- Do we both need to look at one screen?
-
Yes: one Mac, one caption box, two paragraphs. Turn the Mac a little, or hold ⇧ and drag the box to the edge nearest the other person, and take the size up from Text Size and Alignment. On a call the app captions the other side; in the room, both of you.
- Do I have to tell it which language each of us speaks?
-
No. The recogniser works out which of the sixteen is being spoken, sentence by sentence. You choose the one language to translate into; anything else that is said arrives in it, and what is said in it goes back into the last other language it heard. Until it has heard a second language there is nothing to translate back into, and a sentence in the target language stays as it was said.
- Does it record us?
-
No. Nothing is written to disk and nothing leaves the Mac; the ⌥ stack keeps the last few boxes in memory until you quit. macOS shows its orange microphone indicator in the menu bar the whole time the app is listening, as it does for any app.
- What does the other person need?
-
Nothing. No app, no account, no phone on the table. They read the box on your screen, in their language, and hear you in yours.
- Can it do the room and a call at the same time?
-
No. Listen To is one source: the microphone or what the Mac plays. Choosing the microphone again puts it away, and capture goes back to the source it replaced.
- How good is the translation?
-
Apple's own, the one in the Translate app, running on the Mac. Good enough to follow and to answer, not a certified translation, and a sentence that arrives oddly is worth asking about. Names are where it slips most, in both languages.