Back to blog

How to Add Subtitles in OpenShot (SRT and Auto Captions)

Add subtitles in OpenShot by pasting an SRT or VTT file into the Caption effect. Why cue numbers show up, how auto captions work, and a simpler way to get them.

November 15, 2026
Sablate Team

To add subtitles in OpenShot, drag the Caption effect onto a clip and paste the text of an SRT or VTT file into it. There is no command that imports a subtitle file, and whatever a Caption effect holds is drawn into the picture on export, so OpenShot always produces burned-in subtitles, never a track a viewer can switch off.

Automatic subtitles arrived with OpenShot 4.0 in August 2026, but they do not run inside the editor: they need a ComfyUI server on your own machine, set up by hand. For most people the quicker path is to transcribe the clip with another tool and paste the result in.

What the Caption effect can and cannot do

The Caption effect is OpenShot's only subtitle feature. Its documentation says it supports "both VTT (WebVTT) and SubRip (SRT) subtitle file formats", but what it actually takes is their text: a property called caption text, filled with cues in the usual form of a timing line followed by one or two lines of text.

What you get from it:

  • Timed cues from text. Paste a whole subtitle file and every cue appears at its own time.
  • A Captions panel where you edit all cues in one place, and an Insert Caption button that adds a three-second cue at the playhead.
  • Styling for the whole effect: font, size, colour, opacity, stroke, a background box with rounded corners, line spacing, and a fade in and out per caption.
  • Positioning by margins, set by dragging a white rectangle in the preview.

What it does not do:

  • No subtitle file import. You open the file in a text editor and copy its contents.
  • No subtitle output. Captions are burned into the exported frames. There is no SRT export and no subtitle track in the exported file, so YouTube gets nothing it can index or translate, and viewers cannot turn the text off. The difference matters more than it looks; hardsub vs softsub goes through which platforms want which.
  • No speech recognition in the editor itself. That lives in a separate AI workflow, covered below.

How to paste an SRT file into OpenShot

  1. Open the Effects tab and drag Caption onto the clip the subtitles belong to.
  2. Select the effect on the clip. The Captions panel opens with a placeholder cue in it; delete it.
  3. Open your .srt or .vtt in a text editor, select everything, copy, and paste it into the Captions panel.
  4. Scrub through the timeline. Each line should appear while it is spoken.
  5. Set the font, size and colours in the effect's properties, and drag the white rectangle in the preview to where the captions should sit.
  6. Export the video as usual.

A VTT file pastes cleanly. This is the form OpenShot's own documentation uses, a timing line and the text, with no numbers between cues:

WEBVTT

00:00:05.000 --> 00:00:08.200
Welcome back to the channel.

00:00:08.400 --> 00:00:11.000
Today we are fixing the audio first.

An SRT file puts a cue number above every timing line and writes a comma before the milliseconds (00:00:05,000). The comma is not a problem; OpenShot reads either separator. The numbers are, from cue 10 onward. If you only have an SRT, either delete the number lines before pasting or save the file as WebVTT in a subtitle editor first. The differences between the two formats are laid out in SRT vs VTT vs ASS.

Why do numbers or blank captions show up in OpenShot?

Both come from how OpenShot reads the pasted text. In libopenshot's caption code (version 1.1.1, the one in OpenShot 4.1.1), a caption's text is everything from its timing line up to the next thing that looks like a timestamp, and the only lines skipped are those one character long.

That has two side effects:

  • Stray numbers under captions. In an SRT, the number of the next cue sits between this cue's text and the next timing line, so it is read as part of this caption. Cue numbers 1 to 9 are one character long and skipped. From 10 on they are drawn, under the caption before them.
  • Captions cut short or empty. The parser looks for two digits, a colon and two more digits to find the next cue. A clock time in the text, such as 10:30, matches that pattern, and the caption is cut at that point or disappears. Write it as 10.30, or as "half past ten".

