Download
Free and open source · GPL-3.0

Write, time, translate and dub, in one editor.

Subtitld is a subtitle editor for Linux, Windows, macOS and Haiku. Time on the waveform, transcribe and translate with engines that run on your own machine, and dub with synthetic voices.

Download Subtitld See what it does

Linux · Windows · macOS · Haiku

Subtitld editing Tears of Steel: the list of subtitles on the left, the video with the current subtitle over it, and the timeline with the waveform below
Tears of Steel © Blender Foundation, CC BY 3.0
One editor, start to finish

From a silent video to a dubbed one, without leaving the timeline.

  1. 01
    Open

    A video, a subtitle file, or a transcript from a text or Word file.

  2. 02
    Transcribe

    Turn the speech into timed subtitles: all of it, a range, or the selection.

  3. 03
    Time

    Drag, snap and nudge each subtitle on the waveform, frame by frame.

  4. 04
    Translate

    Keep each language's translation alongside the original.

  5. 05
    Dub

    Speak every line with a synthetic voice, one per speaker, or record your own.

  6. 06
    Deliver

    Save to SRT, VTT, TTML and more; export the dub as WAV, FLAC or MP3.

Timing on the waveform

See the speech. Put the words on it.

Drag a subtitle to move it, drag its edges to retime it. It snaps to its neighbours and to a grid of frames or seconds, and the numeric keypad nudges it one step at a time.

The timeline: subtitles as blocks over the audio waveform, with the playhead in red and the player controls above
Snap

To neighbouring subtitles, to frames, to seconds, or to scene changes.

Slice · Merge · Nudge

Set a start or end to the playhead, split at the cursor, join with the next.

One undo per gesture

A whole drag, or a held nudge key, is a single step back.

Read the timing guide
The Speakers panel listing Thom, Celia, Barley, Captain and Old Thom with their faces, colours and speaking time, each on their own timeline lane
Speakers

Every voice gets a face, a colour and a lane.

Name the speakers and take their faces from the video. Each one gets a colour, a track of their own on the timeline if you like, and a count of how long they speak.

When you dub, every speaker keeps their own voice.

Vertical timeline

Work down the subtitles, like reading a script.

The timeline on its side, as a tab beside the video. Each subtitle is as tall as it is long, so gaps and overlaps stand out as you scroll.

The vertical timeline beside the video, with each subtitle drawn as a block whose height is its duration
Plain-text view

Edit everything as one document.

Switch to a text view of all the subtitles, as SRT, Markdown or JSON, and type. The timeline follows.

The plain-text tab showing the subtitles as SRT with line numbers
Quality check

Readable, by the numbers.

Check reading speed, duration, line count and line length. The usual limits come as defaults: 21 characters per second, 140 words per minute, 0.7 to 7 seconds, two lines of 42.

The quality check settings: characters per second, words per minute, minimum and maximum duration, lines and characters per line
Add-ons

Engines that run on your own machine.

Speech recognition, translation, voices and audio separation install from inside Subtitld. Each brings its own models, so the app stays small and you download only what you use.

Speech to text
  • whisper.cpp Offline
  • Vosk Offline
  • RealtimeSTT As you speak
  • AssemblyAI Cloud
Translation
  • Offline translation Offline

OPUS-MT models, no API key.

Voices
  • Piper
  • Kokoro
  • Supertonic
  • sanoTTS
  • Parler-TTS
With voice cloning
  • Qwen3-TTS
  • Coqui XTTS
  • F5-TTS
Audio
  • Audio separator

Split the voices from the music and effects, so a dub can sit on the original background.

Two online voices, Edge and Google, come built in.
Browse all add-ons
subtitld.cc

Someone may have subtitled this file already.

When a video opens, Subtitld can look up subtitles made for that exact file on subtitld.cc, the community hub. It sends a fingerprint, never the video or its name.

Visit subtitld.cc
  1. 01
    Match by fingerprint

    Subtitles for the very file you opened, not one that is merely called the same.

  2. 02
    Search, rate, open

    From a panel inside the app, straight into the editor.

  3. 03
    Publish your own

    Share what you made, and remove it again if you need to.

Formats

Opens what you have. Saves what they asked for.

All formats in detail
Interface languages
17
Operating systems
4
Add-ons in the catalog
14
Commits since 2020
1,560

Get Subtitld

Snap Store, Flathub, the AUR, AppImage, Microsoft Store, PyPI, and soon HaikuPorts.

Download

Keep it free

Subtitld has no ads, no account and no paid tier. Donations, translations and bug reports keep it going.

Support the project