How to Transcribe Audio in DaVinci Resolve 21

First decide what you need: a searchable transcript you can edit from, or timed subtitles your viewers will read. In DaVinci Resolve 21 you transcribe Media Pool clips to find dialogue and assemble selects, and you use Create Subtitles from Audio to build a subtitle track. Both automated workflows require DaVinci Resolve Studio. Free can import, edit and export subtitle files, but it can't generate them from speech natively.

Menu names and limits here follow Blackmagic's latest DaVinci Resolve 21 Reference Manual and hold for Resolve 21.1.2 build 10, the current stable release, whose notes add a fix for selecting multiple titles and subtitles. Older tutorials often show shorter labels from earlier releases, so if a command seems to be missing, look for its full Resolve 21 path. Dialogue cleanup, routing, levels and audio-only delivery live in the DaVinci Resolve Fairlight Audio hub; this page sticks to speech-to-text, transcript editing and captions.

Choose the Right Transcription Workflow

A clip transcript is attached to source media. It helps you find what was said and build an edit from selected words. A timeline transcript represents the edit itself and can drive changes to it. Timeline subtitles are a third thing: separate caption clips you style and deliver. Transcribing a clip won't put a single caption on screen.

• Searching interviews or assembling selects? Transcribe the source clips.
• Captioning a finished edit? Create subtitles from the timeline audio.
• Want a readable text record? Export the transcript as .txt.
• Need timed captions? Export an SRT from the Transcription window, the subtitle track or the Deliver page.

How to Transcribe Audio Clips in DaVinci Resolve 21

1. Open the Media, Cut, Edit or Fairlight page and select one or more clips with speech in the Media Pool.
2. Right-click a selected clip and choose AI Tools > Audio Transcription > Transcribe. The Transcribe Audio icon in the Media Pool toolbar starts the same analysis.
3. Let the Transcribing Audio window finish. It shows how many clips are left, the analysis speed against real time and roughly how long remains, and you can cancel at any point.
4. Open the Transcription window. Each transcribed clip gets a small speech-balloon icon on its Media Pool thumbnail, and Resolve 21 adds a Transcription column in List view. The column shows a truncated preview, yet a search still finds words anywhere in the full transcript.
5. Click a word and the source viewer jumps to the end of it. Drag across text to set a source range, then use Place on Top, Insert or Append to cut that range into the timeline.

Click any word and press X to select its whole sentence as an In and Out range. Two icons in the lower left of the Transcription window keep a selection without cutting it, as the text-based editing section of the manual shows: Create Subclip asks for a name and saves the range as a subclip, and the Marker icon adds a duration marker for a selected range or a standard marker for a single selected word.

The Transcription window searches the selected clips with Contains or Full Word matching. In source-clip mode, deleting text only strikes it through. That range drops out of transcript export and later text-based edits, the recording itself stays intact, and Undelete brings the words back.

Resolve marks detected silence with ellipses, and Remove Silent Portions drops those ranges when you assemble selected dialogue into a timeline. Listen to every join afterward. A pause can hold a breath, a reaction or room tone the edit still needs.

To change an existing edit through its transcript, duplicate the timeline first. Transcribe its source clips in the Media Pool, then click the Timeline icon in the upper left of the Transcription window to enter Timeline mode; from then on Cut, Copy, Paste and Delete change the timeline itself. Check the selected range and play the result after each edit. If a newly added clip shows Transcription Pending, choose Options > Transcribe Missing Clips in Timeline.

Correct, Search and Export a Transcript

I'd use Edit for spelling and proper-name fixes and leave whole paragraphs alone. Search and Replace pays off when the same name is wrong across several clips. Resolve warns that swapping a long passage for an unrelated summary can make text-based editing unpredictable, because every word is still tied to a source time range.

The Export icon saves a .txt transcript. With speaker detection enabled, Resolve Studio's transcript export includes speaker names and timecodes for each speech block. The Options menu also has Export Subtitles, which writes an SRT and asks for a caption preset, maximum characters per line, number of lines and gap between subtitles.

Build a Rough Cut from a Script with IntelliScript

IntelliScript runs the same transcription engine against a script and builds the timeline for you. It's a Studio tool that lines up the take matching each line in script order and stacks the alternates underneath, and because it reads for context, an ad-libbed line with a similar phrase should still find its place. The IntelliScript section of the manual gives three steps:

1. Save the script as plain text (.txt) or a Final Draft (.fdx) file that holds only the spoken lines, with a full stop after each sentence.
2. Select every clip IntelliScript should consider in the Media Pool, right-click, and choose AI Tools > Create New Timeline Using IntelliScript.
3. Load the script file when Resolve asks for it and click Open. Once the transcription finishes, the new timeline holds the best match for each line on track one and the alternate takes on separate tracks, disabled by default.

