Privacy Policy

Overview

Míngbai helps Chinese learners read Chinese subtitles on supported Netflix, YouTube, iQIYI and Viki videos. Intelligent phrase parsing groups each subtitle line into phrase chunks with pinyin and learner cards, episodes become a recap and a short lesson, and optional AI notes appear when a learner explicitly asks for them.

This policy explains how the Míngbai website, account service and browser extension handle data. The data controller and operator is MINGBAI LTD, a company registered in England and Wales under company number 17362232, with registered office at 71–75 Shelton Street, Covent Garden, London WC2H 9JQ. For privacy questions, contact [email protected].

Míngbai does not sell personal information, does not use personal information for advertising, and does not use personal information to determine creditworthiness or for lending purposes.

Míngbai’s use and transfer of information received from Chrome extension APIs complies with the Chrome Web Store User Data Policy, including the Limited Use requirements. Míngbai uses extension data only to provide, secure, support and improve the phrase-first subtitle learning features described in this policy.

The extension uses hosted Míngbai services to provide some features. Subtitle text from supported Netflix, YouTube, iQIYI and Viki text subtitle tracks may be processed by Míngbai’s backend services to provide phrase chunks, pinyin, learner cards, the episode recap and lesson, account status, quota checks, support features and optional AI responses. On YouTube videos without Chinese subtitles, you can also choose to generate AI subtitles, described below.

YouTube

Míngbai uses YouTube API Services to show lists of YouTube videos and to let you search YouTube. When you search, the words you type are sent from our server to YouTube. We keep search results for up to 24 hours and count how many searches each account makes each day, to keep within YouTube’s limits. YouTube’s Terms of Service (https://www.youtube.com/t/terms) apply, and Google’s Privacy Policy (https://policies.google.com/privacy) explains how Google handles data. When you return from the YouTube app, Míngbai checks only whether something was copied, without reading it, and reads your clipboard only when you tap Paste.

Legal basis

We process personal information where it is necessary to provide and support Míngbai, where we have a legitimate interest in operating, securing, supporting and improving the Service, where we need to comply with legal obligations, or where you have given consent, such as when you choose to submit optional contact or support information.

Website and contact

The Míngbai website is a static website. It uses Cloudflare Web Analytics to understand aggregate website traffic and performance. Cloudflare Web Analytics is designed not to use cookies, localStorage or individual visitor fingerprinting for analytics.

Some contact or interest forms may link to Tally. If you submit a form, Tally processes the information you choose to provide under its own terms and privacy policy.

We use contact information to reply to messages, provide onboarding or service information, and understand interest in the product.

Account and sign-in

You can create or sign in to a Míngbai account using an email code, Google or Apple. Supabase handles authentication; Míngbai receives the verified identity needed to maintain your account. Apple may provide a private relay address. Ordinary extension use does not require an account; some account-backed features do.

We process your email, authentication-provider relationship, sessions, account status, entitlements, trial, quota and billing state to provide and protect the service. Website sign-in state is kept in browser storage through our account-session system; credentials are not placed in page URLs.

Subtitle processing in the extension

The extension is designed around supported video subtitle text and selected subtitle chunks. It is not intended to collect general browsing history or build a profile of websites you visit.

When you use Míngbai on a supported Netflix, YouTube, iQIYI or Viki page, the extension may read Chinese text subtitle lines, subtitle timing and limited source information needed to provide the learning feature. Subtitle text may be processed by Míngbai’s backend services to provide phrase chunks, pinyin and learner cards.

Browser-only Saved Materials, Known state, local lookup history, local Context Explanation cache entries, local Scene Q&A cache entries, local account UI state and local diagnostic/debug preferences are stored locally in the browser extension unless a specific feature described in this policy sends information to a hosted service.

The extension may call hosted API endpoints. Those API calls do not mean that the extension loads or executes remote JavaScript or WebAssembly from those services.

Playback time

The Free plan includes a daily allowance of intelligent phrase parsing. To measure it, the extension counts the seconds in which a supported video is playing in a visible tab with Míngbai active and subtitles showing. Paused, hidden or idle time is not counted. While you are signed in, about every 30 seconds the extension sends that number of seconds to Míngbai’s account service, which records it against your account. The count carries no subtitle text, title or page address.

