PulpitCue
Your preacher speaks. PulpitCue finds the Scripture.
How live Scripture detection works, what it needs and where it can go wrong.
The problem
Preachers move between passages, quote verses from memory and paraphrase. A media operator has to find each one in time, often while running songs, slides and the stream.
What PulpitCue does
PulpitCue listens to the pulpit microphone and writes a live transcript. When it recognises a spoken reference, such as "John chapter three, verse sixteen", it suggests John 3:16 to the operator. Suggestions also appear for passages that are being read or quoted.
- The operator sees each suggestion with the evidence behind it, and chooses when to show it.
- Hands-free mode is optional. It can advance Scripture during a reading, with guardrails, and stops where the preacher stops.
- Speech recognition and Bible matching run on your own computer, so sermon audio is not sent to a cloud transcription service and there are no paid speech APIs.
Honest limits
- Speech recognition is English only.
- Suggestions can be wrong. Loud music or a quiet microphone makes recognition less reliable, so keep watching the screen.
- The models download on first setup. After that a service can run offline, though licence check-ins still need an occasional connection.
- Live captions and Scripture detection need a licence after the trial.
Set it up
Read the guide to Bible and live Scripture detection and hands-free mode, then run a test with your own microphone feed before Sunday.