Prefer close, consistent speech
Headset, lavalier, lectern, or handheld microphones usually provide a more useful signal than a distant room microphone. Placement and gain still need to be tested with every speaker.
Live-caption audio setup
Microphone Setup for More Reliable Live Captions
Live captions begin with the audio you send. Pockitle listens to one spoken source language and can add up to three translations, but no software can restore words that never reach the input clearly.
A stronger live-caption input
- A post-fader or dedicated dialogue feed containing the microphones people will actually use.
- Consistent close microphones for moderators, performers, presenters, and audience questions.
- A preflight using the real browser, device, cables, room, and speaking positions.
- One spoken source language with the selected original and translated output lanes tested together.
Where this may not fit
- Assuming a distant room microphone will isolate every voice in a reflective or noisy venue.
- Sending music, applause, playback, and effects at the same level as speech when a dialogue-only feed is available.
- Changing input devices or the spoken source language after listening begins.
- Treating a successful input test as a guarantee of caption or translation accuracy.
Input before interface
A controlled dialogue feed is the safest starting point
Use the cleanest speech-focused feed the venue can provide. The exact mixer routing depends on the production, but the caption input should follow the people speaking, avoid unnecessary programme audio, and remain stable throughout the run.
Audio preflight
- 1List every person and microphone that must reach the caption feed.
- 2Choose a stable dialogue-focused mixer or microphone output.
- 3Select that exact browser input and the event's one spoken language.
- 4Run the input test with representative speech and room noise.
- 5Check the original captions and every selected translation before doors open.
Build one deliberate feed
A mixer can combine the microphones needed for the event while keeping unrelated music and effects lower or out of the caption feed. Confirm the route with the person responsible for sound.
Test the difficult moments
Do not test only one clear sentence. Include quiet speech, fast handovers, names, numbers, audience questions, applause, and the loudest expected room conditions.
Related planning
Read before choosing a workflow
Common questions
Should I use a room microphone or a mixer feed?
Use the clearest stable speech-focused feed available. A controlled mixer feed is generally safer when it can include every required microphone without unnecessary programme audio.
Does an input test guarantee accurate captions?
No. It confirms that the selected device is providing a usable signal. Speech, names, overlap, noise, connectivity, and translation can still affect the result.
Can I change the microphone during a run?
Treat the selected input as part of the run setup. Pause and follow the documented recovery path rather than switching devices without testing.
What should the sound check include?
Test every speaking position, quiet and fast speech, handovers, names, numbers, audience questions, overlap, applause, and the selected translation lanes.