sounds which would probably ruin the effect. It's probably not even feasible in a setting like paying for coffee at a starbucks due to too many variables listed above. Humans are bad at following procedures precisely unless they are conscious of the requisite conditions and why they are needed, and I think #1 and #2 would trip normal people up.

E.g. I suspect people would try to pay with headphones plugged in and then not understand why it's not working. Even without headphones, some users would probably turn the volume down if the payment app turned it up without them understanding why. And you can't rely on "just telling them why" because people don't read what their apps tell them, they just complain when they don't work the way they wish things worked.