If the captions show up at the wrong moment instead, the cause is usually timing rather than parsing; see the section on trimming at the end.

Does OpenShot have auto subtitles?

Yes, since OpenShot 4, but not in a form most people will run. The feature is Add Captions from Speech under Enhance with AI, and it hands the work to ComfyUI, a separate AI workflow server you install on your own machine. OpenShot's AI documentation lists the setup:

  1. Install ComfyUI and confirm it starts.
  2. Install the custom nodes OpenShot needs, among them ComfyUI-Whisper.
  3. Download the model files into ComfyUI's model folders.
  4. Start ComfyUI, then in OpenShot open Edit → Preferences → Advanced, enter the ComfyUI URL and click Check.

The same page calls the AI features experimental and recommends, for the set as a whole, an NVIDIA 5070 with 12 GB or more, 64 GB of RAM and 200 GB of free space, warning that jobs will stall or fail below that. Speech recognition alone asks far less: the Whisper medium model this workflow uses needs about 5 GB of graphics memory, according to OpenAI's Whisper README. But you still install and maintain the whole server to reach it.

When it runs, the workflow transcribes the selected video with Whisper medium, language set to automatic, and saves an SRT. OpenShot stores that text on the file, and every clip you then add to the timeline from that file gets a Caption effect holding it. The SRT the ComfyUI-Whisper node writes is numbered, so the stray-number issue above applies to these captions too; check a long clip past its tenth cue.

Getting the subtitle text from another tool

Any tool that writes a subtitle file can feed the Caption effect. Two that work well, depending on your system:

Whisper on the command line

Free, open source, and on Windows, Mac and Linux. It needs Python and FFmpeg installed. Run it on the clip's source file and ask for VTT, which pastes without the number problem:

pip install -U openai-whisper
whisper interview.mp4 --model turbo --language en --output_format vtt

That writes interview.vtt into the folder you ran the command from. Change en to the spoken language, or drop --language to let Whisper detect it. The trade-offs between turbo and the other model sizes are in our Whisper model comparison. There is no editor here: correct mistakes in the text file or in OpenShot's Captions panel.

A subtitle app

A desktop subtitle app does the transcription, the corrections and the styling in one window. In Sablate, which we make:

  1. Drop the video or audio file in and pick a model: Fast or Balanced on the free plan, Accurate (large-v3-turbo) or Best (large-v3) on Pro.
  2. Correct names and punctuation in the editor while the video plays beside the text.
  3. Export a .vtt and paste it into OpenShot, or skip OpenShot's captions and burn the subtitles in from Sablate with one of its 13 styles, at 16:9, 9:16 or 1:1.

Sablate's .vtt export carries no cue numbers. On the free plan the .srt and .vtt exports end with a short credit cue, and you will see it as the last caption after pasting. The free plan needs no account and covers videos up to ten minutes, three jobs a month, the Fast and Balanced models and one offline translation layer; its burn-in carries a small corner mark. Once the speech model is downloaded, transcription runs without an internet connection. On Windows an NVIDIA graphics card speeds transcription up; Macs transcribe on the CPU. There is no Linux build, so on Linux the command line or OpenShot's own workflow is the way.

OpenShot captions compared with the other routes

Checked on 2026-11-15 against OpenShot 4.1.1 (released 2 November 2026), its documentation and source code, and OpenAI's Whisper README.

OpenShot Caption effectOpenShot Add Captions from SpeechWhisper command lineSablate
PriceFreeFreeFreeFree plan; Pro $29 one-time or $5.99 a month
PlatformsWindows, Mac and LinuxWindows, Mac and Linux, plus a ComfyUI serverWindows, Mac and LinuxWindows and Mac
Makes subtitles from speechNoYes, Whisper mediumYes, any Whisper modelYes, four Whisper tiers
SetupNoneComfyUI, custom nodes, model filesPython and FFmpegInstaller
Edits the video itselfYesYesNoNo
Subtitle file outNoSRTSRT, VTT, TXT and moreSRT, VTT, TXT; ASS and MKV tracks on Pro
Burned-in subtitlesYes, at exportYes, at exportNoYes; small corner mark on the free plan
TranslationNoNoInto English onlyInto other languages, as layers

OpenShot is the only one here that edits video, and Sablate the only one without a Linux build. Sablate is there for the part OpenShot leaves out: making the subtitles and their files.

