Media
5 items
Description
Context-aware AI subtitles for streaming video, synchronized with the stream's official caption track. AI Context-Aware Subtitle Translator Watch streaming shows with AI-translated subtitles that actually understand the show - not just the current line. Re-translates the show's own caption track rather than transcribing it from scratch. The source is whatever caption track the platform provides (English, Japanese, Korean, Spanish and more); the target is any of 20 languages, English included. Watching a Japanese drama with only Japanese captions? Get English subtitles that keep the characters' names and how they address each other straight. FREE EXTENSION - BRING YOUR OWN AI The extension itself is free to install and use, with no in-extension payments or paywalled features. AI translation requires access to a model. Use whichever option works for you: • An eligible ChatGPT plan through the Codex CLI bridge, or a Claude Pro or Max plan through the Claude Code bridge. • An API key for Anthropic (Claude), OpenAI, or Google (Gemini) - or an OpenAI-compatible endpoint you provide. • A local model you already run, such as Ollama or LM Studio. Translation requests go directly to it. • A DeepL API key, including a free one, as a lighter alternative to the AI backends. Switch engines from the popup. Your selected AI service's usage limits, pricing, and terms apply. The extension never bills you directly. WHY CONTEXT-AWARE TRANSLATION MATTERS Most subtitle translators work line by line, with no memory of who's speaking, how characters address each other, or what happened a few lines ago. That produces stilted, inconsistent dialogue: an honorific that changes mid-scene, a nickname mistranslated as a stranger's name, a pronoun sent to the wrong person. This extension corrects for that in two ways: 1. When a recognized title starts playing, it looks up character names and relationships from public reference sources (Wikidata, Wikipedia, TVMaze) and builds a "Series knowledge" profile per title and target language, so translations use the right name, honorific, and register from the very first line. 2. It also reads the surrounding dialogue, not just the current line, so ambiguous replies, callbacks, and continued sentences are translated in the context of the actual conversation. KEY FEATURES • Synchronized to the platform's own official caption track, not a separate, estimated timeline. • Bilingual overlay (original + translation) or translation-only, your choice, with adjustable subtitle size and position. • Works automatically on major subscription streaming platforms. • On YouTube and other sites with a video playing directly on the page, a small button appears on the video when captions are detected - click it, approve the site once, and translation turns on for that site. The toolbar icon does the same thing. • Choose whether to translate dialogue only, or to include sounds, actions, and song lyrics. • Export the AI translation as a source-timed SRT file for offline use, or import a subtitle file you already have and adjust its sync by hand. FOR VIDEOS WITH NO CAPTIONS AT ALL Optionally, the extension can generate subtitles from the video's audio instead. You start this yourself for a specific tab; it captures that tab's audio and sends it to the speech-to-text endpoint you configure - OpenAI's Whisper API by default, or any OpenAI-compatible endpoint, including one you host yourself. It never starts on its own and never runs in the background. Everywhere else, the extension works from the platform's existing caption track and does not touch the audio. WANT MORE SITES, AI PROVIDERS, OR LANGUAGES? Automatic support today covers major streaming platforms and the AI backends listed above. If you'd like another site, provider, or language supported, use the feedback link in the extension (or the Support URL) to request it. YOUR DATA This extension has no server of its own. Subtitle text, episode context, and translation context are sent directly from your browser, or through the local helper you choose, to the AI service you configure - never to the developer. If you turn on audio subtitles for a tab, that tab's audio goes only to the speech-to-text endpoint you configured, on the same terms. When a title is recognized, the title alone may also be queried on public reference sources (Wikidata, Wikipedia, and TVMaze) to build Series knowledge. Subtitle and dialogue text are never sent to those sources. See the privacy policy for full details. Recorded (on-demand) video only. Live streams and video calls are not supported.