Start Free Trial
← All posts
Care through a speaker

Live captions and translation for telehealth visits

The visit moved online; the interpreter and the lip-reading didn't come along. How to add a captioning and translation layer beside whatever platform the visit already uses.

60+ languages, side by side 7 min read Published July 2026

A telehealth visit strips out most of what makes clinical conversation work: room presence, clear sightlines for lip-reading, an interpreter physically beside the patient. What's left is compressed audio through consumer hardware — and for a patient with limited English proficiency or hearing loss, that is often not enough to be safe. The fix is not a better telehealth platform; it is a captioning and translation layer running beside whichever platform the visit already uses. Here is how clinicians, interpreters, and patients set that up.

Two patients, two different gaps

Telehealth access problems concentrate in two groups whose needs overlap but are not the same:

  • Patients with limited English proficiency (LEP) have a language gap. They may hear the audio perfectly and still leave the visit unsure which medication changed. What helps them most is live translation — the clinician's words in their own language, next to the original.
  • Deaf and hard-of-hearing patients have an audio-access gap. Telehealth removes the lip-reading and facial cues that carry much of an in-person conversation, and platform audio compression takes another bite. What helps them most is fast, accurate captions of exactly what was said.
  • Many patients are both. Older LEP patients with age-related hearing loss sit in the intersection — and they are heavily represented in exactly the populations telehealth was supposed to reach.

Add a third participant with the same problem: the remote medical interpreter, often joining by phone as a third leg of the call, working from the worst audio in the room.

Audio-only visits are the worst case

A meaningful share of telehealth is not video at all — it is a phone call. Patients without reliable broadband, without a camera-equipped device, or without the patience for a portal login get the audio-only visit, and it removes every fallback at once: no faces, no lips, no chat window, no screen-shared instructions. Mishearing a dosage on a phone visit has no visual channel to catch it.

The good news is that a phone visit is still just audio reaching a device, and audio reaching a device can be captioned. The general patterns — same-device system audio, or speakerphone plus a second device — are covered in our guide to captioning phone calls in real time; everything there applies directly to audio-only telehealth.

The setup: a captioning layer beside the platform

The reason this works with every telehealth platform — the big EHR-integrated ones, the boutique ones, plain Zoom — is that nothing integrates with anything. The captioning tool reads the computer's audio; the platform never knows it is there.

Open Unicaption in a browser tab beside the visit

No install, no IT ticket, no bot joining the visit. Unicaption runs in the browser (or as a desktop app) in a window next to the telehealth platform and the chart.

Choose system or tab audio as the source

This is the step that makes the quality: the tool captions the far end's audio as transmitted, not a microphone's impression of your speakers. Room noise, typing, and echo never touch the transcript.

Set languages — or let auto-detect decide

For an LEP visit, set the patient's language and yours; the original and the live translation appear side by side. Patients who code-switch mid-sentence are handled by automatic language detection.

Preload the visit's vocabulary

Medication names, conditions, and the patient's name go into the custom dictionary before the visit. Built-in medical terminology covers the rest — drug names are exactly the words generic transcription mangles most.

Who runs it, and what it solves for each seat

The same tool serves three different seats in a telehealth visit, and it matters which seat you are setting up:

WhoSetupWhat it solves
ClinicianUnicaption beside the platform, system audioSoft-spoken and accented patients become readable; nothing said gets lost while eyes are on the chart
Remote interpreterTranscript running during the interpreted visitA verification layer for dosages, spellings, and numbers — the details phone audio drops first
Patient (HoH or LEP)Browser tab on their own device, captioning the visit audioCaptions of the clinician, or live translation into their own language, moments behind the speech
Interpreted visits are not either/or. A qualified medical interpreter plus a live transcript is strictly better than either alone: the human carries meaning, culture, and clarification; the text holds the milligrams still.

The privacy bar for anything that hears a visit

Any tool that processes visit audio is handling protected health information, and it should be interrogated like any other vendor in the care pathway: Is it HIPAA compliant? Is audio stored, and for how long? Are transcripts retained? Is session content used to train models? A tool with wonderful captions and vague retention answers does not belong in a clinical call.

Unicaption's answers are the short kind: audio is never stored, transcripts are deleted when the session ends, sessions are end-to-end encrypted, and content never trains models. The full checklist for evaluating captioning tools in clinical settings — BAAs, retention, the questions to put to any vendor — is in our HIPAA captioning guide. Treat this article as practical orientation, not legal or compliance advice for your organization.

Frequently asked

Can you add live captions to a telehealth visit?

Yes, without changing platforms. Run a captioning tool beside the telehealth platform and set its audio source to system or tab audio — it captions the visit's audio directly, and nothing joins the call. Unicaption does this in the browser with no install, at state-of-the-art streaming accuracy and real-time latency, which keeps captions usable in real conversation.

How do you translate a telehealth call in real time?

Set a source and target language in Unicaption (or leave automatic detection on) and it shows the original transcription and a live translation side by side, in 60+ languages. Clinicians use it to be understood by LEP patients; patients use it to read the clinician in their own language; interpreters use it as a verification layer during interpreted visits.

Is live captioning HIPAA compliant?

It depends entirely on the tool, and 'HIPAA compliant' on a marketing page is not the test. The questions that matter: is audio stored, are transcripts retained, is content used for model training, and will the vendor state all of it in writing. Unicaption's answers: end-to-end encrypted, audio never stored, transcripts deleted when the session ends, never used for training. Verify any tool — including us — against your organization's own compliance requirements.

What about audio-only telehealth visits — can a phone visit be captioned?

Yes, and phone visits are where captions matter most, because there is no visual channel to catch a mishearing. If the call reaches a computer (softphone, VoIP), caption its system audio directly. If it lives on a phone, put the call on speaker and let a second device running Unicaption listen. Keep the devices close and the room quiet.

Do captions replace a medical interpreter?

No. For consequential clinical conversations with LEP patients, a qualified medical interpreter carries meaning, culture, and clarification in both directions — captions and machine translation don't. The strongest setup is both: the interpreter interprets, and a live transcript holds dosages, names, and numbers still for everyone to verify.

Every visit, understood

Live captions and translation beside any telehealth platform — end-to-end encrypted, nothing stored. 30 free minutes every week, no credit card.

Start Free Trial →