Live captions fail for a boring reason: the tool is listening to the wrong thing.
Your headset works. The caller hears you. The agency portal shows an active call. But the caption tool hears silence because it has access to your microphone, not the call audio. Or it captions your voice and misses the provider.
OPI interpreters need a clean way to share call audio with the caption or transcription tool while keeping their normal call setup intact.
For consumer call-recording context, the Reporters Committee maintains a practical recording guide.
Captions cannot help if the tool hears only half the call.
Know the two audio paths
Most interpreting setups have two paths:
TIP
Test audio routing with a harmless sample before a shift. If the caption tool cannot hear both parties, fix that before the real call starts.
- Your microphone, which sends your voice into the call
- System or call audio, which carries the voices you hear
Live captions need the audio you hear. If the tool listens only to your microphone, it may caption your interpretation rather than the source speaker. That can confuse your workflow and expose the wrong text on screen.
Before you blame the caption tool, check the input. Make sure it can hear the agency portal, softphone, browser tab, meeting app, or routed phone audio.
Browser-based calls
Browser-based OPI portals are often the easiest place to start.
If your call runs in Chrome or Edge, a tool may capture tab audio or system audio through screen sharing. In Interpreter, you can start a session, choose the correct audio source, and keep the transcript visible while you work.
Use this checklist:
- Open the agency portal in the browser
- Confirm you can hear the caller through your headset
- Start your transcription or caption session
- Select the tab, window, or screen that includes audio
- Watch for movement in the transcript before the live call gets complex
If the browser asks whether to share audio, enable it. If you share a window that does not include audio, the captions may stay blank.
Softphones and desktop apps
Softphones can be trickier because the audio may never touch the browser.
Apps like desktop phone clients, call center tools, or remote desktops route sound through the operating system. A browser caption feature may not hear them unless you share full system audio or route the audio into a capture source.
Start with the simple path. Use your transcription tool’s system-audio capture option if it has one. Then test with a non-sensitive call or a test audio source before work.
If the tool can capture only browser audio, move the call into a browser version of the softphone if your agency allows it. Do not install drivers, virtual cables, or routing tools on an agency computer without permission.
For a deeper setup guide, read Audio Routing for OPI Interpreters.
Video meeting calls
Zoom, Teams, Meet, and Webex each handle captions in their own way.
Built-in meeting captions may work for meeting audio, but they usually do not give interpreters the full OPI workspace: speaker-aware transcript, two-way translation, quick lookup, notes, and custom terms.
If you use a separate interpreter caption tool, make sure it hears the meeting audio. On many systems, that means sharing system audio or selecting the meeting app as the source.
Also check policy. Some meetings restrict recording, transcripts, or external tools. Captioning during a meeting can still raise approval questions even when you do not save the output.
For platform-specific limits, see Zoom captions for interpreters, Microsoft Teams captions, and Webex captions.
Phone line through a handset
If you interpret through a physical phone that never connects to your computer, live captions cannot hear the call unless you route the audio.
Common options include:
- A headset base connected to both phone and computer
- A softphone instead of a desk phone
- A compliant audio interface approved by your agency
- A speakerphone in a private room, if policy allows and audio quality holds
Avoid improvised setups on sensitive calls. Do not put protected health information on a smart speaker, personal recorder, or unapproved device. If the setup feels like a privacy workaround, stop and ask.
Test before a real call
Use a simple test:
- Play sample speech through the same app or device that carries your calls.
- Start the caption or transcription tool.
- Confirm the tool captions that audio, not your microphone.
- Switch speaker sides if your setup has separate channels.
- End the session and confirm nothing stayed open.
The test should take a few minutes. It can save you from discovering mid-call that your transcript has captured only your own voice.
Privacy and permission
Sharing call audio with a tool can count as sending sensitive content to another service. Check agency and client policy before using live captions on professional calls.
For medical calls, use a HIPAA-approved workflow. For legal calls, follow the court, client, or agency rule. For no-recording workflows, confirm whether the tool stores audio, text, notes, or logs. No-recording interpreter notes and privacy covers the practical checklist.
The simple rule
Live captions need both the audio you hear and the audio you speak.
Pick the source with care. Test it before the shift. Keep the setup approved, private, and simple enough that you can operate it while interpreting.
Related reading: