Shinkansen - Blazing fast web & YouTube translator par Jimmy Su
Shinkansen uses AI models to translate web pages and YouTube subtitles. Text is replaced in place to preserve the page layout, and connects directly to the AI model â your data never passes through any third party.
108 utilisateurs·trices108 utilisateurs·trices
MĂ©tadonnĂ©es de lâextension
Captures dâĂ©cran
Ă propos de cette extension
Recently added:
âą New floating button: tap to translate the page, long-press to switch translation engines or open the menu
âą New Send to Instapaper: save the entire translated article to Instapaper in one click for later re-reading
âą iOS / iPadOS version coming soon to the App Store â stay tuned
Highlights:
âą Privacy-first â connects directly to Google, no third-party servers in between
âą Open source â the code is public on GitHub, so anyone can audit its security
âą Multilingual translation + multilingual UI â Traditional Chinese / Simplified Chinese / English / Japanese / Korean / Spanish / French / German
Web page translation
âą Instant streaming: the page starts changing into your language within 1 second of pressing translate â no waiting for the full API batch to return (Gemini only)
âą Translated paragraphs are cached automatically, so re-translating costs nothing
âą Two display modes: "Monolingual" replaces text in place; "Bilingual" shows the original and translation side by side
âą Edit translations: after translation completes, edit the translated text directly on the page to polish rough spots before exporting to PDF
âą Multilingual translation + multilingual UI â Traditional Chinese / Simplified Chinese / English / Japanese / Korean / Spanish / French / German
âą Translate-the-opening option: translate only the first N paragraphs as a preview, saving a large amount of tokens
PDF translation (beta)
âą Analyzes the layout locally, sends the text for translation, and presents the result in a side-by-side two-page comparison view
âą The entire document is processed inside your browser â nothing is ever uploaded to any server
âą Edit PDF translations: edit the translated text paragraph by paragraph inside the reader, then save to regenerate the two-page preview
âą Document glossary: before translating, ask the AI to extract a glossary of names and proper nouns from the document; confirm or edit it, then translate â keeping terminology consistent throughout the entire document
YouTube caption translation
âą Translates YouTube captions in real time, with caption styling that matches native YouTube
âą AI smart sentence segmentation: auto-generated (ASR) captions are re-segmented by meaning, turning fragmented short lines into readable full sentences
âą Bilingual captions: original + translation shown on two lines at once â great for listening practice
Three translation shortcuts / three engines
âą Alt+A / Alt+S / Alt+D each bind to a preset translation engine
âą Gemini AI (best quality, requires an API key)
âą Google Translate (free machine translation, no API key needed, faster)
âą Custom AI model (OpenAI-compatible â connect to hundreds of models via OpenRouter / Claude / DeepSeek / Groq / local Ollama, and more)
Translation quality control
âą Fixed glossary: force specific terms to always be translated the way you specify (global + per-domain layers)
âą Customizable banned-word list: explicitly tell the LLM which words it must never use
âą Document-wide terminology consistency: names and proper nouns stay consistent throughout long articles
âą Translate-the-opening option: translate the first few paragraphs as a preview before deciding whether to read on, saving a large amount of tokens
Auto-translate and cross-page continuation
âą Auto-translate specified sites: add a domain and pages are translated automatically on entry
âą Cross-tab continuation: open a link from a translated page in a new tab, and the new tab is translated automatically with the same preset
Usage tracking and savings reports
âą Records token counts, cost, and cache hit rate for every translation, with charts and CSV export
âą Two-layer caching (local cache + Gemini implicit context cache), with cache hit rate and actual savings shown right after each translation
âą New floating button: tap to translate the page, long-press to switch translation engines or open the menu
âą New Send to Instapaper: save the entire translated article to Instapaper in one click for later re-reading
âą iOS / iPadOS version coming soon to the App Store â stay tuned
Highlights:
âą Privacy-first â connects directly to Google, no third-party servers in between
âą Open source â the code is public on GitHub, so anyone can audit its security
âą Multilingual translation + multilingual UI â Traditional Chinese / Simplified Chinese / English / Japanese / Korean / Spanish / French / German
Web page translation
âą Instant streaming: the page starts changing into your language within 1 second of pressing translate â no waiting for the full API batch to return (Gemini only)
âą Translated paragraphs are cached automatically, so re-translating costs nothing
âą Two display modes: "Monolingual" replaces text in place; "Bilingual" shows the original and translation side by side
âą Edit translations: after translation completes, edit the translated text directly on the page to polish rough spots before exporting to PDF
âą Multilingual translation + multilingual UI â Traditional Chinese / Simplified Chinese / English / Japanese / Korean / Spanish / French / German
âą Translate-the-opening option: translate only the first N paragraphs as a preview, saving a large amount of tokens
PDF translation (beta)
âą Analyzes the layout locally, sends the text for translation, and presents the result in a side-by-side two-page comparison view
âą The entire document is processed inside your browser â nothing is ever uploaded to any server
âą Edit PDF translations: edit the translated text paragraph by paragraph inside the reader, then save to regenerate the two-page preview
âą Document glossary: before translating, ask the AI to extract a glossary of names and proper nouns from the document; confirm or edit it, then translate â keeping terminology consistent throughout the entire document
YouTube caption translation
âą Translates YouTube captions in real time, with caption styling that matches native YouTube
âą AI smart sentence segmentation: auto-generated (ASR) captions are re-segmented by meaning, turning fragmented short lines into readable full sentences
âą Bilingual captions: original + translation shown on two lines at once â great for listening practice
Three translation shortcuts / three engines
âą Alt+A / Alt+S / Alt+D each bind to a preset translation engine
âą Gemini AI (best quality, requires an API key)
âą Google Translate (free machine translation, no API key needed, faster)
âą Custom AI model (OpenAI-compatible â connect to hundreds of models via OpenRouter / Claude / DeepSeek / Groq / local Ollama, and more)
Translation quality control
âą Fixed glossary: force specific terms to always be translated the way you specify (global + per-domain layers)
âą Customizable banned-word list: explicitly tell the LLM which words it must never use
âą Document-wide terminology consistency: names and proper nouns stay consistent throughout long articles
âą Translate-the-opening option: translate the first few paragraphs as a preview before deciding whether to read on, saving a large amount of tokens
Auto-translate and cross-page continuation
âą Auto-translate specified sites: add a domain and pages are translated automatically on entry
âą Cross-tab continuation: open a link from a translated page in a new tab, and the new tab is translated automatically with the same preset
Usage tracking and savings reports
âą Records token counts, cost, and cache hit rate for every translation, with charts and CSV export
âą Two-layer caching (local cache + Gemini implicit context cache), with cache hit rate and actual savings shown right after each translation
Noté 5 par 4 personnes
Autorisations et données
Autorisations nécessaires :
- AccĂ©der Ă lâactivitĂ© du navigateur pendant la navigation
- Accéder à vos données pour tous les sites web
Autorisations facultatives :
- Accéder à vos données pour tous les sites web
Collecte de données :
- Le dĂ©veloppeur indique que cette extension nâa pas besoin de collecter de donnĂ©es.
Plus dâinformations
- Version
- 2.0.9
- Taille
- 10,25Â Mo
- DerniĂšre mise Ă jour
- il y a 3 jours (9 juil. 2026)
- Catégories associées
- Licence
- Licence MIT
- Historique des versions
- Ajouter Ă la collection