Supported recordings
MP3, M4A, WAV, WebM, and OGG files up to 15 MB, sent for transcription while you are signed in. HomilyForge does not store the original audio.
sermon transcription
Sermon transcription that stays attached to its source
Upload a recording and get back editable text split into numbered paragraphs, so every note you make later can point at the words that produced it.
You already have the recording. What you need is text you can work from, in a workspace that still knows which paragraph each idea came from.
What HomilyForge actually does here
Audio/text input, transient transcription, supported formats, and privacy.
Text works too
Paste a transcript directly, or upload a UTF-8 .txt or .md file up to 2 MB. The unsigned local editor needs no account at all.
Numbered from the start
The transcript is split into up to 24 numbered paragraphs. Those numbers are what later outlines, notes, and exports cite.
A worked example
Here is the beginning of a transcript after paragraph numbering, taken from original demonstration material written for this page.
- §1
Preparation begins long before the room fills. A message becomes useful when someone has already sat with it, argued with it, and written down the part they still do not understand. Philippians 4:6-7 keeps surfacing in this kind of preparation, because anxiety is usually the first thing that crowds out attention.
- §2
Notes made in a hurry rarely survive the week. They record what was said without recording where it was said, so no later review can check a single claim. A paragraph number is a small thing, and it is what separates a quotation from a memory.
- §3
Listening well is slower than agreeing quickly. A talk that moves fast invites us to keep the conclusions and quietly drop the reasoning that produced them. Psalm 34:18 was read twice here, and attention to that repetition changes how the middle of the argument should be weighed.
Extracted references from the full transcript, normalised and de-duplicated:
- Philippians 4:6-7
- Psalm 34:18
Demonstration material written for HomilyForge and processed by the same engine the product uses. No supplied recording or transcript is ever published as an example.
How the workflow works
Step 1
Supply the message
Paste a transcript or upload a supported audio file for transient transcription in a private workspace.
Step 2
Structure the source
Number paragraphs and derive summaries, outlines, explicit references, questions, and applications with visible source links.
Step 3
Review and export
Edit the material, verify interpretation and permissions with a qualified human, then export Markdown or a print-ready view.
Limits that apply before you start
Useful search results begin with an accurate promise. These are the current bounds on this job.
- Audio transcription requires Google sign-in and counts against a daily allowance of eight recordings.
- There are no speaker labels, timestamps, or diarisation. One recording becomes one continuous transcript.
- Words the model cannot make out are marked rather than guessed, so the transcript needs a read-through before use.
Questions about sermon transcription
Is my sermon audio stored after transcription?
No. Audio is sent for transient transcription and is not persisted by HomilyForge. The resulting text is what you keep, edit, and can delete at any time.
How long a recording can I transcribe?
Uploads are capped at 15 MB per file, which is the current bound on a single transcription. Longer recordings are on the backlog and are not available today.
Do I need an account to use the transcript editor?
Only for audio. Pasting or uploading text works without an account in a browser-local editor; sign-in adds a private cloud library, generation, and transcription.