How to translate video calls without making everyone miserable

You have a meeting with someone who speaks a different language. It's in thirty minutes. What do you do?

Five years ago, the honest answer was: find a human interpreter or reschedule. Today there are real options, but most of them create their own set of problems. Subtitles that move too fast to read. Robot voices that strip all personality from the conversation. Awkward delays that turn a discussion into a game of wait-and-respond.

Here's what I've found actually works, and what doesn't.

Option 1: Translated captions (the obvious choice)

Zoom, Teams, and Google Meet all offer some form of translated captions. You speak English, the other person sees Italian subtitles on their screen. Seems reasonable.

In practice, it's a reading exercise disguised as a conversation. The other person is looking at text instead of looking at you. Their responses are slower because they need to finish reading before they can react. And if two people start talking at once — which happens in every meeting — the caption system chokes.

I've watched a sales call go sideways because the prospect kept looking down at captions and missing the visual cues the rep was giving. The deal didn't die because of bad translation. It died because nobody was making eye contact.

Captions work for simple, structured calls where one person talks and the other listens. A company all-hands, maybe. But anything interactive — negotiations, brainstorms, interviews — needs something better.

Option 2: Hire an interpreter

Still the gold standard for high-stakes situations. A good interpreter catches nuance, reads the room, and adapts in real time. The problem is availability and cost.

Professional interpreters charge by the hour and need advance booking. Finding one for a specific language pair on short notice ranges from difficult to impossible. And for the daily operational calls that international teams run — standup meetings, client check-ins, project reviews — the math doesn't work. You're not going to hire an interpreter for a 15-minute sync.

There's also the third-person dynamic. When an interpreter is in the room, the conversation changes. People talk to the interpreter instead of each other. The directness is gone.

Option 3: Voice translation

This is the approach that solves the biggest problems with the first two options. You speak, and the other person hears you in their language. Not reading subtitles. Hearing a voice.

Hitoo does this with an AI model called AURIS. It processes speech and produces translated speech in under 300 milliseconds. The output keeps your voice — your tone, your rhythm, your inflection — instead of replacing it with a synthetic voice.

The difference this makes is hard to appreciate until you experience it. In a caption-based call, there's always a distance between participants. In a voice-translated call, that distance shrinks dramatically. People respond faster, they interrupt naturally, they laugh at the right moments. The meeting feels like a meeting.

What actually matters when you set this up

A few things I've learned from doing this hundreds of times:

Test before the real call. Spend five minutes on a test call. Check that your microphone is picking up clearly, that the language pair works, and that you're comfortable with the pace. Don't learn the tool during the meeting that counts.

Tell the other person what's happening. A quick "I'm using real time voice translation so we can each speak our own language" at the start of the call sets expectations and prevents confusion. Most people are curious about it, not skeptical.

Speak at a normal pace. You don't need to slow down or over-enunciate. Good voice translation handles natural speech. If anything, speaking artificially slowly makes the output less natural because the AI is trained on how people actually talk.

Don't mix languages mid-sentence. Code-switching (jumping between languages) trips up most translation systems. Pick one language and stick with it for each thought.

Use a quiet room. Background noise affects translation accuracy the same way it affects any voice recognition. A decent microphone and a closed door make a bigger difference than any software setting.

When to use what

Not every call needs voice translation. If you're reviewing a document with someone and mostly pointing at things on screen, captions might be fine. If you're having a conversation that depends on rapport, persuasion, or nuanced discussion, voice translation is the right tool.

The question isn't whether to translate. It's whether the translation method lets both people show up as themselves. If the answer is yes, you're set. If the answer is "sort of, but with some friction," that friction will cost you — in longer meetings, missed signals, and relationships that don't quite develop.