Episode recap and lesson

When you watch an episode while signed in, Míngbai builds that episode’s recap and lesson on its servers. Usually after about 30 seconds of viewing, the extension sends the episode’s Chinese subtitle transcript with its timing, matching English subtitle lines where available, and the show and episode title, page address, platform and video identifier. Míngbai’s servers analyse the transcript, using MiniMax for part of the language analysis, and keep the results with your account.

Your lesson is planned from your own Saved words, Known words and review history, and your lesson answers are recorded so later lessons and reviews can adapt to you.

AI subtitles

On a YouTube video that has no Chinese subtitles, Míngbai can make AI subtitles from the video’s audio. When the extension finds no Chinese subtitles, it sends the YouTube video identifier to Míngbai’s account service to check whether AI subtitles are available for your account and whether a transcript already exists.

Audio is transcribed only when you choose Generate. Míngbai’s servers then retrieve the video’s audio from YouTube and transcribe it with MiniMax; the result may include a machine English translation, which is labelled as such. A transcript belongs to the video rather than to you: it is stored and can be reused when anyone generates AI subtitles for the same video. Each generation counts towards your account’s daily AI-subtitle allowance.

Pronunciation

When you are signed in, the speaker button beside a word or phrase on its card plays it aloud. Only when you click it, the extension sends that word or phrase’s Chinese text to Míngbai’s subtitle parsing service. If no recording of that text exists yet, Míngbai’s servers send only the text to MiniMax, our voice provider, to make the audio; no account details go to MiniMax. Recordings are stored by their text, not by who asked, so everyone who hears the same word gets the same recording. Your browser keeps recent recordings in memory until you sign out or leave the page.

Context Explanation and Scene Q&A

Context Explanation and Scene Q&A run only when you deliberately request those features. They are not automatic background AI processing.

When you request Context Explanation, Míngbai may send the selected word or phrase, the current subtitle line, limited nearby subtitle context, relevant learner-card context and account/quota request metadata to Míngbai’s backend. The backend may send a prompt to MiniMax, the current AI provider, and return the response to the extension.

When you request Scene Q&A, Míngbai may send your question, the current subtitle moment, limited nearby subtitle context, selected phrase or learner-card context and, if enabled in the UI, available show title or episode information. Scene Q&A is text-first; screenshot or frame upload is not part of the current flow.

Successful Context Explanation and Scene Q&A results can be cached locally in your browser so the same answer can be shown again without another provider request. Request handling, error handling and quota outcomes may also create operational metadata in Míngbai’s backend services.

If Míngbai changes its AI provider, this policy will be updated where necessary.

Saved Materials and exports

When you use account-backed Saved, Recall, Study and Progress, Míngbai stores learner data on its servers. This includes saved learning material and source context, recall units and review outcomes, Study sessions and progress, and information used to adapt your own learning. Episode features also process subtitle material and retain results and learner context needed to provide the lesson.

Browser-only Saved, Known and recent lookup history, display preferences and local AI answer caches remain on that browser profile. Account-backed and local data are separate. Signing in does not copy local items to your account automatically: the extension asks whether to add your browser-only Saved items to your account, and copies them only if you choose to. Known state, lookup history, display preferences and local AI answer caches stay on that browser.

Saved material may include Chinese text, pinyin, gloss, learner-card details and available source title, URL, sentence and timing. Extension CSV exports can include these fields. Under Account → Data & privacy, you can download the currently supported account export: account details, creation acknowledgement, access and billing summaries, usage and support reports. That export does not currently include account-backed Saved, Recall, Study or Progress; contact us for help accessing those records.

Product analytics and your choices

We use PostHog in the EU for limited product usage analytics. Account-linked analytics connects permitted usage and commercial events to a pseudonymous account so we can understand product journeys and returning use. Anonymous usage statistics count limited learning interactions without linking those events to an account or building a persistent personal usage history.

After signing in, go to Account → Data & privacy on the website. Improve Míngbai for everyone controls account-linked analytics. Share anonymous usage statistics separately controls personless collection for your authenticated account. Both default to On. If account-linked analytics is Off and anonymous statistics is On, limited learning-use events remain anonymous and commercial analytics stops. If account-linked analytics is On, permitted linked analytics continues even when anonymous statistics is Off. Turning both Off stops future discretionary Product Analytics for that authenticated account.

