Privacy Policy

Overview

AI Voice Reader is a browser extension that reads web pages and selected text aloud using neural AI voices. It is designed with privacy as a core principle.

Data Collection

AI Voice Reader does not collect, store, transmit, or share personal data from your reading activity.

• No personal information is collected.
• No analytics or tracking inside the extension.
• On the free tier, all voice synthesis happens entirely on your device.
• No reading data is sent to any server we operate at any time.

Free Tier (Local Voice Synthesis)

The free tier uses an open-source neural voice model (Kokoro) that downloads to your computer the first time you install the extension. After download:

• Page text and selected text are processed entirely within your browser.
• Voice synthesis runs locally using your CPU or GPU.
• No page content is transmitted over the network for synthesis.

The one-time model download fetches the voice model file from Hugging Face’s content-delivery network, the standard public host for open-source AI models. After that, the extension runs offline.

Pro Tier (Bring Your Own API Key)

If you upgrade to Pro and add an API key for a third-party voice provider (ElevenLabs, OpenAI, Cartesia, Deepgram, Microsoft Azure, Google Cloud, or AWS Polly), the following happens:

• The text you choose to read is sent directly from your browser to your chosen provider’s API, using your API key.
• The provider returns synthesized audio, which the extension plays locally.
• This network request does not pass through any server we operate.
• Each provider has its own privacy policy and data-retention practices; please review your provider’s policy directly. We are not involved in that data relationship.

You can remove an API key at any time in Settings; the key is stored locally in your browser’s extension storage and is never transmitted to us.

Permissions

The extension requests the following permissions, each used solely for its core reading functionality:

• activeTab. To read the text content of the page you have open when you trigger the extension.
• contextMenus. To provide the “Read aloud” right-click menu item on selected text.
• offscreen. To create a background document for voice synthesis and audio playback (required by Chrome’s Manifest V3 architecture).
• storage. To remember your voice, speed, and reading-position preferences, and your Pro license key. All stored locally in your browser.
• unlimitedStorage. To cache the downloaded neural voice model in IndexedDB without risk of LRU eviction.
• scripting. To inject the content script that extracts readable text from the current page.

Pro License Activation

If you purchase the optional Pro upgrade, the following user-initiated network requests are made:

• Clicking “Upgrade to Pro” opens a Stripe checkout page in a new browser tab (outside the extension). Payment data is handled entirely by Stripe and never passes through the extension.
• Entering a license key sends it to simple-ai-capture.com for one-time validation. Only the license key string is transmitted; no other data.
• The “Lost your key?” feature opens the AI Voice Reader website in a new tab where you can look up your key by email. This happens on the website, not within the extension.

No network requests for synthesis are made on the free tier, except the initial one-time voice-model download from Hugging Face.

Third-Party Services

The neural voice model used on the free tier is Kokoro, an open-source model licensed under Apache 2.0. It is hosted at Hugging Face and downloaded once on first use, then cached locally. If you configure a Pro provider, audio synthesis is performed by that provider’s API on its servers. No analytics, tracking, or advertising SDKs are included in the extension.

Changes to This Policy

If this privacy policy is updated, the changes will be reflected on this page.

Contact

If you have questions about this privacy policy, please email [email protected] or visit simple-ai-capture.com.