Oyelo — Privacy Policy

Last updated October 13, 2026

Oyelo (formerly Video Dub Translator) is a Chrome extension that translates the speech of any video, tab or microphone into the language you choose, as a voice and as text. This page explains, plainly, what data it touches and where it goes.

In short: the extension has no server of its own. It never collects, stores, or sells your data. To do its job, it sends the video's subtitle text, recognized speech, or audio directly from your browser to the third-party service performing that one step — nothing more. Two things can reach the developer, and only if you send them: the anonymous answer from the page that opens after you remove Oyelo, and your email if you join the Pro waitlist on oyelo.online (see below).

What the extension does not do

It does not collect or transmit any personal information to its developer.
It does not track or record your browsing history.
It does not use analytics, advertising IDs, or third-party trackers.
It does not sell or share user data with anyone, for any purpose.

What leaves your device, and where

Each step of a dub is a separate, narrow request sent straight from your browser to the service that performs it. None of it passes through, or is stored on, any server operated by this extension's developer — there isn't one.

Data	Sent to	When	Why
Subtitle or recognized speech text	Google Translate (translate.googleapis.com)	Default Translate Service	Translate the line into your target language
Text to be spoken	Google Translate TTS, or Soniox	Every dubbed line, with a free voice	Synthesize the dubbed voice
Text to be spoken, and your key	Google Cloud Text-to-Speech, or Soniox TTS	Only if you choose the Google Cloud HD or Soniox TTS voice and add your own key	Synthesize the dubbed voice with that provider
Tab, captured or microphone audio	Soniox (soniox.com)	"Video speech" source, or the "Translate Any Audio" window (tab, window, screen or microphone)	Real-time speech recognition and translation
The video's audio, and your key	OpenAI (api.openai.com)	Only if you choose the "OpenAI Live (voice)" Speech Service and have an OpenAI key	Translate the speech and speak the translation (real-time speech translation)
The video's audio, and your key	Google Gemini (generativelanguage.googleapis.com)	Only if you choose the "Gemini Live (voice)" Speech Service and have a Gemini key	Translate the speech and speak the translation in the voice you chose
Subtitle lines to translate, up to ten at a time, with the three lines before them (and their translations), the line after, the video's title (on Coursera, also the course's name), short notes on the forms of address and names used so far, and your key	OpenAI, Gemini, DeepSeek, or Claude	Only if you choose that engine as your Translate Service and add your own key	Translate using that provider's model, with the conversation around each line as context so names, terms and forms of address stay consistent
Recognized speech lines, a few at a time, the context already learned in that session, and your key	OpenAI, Gemini, DeepSeek, or Claude	Only during a "Video speech" dub or a "Translate Any Audio" session, and only if your Translate Service is one of these engines with your own key	Learn the topic, names and terms of the conversation, so they are recognized and translated correctly; in "Translate Any Audio", also tidy a spoken sentence before it is translated. What is learned is kept for that session only
The page's title, and its description when the page is the video's own page	Soniox, or Google Gemini for the "Gemini Live (voice)" Speech Service	When a "Video speech" dub starts	Tell the speech recognizer what the video is about, so names and terms in it are recognized correctly

Any API key you type into the popup is stored only in your own browser (chrome.storage.local) and is sent only to the provider it belongs to, exactly as you entered it — never anywhere else, and never to the extension's developer.

What's stored, and where

Your settings — dubbing source, languages, translate service, voice, volume mix, subtitle options, interface language, and any key you add — live locally in your browser via Chrome's extension storage. None of it syncs to a server the developer controls, because none exists.

Transcripts — from the "Translate Any Audio" window and from Subtitles and Video speech dubs: the recognised or caption text, its translation, when each line was said, and for a video the video's title and address — are kept in a database inside your browser, on your device only, so you can read them again from the History page. They are never sent anywhere. You can delete any of them, or all of them, from the History page; the extension keeps at most 100 sessions, deleting on its own the oldest ones you have not pinned; and removing the extension removes them all.

