If you rebroadcast recorded sermons as a continuous YouTube Live stream, captions must be configured for the live broadcast or supplied by an eligible live automatic-caption feature. Uploading captions to the sermon recording in YouTube Studio is a separate workflow; it does not configure captions for the live stream.
YouTube documents two routes for sending captions to a live broadcast: a supported caption feed sent to a Captions Ingestion URL, or EIA-608/CEA-708 captions embedded in the outgoing video. Automatic live captions may also be available for eligible English channels at normal latency, but they are conditional and need review. Decide whether viewers need captions during the broadcast, on the archive, or both before choosing a route.
Decide whether captions are for live or VOD
A live broadcast and its video-on-demand recording (VOD) are different captioning targets. If you add a timed caption file, transcript, or manually entered captions to an uploaded video in Studio, those captions belong to that video. They are not a live caption feed and do not automatically appear when you play the same sermon as part of a continuous broadcast.
For live captions, YouTube needs captions to arrive with the live event, either as a separate feed or embedded in the outgoing video. That means planning the caption route alongside the stream setup. A caption file already prepared for the sermon might help, but only if the software or encoder sending your broadcast can deliver it through a supported live route. Do not assume that the file can be attached to a playlist or loop without checking the relevant software documentation.
There may be two jobs to complete. First, supply captions while the sermon is being broadcast. Then, after the live stream ends, review the archive and ensure it has useful captions of its own. YouTube says captions created by live automatic captioning do not remain as the VOD caption track; new captions are generated for the recording and can differ from what viewers saw live. A live-caption check is therefore not an archive-caption check.
For a continuous channel, consider what a viewer joining midway needs. If a sermon is already under way, live captions should reflect the current speech rather than depend on a viewer finding an earlier point in the recording. YouTube’s DVR guidance for live streams explains the separate rewind behaviour: DVR may be limited or unavailable on streams longer than 12 hours, and viewers cannot seek to before the broadcast began. A continuous stream should not be treated as a convenient captioned archive with guaranteed rewind access.
Choose a live-caption route
Compare the routes by what you already have and what the stream requires. A live caption feed is suited to a workflow with supported captioning software or an operator supplying captions. Embedded captions require an encoder capable of putting EIA-608/CEA-708 captions into the outgoing video. Automatic captions avoid preparing a live caption track, but availability, language and latency constraints apply, and recognition can be wrong.
| Route | What must be available | Main checks | Useful when |
|---|---|---|---|
| Caption ingestion URL | Supported caption software and a caption feed | Correct signed URL, one feed per stream entry point, appropriate delay | A captioner or caption software can provide text during the sermon |
| Embedded captions | A compatible encoder outputting EIA-608/CEA-708 | YouTube caption mode and encoder output agree; check track and language support | The existing broadcast encoder can carry captions |
| Live automatic captions | An eligible channel, English audio and normal latency | Feature availability, clear speech, generated text review | YouTube offers the feature and its limits suit the broadcast |
There is no universally best route. If a church already works with a captioner, a feed may be the most direct arrangement. If its encoder and caption workflow already support embedded captions, that can avoid a separate incoming feed. If neither is in place, check automatic-caption eligibility before building a workflow around it, and do not treat it as a sure fallback.
Latency matters. YouTube’s automatic-live-caption instructions specify normal latency, and the Live Streaming API says ultra-low latency does not support closed captions. The caption-feed guidance also describes a broadcast delay for its event setup. This is not a reason to add delay without checking your event configuration: follow the instructions shown for the specific route and account for the effect on viewer timing.
If your channel also loops titles or programme details on screen, keep those overlays distinct from captions. An episode label tells viewers what is playing; captions represent spoken content. A separate guide on showing episode titles during a 24/7 stream can help with the display layer, but it does not replace the live caption configuration.