Which route fits your video?

"Five lines on a short clip." Type them in OpenShot's Captions panel with Insert Caption. Nothing else to install.

"I already have an SRT." Paste it into a Caption effect, after deleting the cue numbers or saving it as VTT.

"Linux, a big NVIDIA card, and ComfyUI already running." Try Add Captions from Speech, and check a long clip for stray numbers.

"Linux, ordinary hardware." Whisper on the command line with --output_format vtt, then paste.

"Windows or Mac, a long video, a styled look, or subtitles in a second language." Make the subtitles in a subtitle app and either paste the .vtt or burn them in there. The full set of burn-in options is in how to burn subtitles into a video.

"The subtitles must be switchable on YouTube." Keep them out of OpenShot. Export the video clean and upload the .srt or .vtt beside it; how to make an SRT file covers writing one.

Keeping captions in sync when you trim

OpenShot times a Caption effect against the clip's source file, not against the timeline. In libopenshot, the frame number an effect receives is the frame of the source file, so a caption timed at 0:42 in the original recording stays on the words spoken at 0:42, wherever the clip sits on the timeline and however much you trim from its start.

That decides which file to transcribe:

  • One long clip, trimmed (a talk, an interview, a screen recording): transcribe the source file and paste into a Caption effect on that clip. Trims and moves keep the sync.
  • Many clips cut together: a subtitle file made from the exported video counts from the start of the finished cut and fits none of the clips inside the project. Either burn it into the exported video with another tool, or import the exported video into a new OpenShot project, add the Caption effect there and export again, which costs a second encode.

Finish the cut before you make subtitles from the exported video. A trim after that point shifts every caption after it.

OpenShot is a capable free editor with a caption effect that does exactly what it promises, provided the text arrives in the right shape. If you would rather not set up ComfyUI or a command line to make that text, download Sablate for Windows or Mac, drop in the clip, and paste the .vtt into OpenShot or burn the subtitles in directly.

Frequently asked questions

Can OpenShot import SRT files?

Not as a file. OpenShot has no command that imports a subtitle file, but its Caption effect reads SRT and VTT text. Open the file in a text editor, copy everything, and paste it into the Caption effect on your clip. The captions are then drawn into the picture when you export the video.

Does OpenShot have auto subtitles?

Yes, since OpenShot 4, through Add Captions from Speech in the Enhance with AI menu. It runs Whisper on a local ComfyUI server that you install and connect yourself, with the ComfyUI-Whisper nodes and model files. OpenShot calls its AI features experimental and recommends a high-end NVIDIA graphics card for them.

Why do numbers appear in my OpenShot captions?

They are the cue numbers from the SRT file. OpenShot's caption parser treats every line up to the next timing line as caption text and only skips lines one character long, so from cue 10 onward a number can be drawn under the previous caption. Delete the number lines or paste a VTT file, which has none.

Can OpenShot export subtitles as a separate track or SRT file?

No. A Caption effect is drawn into the video frames on export, so OpenShot subtitles are always burned in and cannot be switched off. For switchable subtitles, export the video without captions and upload the SRT or VTT file next to it, or put the file into an MKV as a subtitle track with another tool.

How do I change the font and position of captions in OpenShot?

Select the Caption effect and edit its properties: font, font size in points, text colour, stroke colour and width, background colour and corner radius, and a fade in and out per caption. To move the captions, drag the white rectangle in the video preview, which sets the effect's margins. The style applies to every caption in that effect.