THE DUO JOURNAL ·
Add captions to an iPhone video with Teleprompter Duo
In Teleprompter Duo, open a recorded take, expand Trim and edit, choose Edit captions, pace and branding, then generate captions from the recorded speech. With Pro and supported on-device speech recognition, you can review the words and their timing before enabling captions in the export. Finish by checking the prepared video, not only the text list.
This guide covers captions for a take recorded in the app. It does not describe importing arbitrary videos, translating speech, or exporting a separate subtitle file. Recording, script reading and caption generation are different steps; your teleprompter script is not automatically a transcript of what you actually said.
Before generating captions.
Listen to the saved take and decide whether its speech is clear enough to use. If a sentence is wrong in the audio, correcting only the caption can create a different mismatch. Decide whether to record that explanation again before spending time polishing its text.
Captions and caption corrections are Pro editing features. Free includes timed rehearsal, one completed recording, trimming and watermarked exports; further recordings require Pro. Check the current Free and Pro details before planning multiple takes or a captioned export.
The W3C captions guide explains that captions include speech and meaningful non-speech audio. Generated words still need review; a caption button alone does not establish accessibility.
Generate captions from the saved recording.
The following labels match the app’s English controls. A translated website page does not mean every speech language or localized app interface is supported on your device.
- Open the saved take in its review screen. Expand Trim and edit, then choose Edit captions, pace and branding.
- In Captions from your recording, choose Spoken language to match the language actually spoken in the take. Do not choose the language you want to translate into.
- Choose Generate captions on device. Allow Speech Recognition if you want to use this feature. Generation requires supported, currently available on-device recognition for the selected language.
- Keep the app active while it works. If you need to stop, use Cancel captions. Moving the app into the background cancels this generation task.
- When generation finishes, open Review words and timing. Treat the result as a draft, not a verified transcript.
If the selected language is unavailable, the app reports that condition and does not upload audio as a fallback. Choosing another language is useful only if it matches the recording. An English website or English script does not make another spoken language recognizable.
Correct the speech you recorded, not the script you planned.
Play a short passage, compare it with the generated text and fix names, dates, quantities and words that change the meaning. The correction screen provides a Word or phrase field for each existing entry. Do not assume that editing the script after recording changes the captured audio or these captions.
Here is an invented review example, not output from a test. Your planned line is “Share draft three on Monday.” In the recording, you instead say “Share draft four on Tuesday.” Suppose the caption draft reads “Share draft for on Tuesday.” The transcription correction is “four,” because that matches the audio. If Tuesday itself is the wrong instruction, resolve the spoken instruction rather than captioning Monday over an audible Tuesday.
Use your own recording as the reference. A clean script can help you notice differences, but it cannot establish which words survived an interruption, an improvised sentence or a skipped paragraph. Our script-formatting guide helps prepare the reader text; this review happens after capture.
Review timing on the original recording.
The correction screen exposes Start and End values in seconds in the original take. Listen near the beginning and end of the entry before changing them. Keep times ordered, within the recording, and with an end later than the start. If Apply reports a timing problem, return to the affected entries rather than guessing an arbitrary offset.
Trim, quiet-pause removal and playback-speed edits retime retained caption words with the exported video. Do not subtract the trimmed opening from every source timestamp by hand. Review the prepared export again after these changes, especially around a cut or a quiet word.
For a first caption review, simplify the job: check the wording and source timing before combining several pace edits. Then inspect the edited result. That makes it easier to identify whether a mismatch came from the transcription, a correction or the final cut.
Include captions in the video you share.
Enable Show captions in export and choose Apply. Back in the review screen, prepare the Export, then use Preview export in Trim and edit when that prepared export is available. You can also save to Photos and inspect the saved file before posting.
These captions become part of the exported picture. Viewers cannot turn that rendered text off as if it were a separate closed-caption track. This app workflow does not provide an SRT or VTT subtitle-file export step. If your destination needs an editable caption track, prepare it separately with a tool that supports that format.
Choose the final crop before judging readability. Check whether captions obscure a face, a product label or a teaching example. The portrait and landscape guide separates reading layout, recording orientation and export crop. A supported up-to-4K recording does not guarantee readable text in every smaller viewing context.
If you publish on YouTube, inspect its caption track as a separate task. YouTube’s automatic-caption guidance tells creators to review machine-generated text. A reviewed app export does not verify a later platform-generated track.
Use a three-pass caption check.
- With sound: compare words with the actual recording, especially names, numbers, negation and improvised phrases.
- Without sound: check whether the text communicates the explanation at a useful viewing size and whether important sound context is missing.
- In the final export: inspect the opening, closing and edited transitions, plus text visibility in the intended crop.
Download the English caption-review checklist. This is a planning worksheet, not a subtitle file and not an assertion that any video has passed an accessibility audit.
If generation repeatedly fails, check the permission, selected spoken language and recording clarity before retrying. A timeout or unverifiable timing should not be treated as successful captioning. Keep the original take and resolve the issue before sharing an inaccurate result.
For the recording stage, follow the Duo setup guide and rear-camera recording guide. To compare the complete workflow, visit the iPhone Duo teleprompter buying guide or Teleprompter Duo app destination.