I'd strip a screenplay down to its dialogue before loading it, because the manual warns that scene descriptions, parentheticals and character names will likely cause transcription issues. Resolve 21 added Final Draft (.fdx) import, but the file should still hold only what people say. Track one is a first pass. IntelliScript picks the "best" take by matching dialogue only, with no good-take metadata behind it, so enable the alternates and audition them on the lines that carry the scene.

How to Create Subtitles from Timeline Audio

1. Open the timeline on the Edit page. Set In and Out points if only part of it needs captions; leave them unset and Resolve analyzes the whole timeline.
2. Choose Timeline > AI Tools > Create Subtitles from Audio.
3. Pick the spoken language or Auto, select a Caption Preset, set Max Characters per Line, choose one or two lines, and set the frame gap. The documented default gap is 0 frames.
4. Click Create. Resolve writes the result to a subtitle track: it creates one if none exists, and if you already have several, it uses the highest.
5. Read every caption while you watch the program. Fix words, cue timing, line breaks, speaker changes and non-speech labels before export.

Overlapping dialogue on multiple audio tracks is the failure Blackmagic calls out by name. Mute any track that shouldn't be analyzed, generate the subtitles, then unmute it. If dialogue and music are hard to pull apart, work through the routing steps in Mute and Split Audio in DaVinci Resolve and generate the subtitles again. Blackmagic's manual says Create Subtitles from Audio always writes to the highest subtitle track, so if you've already corrected captions there, duplicate the timeline or export them first: right-click that track header and choose Export Subtitle.

Edit and Style the Generated Subtitles

Select a subtitle clip and fix its text in the Caption tab of the Inspector. Track controls set the shared typeface, size, color, position and outline. Turn on Customize Caption only when one subtitle or a few selected words need a local override; everywhere else, let the track style do the work.

Resolve 21 adds Find and Replace in the Subtitle Inspector and spell checking for text elements. The spell checker flags likely typos. It can't tell you whether names, numbers and technical terms are right, or whether a caption matches what the speaker said. That still takes a listening pass.

For word-level animated captions, Resolve 21 can analyze captions you typed or imported. Select a subtitle and click Update Word Timings in the Caption tab, or right-click a subtitle track header and choose Synchronize Word Timings to retime the whole track. Only the word highlighting moves, so a wrong word stays wrong.

Resolve 21 Transcription Changes That Matter

Blackmagic's Resolve 21 New Features Guide documents four additions that matter here: background analysis for transcription, transcript search inside IntelliSearch, word timing for external subtitles and the Media Pool Transcription column. The 21.1 update adds scripting access to clip transcriptions with speaker and timing data, which mostly matters if you automate exports.

Background Rendering is off by default. In DaVinci Resolve > Preferences > System > Background Tasks, enable background renders and select the Transcription task. On a machine that's short of memory, I'd leave it off. Pause background renders on foreground activity gives interactive work priority, so background processing is a scheduling choice; it won't make the analysis itself faster.

IntelliSearch needs its own model download and media analysis. Search Transcript Only limits results to words in audio transcripts, which suits exact phrases; Search All can mix transcript, visual and metadata results. A clip transcript doesn't mean the picture has been analyzed for IntelliSearch.

Languages and Speaker Detection

Project Settings > Subtitle and Transcription > Transcription Setup sets the transcription language. The Resolve 21 manual lists 14 base language entries: Danish, Dutch, English, French, German, Italian, Japanese, Korean, Mandarin, Norwegian, Portuguese, Russian, Spanish and Swedish. Mandarin comes with simplified and traditional models. Auto lets Resolve pick a model for you.

The Extended AI Transcription Support Package in Extras Downloads Manager adds 43 more named languages, among them Arabic, Finnish, Greek, Hindi, Polish, Romanian, Thai, Turkish, Ukrainian and Vietnamese. Create Subtitles from Audio uses the same base list, except the manual names Chinese as Simplified Mandarin, and the extended package supplies the additional languages there as well.

Speaker Detection is project-specific. Turn it on from AI Tools > Audio Transcription > Speaker Detection or from the Transcription window Options menu before you transcribe. Resolve then remembers each voice signature across clips in the project. Rename Speaker 1, Speaker 2 and the rest, and use Assign To when a segment lands on the wrong person. Reset Speaker Data clears the saved project-level voice assignments.

However confident a transcript looks, I'd still listen through the whole thing. Check names, numbers, negations and speaker changes against the recording. Noise, accents, overlapping voices and microphone distance all shift the result, and when Auto picks the wrong language, set the known language and review the regenerated text.