Usage counts — for each dubbing or "Translate Any Audio" session, how many characters were spoken by each voice, how many were sent to each translation service, how many seconds of audio were sent for recognition, and how long the session played — are kept in your browser's extension storage, on your device only. They hold numbers and the names of the services, never any text, address or title, and they are never sent anywhere. The extension keeps the latest 100 sessions, and removing the extension removes them.

Permissions this extension uses

The "Translate Any Audio" window uses no permission of its own: it opens Chrome's standard "Choose what to share" dialog, and hears only the tab, window or screen you pick there. If you press its Microphone button instead, Chrome asks for your permission to use the microphone, and the window hears that microphone only while it is listening. Microphone audio goes to the same speech service as any other captured audio, and nowhere else; it is never recorded or played back.

storage Remembers your settings and any key you enter, locally on your device.
offscreen Plays the dubbed audio when a page's own security policy blocks playback in the visible tab.
tabCapture Reads the current tab's audio for the "Video speech" dubbing source, when there's no caption track to read instead.
declarativeNetRequestWithHostAccess Sets one request header on one endpoint, so the extension can fetch a free, temporary Soniox key. It does not read or alter any other traffic.
alarms Refreshes that temporary Soniox key on a schedule, in the background.
contextMenus Adds Oyelo's items to the page's right-click menu: dub this video, open "Translate Any Audio", and open Oyelo. It reads nothing from the page you right-click.
host permissions (all sites) The content script checks any page for a video element and, on YouTube, Udemy, Coursera or Bilibili, a caption track. On a page or frame with no video, it does one check and nothing further.

Third-party services

Because requests go straight from your browser to these providers, their own privacy policies govern the data described above:

Google (Translate and its text-to-speech endpoint)
Soniox — speech recognition and its optional text-to-speech voice, including Soniox TTS with your own key
Google Cloud Text-to-Speech — only if you choose the Google Cloud HD voice and add your own key
OpenAI, Google Gemini, DeepSeek, or Anthropic — only if you configure one as your Translate Service
OpenAI — also for real-time speech translation, only if you choose the "OpenAI Live (voice)" Speech Service
Google Gemini — also for real-time speech translation, only if you choose the "Gemini Live (voice)" Speech Service

If you remove Oyelo

When the extension is removed, Chrome opens a short page on oyelo.online that asks why. The address of that page carries only Oyelo's version and the language of its interface; it carries no identifier.

Answering is optional, and nothing is sent unless you press Send. If you do, the page sends four things: the reason you picked, your comment if you wrote one (at most 500 characters), the version and the interface language. They are stored on Oyelo's server (Amazon Web Services, Singapore) with the time they arrived, and deleted automatically after 12 months. No IP address, browser details, cookie or identifier is stored with them, so an answer cannot be linked to you. The answers are read only in aggregate, to learn what to improve.

The Pro waitlist

The page at oyelo.online/pro/ lets you join a waitlist for a planned paid plan. Joining is optional and happens only on the website, never from inside the extension. If you join, the page sends your email address, the answers you chose (which offer, the most you would pay, how many hours you use Oyelo, how much natural voices matter — each optional), the page's language and your consent. They are stored on Oyelo's server (Amazon Web Services, Singapore), one entry per address, and deleted automatically after 12 months. No IP address, browser details or cookie is stored with them.

Your address is used for one thing: a single email when Pro opens. It is never sold or shared, and it is never added to a newsletter. The answers are read only in aggregate, to set the plan's price and hours. To be removed sooner, send the address to the project's Facebook page (below) and it will be deleted.

Children's privacy

This extension is not directed at children and does not knowingly collect data from anyone.

Changes to this policy

If this policy changes, the "last updated" date above will change with it.

Contact

Questions, feedback, or a bug to report: reach out on the project's Facebook page.