Ask around and you will hear that live translation needs a small fortune in gear: transmitters, receivers, booths, headsets, a technician to run it all. That was true once, and for a long time it was the only answer available. Today the honest list is much shorter — and the money that remains should nearly all go on one thing.
What you genuinely need
- A device you already own. A laptop, tablet or phone with a modern browser is enough to run the broadcast. It does not need to be fast or new.
- Clean audio of the speaker. This is the one place worth care and, if necessary, money.
- Wi-Fi or mobile data for the broadcasting device, and enough signal in the room for attendees' phones.
That is the whole list. Everything else is optional, and most of it is unnecessary.
Audio capture is the entire game
If you take one thing from this article, take this: the quality of your translation is decided almost entirely before any software is involved. Everything downstream — transcription, translation, subtitles, read-aloud — is working from whatever the microphone heard. Good audio makes an ordinary system look excellent. Poor audio makes an excellent system look broken.
The size of the difference surprises people. On our own recordings, a feed taken from the sound desk ran several times the signal level of a device placed across the room from the same speaker — and the gap matters far more than the ratio suggests, because of what happens at the bottom end.
The failure mode nobody expects
Intuition says quiet audio produces gaps: missing words, blank lines, obvious failure. It does not. Below a certain level, speech recognition produces confident, fluent, grammatical sentences that nobody said. It fills in.
This is worth dwelling on, because it changes how you test. A system that fails obviously is annoying but safe — everyone can see it. A system that fails fluently is dangerous, because the congregation reading in Ukrainian has no way to know that the paragraph in front of them was invented. They will assume the preacher said it.
So the goal is not "audible". The goal is a signal strong enough that the engine never has to guess. That is a hardware problem, and it is solvable for very little money.
Three ways to get the audio, in order of preference
| Method | Quality | Cost | Notes |
|---|---|---|---|
| Feed from the sound desk | Best | Small | Needs a cable and usually a simple USB interface. The speaker is already mic'd; you are just borrowing the signal. |
| Dedicated microphone on the speaker | Good | Moderate | A headset or lapel microphone into the broadcasting device. Best where there is no mixer. |
| Laptop microphone in the room | Poor to unusable | None | Fine for a test, unwise for a service. This is where invented sentences come from. |
The middle row is the honest answer for most small churches and every home group. The top row is the answer for anyone who already has a sound desk, which is most churches with more than about fifty people.
"It sounds fine to me" is not the test
Human hearing is extraordinary at reconstructing speech from a poor signal. You know the preacher, you know the topic, you know the vocabulary, and your brain fills the gaps without telling you it did. A speech engine has none of that context.
So do not judge the feed by listening to the room. Judge it by reading the transcript of five minutes of real preaching. If the text is accurate, the audio is fine. If it is fluent but wrong in places, the audio is not fine — no matter how clear it sounded standing at the back.
Three budgets
- Nothing at all. Use a laptop with the speaker wearing an existing radio microphone, and place the laptop near the sound desk output if there is a headphone socket. Many churches can borrow a usable feed from a headphone or monitor output with a cable that costs less than a coffee.
- The one upgrade worth making. A simple USB audio interface that takes a line feed from your mixer. This is the single highest-value purchase available and it is not expensive. It removes room echo, congregation noise and distance in one step.
- If there is no mixer at all. A dedicated USB headset or lapel microphone for the speaker. Less elegant than a desk feed and enormously better than a room microphone.
What you do not need
- Receiver headsets. Attendees use their own phones, so there is nothing to buy, charge, sanitise, count back in, or replace when somebody walks off with one.
- Interpreter booths. No soundproof boxes, no floor space, no rental, no argument about where to put them.
- A dedicated technician. If someone can start a video call, they can start a translated broadcast. Train two people so it does not depend on one.
- A new computer. The work happens on a server, not on your laptop.
What about the network?
Less than people fear. Sending speech to be transcribed is a modest, steady upload — far lighter than streaming video, which many churches already do. Subtitles arriving on phones are text, which is close to nothing. The read-aloud option is the heaviest part and is still comparable to a low-quality podcast.
The realistic risk is not bandwidth but coverage: a stone building where one corner has no signal, or guest Wi-Fi that limits how many devices can connect at once. Test in the corners of the room, not next to the router.
A ten-minute check before your first service
- Start a broadcast and speak into the actual microphone from the actual position, for two minutes, at normal preaching volume.
- Read the transcript. Not skim — read it.
- Walk to the four corners of the room with a phone and check the subtitles are arriving.
- Have somebody scan the code cold, without help, and see how long it takes them.
- Do it again with the room full, on a Sunday, before you rely on it.
Why the old model persisted
For years the hardware model was the only option, so churches assumed the cost was unavoidable and budgeted accordingly. It no longer is. Before you price up receivers and booths, check whether the phones already in the room can do the job — and put the money you save into the one thing that actually determines quality, which is the microphone.
Try it at your next service
InMyTongue streams live subtitles and spoken translation to every attendee's phone, in the language they choose, straight from a browser with nothing to install. New accounts include free credit and no card is needed to try it. See how it works.