Keep translation in sync with video Beta

Last updated October 1, 2026 For: Tech volunteers fine-tuning translated streams and recordings in OBS

On this page

A translation always runs a few seconds behind the speaker, like a live interpreter. This article explains the controls that decide how that delay is handled, how loud the original sound is under the translation, and how to play the translation through a speaker in the room.

Who it's for#

Anyone using translated streams or translated recordings who wants them to sound and look right.

Before you start#

  • At least one language is added under Languages in the dock.

How far behind is the translation?#

Without any delay, each translated sentence follows the original by about 3–6 seconds. That's fine for the listener page and for the in-room speaker. On a video stream, though, viewers see the pastor's lips move and hear the translation later.

There are two answers:

  • For translated streams: Sync to video delays that one stream so the translation lands on the right moment.
  • For translated recordings: Re-sync recording after stop fixes it after the service. See Record a translated copy in OBS.

Sync to video (translated streams)#

In a language's panel, Sync to video (delay this stream by 8 s) is on by default for YouTube and Facebook, and off for a custom server.

When it's on, that stream shows the picture, the original sound and the captions 8 seconds late, so every translated sentence starts when it is spoken on screen. Only this translated stream is delayed. Your main stream and recording, the in-room speaker and the listener page stay live.

While it runs, the panel shows something like Delay 8 s · lag 4.6 s ✓: the stream is delayed 8 s and the translation is currently 4.6 s behind, so it fits. When the stream starts, it shows "Buffering 8 s of video before going live…" for about the length of the delay. When the main stream stops, it sends its last seconds, then stops.

SermonWave dock language panel with "Sync to video (delay this stream by 8 s)" ticked and the "Delay 8 s · lag 4.6 s" readout

Set the sync delay#

Under Languages, Sync delay sets the delay for all "Sync to video" streams of this device: 4–15 seconds, default 8. Change it and click Apply. It's the device's setting, so you can also change it in the Dashboard's Device settings.

How to tune it#

  1. Start with the default 8 s.
  2. Watch the readout during a service. If the lag stays well below the delay, you're fine.
  3. If the panel says "… sentence(s) later than the delay (up to … s): raise the sync delay", a sentence was not ready in time; it played as soon as it was ready. Raise the delay by 1–2 seconds.
  4. If it never warns and you want viewers closer to live, you can lower it a little. A shorter delay means more risk of late sentences.

With Sync to video off, the stream is not delayed and the translation simply follows by 3–6 s.

Original audio and Under translation#

Each language's stream and recording use their own mix: the translated voice over the original sound of the service (OBS Track 1: pastor, music, room). The original is turned down while the translation speaks, like an interpreter on TV. Two sliders in the language's panel control it, both relative to the original level:

SliderControlsRangeDefault
Original audioHow loud the original is between translated sentences−40 to +6 dB; far left = Off0 dB
Under translationHow loud the original is while the translation speaks−40 to 0 dB−18 dB

Changes apply at once to that language's stream and recording, and to the re-synced recording.

Language panel with the "Original audio" and "Under translation" sliders

Tips:

  • Viewers struggle to follow the translation? Lower Under translation (for example −25 dB).
  • Want worship songs to come through fully between sentences? Keep Original audio at 0 dB.
  • Want only the translated voice, nothing else? Move Original audio all the way left to Off.

Play the translation in the room#

Tick Play … in the room in a language's panel to play the translated voice through a speaker in the building, for example for a section of seats or an overflow room.

  • It plays only the translated voice, never the original sound, so there's no feedback loop.
  • It plays on OBS's Monitoring Device. Choose the speaker or headphone output in Settings → Audio → Advanced → Monitoring Device.
  • It plays whatever scene is showing, and it is not in the stream, the recording or the Audio Mixer.

The panel shows "Playing on the monitoring device (Settings → Audio → Advanced)." while it runs.

Troubleshooting#

Nothing plays in the room. Check the Monitoring Device in OBS audio settings and the volume of that output. Only one monitoring device can be used at a time in OBS.

The translated stream takes a while to start. With Sync to video on, it first buffers the delay ("Buffering 8 s of video before going live…"). Also set the keyframe interval to 2 s; see Stream a translated version from OBS.

The original drowns out the translation. Lower Under translation.

FAQ#

Does Sync to video delay my main stream?#

No. Only the translated stream it's ticked on.

Does Sync to video use more CPU?#

Hardly. The video isn't encoded again; the main stream's frames are sent later.