How to transcribe a sermon, from Sunday service to searchable text

Sermon transcription means turning everything said in the message into text, in order, with the scripture references and a timestamp for each section. The transcript serves the pastor reviewing their own preaching, small groups studying during the week, members who missed Sunday, the church website and blog, and captions on the YouTube video.

In FilmScribe, the route is short: you upload the service or paste the livestream link, mark just the message and get the text back with timestamps, speakers and chapters by theme. This guide walks through that workflow and what makes the difference in the result: the right audio, proofreading scripture and names, chapters by passage and the right format for each use.

The short version

• Upload the service or paste the livestream link, then mark only the message: only that range is processed and counts toward your hours.
• We transcribe word by word, with timestamps, speakers separated and chapters by theme.
• Transcript quality depends on audio. The soundboard feed beats the camera mic every time.
• Proofreading is your part: references, biblical names and people’s names. One correction carries over to every export.
• Each destination has its format: PDF to read and print, TXT to publish, SRT for captions.

How to transcribe a sermon with FilmScribe

1. Upload the service or paste the livestream link

Upload the audio or video file (MP4, MOV, MP3, WAV, M4A and more) or paste a YouTube link. Livestreams work too: we process them once the replay is available on the channel. There is no length limit per file; what counts is your plan’s hours.

2. Mark just the message

A Sunday service can run two or three hours. Before processing, set the start and end of the sermon, say from 1:12:40 to 2:05:10. Only that range is processed and only it counts toward your hours. Transcribed worship turns into a block of song lyrics in the middle of the text, and announcements and the offering don’t help anyone studying.

3. Get the text with timestamps and speakers

We transcribe word by word, with a timestamp on every word, and separate the speakers: the pastor, the scripture reader and whoever led prayer show up apart. The language is detected on its own, among 90+ languages. Click any sentence and the player jumps to that moment. See transcription and speakers.

4. Proofread right in the text

Search for each book mentioned, check the reference and fix it in place. The correction carries over to the transcript, the captions and every export. Mark as favorites the lines your media team will want later.

5. Use the chapters by theme

We split the message by topic, with titles and timestamps, in the format YouTube descriptions accept. The chapters become the table of contents of your PDF and the skeleton of your sermon notes. See chapters and moments.

6. Download the right format for each use

TXT and PDF of the full transcript, SRT with the YouTube preset, and an MP3 of the whole message or just one passage for the church podcast (download snippets). Keep each series in folders and projects, so finding “that sermon on Joseph” months later is a search.

To try it on your own service, the free audio to text page leads to the trial of 3 days with 1 hour of material, with a card, and the result shows up on screen.

What Studio and Agency add

• Named speakers. Rename “Person 1” to “Pastor Mark” and ask us to identify names, with the excerpt that proves it.
• Moments inside each chapter, to find the illustration or the altar call without reading everything.
• Diagrams: a mind map, flow and timeline of the message, in PDF, Markdown or TXT, handy for small group leaders.
• Translation of the transcript and captions into 9 languages.
• Connector: ask ChatGPT, Claude or Gemini CLI for a summary or discussion questions based on the transcript, in conversation.

What we don’t do: we don’t write the sermon outline for you. Inside the app there is only a short synopsis at the top of the transcript; chapters and diagrams are the skeleton, and the final wording is your team’s.

Before you upload: the right audio

Transcript quality depends on the audio more than anything else. Two things to watch:

1. Use the soundboard feed. A camera mic at the back of the sanctuary picks up echo, fans, kids and chatter. The feed from the soundboard, or the livestream itself, carries a clean voice. Better audio means less proofreading.
2. Interpreters. If the service has a sign language interpreter, the transcript still follows the preacher’s speech. If it has live spoken interpretation, record each language on its own channel if you can and upload the original.

Proofreading: scripture, names and filler words

Proofreading is what separates a useful transcript from one that embarrasses the church. In FilmScribe it happens in the text itself, with the audio one click away. Four passes:

• References. Search each book name (“Romans”, “Hebrews”, “Psalms”) and check chapter and verse. Pick one style, “Romans 8:28”, and use it everywhere.
• Names. Uncommon biblical names (Mephibosheth, Zerubbabel, Nebuchadnezzar), members’ names and guest speakers’ last names. Fix once, and captions and exports follow.
• Scripture read aloud. When the pastor reads a verse, the transcript records what was read. If the church wants a specific translation (NIV, ESV, KJV), adjust the text and name the version.
• Filler words. For study PDFs and the website, drop “um”, “you know” and repeated words. For captions and the archive, keep what was said.

