Privacy Policy

1. Who we are

Video Transcriber (the “extension”) and this website are operated by Uladzislau Karzo (“we”, “us”). For privacy questions and requests, write to support@video-transcriber.app.

2. Short version

• Transcripts, summaries, your questions to a video with their answers, and translations are stored only in your browser.
• For a cloud summary (the Summary tab) or a question to the video (Ask), we send the video title and the transcript text (not the audio) to Google Gemini through our server. We do not store or log them.
• For an AI transcript in fast mode (of a whole video, of a live stream that Chrome speech recognition cannot transcribe, or of a file on the Files page), the extension compresses the speech audio in your browser into short pieces. Our server sends them, with the likely language, the video or file title and the last words already transcribed, to Google Gemini to turn speech into text. It also sends each piece, with the language and the title, to DeepInfra, which runs the open Whisper speech model, for precise timestamps and as a backup if Gemini is busy or fails. A file from your computer is never uploaded. Neither we nor DeepInfra stores the audio or the text.
• To translate a transcript or remove filler words from it, we send its text to Google Gemini through our server. We do not store or log it.
• When an AI transcript uses Chrome speech recognition, we send the recognised text (not the audio) to Google Gemini through our server to add punctuation and capital letters. We do not store or log it.
• Chrome speech recognition runs on your device when Chrome has the language installed. Otherwise Chrome sends the video’s audio to Google’s speech service. Local Whisper, which the extension switches to by itself when Chrome’s recognition is unavailable, never sends audio anywhere.
• To apply daily limits, our server keeps a small account record: an anonymous id or your Google e-mail, and hashed IP addresses for abuse limits.
• We have no ads, no analytics and no tracking, and we do not sell data.

3. Data processed on your device

The extension stores the following in your browser’s extension storage. We cannot access it.

• transcripts (from the Captions or AI transcript tab), with the video title, page address and thumbnail link, any summaries you generated, your questions to the video with their answers, and translations of the transcript;
• the history of the Files page (My files): for each file from your computer that you transcribe, its name, size, format and duration, a small preview frame and its transcript (with the summaries, answers and translations above), and, where Chrome provides one, a reference to the file so you can reopen it with Chrome’s permission (the file itself is not copied); it can also show the site videos you worked on (an AI transcript, a summary, a question or a translation). You can remove items, clear the history or have old items deleted automatically;
• your settings (theme, font size, languages, engine choices, disabled sites, panel position) and the colour scheme last received from our server (section 4j);
• a random device identifier and your sign-in session;
• Local Whisper model files, if you use Local Whisper.

This data is removed when you uninstall the extension.

To show the panel and read captions, the extension runs on the pages you visit and reads the page’s video player and caption data. Caption files are loaded from the video site the same way its player loads them. This page data is not sent to us, except as described in section 4.

4. Data processed on our servers

Our backend runs on Supabase (database, authentication and server functions, hosted in the EU (Ireland)).

a) Account. The first time the extension needs our server (for example, the first AI transcript, summary or contact message), it creates an anonymous account (a random id) so daily limits can be counted. Opening the panel alone does not create one. Signing in with Google is optional, and no feature needs it: it keeps the same account, with the same daily limits, in every browser and on every device where you sign in. When you sign in, we receive your e-mail address, name, profile picture link and Google account id (scopes: openid, email, profile). The extension uses the e-mail to show who is signed in and to fill in the contact form.

b) Usage records. Video Transcriber is currently free: we do not count uses per video and keep no record of which videos you use it on. Daily limits are counted as described below, per account id and per salted hash of your IP address. We do not store raw IP addresses.

c) Cloud summaries and questions. When you generate a summary (short, detailed or by chapters) or ask the video a question, the extension sends our server the transcript text (the whole transcript up to about 380,000 characters; a longer one is shortened on your device first), the video title, the duration, the summary language, the video key and the device id, and for a question also the question and up to three earlier questions with their answers. The server checks your daily limits, then sends only the title, duration, time range, language, transcript text and, for a question, the question and those earlier questions and answers to the Google Gemini API and streams the answer back. We do not store or log the transcript, the title, the questions, the summary or the answers. For daily limits, we record the time of each summary or question with your account id and a salted IP hash.

d) Payments. Video Transcriber is currently free, and we do not take payments. If we introduce paid plans, we will update this policy before anything is charged.

e) Contact form. If you write to us, we store your e-mail address (as you typed it), your message, the extension version, your account id and a salted IP hash. If you tick “Attach technical details”, we also receive your browser name and major version, the interface language and the domain of the current site. We never receive the page address or the transcript. Resend also e-mails the message, with your e-mail address, your account id and these details (not the IP hash), to our mailbox, so we can read and answer it. Our mailbox is hosted by our e-mail provider, which also receives the e-mails you send to our support address (section 7).

