A teleprompter for online course creators — forty modules, one voice

A course is not a video. It is thirty or forty of them, filmed in batches over several weeks, and sold as one coherent thing. The enemy is drift. In module 2 you called it the dashboard; by module 19 it has become the control panel. The energy you had at nine in the morning is not the energy you had at six that evening, and the phrasing you improvised in October is not the phrasing you improvised in November. Students binge modules back to back, so every inconsistency sits side by side on their screen — and turns up in reviews as ‘feels stitched together’.

Then there is the re-record. The tool you teach ships an update and module 3 is suddenly wrong, or a student question exposes a gap in module 11. Now you are trying to say the same thing the same way, a month later — same terms, same order, same pace — with no reliable record of what you actually said the first time. The published video is the only source, and it is the edited version of a script you have since overwritten.

And the economics are unforgiving. Recording days are batch days, blocked out around a day job or a launch date. Every take lost to a fumble on module 7 is time taken from module 8, and a course that slips a week slips its launch — with an email list that was already told the date.

The usual system is a folder of documents and a prompter app that holds one pasted script at a time. Paste module 9, record it, paste module 10, and hope you did not paste the draft instead of the final. The running order lives in your head. The version you actually read into the camera lives nowhere, because the document kept changing after the shoot.

Batch days leak time in small ways too: stopping the take to check the next section, trimming out the sip of water, working out afterwards which file was which. Nine takes into a Sunday, the camera roll is a wall of identical thumbnails, and the wrong one gets edited.

Re-recording an old module means archaeology. You scrub through the published video, transcribe yourself, and hope the wording you recover matches the modules either side of it. Some creators keep a spreadsheet of what changed where; most just re-shoot and accept the seam.

Captions are their own job. The course platform’s auto-transcription spells your product’s name three different ways across the series, and fixing it means opening every module’s caption file in turn — or paying a captioning service by the minute, per module, per revision.

Talk2Camera keeps the whole course in the script library: a folder per course, one script per module, favourites for the ones still being written. Every script shows a word count and an estimated read time, so you can see before the batch day that module 6 is a four-minute read and module 7 is a twelve-minute one — and fix the imbalance in the text, where it costs minutes, not on camera, where it costs a day.

On the day, you work down the folder in order. Pause mid-take between sections: check the next paragraph, drink the water, pick the take straight back up. The paused time is excluded from the file rather than sitting there as dead air, so a module recorded in five sittings still comes out as one clean take.

Every recording lands in the takes library with its own poster frame, length, size and date, so on Sunday evening you can still tell which of nine takes was module 12 without playing them. More importantly, each take remembers its script: the exact text you read is saved beside the recording — permanently, not just in the session you shot it.

That sidecar is what makes the month-later re-record routine instead of painful. Open the old take of module 3 and the script you actually read is right there. Change the paragraph the update broke, leave the rest, and record the new version in the same words, the same order and the same voice as the series around it.

Captions per module come from that same saved script rather than a recognition engine’s guess at it, and export as a separate .srt file — the format course platforms prefer, because learners can switch subtitles off or search inside them. One honest note: on the free plan, exported video carries a small Talk2Camera mark; the paid plans remove it.

A creator is building a 22-module course on a photo-editing app. Saturday is a batch day: modules 9 to 14 sit in the course folder, ordered, with read times that already add up to the planned lesson lengths. She records them in sequence, pausing between sections instead of cutting, and by the evening the takes library shows six takes with six poster frames — one per module, each still carrying its script.

Six weeks after launch, the app she teaches moves a menu, and module 11 is wrong. She opens the take from launch week. The script is beside it — not the draft, the words she actually read. Two sentences change. She re-records the module in the same terms the rest of the series uses, exports it with a fresh .srt, and swaps both files on the course platform. Students who binge the course never feel the seam, and the other 21 modules were never touched.