Language · English
Live English captions and translation
TanLens turns English speech into captions and translates it into any of 20 languages while you watch. It works with a browser tab, your microphone or an uploaded recording, and the transcript keeps both the original line and the translation.
Native name
English
English name
English
Speakers (approx.)
1.5B
Script direction
Left to right
Pipeline
Tuned prompt set
How to caption it
- 1Open TanLens in Chrome or Edge and choose the language as the source (or leave it on auto-detect).
- 2Start microphone captions, or install the extension and capture the tab that plays the meeting or video.
- 3Pick a target language. Captions show the original line and the translation underneath.
- 4When you stop, save the transcript and export SRT, VTT, ASS or TXT if you need a subtitle file.
Popular pairs
Spanish → EnglishSpanish speech is fast and regional accents vary widely; we keep the last ten sentences of context so names and tense stay consistent.English → SpanishEnglish idioms usually need rewriting rather than word-for-word translation; our prompts ask for natural Latin American Spanish by default.English → Chinese (Simplified)Chinese output is kept short enough to read at speaking speed; we avoid translating filler words that clutter captions.Chinese (Simplified) → EnglishChinese has no spaces, so segmentation happens before translation; we keep product and place names in their original form.Japanese → EnglishJapanese drops subjects often; the translator reconstructs them from the previous lines instead of producing ambiguous pronouns.English → JapaneseJapanese captions are formatted for vertical reading with short clauses, so they stay readable on mobile.Portuguese → EnglishBrazilian and European Portuguese differ in rhythm and vocabulary; pick the variant in settings for better names and numbers.Indonesian → EnglishIndonesian informal speech mixes in English loanwords; we keep them as-is instead of double-translating.Vietnamese → EnglishVietnamese tones matter for names; the ASR layer is tuned on Northern and Southern accents separately.Korean → EnglishKorean honorifics have no direct English equivalent; the translation keeps the register neutral and consistent.French → EnglishFrench liaison makes word boundaries hard for ASR; streaming partials help the recognizer recover mid-sentence.German → EnglishGerman separable verbs can finish a sentence late; we wait for the clause boundary before translating.
Honest limits
- English accuracy depends on the microphone, accent and background noise, and on the speech provider you configure.
- DRM-protected players can block tab audio capture. Test a specific title before relying on it.
- Captions are assistive. Do not use them as the only channel for legal or medical interpretation.