How to auto caption videos in DaVinci Resolve

Generate captions after the picture edit is stable enough that cue timing will not immediately drift. Duplicate the timeline or preserve a clearly named pre-caption version, then confirm the frame rate, start timecode, language, and delivery range. Captioning the wrong timeline cleanly is still the wrong result.

Auto Caption analyzes timeline speech. Blackmagic Design's original Auto Caption feature guide warns that overlapping dialogue across audio tracks is a common failure case and recommends muting tracks that should not be part of the analysis. Before running it, mute scratch mics, duplicate channels, guide tracks, or translated dialogue that would make two versions of the same sentence compete. Keep the actual program mix unchanged by recording what you muted and restoring it after generation.

Open the target timeline on the Edit page. Set In and Out marks if you want to caption only a test or delivery range. Open the Timeline menu and choose Create Subtitles from Audio; some Blackmagic Design documentation calls the same workflow Auto Caption. If that command is absent, first confirm that you are running DaVinci Resolve Studio and that the correct timeline is active.

1. Select the spoken language rather than relying on a guess when the dialog language is known.
2. Choose the caption preset that is closest to the required delivery format. Styling can be corrected later; the text and timing come first.
3. Set the maximum characters per line for the frame and expected reading distance. Treat it as a starting constraint, not proof that every cue will read well.
4. Choose the available line-break and gap controls for your DaVinci Resolve version, then start generation.
5. Wait for analysis to finish and confirm that a subtitle track with actual cues appears over the intended range.

Blackmagic Design documents language, caption preset, and maximum characters per line as core Auto Caption controls. Labels and additional controls have changed across releases, so use the terms visible in your installed version rather than copying an old dialog setting blindly.

Play the test range at normal speed with the subtitle track visible. Do not begin by polishing font and color. First ask whether the correct speech became the correct words at the correct time. A visually polished track built from the wrong microphone is expensive to unwind.

If the source and settings are wrong, delete the test track and regenerate. If they are right and only a few cues need repair, keep the track and edit those cues. Repeatedly regenerating a nearly correct full timeline can erase useful manual fixes and create another source of truth.

Review in passes because each pass catches a different failure. Read once for factual text: names, numbers, negation, technical terms, and punctuation that changes meaning. Then watch the cut for entrances, exits, shot changes, speaker changes, and cues that disappear before they can be read. Finish with line breaks and placement against faces, lower thirds, and platform overlays.

• Search every occurrence of a corrected name or recurring product term; one accurate cue does not fix the rest.
• Split cues at a sentence, clause, pause, or speaker boundary when the automatic break hides meaning.
• Merge fragments that flash too briefly when the combined cue remains readable.
• Listen around every overlap, interruption, and off-camera voice instead of trusting the transcript view alone.
• Check captions over the busiest and brightest frames after the track style is applied.

DaVinci Resolve 21 adds word-level timing corrections. Blackmagic Design's DaVinci Resolve 21 New Features Guide documents Update Word Timings for a selected subtitle and Synchronize Word Timings for a subtitle track. Use those commands when text is correct but the highlighted or word-timed caption has drifted. They do not replace a playback check.

Native speech-to-caption generation is a DaVinci Resolve Studio feature. In DaVinci Resolve Free, create or receive a timed subtitle file outside the app, import it, place it on a subtitle track, and run the same text, timing, picture, and export review. Blackmagic Design's Edit page documentation lists timed-text support including SRT, TTML, XML, and embedded MXF/IMF subtitles, plus separate SRT, TTML, and VTT export. Exact format availability can depend on the installed release and delivery context.

If the project needs animated, branded captions rather than a native subtitle track, use the separate subtitle track versus Text+ guide. That format decision should happen before you spend time correcting a cue set in the wrong system.

A correct subtitle track can still disappear at delivery. On the Deliver page, confirm whether the job should burn captions into the picture, embed them where the format permits, or export a separate subtitle file. Select the intended track and language, render a short section, and inspect the actual output in a player that can show the chosen caption format.

Keep the corrected subtitle track and delivered file tied to the same timeline revision. If picture changes later, regenerate or retime only after you know which edit points moved. The next useful action is a short rendered proof from the densest dialogue section, not another pass through the generation dialog.