Sound-off scroll is the default behavior on Instagram, TikTok, and YouTube Shorts — and the people who most need to hear your sermon are the ones who'll never unmute. Captions are not a polish step. They are the entire reason your clip reaches beyond your existing audience. Sermon Clips generates Whisper-grade transcripts tuned for theological vocabulary, breaks them at natural phrase boundaries, and burns them into the video so they survive every cross-platform repost — without an SRT export, a DaVinci subtitle layer, or a 20-minute manual proofread per clip.
Most caption tools were trained on lifestyle vloggers, gaming streams, and recipe videos. Sermon content breaks them in three specific ways.
Auto-captions from CapCut, Premiere, and TikTok mis-spell 'covenant,' 'sanctification,' 'eschatology,' and every Bible book past Job. Every error is a credibility hit your audience reads in real time.
Separate caption files don't travel when someone reshares your clip from Instagram to Reddit, from TikTok to a group chat, or downloads and reposts it. Burned-in captions are the only kind that survive the journey.
Asking a volunteer to verify whether the AI heard 'justification' vs. 'justification by faith' correctly requires theological training they don't have. Proofreading captions becomes a pastor bottleneck — or it doesn't happen at all.
Five automated steps between "upload the sermon" and "ready-to-post clip with burned-in captions." You don't touch any of them — but here's what's happening.
Upload the full sermon
Drop in a video file, paste a YouTube link, or send a Vimeo URL. The same source the AI uses to find clip-worthy moments is the source we transcribe — no separate audio prep.
AI transcribes with theology-aware vocabulary
A Whisper-grade transcription model with a theological vocabulary overlay processes the audio. Books of the Bible, doctrinal terms, denominational language, and common proper nouns are recognized at 95%+ accuracy on clear audio.
Captions are broken at natural phrasing
Instead of mid-clause line breaks ('the grace of / God is sufficient') the engine breaks at clause and sentence boundaries — 1 to 2 lines per caption, readable at a glance, no 4-line walls of text.
Captions are burned into the 9:16 export
No SRT file. No subtitle track. The captions are rendered directly into the video pixels, which means they survive every download, repost, and platform conversion. Once the clip leaves your account, the captions go with it.
Position is safe-zone aware
Captions sit in the upper-middle third of the frame — clear of the speaker's face (tracked) and clear of the platform UI overlays on Instagram, TikTok, and YouTube Shorts. No more captions buried behind the like-button column.
These are the caption choices that consistently outperform on sermon-style content — the defaults Sermon Clips ships, and the reasoning behind each one. For the workflow side of captioning across all your weekly content, see our captions & subtitles workflow guide.
1 to 2 lines per caption
Not 4. Viewers read caption blocks in roughly 1.5 seconds — anything longer than 2 lines competes with the speaker's next thought and the clip stops landing.
Readable font size
Sized for a phone held at arm's length, not for a desktop preview. The most common DIY mistake is captions that look fine in a 16-inch editor and become illegible on a 6-inch screen.
White text on a soft drop shadow
High contrast against any background — light sanctuary stage, dark backdrop, busy outdoor shot. Plain white with no shadow disappears the moment the speaker walks past a window.
Upper-middle position
Lower-third is where Instagram and TikTok render their own UI — handle, caption text, like buttons. Anything in the lower third gets fought over. Upper-middle is the safe zone.
Break at clause boundaries
Splitting 'I will never / leave you nor forsake you' loses the rhythm of the verse. The engine breaks where a natural reader would pause — at commas, conjunctions, and sentence ends.
Emoji sparingly
One emoji per highlight moment, not three per clip. Excess emoji in religious content reads as desperate-for-attention and undercuts the gravity of the message.
Theology-aware vocabulary
The model recognizes the full canon — every book of the Bible — plus the doctrinal terms (justification, sanctification, propitiation, eschatology, ecclesiology) that general-purpose captioners routinely butcher.
Bible reference auto-formatting
Spoken 'John three sixteen' renders as 'John 3:16.' Spoken 'first Corinthians thirteen' becomes '1 Corinthians 13.' References stay readable instead of as transcribed phonetics.
Proper-noun capitalization
Names of people, places, churches, and ministries are capitalized correctly. The Holy Spirit doesn't come through as 'the holy spirit.' Jesus stays Jesus, not jesus.
Automatic line-break logic
Captions break at clause boundaries and natural reading pauses — never mid-phrase, never with one orphan word on a second line. Optimized for the phone screen, not the editor preview.
Configurable style on paid plans
Match the highlight color to your church brand, choose from a curated set of fonts, and adjust position within the safe zone. Set it once at the account level and every clip ships consistent.
No SRT files — burned in
The captions are rendered into the video pixels, not delivered as a separate subtitle file. They survive every reshare, repost, download, and platform conversion — which means your message reaches viewers exactly as you shipped it.
Upload your sermon. Get clips back with burned-in, theology-tuned captions positioned in the safe zone — ready to post. No SRT exports, no manual proofread, no DaVinci layer.
Get Free ClipsHow accurate are the auto-generated sermon captions?
95%+ accuracy on clear sermon audio. We use a Whisper-grade transcription model with additional theological vocabulary tuning — meaning words like 'covenant,' 'sanctification,' 'eschatology,' 'propitiation,' and every book of the Bible are recognized correctly out of the box. Generic auto-captioners (CapCut, Premiere, TikTok built-in) routinely mis-spell these terms because their training data is dominated by general internet content, not sermon content. We also apply proper-noun capitalization and Bible-reference formatting so 'John 3:16' renders cleanly instead of as 'john three sixteen.'
Can I edit captions before they're burned into the clip?
Yes. Every clip ships with a caption preview before final render. You can read through the transcript, fix any word that came through wrong, adjust line breaks, or tweak emphasis — and only then export. For paid plans, you can also configure caption styling (font, highlight color matched to your church brand, position) on a per-account basis so every clip you ship looks consistent.
Do you support languages other than English?
Yes — 30+ languages including Spanish, Portuguese (Brazilian and European), Korean, Mandarin, French, German, Tagalog, Swahili, and Arabic. The same theological vocabulary tuning carries across languages where the training data supports it. This matters for multilingual congregations and for churches whose sermon clips travel internationally via reshares. For Spanish specifically, we maintain a dedicated /es subdirectory of the marketing site — same product, Spanish-language landing pages.
Will the captions cover the speaker's face?
No. Captions are positioned in a safe zone — typically the upper-middle third of the 9:16 frame — that stays clear of both the speaker's face (tracked via the same face-detection used for vertical reframing) and the platform UI overlays from Instagram, TikTok, and YouTube Shorts (like-button column on the right, caption/handle text on the lower-left). This is a deliberate design choice. Captions buried behind the IG comment overlay or floating across the speaker's mouth are the most common reason DIY sermon clips look amateur.