Verbatim or clean? Verbatim keeps everything, hesitations included. Clean edits speech for reading without changing the meaning. For the pastor’s review and for captions, go verbatim. For the website, newsletter and study guide, clean reads better. If it will become a book or devotional, treat it as edited prose with the pastor’s sign-off.

From transcript to sermon notes and small group guide

A raw transcript has a beginning, middle and end, but no visible structure. With the chapters we generate, three steps turn it into study material:

1. Chapters by passage. Adjust the titles where the pastor changes text or idea: “Reading: Philippians 4:4-7”, “The anxiety nobody sees”, “The peace that guards your heart”, “Invitation and prayer”. With title and time, each chapter becomes an entry in the PDF’s contents and, pasted into the video description, a YouTube chapter. Our guide on how to add YouTube chapters shows the format.
2. Sermon notes. For each chapter, write the main idea in one line and note the key verse. Theme, text, points and application: your outline is done, with the pastor’s own words alongside. On Studio and Agency, the mind map of the message gets you there faster.
3. Discussion questions. End each chapter with one application question: “Where have you been trying to fix on your own what you should bring to God in prayer?” Three or four questions fill a small group meeting.

What a sermon transcript is for

One transcript covers every use below. You proofread once and download in several formats.

How long is a sermon transcript?

At a typical preaching pace of about 130 words per minute:

Pages assume about 500 words per single-spaced page in a 12-point font. Faster preachers pass 150 words per minute. To run the numbers at your pastor’s pace, use our free words to minutes calculator.

That volume is why a small group PDF almost always needs sections: nobody reads twelve pages without a table of contents.

Formatting the sermon PDF

Download the TXT to lay out the handout in your church’s word processor, or send the PDF as it is for the pastor to review. A sermon PDF people actually finish usually has:

• A header: sermon title, preacher, date and main passage.
• Contents: the chapters with timestamps, for anyone who wants to hear the original.
• Paragraphs, not loose caption-sized lines.
• Speaker names when there is more than one voice, like a scripture reader or a testimony.
• A link to the full service at the end.

Captions for the sermon on YouTube

The same proofread transcript becomes the caption file for the full video. Download the SRT with the YouTube preset and upload it as the video’s captions on your channel. Viewers watching muted, members who are hard of hearing and people following along on the bus get the whole message. If your church already has captions in another format, the free SRT, VTT and TXT converter switches formats without reprocessing anything. To take the message into another language, see how to translate subtitles.

Other routes: by hand and YouTube’s transcript

For one quote or a few minutes, typing by hand is fine. If the sermon is already posted and the video has automatic captions, the Show transcript button in the description gives you a free rough draft. It comes in short caption-sized lines, with uneven punctuation, no split between the pastor and the scripture reader and mangled names, so it works as a starting point, not as small group material. For a weekly routine with a PDF, captions and chapters, the workflow above saves the time that cleanup would take. Our guide on how to transcribe a YouTube video covers the link route in detail.

From text to clips

If the service is filmed, the same upload also yields captioned vertical clips for social media, picked from the transcript. The workflow is in our guide on how to make sermon clips and on the sermon clips page. For Bible studies recorded on a phone, see Bible study to text. For other kinds of recordings, see how to transcribe a video to text.

Common mistakes

1. Processing the whole service. Worship becomes a block of lyrics mid-text, proofreading doubles and your hours go with it. Mark just the message.
2. Publishing without proofreading. One wrong reference in the small group PDF becomes a question for the whole group.
3. Using the camera audio. Sanctuary echo and room noise cause most errors.
4. Handing out a wall of text. Twelve pages without contents won’t get read.
5. Mixing verbatim and clean. Pick a style before proofreading and stick with it.
6. Skipping the pastor’s sign-off. Anything going on the website or into a book needs the preacher’s approval.