Live church translation went quiet: what to check, in order

The translation has gone quiet in the middle of a service, and the first move is not to restart anything. Look at the level of the audio the system is being given. In almost every case the sentence stopped arriving because the sound stopped arriving, and that is visible in a few seconds from wherever you are standing. Restarting throws away the only evidence you have and costs you the next two minutes of the sermon.

The reason this one is worth rehearsing is that it looks like a software fault and usually is not. From the listener's seat a dead translation and a dead microphone are the same experience: a phone with nothing coming out of it. From the desk they are almost the same too, which is why the team's instinct, to reach for the thing they can restart, is the expensive instinct.

Why not just restart it?

Because a restart is a thirty-second gap in the middle of a sermon, and it fixes the one cause that is least likely. If the source is silent, the restart ends with a silent feed again and now you have lost a paragraph. If the source is fine, you have interrupted a listener who was following along.

Three honest exceptions. Restart when the phone running the capture has gone to sleep or shows a crashed app, when a cable has clearly come out, or when you have already confirmed a healthy level and the output still has not moved for a good half minute. Everything else is a signal problem.

Is the system being given any sound at all?

This is the whole check, and it takes one look. Find the meter on whatever is feeding the translation and watch it through a full sentence of speech.

A healthy speech level moves constantly and unevenly. It climbs on stressed words, drops in the pauses between sentences, and never sits at the top. The part people miss is the quiet half: during a soft passage the meter should still be visibly moving, just lower. If the quiet passages are flat, the quiet passages are not reaching the system either, and names and verse references are exactly what gets said quietly.

A feed that is connected and carrying nothing looks different from a broken one. It is not an error. It sits at the floor, perfectly still, with the connection reported as fine, because a muted channel and an unplugged channel sound the same to software. That is the state to go looking for.

Where is the phone or the capture device standing?

If you are capturing with a phone or a small device in the room rather than from the desk, distance decides everything. Within about a metre of the preacher it hears the preacher. At three metres in a stone-walled hall it hears the hall, the air handling and the second row, and the voice you want is one more thing in a mix.

The other place that works is immediately next to a loudspeaker that is carrying the desk feed, which is a reasonable answer in a building where you cannot reach the desk. What it cannot do is sit at the back of the room on a windowsill and work.

A clip-on microphone on the preacher changes the problem rather than solving it. It is the most reliable input you can give any system, and it introduces one new failure the room does not have: a battery, a mute switch under a jacket and a transmitter that walks offstage during the songs. If you use one, the pre-service check below matters more, not less.

Is it the gap between songs rather than a fault?

Churches report a dead translation that was never dead. The band finishes, the desk pulls the speaking output down while the stage resets, the preacher walks to the lectern, and for forty or fifty seconds the system is being handed silence and faithfully producing nothing. Anyone glancing at a phone in that window sees a failure.

Learn what your own service looks like in those forty seconds, because it happens every week in the same place. If a muted send is the normal state between songs, then that gap is not information and nobody should act on it. The same applies to the walk to the lectern and to anything your team routinely mutes during communion or prayer.

What do you say to the room while you are checking?

One sentence, from the front, in the service language and then in the language that is waiting if you can: the translation has stopped and someone is looking at it, so hold on for a moment. It costs eight seconds and it keeps a dozen people from poking at their phones, closing the tab, giving up and leaving before anyone can help them.

The listener who needs the translation most is usually the one least likely to put a hand up and ask. Say it from the front and nobody has to.

The one-minute check that prevents most of this

Do this at the same point every Sunday, after the sound check and before anyone sits down.

1. Have whoever is preaching say two normal sentences into the microphone they will actually use. Not a tap, not a count: sentences.
2. Watch the meter on the translation feed during those sentences. Confirm it moves, and confirm it still moves on the quieter words.
3. Open the listener side on one phone that is not the capture phone and listen to one sentence arrive. This is the step that catches everything else.
4. Check the battery and the mute switch on any clip-on microphone, and say out loud who is responsible for it after the songs.
5. Note what the feed does while the band plays, so the gap never surprises you.

Four of those five are about the audio, which is the honest proportion. Which audio feed the translation needs covers the two ways of producing a clean one, and if you are still choosing between systems, what to test on a real sermon is the list we would hand a church before a decision.

We run four languages on a Sunday in Vienna, and the pre-service listen on a second phone is the step we added after learning this the slow way. Everything upstream can report itself healthy while the one thing that matters, a sentence arriving in someone's ear, is not happening.

In the room, Sanctavo takes that one feed and carries the service on each listener's own phone from a QR code: a natural voice that keeps the speaker's tone and rhythm, under a second behind the speaker, with captions alongside. The same recording becomes the transcript, the summary, the full message and a small-group guide afterwards, in every language the church serves, so the feed you check on Sunday morning is also what produces the week's written record. If you want to try the checks above against your own building first, record two minutes of that feed on a normal Sunday and judge any system, ours included, on that file rather than on a demo.