Read what’s visible.
Select a screen area or player window. Kotoba uses text recognition and Apple Translation to translate visible text, including on-screen captions.
Project 03 · Kotoba Overlay
Follow the words.
Stay in the moment.
Translate the text on your screen, or turn application audio into translated subtitles. A native Mac project that keeps the words close to what you’re watching.
In development · macOS 27.0 required by the current project
KOTOBA OVERLAY日本語 → ENGLISH
Screen text + application audioOn-device recognitionTXT & timed SRT export
What it does
Two ways to listen.
More ways to understand.
Follow what’s spoken.
Select a browser or media-player app to create translated audio subtitles. Speech recognition runs on-device for supported languages and devices; microphone capture is disabled.
Keep the words.
Export text as TXT or audio subtitle cues as timed SRT. Choose original speech, translation, or both, and adjust the first subtitle timestamp for your export.
The workflow
Choose. Translate. Keep.
Choose your source.
Select a visible text area or an application’s audio. For browser audio, stop other playing tabs to avoid mixing sources.
Prepare your languages.
Choose source and target languages, allow the required capture permissions, and download Apple’s language models when prompted.
Follow along.
Read the translated overlay as results arrive. For audio, stop and finish subtitle processing before exporting. Live subtitles will lag playback.
Built around your Mac
Local by design.
Recognition runs on-device, with Apple Translation handling the language layer. Language models may need an internet download. Transcripts and translations stay in local session memory until you export them; raw audio is not saved.
Select a capture area that excludes sensitive content, and export anything you want to keep before quitting.