Send captions through a caption ingestion URL
For the caption-feed route, YouTube’s live caption requirements describe enabling Closed captions in the stream settings, copying the signed Captions Ingestion URL, and giving that URL to the caption operator to enter in supported software. The operator’s software sends captions to YouTube while the broadcast is running. The URL is part of the live setup, not a place to upload a transcript in Studio.
Treat the URL as a credential. Share it only with the intended caption operator, and avoid putting it in public notes, a shared document with broad access, or a stream description. If it is exposed, review YouTube’s current instructions for replacing or regenerating it before relying on the same event setup. Check that the operator has the correct URL for the broadcast you are about to run; a working caption workflow pointed at a different event will not help viewers on this one.
YouTube’s event guidance says that each stream entry point can have one caption feed. If your continuous channel changes its video source or event configuration, confirm which stream entry point the feed is attached to and whether the arrangement remains valid for the next broadcast. The guidance also recommends a 30- or 60-second broadcast delay for the described event setup. Apply that note to the matching setup rather than assuming it is a universal requirement for every channel or caption system.
The captioner and the video operator should agree on practical hand-off details before going live: who starts the feed, which event it targets, and how they will notice if captions stop arriving. A prerecorded sermon does not necessarily make the captioning work automatic. If captions are being entered or corrected in real time, someone still needs to follow the speech and handle names or phrases that recognition or preparation may not capture correctly.
YouTube’s live caption requirements are the primary reference for the ingestion route, signed URL and event settings. Use them alongside the caption software’s own documentation. The available official guidance does not establish a universal end-to-end procedure for pairing any pre-timed file with any continuously looped video source, so verify that pairing with the software provider before depending on it overnight.
Embed captions in the outgoing video
The second documented live route is to embed captions in the video signal. In YouTube’s setup, enable Closed captions and select Embedded 608/708. Your encoder must output captions in EIA-608 or CEA-708 format; choosing the option in YouTube alone does not create captions or turn an ordinary subtitle file into an embedded track.
Confirm the arrangement from both ends. Check that the encoder can output the required caption format, that captions are enabled for the relevant event, and that the language track is the intended one. YouTube notes that the 608/708 standard can support up to four language tracks, while its guidance currently supports one caption track. Do not plan on multiple tracks simply because the underlying standard permits them; check YouTube’s current requirements for your setup.
For a prerecorded sermon, ask the encoder or captioning provider how a prepared caption track follows the sermon as it repeats. The key question is whether the outgoing signal carries correctly timed captions through the full clip and back to the beginning on each loop. A caption track that runs once and then drifts, disappears, or continues over a transition will confuse viewers even when its format is technically accepted.
If you are considering changing the broadcast software or encoder, make the caption requirement part of that decision rather than buying equipment on assumption. YouTube’s streaming tips describe software encoding and professional-grade encoders as possible approaches and recommend testing and monitoring. They do not say that every channel needs new hardware to add captions. Your current setup may be sufficient, but its documentation should confirm caption output before you depend on it.
Check whether automatic live captions are available
YouTube’s live automatic-caption instructions describe a feature that is English-only, individually enabled, and gradually rolled out to eligible English channels. They specify normal latency. These conditions mean you should check the feature on your own channel and for the intended broadcast rather than promising viewers that every stream will receive automatic captions.
Where available, automatic captions may reduce the need to supply a live caption feed, but they are speech recognition, not a reviewed transcript. YouTube warns that poor sound, unsupported language and overlapping speech can interfere. A sermon can also include personal names, place names, scripture references, Sanskrit or other liturgical terms that are unfamiliar to a general-purpose recogniser. Check actual output; do not assume that a familiar speaker or a clear recording makes every phrase accurate.
If the channel is not eligible, the feature is not enabled, the sermon is not in English, or the stream must use a latency mode that does not meet the requirements, choose a supported caption feed or embedded captions instead. A recorded source does not change these live-feature constraints. Nor should automatic captions be treated as a substitute for the separate captions needed on the VOD.
