Dictozy: Voice Dictation

Privacy Policy

This policy explains what Dictozy handles to provide voice dictation, where that data goes, what stays in Chrome, and the controls available to you.

Effective date: November 1, 2026

Dictozy: Voice Dictation helps you dictate short text into supported web fields. Recording starts only when you click the visible microphone button or press the assigned browser shortcut. Dictozy returns the transcript for the field you selected.

Versions 0.1.12 and 0.1.13 use the same transcription backend and temporary upload handling described below. The practice page, temporary transcript recovery and personal vocabulary are features of 0.1.13. If you have 0.1.12 installed, your extension does not send vocabulary terms.

Data Handled

Dictozy handles the following data only to provide voice dictation:

  • Audio recorded after the user clicks the visible microphone button or presses the assigned browser shortcut.
  • The transcript returned from the speech-to-text service.
  • The enabled state, recording-duration preference, language-formatting preference, and origins the user explicitly disables stored locally with chrome.storage.local.
  • Optional personal-vocabulary terms entered by the user and a separate sharing preference, stored locally with chrome.storage.local. Vocabulary sharing is off by default.
  • Page field information inspected locally to determine whether the focused field is supported. The extension does not transmit the page URL, browsing history, existing field contents, or surrounding page content to the backend.

Sites are enabled by default. Dictozy stores an exact origin only when the user explicitly disables that origin. It does not retain a record for each site visited, and it does not use site preferences as browsing history. Dictozy does not activate on password or payment fields. Recording starts only after an explicit user action. In version 0.1.13, hiding the tab cancels the active recording.

How Data Is Used

Recorded audio is sent to the fixed Dictozy production backend solely to generate a transcript. The selected language-formatting code is sent with that user-triggered audio request. For an explicit language, the backend passes the code to xAI Speech-to-Text to guide written formatting such as numbers, currencies, and units. In Automatic mode, the backend omits the provider language and formatting parameters. The backend returns the transcript to the extension for insertion into the selected field.

Opening the optional practice page in 0.1.13 does not record or send audio. If you choose to dictate there, the same backend, xAI service and vocabulary preference apply. The practice field is not submitted as a form.

Saving vocabulary does not enable sharing. If the user separately enables personal vocabulary, saved terms are sent with subsequent dictation requests to the Dictozy backend and then to xAI as transcription hints. Terms may include names, organizations or other personal information the user chooses to enter. Dictozy does not extract terms from page contents or transcripts. Hints do not guarantee a particular spelling or improved recognition.

Site preferences, origins, URLs, and hostnames are not sent to Dictozy's backend, xAI, or an analytics service.

Data is not used for advertising, profiling, credit decisions, or sale to third parties.

Data Sharing

Audio, any selected explicit language code, explicitly enabled vocabulary terms, and resulting transcript data are processed by:

  • The Dictozy FastAPI backend hosted on Render.
  • xAI, which provides the Speech-to-Text service.

Render receives the user's connection IP address. xAI receives the backend's connection IP address; application code does not forward the user's IP to xAI. Providers may process technical request metadata under their respective policies. The extension does not request or collect GPS coordinates or precise device location. Dictozy does not send audio or transcripts to any other service for transcription. Text inserted into a website is available to that website.

Storage And Retention

The extension stores the enabled state, recording-duration preference, language-formatting preference, and exact origins the user explicitly disables in Chrome local extension storage. Site preferences are not synchronized to a backend and are not used for analytics, advertising, profiling, or browsing-history collection. These settings remain until the user changes or resets them, clears extension data, or removes the extension.

Your saved vocabulary list stays in Chrome local extension storage, without Chrome Sync. It can contain up to 100 terms of up to 50 Unicode characters each. Turning sharing off keeps the saved list but leaves it out of future requests. Clearing vocabulary deletes the local list and turns sharing off; removing the extension or clearing its data also deletes the list. None of these actions recalls a request already sent. The backend does not keep a vocabulary list or log the terms. Terms sent to providers are subject to their retention policies.

The extension and backend application code do not intentionally persist raw audio or transcripts as history. The backend may write audio uploads larger than 1 MiB to temporary files while processing a request. It closes and removes these files when the request ends, including on handled errors. If the server stops abruptly, cleanup follows the hosting platform's temporary-storage lifecycle. Dictozy does not create an audio archive. Render and xAI may retain data they process according to their own policies.

In 0.1.13, Dictozy compares the selected field's contents in memory to avoid overwriting your edits. It does not send or save that comparison. If insertion is unsafe, Dictozy can keep one completed transcript in the active tab's memory for up to 60 seconds. The recovery notice offers Copy and Discard without displaying the text. A successful copy, discarding, expiry, starting another recording, disabling Dictozy, hiding the tab or leaving the page clears that result. If copying fails, you can retry until the original deadline. Cancelled dictation is not recoverable. The 60-second limit applies to the recovery copy only: the destination site controls text inserted there, and your system's clipboard settings govern text you choose to copy.

To limit abuse, the backend temporarily counts requests by connection address and tracks overall audio usage in memory. Application request logs record a request identifier, method, route category, status and response time. They do not include audio, transcripts, vocabulary terms, field contents or client IP addresses. Hosting-provider logs are governed separately.

Security

Production audio requests use HTTPS. The xAI API key is stored only in backend environment variables and is never included in extension code. Browser permissions are limited to storage, supported webpage access required for dictation, and the fixed Dictozy production backend host.

User Controls

Recording begins only after the user clicks the microphone button or presses the assigned browser shortcut. The user can stop recording immediately with the visible control or shortcut. Users may turn Dictozy off globally, disable or re-enable the current supported site, reset all site preferences without changing unrelated settings, remap or remove the shortcut in Chrome, choose language formatting, change the recording limit, or clear extension data in Chrome.

Users can add, edit or clear personal vocabulary in the popup and independently enable or disable its transmission. Saving, editing or enabling vocabulary does not start recording or request microphone access.

Limited Use

The use of information received from Google APIs adheres to the Chrome Web Store User Data Policy, including the Limited Use requirements.

Changes And Contact

This policy will be updated if the extension's data handling changes. Questions or privacy requests can be submitted through Dictozy's first-party support page.