Local-First Architecture & BYOK Disclosures

Privacy Policy

Last updated: August 27, 2026 • Effective immediately

Summary of Core Privacy Principles

Local-First Architecture: By default, your video clips, voice audio files, scripts, and generated captions never leave your local computer. If you choose to enable optional third-party cloud AI features, only the specific text you submit is transmitted.
Zero Cloud Model Inference by Default: Core AI engines (Gemma 4 E2B, Kokoro/Piper TTS, Whisper WASM, HY-MT1.5) execute entirely on your local hardware.
No Data Monetization: We do not collect your personal content, sell your data, or license information to advertisers or AI training brokers.

1. Introduction

JustAddVoice ("we," "us," or "our") is dedicated to protecting your privacy. This Privacy Policy details how data is handled when you download, install, and use the JustAddVoice desktop software application (the "Software") and visit our website at JustAddVoice.com (the "Site").

JustAddVoice is engineered from the ground up as a local-first desktop software application. Unlike web-based video and AI generation platforms, our core architecture processes your media files entirely on your personal device.

2. Data We Do NOT Collect

Because JustAddVoice runs on your local workstation, we have no access to your media assets. Specifically, we do not upload, collect, intercept, or store:

Video and Audio Files: Any MP4, MP3, WAV, or other media files you import into the Software.
Scripts & Narration Text: Any text, prompt, or book chapter you type or paste into the Software.
Generated Subtitles & Transcripts: Word-level timestamps, SRT files, or burned captions produced by the Whisper engine.
Voice Model Outputs: Synthetic voice waveforms created by the Kokoro or Piper neural engines.
Rendered Output Videos: High-bitrate MP4 video files composited on your local GPU/CPU.

3. Local Artificial Intelligence Processing

All default artificial intelligence capabilities in JustAddVoice execute offline on your device using bundled or locally downloaded model weights:

Google Gemma 4 E2B: Runs locally via node-llama-cpp to structure scenes and extract visual keywords without sending text to cloud LLMs.
Kokoro-82M & Piper TTS: Neural voice inference runs directly on your CPU/GPU.
WASM Whisper: Audio-to-text transcription executes in local sandboxed memory.
Tencent HY-MT1.5: Multilingual translation executes entirely offline.

4. Network Connections & Optional Cloud Features

The Software only initiates network requests in the following specific scenarios:

Stock Footage Discovery (Pexels & Pixabay): If you choose to search for external B-roll footage, the Software transmits your search keyword to the respective API (Pexels or Pixabay) to retrieve candidate video thumbnails and download selected video files. Video files are stored directly in your local disk workspace.
License Verification: To activate your commercial license, the Software communicates with our Merchant of Record, Lemon Squeezy, to validate your license key.
Software Update Checks: The Software periodically queries our public GitHub releases repository to determine whether a newer version of JustAddVoice is available for download.
Crash Reports (Optional): If an unexpected application crash occurs, anonymized technical telemetry (stack traces, operating system version) may be transmitted to assist our development team in resolving bugs.

4.1 Optional Third-Party AI Integrations (Bring Your Own Key)

JustAddVoice offers the option to connect third-party cloud AI models (such as OpenAI, Anthropic Claude, XAi, or Google Gemini) to accelerate specific text-generation tasks. These integrations are strictly optional and disabled by default.

If you choose to use these features by providing your own API key (a "Bring Your Own Key" or BYOK model):

Data Transmission: Only the specific text prompt you submit within that feature is transmitted directly from your local machine to the chosen third-party AI provider's API endpoint.
Zero App-Side Telemetry: We do not route these requests through our servers. We do not intercept, log, or store your prompts or the AI provider's responses.
API Key Storage: Your API keys are stored securely on your local operating system. They are never transmitted to JustAddVoice servers.
Third-Party Privacy Policies: When you use a third-party AI provider, the processing and retention of your text prompts are governed exclusively by the terms of service and privacy policy of that provider (e.g., OpenAI, Google, Anthropic, or XAi), not by JustAddVoice. We encourage you to review their specific data retention policies regarding API usage.

5. Payments and Billing

All purchases, payments, and invoice processing are managed by our Merchant of Record, Lemon Squeezy (Lemon Squeezy LLC). We do not collect, process, or store your credit card details or payment credentials on our servers. When you make a purchase, you provide payment information directly to Lemon Squeezy, governed by their Privacy Policy.

6. Information Sharing & Third Parties

We do not sell, rent, trade, or monetize any user data. We do not partner with advertising networks or data brokers. Data shared with service providers (such as Lemon Squeezy for licensing validation or GitHub for software downloads) is strictly limited to fulfilling the operational needs of the Software.

7. Your Rights and Inquiries

Depending on your location (including the European Economic Area under GDPR and California under CCPA), you have rights regarding your personal data. Because we do not store your media files or account content on cloud servers, the vast majority of your data is entirely under your control on your local device.

For questions regarding this Privacy Policy or your license data, please contact our support team at [email protected].