The automatic-caption help page explains YouTube’s eligibility, language, latency and review caveats. Check the current page because availability and interface details can change. If live captions are essential to the service, establish a plan that you can test and operate without depending on a feature that may not be available to your channel.
Review caption accuracy and timing
Reviewing captions means listening and watching them against the sermon, not merely confirming that some text appeared. Pay particular attention to speaker names, scripture references, place names, repeated refrains, and words likely to be unfamiliar to a general recogniser. A small change in a name or negation can alter meaning. For a live feed, arrange who can see the output and who can correct it if the route allows corrections.
Timing matters as much as wording. Captions that arrive well after speech are difficult to follow; captions that appear too early can reveal the next sentence before it is spoken. Check pauses, transitions, and the start and end of the recording. If your programme moves from an introduction into a sermon, make sure the caption track follows the actual audio rather than starting part-way through or carrying text into silence.
For a continuous loop, inspect the boundary between the end and beginning of the clip. Confirm that the final caption clears and the opening caption appears at the right speech, each time the sermon restarts. If the stream rotates between sermons, verify each recording’s timing and track rather than assuming one caption file will fit all items. The title card or episode label can identify the sermon, but it cannot tell viewers what the speaker said.
After the live event, review the VOD independently. YouTube generates new captions for the archived video after live automatic captioning, and they may differ from the live captions. If the archive needs correction, use the Studio caption workflow applicable to uploaded videos, such as supplying a timed file, working from a transcript, or editing captions. YouTube cautions that automatic syncing may not suit very long or poor-quality audio; inspect the result before treating it as finished.
Test captions before the broadcast
Run a private or otherwise non-public test that exercises the same caption path you intend to use. Include the actual sermon audio, event settings, encoder or caption software, and the loop boundary. Watch from a viewer account or device where practical, since a preview in the operator’s own software does not prove that captions are reaching YouTube in a usable form.
For a caption feed, confirm the operator has the correct signed URL, that the feed reaches the intended stream entry point, and that captions arrive in step with speech. For embedded captions, verify the encoder’s 608/708 output and the selected YouTube caption mode. For automatic captions, check that the feature is present for the channel and event, then inspect the generated text against the recording. A test of one route does not validate another.
Keep a short checklist with the stream notes: caption route, responsible person, event or entry point, language, latency setting, and what to check if text stops or drifts. Avoid putting the signed URL in that checklist. If the stream is intended to run while the operator is away, arrange a way to notice a caption failure and know whether captions can be restarted without interrupting the video.
The same principle applies to overnight operation. The common streaming mistakes checklist is useful for thinking about source, monitoring and recovery, but captions need their own check. If the video source itself is intended to keep running without the local computer, a guide to looping a YouTube Live programme covers the broader continuous-playback problem. Neither removes the need to verify caption delivery for the actual broadcast.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Do captions uploaded in Studio appear during a live sermon stream?
No. Captions added to an uploaded video in Studio are for that VOD workflow and do not configure the live broadcast. For the live sermon, use a supported caption feed, embedded captions, or live automatic captions if YouTube makes that feature available for your channel and stream.
Will YouTube keep live automatic captions on the sermon recording?
YouTube says live automatic captions do not remain as the VOD caption track. New captions are generated for the recording and may differ, so review the archive separately and correct it if needed.
Can I use automatic captions on every continuous stream?
No. YouTube documents live automatic captions as English-only, dependent on channel eligibility and availability, and requiring normal latency. Even when enabled, recognition can be affected by sound quality, overlapping speech and vocabulary, so check the output rather than assuming it is accurate.
Does an embedded caption track work automatically with any sermon loop?
Not necessarily. You need a compatible encoder that outputs EIA-608/CEA-708 captions, and the track must stay in time with the sermon and its repeat boundary. Confirm file and encoder compatibility with their documentation, then test the complete loop before relying on it.