Subtitles

One module for every place a character talks. Try the six styles, move the playhead, load or export a subtitle file, and watch the reading speed.

0:00.0 / 0:00.0

Bars show each line by speaker (shape marks). Red-outlined bars read too fast.

Style
Language
Place
Labels
100%
default
Captions

Size, background and captions on/off are saved in this browser only (localStorage). Your system "reduce motion" setting is honoured automatically.

Now showing
-
Scene check
-

    Add a line

    Import and export

    Storyboard thumbnails

    The same lines drawn on 2D canvases with drawSubtitleToCanvas, no DOM: what the Previs storyboard strip would use.

    What is real and what is simulated

    • Real: line splitting, timing, the styles, reading-speed checks, SRT/WebVTT/JSON import and export, the accessibility transcript, saved settings, canvas drawing.
    • Simulated: there is no recorded audio. Word times are estimated from text length (syllable-like weights), except for lines you load from a WebVTT file that carries its own word stamps. The two faces are plain SVG stand-ins whose mouths follow the talking event; they are not the 3D avatar.
    • Play with voice uses your browser's built-in system voice only (speechSynthesis, no network). If the voice sends word boundary events the highlight follows them; if it does not, the module falls back to the estimated timing.
    • Not done: translation (the Japanese and Arabic lines are hand-written samples), speech recognition, and wiring into Previs, the chat widget or the avatar. See the README for the integration plan.