These account settings apply when Míngbai can recognise your authenticated account. Signed-out installations retain default anonymous collection. A signed-in choice does not change unrelated signed-out devices, erase earlier analytics or link anonymous history back to you. Re-enabling collection does not replay events suppressed while Off. For privacy requests or objections, including signed-out statistics, contact [email protected].

Product analytics excludes subtitle and phrase content, translations, AI prompts, programme titles, video identifiers, source URLs, email and raw account identifiers. We do not use session replay, DOM autocapture, advertising profiles or device fingerprinting for Product Analytics.

Essential processing

Analytics choices do not disable authentication, billing, access and quota enforcement, security, narrowly bounded service reliability diagnostics or learner data needed for your own Saved, Recall, Study, Progress and Episode experience. Those systems remain separate from discretionary analytics. Support reports are sent when you explicitly submit them.

Bug reports and diagnostics

If you submit a bug report, Míngbai may store the report with your account so we can investigate and respond. A report may include category, severity, platform, feature, what happened, optional expected behaviour, timestamps, safe diagnostics and support status.

Bug reports do not include subtitle text by default. The form may offer separate options to attach the current Chinese line or selected phrase. If you choose those options, that subtitle or phrase content is stored with the report. Screenshots are not part of the current bug-report upload path.

Diagnostics are designed to help identify extension version, browser class, feature status, request/error class, account/access state class, quota status and similar reliability information. Diagnostics and support logs are designed to avoid access codes, tokens, raw provider payloads, provider keys, database URLs, full prompts, provider outputs and screenshots.

Third-party services

Míngbai may use third-party service providers only as needed to operate the website, extension, billing and support process.

Current service providers include Tally for forms, Cloudflare for website hosting and website analytics, Supabase for authentication and account-related storage, Google and Apple for their sign-in services, PostHog for limited product analytics, Stripe for payment processing, billing management and transaction support, MiniMax for requested AI help, AI subtitles, pronunciation audio and server-side Episode recap and lesson analysis, and hosting/infrastructure providers that operate Míngbai’s backend services.

Stripe-hosted Checkout collects payment and billing-address information. Míngbai receives payment, subscription and billing-country status needed to provide and reconcile paid access; the Míngbai website does not collect card details.

These providers may process technical request metadata such as IP address, browser or device information, timestamps, request metadata and error data when their services are used. This information is used for reliability, security, debugging, support and abuse prevention, not advertising.

Míngbai is independent and is not affiliated with, endorsed by, or sponsored by Netflix, YouTube, iQIYI, Rakuten Viki, MiniMax, Tally, Cloudflare, Supabase or Stripe.

Retention and deletion

We retain account, learner, billing, usage, support and operational records as needed to provide the service, respond to requests, protect access and meet applicable obligations. Analytics retention and historical erasure are separate from switching future collection off.

You can request account deletion on the signed-in website Account page after identity verification. Deletion revokes existing sessions and stops future subscription renewal. You have seven days to cancel the request; the account page shows the exact deadline. After that deadline, account and server learner-data erasure proceeds irreversibly. Cancelling deletion does not restore automatic renewal, and unused paid time is not refunded.

Some records may need to remain for accounting, security or other applicable obligations. Where associated analytics-vendor deletion is required, it is handled separately and may remain pending; a request does not mean vendor erasure is already complete. Contact [email protected] to ask about access, export or erasure.

Deleting an account does not clear browser-only Saved, Known, history or answer caches. Export or clear those separately using browser or extension controls.

Your rights

You can ask to access, correct or delete personal information we hold about you, and you can object to certain processing. Contact [email protected].

If you are in the UK and are unhappy with how we handle your request, you can also complain to the Information Commissioner’s Office.

Some providers used for hosting, forms, email, authentication, AI services or infrastructure may process information outside the UK or European Economic Area. Where UK data-protection law requires safeguards for a restricted transfer, MINGBAI LTD uses an applicable international-transfer mechanism.

Contact

For privacy questions or deletion requests, contact Míngbai using the address below.

[email protected]