f) Security logs. Supabase keeps technical logs (such as IP address and time of requests to its authentication service) for security and operations, under its own retention.

g) Punctuation for the AI transcript. When an AI transcript uses Chrome speech recognition, the extension sends our server the recognised text a few lines at a time, with up to about 600 characters of the text before them and 300 after (for context), and its language. The server sends only this text and the language to the Google Gemini API to add punctuation and capital letters, and returns the result. We do not store or log the text. For daily limits, we keep a count of calls per day for your account id and for a salted IP hash, deleted automatically after a few days.

h) AI transcript in fast mode. When you press “Transcribe audio” in the AI transcript tab on a video that is not live, or on a live stream that Chrome speech recognition cannot transcribe (it is missing or fails, or your device or browser cannot use it for a stream), the extension reads that video’s soundtrack in your browser (it fetches the audio from the video site the same way the site’s player does, and notes the addresses of the streams the player loads, in memory only), compresses it into short pieces of about two minutes (about 30 seconds on a live stream, from the moment you press the button) and sends them to our server, together with the video key, the device id, the likely spoken language and two short text hints that help spell names and join the pieces: the video or file title (with a live stream’s category when the page shows one; up to 200 characters) and the last words already transcribed before that piece (up to about 600 characters). The server forwards only the audio pieces, the language hint and these text hints to Google’s Gemini API to turn speech into text, and each audio piece with the language hint and the title hint to DeepInfra, which runs the open Whisper speech model. Whisper’s word timings make the timestamps precise and add any speech Gemini skipped, and its text is used instead when Gemini is busy or fails. The server then returns the text. We do not store the audio, the hints or the text; our server keeps only counters (seconds transcribed per day for your account id, for a salted hash of your IP address and in total) to enforce limits, deleted automatically after a few days. The resulting text is stored only in your browser. The extension never saves the video or its audio. For YouTube videos, our server sends Google’s Gemini API the public video address and a time range instead of audio, with the same hints; this path uses only Gemini.

On the Files page you can transcribe a video or audio file from your computer. The file is read in your browser and is never uploaded: only its compressed speech audio is sent, the same way and to the same providers, with the file name (without its extension, or the name you gave it) as the title hint and, in place of the video key, a key computed from a hash of the file that contains neither its name nor its content. Summaries, questions, translation and filler-word removal of a file’s transcript work as described for videos.

i) Translation. When you translate a transcript, the extension sends our server the text of its rows a batch at a time (with a little of the neighbouring text for context), the target language, the video key and the device id. The server sends only the text, the context and the target language to the Google Gemini API and returns the translation. We do not store or log the text. For daily limits, we keep a count of translated characters per day for your account id and for a salted IP hash, deleted automatically after a few days.

j) Colour scheme. The extension (when the browser starts, when you open its menu, and otherwise at most every 15 minutes) and this website (when a page opens) ask our server which colour scheme to show. The request contains no information about you or how you use the extension or the website; like any request, it shows Supabase your IP address. The answer is kept in your browser.

k) Filler-word removal. When you choose “Remove filler words”, the extension sends our server the text of the transcript’s rows a batch at a time (with a little of the neighbouring text for context) and its language. The server sends only the text, the context and the language to the Google Gemini API, which deletes filler words, stutters and repeats, and returns the result. We do not store or log the text. For daily limits, we keep a count of characters per day for your account id and for a salted IP hash, deleted automatically after a few days.

5. Chrome speech recognition and Chrome built-in AI

• The “Chrome speech recognition” engine uses the Web Speech API built into Chrome. When Chrome has an on-device language pack, recognition happens on your computer. Otherwise Chrome sends the tab’s audio to Google for recognition. This happens between Chrome and Google, under Google’s terms and privacy policy; we do not receive the audio.
• When Chrome speech recognition transcribes a live stream and hears sound but returns no words for a while, the extension checks whether Google’s server can be reached: one request to www.google.com that carries no data and no cookies. Only when it cannot be reached does the AI transcript of the stream switch to fast mode (section 4h).
• “Chrome built-in AI” (Gemini Nano) runs entirely on your computer, and nothing is sent to us or to Google.
• “Local Whisper” runs entirely on your computer. On first use, it downloads model files from Hugging Face, which sees a normal download request (such as your IP address) but no content from you.

6. Why we process data (legal bases)

• To provide the service you ask for (contract): accounts, AI transcripts (fast mode and punctuation), summaries and answers to your questions, translation, filler-word removal, support.
• Legitimate interests: preventing abuse (usage limits, hashed IPs), security, and delivering this website and e-mail sent to our support address.
• Consent: the optional technical details in the contact form, and signing in with Google.