Why Transcribe Audio Is Missing or Produces Bad Results

• The command is missing: confirm the installed application says DaVinci Resolve Studio, then use the exact AI Tools > Audio Transcription menu path on a Media Pool clip.
• The transcript is mostly ellipses: play the selected clip and confirm the source clip itself carries audible dialogue. Speech you hear elsewhere in the timeline doesn't count.
• The wrong language appears: set the language manually in Project Settings > Subtitle and Transcription > Transcription Setup, then clear and regenerate the transcription.
• One speaker or track is ignored: check the actual source clip being transcribed. A timeline transcript analyzes the underlying media, and that input may differ from the finished mix you hear. For captions, use the separate subtitle-generation workflow and test a short range.
• Subtitle generation misses overlapping speech: mute the tracks Resolve should ignore and run Create Subtitles from Audio again, as Blackmagic recommends.
• An imported transcript is out of sync: the SRT timecode must match the clip. Resolve 21 can also align it for you with AI Tools > Audio Transcription > Import and Detect Timing on a Media Pool clip.

Before you clear or regenerate a corrected transcript, export a copy of its text and save the project. Clear Transcription deletes the saved transcript along with every correction, though the original recording is untouched. Pin down the source, the language and the workflow you picked before you touch routing or channel formats.

On an iPad, the chip decides whether these commands exist at all. Blackmagic's Studio and iPad features list gives iPads with M-series chips IntelliScript, Speaker Detection from Audio and Subtitles and Transcriptions from Audio through the Studio in-app purchase, and none of the three on A-series models. Its page and panel table marks Subtitles as desktop only, so I'd transcribe and pull selects on an M-series iPad and build, check and export the subtitle track on a computer. Setup and the Studio upgrade on a tablet are in the DaVinci Resolve for iPad guide.

How to Export SRT Subtitles

For a transcript attached to a clip, open the Transcription window Options menu and choose Export Subtitles. Resolve writes an SRT once you set the caption preset, line length, line count and gap. The Export icon does a different job: it saves a readable .txt transcript.

For a finished program, the Deliver page can write a separate subtitle file, burn captions into the rendered video or embed them in supported formats; since 21.1, QuickTime and MXF renders can carry multiple embedded captions. Pick whatever the destination asks for, render a short test and open the exported file before the full delivery. The complete subtitle-track workflow, import formats and styling controls are in DaVinci Resolve Subtitles and Captions, and final video settings are in the Resolve export guide.

Resolve Transcription vs. Premiere Pro

Both applications can assemble edits from transcripts, and in both, captions stay a separate delivery step. Premiere's Text-Based Editing guide documents transcript editing, filler-word handling and speaker filtering. I'd give a list of supported languages no weight when judging accuracy for your speakers.

Run the same sample through both and compare wording, speaker overlap and export requirements before you pick an editor. Licensing, collaboration, color and handoff differences are in the DaVinci Resolve vs. Premiere Pro comparison.

DaVinci Resolve 21 Transcription FAQ

Is audio transcription available in free DaVinci Resolve?

No. Native audio transcription, text-based editing and Create Subtitles from Audio require DaVinci Resolve Studio. In Free, type captions by hand or make a timed SRT with an external transcription tool, then import and review it on a subtitle track. An imported SRT still won't give you Studio text-based editing.

Can I transcribe one clip instead of the whole timeline?

Yes. Select the source clip in the Media Pool and choose AI Tools > Audio Transcription > Transcribe. To work with an existing edit, select its timeline in the Media Pool and transcribe it. Create Subtitles from Audio is a separate command for a subtitle track.

How many languages does Resolve 21 transcription support?

The manual lists 14 base language entries, with simplified and traditional Mandarin models, plus 43 named languages in the optional Extended AI Transcription Support Package. Choose the known source language manually when Auto selects the wrong model.

Can Resolve identify multiple speakers?

Yes. Speaker Detection assigns project-level speaker labels and remembers voice signatures across clips. Rename the labels and fix wrong segments with Assign To. Overlapping voices still need a careful listen.

Can I export a Resolve transcript as SRT?

Yes. In the Transcription window Options menu, choose Export Subtitles. For a finished timeline, export the subtitle track or use the Deliver page. The separate Export icon writes a .txt transcript instead.

Why is Create Subtitles from Audio missing?

It's a Studio-only command. Confirm the installed app is DaVinci Resolve Studio, open a timeline on the Edit page, and use Timeline > AI Tools > Create Subtitles from Audio.

Does Resolve transcription need internet access?

Native transcription runs on your computer through the DaVinci Resolve Neural Engine. Download any language or IntelliSearch packages you need before going offline. That covers transcription only: licensing, cloud projects, shared storage and third-party tools have their own network requirements.

Jason Miller I run DaVinci Resolve Club, an independent publication. My guides explain editing, color, Fairlight, Fusion and delivery, with clear Free vs Studio limits.