Doctor's appointments, front desks, family dinners, site visits. The microphone setup that works, the placement that ruins it, and translation for two-language talks.
Every video platform now has a caption button. The physical world does not. The conversations that carry the most weight — the doctor explaining a dosage, the pharmacist at the counter, the contractor walking the site, the family dinner where everyone talks at once — happen in rooms, with no software in the loop. Captioning an in-person conversation means one thing: a device with a microphone, placed where it can actually hear. The setup takes a minute. The placement is where people go wrong, so this guide is mostly about that.
It runs in the browser — no install, which matters when you're setting up in a waiting room. Desktop apps for macOS and Windows work the same way.
For a room conversation, the mic is the source — it hears everyone present. (System and tab audio are for captioning calls and meetings; for face-to-face talk, the microphone is the whole game.)
Flat on the table, roughly centered between you and the other person. This one action determines most of your caption quality — more below.
Pick the language you expect, or leave automatic detection on. Add a target language if you want live translation running beside the original.
That's the whole procedure. Captions appear moments behind the speaker, and you glance at the screen the way you'd glance at subtitles — the conversation keeps its rhythm.
A transcription engine can only be as good as the audio reaching it, and with a microphone the audio is governed by distance, obstruction, and competing noise. The failure people report most — "it worked at home but not at the appointment" — is almost always placement, not the engine.
The same table setup handles a conversation that crosses languages — a parent who speaks Spanish at an English-speaking clinic, a vendor negotiation drifting between Mandarin and English. The transcript shows the original with a live translation beside it, so both people can read the line the other one meant.
Automatic language detection carries the code-switching: when a speaker changes language mid-conversation, the captions follow without anyone touching a setting. And with the device on the table between you, the screen becomes a shared surface — both parties glance at the same two columns, which quietly turns the tool from an aid for one person into common ground for two.
Unicaption is built for exactly this off-platform work — conversations no meeting software will ever host:
The free tier gives you 30 minutes every week, no credit card — enough to test-drive one real appointment and see the transcript quality for yourself.
A phone appearing on the table mid-conversation reads, to most people, as recording. So say what it is before it appears: "I use live captions to follow conversation — this shows me what you're saying as text, and it doesn't record anything." One sentence, delivered while you set the device down.
In practice the reaction is usually curiosity, and often accommodation — people face the device, speak one at a time, repeat things without being asked. The sentence is honest, too: a captioning tool that stores no audio and deletes transcripts at session end is not a recording device, and being able to say so plainly is precisely why the no-storage standard matters. In clinical settings covered by privacy rules, that standard is table stakes — see our HIPAA captioning guide for what to verify.
Open a captioning tool like Unicaption in the browser on a laptop or phone, choose the microphone as the audio source, and lay the device flat on the table between you and the other person. Captions appear moments behind the speaker. Keep the device out of pockets and bags, within a few feet of whoever is talking, and test for thirty seconds before the conversation that matters.
Yes. Bring a phone or laptop, open Unicaption in the browser, select the microphone source, and place the device on the desk between you and the clinician. Add medication names to the custom dictionary beforehand if you know them. Unicaption is HIPAA compliant, audio is never stored, and transcripts are deleted when the session ends — worth mentioning to the clinician when you set the device down.
Fabric over the microphone muffles the high-frequency consonants that distinguish words, so the engine receives smeared audio and produces guesswork. Placement is the fix, not settings: flat on a table between speakers, or hand-held at chest height while walking. Distance and background noise matter for the same reason — a transcription engine can only transcribe what actually reaches the microphone.
Yes. Unicaption shows the original transcription and a live translation side by side, in 60+ languages, with automatic detection when speakers switch languages mid-conversation. Put the device on the table between both people and the screen becomes a shared surface — each person reads the other's words in their own language, in real time.
Appointments, counters, dinners, site visits — live captions and translation from the device in your bag. 30 free minutes every week, no credit card.
Start Free Trial →