7. Who receives data

We use the Gemini API on Google’s paid tier, under which Google does not use prompts and responses to improve its products. Under the Gemini API terms, Google keeps prompts and responses for a limited time only to detect and prevent abuse (violations of its Prohibited Use Policy) and for any legally required disclosures. DeepInfra, Inc. (United States) does not store or log the audio and text we send it, and does not use them for training. We use Cloudflare only to deliver and secure this website and to route e-mail to our support address, not for advertising or analytics. We do not sell personal data or share it for advertising.

8. International transfers

Our providers may process data in the United States, Singapore and other countries. Where required, transfers rely on the providers’ Standard Contractual Clauses or equivalent safeguards.

9. How long we keep data

• Data in your browser: until you delete it or uninstall the extension (the Files page’s history can also delete old items automatically).
• Account and summary-limit records: while your account exists, or until you ask us to delete them.
• Contact messages, uninstall feedback and e-mails to our support address: as long as needed to handle your request and follow-ups, or until you ask us to delete them.
• Transcripts, summaries, questions and answers of cloud summaries, text sent for punctuation, translation or filler-word removal, and audio pieces and text hints sent for an AI transcript in fast mode: not stored on our servers (Google keeps Gemini API prompts and responses for a limited time for abuse detection, section 7).
• Daily counters for punctuation (calls), AI transcripts in fast mode (seconds), translation and filler-word removal (characters), per account id and per salted IP hash: deleted automatically after a few days.

10. Security

Data is encrypted in transit (HTTPS). Database access is restricted by row-level security, so you can read only your own records, and only our server functions can write them. IP addresses are stored only as salted hashes. Your sign-in session is kept in extension storage that web pages cannot read.

11. Your rights

Depending on where you live (for example under the GDPR, the UK GDPR or US state laws), you may have the right to access, correct, delete or export your data, and to object to or restrict processing. You can also withdraw consent at any time. To use these rights, including deleting your account, write to support@video-transcriber.app from the e-mail address you signed in with, or describe your device if you never signed in. We answer within 30 days. You may also complain to your local data protection authority.

You can also act yourself: uninstall the extension to delete local data, sign out in the account menu, or use only the Captions tab, which never sends us audio or text.

12. Children

You must be at least 13 years old to use Video Transcriber. If you are under the age of digital consent in your country (up to 16 in some EU countries), use it only with your parent’s or guardian’s permission. Video Transcriber is not directed at children under 13, and we do not knowingly collect their data. If you believe a child under 13 has sent us personal data, write to support@video-transcriber.app and we will delete it.

13. Chrome Web Store User Data Policy and Google API Services

Video Transcriber’s use of information received from Google APIs, including Google Sign-In, adheres to the Chrome Web Store User Data Policy, including the Limited Use requirements, and to the Google API Services User Data Policy. In particular, we use this data only to provide and improve the extension’s single purpose (transcripts, transcription and summaries of videos). We do not transfer it except as needed for that purpose, for security, or to comply with law. We do not use or transfer it for advertising, and we do not sell it. No human reads it except with your consent, for security, or to comply with law.

14. This website

This website has no analytics, no ads and no trackers, and it sets no cookies. The only things it stores in your browser (in local storage) are your theme choice (light or dark) and the current colour scheme (section 4j). Our hosting provider (Cloudflare) processes standard server logs (such as IP address and pages requested) to deliver and secure the site (see section 7).

When you install the extension, it opens our Welcome Page (/welcome/ on this website) once. The page’s address carries the extension’s id and version, its interface language, whether its toolbar icon is pinned and its colour scheme, so the page can show the right step. While the page is open, it asks the installed extension directly, through Chrome’s extension messaging, whether the icon is pinned and which version and colour scheme it has; the answer stays in the page and is not sent to us. If you press “Transcribe a file”, the page asks the extension to open its Files page. The Welcome Page has no analytics, loads nothing from other sites and sends nothing to our backend; it only remembers, for that tab, that it has already played its welcome animation.

If you use the contact form on this website, it creates an anonymous account (a random id) with our backend; its session is kept in memory only and is not saved in your browser. The form sends your e-mail address, your message, the page language and the page path, which we store with that account id as described in section 4e.

After you uninstall the extension, Chrome may open our feedback page. If you answer there, the page creates an anonymous account (a random id) with our backend in the same way, and we receive the reason you pick, your comment, the extension version and interface language taken from the page address, the page path, and your e-mail address only if you enter one. It is stored like a contact message, with that anonymous account id (section 4e). If you don’t send the form, none of this reaches our backend.

15. Changes

If we change this policy, we will update the effective date above. If the changes are significant, we will tell you in the extension or on this website.

16. Contact

Operator: Uladzislau Karzo. Contact: support@video-transcriber.app.