Privacy Policy

Vampro Voice Generator Text-to-Speech

Last Updated: June 2026

Vampro Voice Generator is a local text-to-speech tool for Adobe Premiere Pro. We designed it to keep your content on your own machine. This policy explains what data stays local, what uses the network, and the controls available to you.

What Stays on Your Device

Everything that matters stays local — your text, your audio, and your projects never leave your computer.

All voice synthesis happens locally. The text you enter and the audio you generate are processed entirely on your computer by the companion service running on localhost (127.0.0.1). Your text and audio are never transmitted to us or any third party.

Generated audio stored locally. Audio files are saved in the app's generated_audio folder on your device and are never uploaded anywhere.

Voice settings & project metadata. All configuration, voice presets, speed/pitch settings, and project metadata remain on your local machine.

What Uses the Network

The app makes limited, transparent network requests — none of which involve your content.

Update Checks

The app periodically contacts vampro.in to check whether a newer version or new voice models are available. These requests reveal your IP address, approximate time, and app version — like any web request. No content from your projects is sent.

Model Downloads

If you accept (or have auto-update enabled), the service may download new voice models from vampro.in and/or Hugging Face. Only model files are downloaded; nothing is uploaded.

First-Run Fallback

If the offline model bundle is not present, the service downloads the default voice model once from Hugging Face. Standard installations ship the model offline and skip this step entirely.

What We Do Not Collect

No accounts or sign-ups

No telemetry or analytics

No advertising identifiers

No recording of your text or audio

No usage tracking

No project data collection

Your Controls

Disable network access

Set the environment variable VAMPRO_FORCE_OFFLINE=1 — the service then makes zero network calls, including update checks and model downloads.

Delete generated audio

Remove files from the generated_audio folder at any time. The app never auto-deletes your files.

Local diagnostic logs

The service writes logs to its logs folder with automatic rotation and 14-day retention. Logs stay on your device and are never uploaded.

Data Storage & Sharing

Data Storage

Generated audio, voice settings, and metadata are stored locally on your device. We have no access to this data.

Data Sharing

We do not sell or rent personal information. Disclosure occurs only when required by law.

Third-Party Services

Model downloads are served by Hugging Face; their privacy practices apply to those requests.

Children's Privacy

This software is not intended for children under 13.

International Users

Users are responsible for compliance with their local data protection laws.

Policy Changes

This policy may be updated. New versions will be published on the official website.

Security & Verification

Vampro Voice Generator runs entirely on your computer, processes your text locally, and does not upload your content.

What the App Does on Your Machine

Runs a local service on 127.0.0.1:8000 — not reachable from the internet
Processes text → audio locally; your text and audio never leave the device
Network is used only to check for and download updates and models
Reasonable security measures are implemented across the service

Third-Party Open-Source Components

Vampro Voice Generator is built on open-source software and open voice models. Below is a summary of the key components and their respective licenses.

Voice Model & Speech Engine

Kokoro-82M — Neural text-to-speech model

Apache-2.0

kokoro — Python TTS pipeline

Apache-2.0

misaki — Grapheme-to-phoneme front end

MIT

Text-to-Phoneme (Permissive Replacements)

DeepPhonemizer — Neural G2P engine for out-of-dictionary words

MIT

OpenPhonemizer checkpoint — G2P model weights

BSD-3-Clause-Clear

inflect — Number/ordinal/year spelling

MIT

spaCy + en_core_web_sm — NLP processing

MIT

Runtime & Libraries

Component	License
PyTorch (torch)	BSD-3-Clause
NumPy / SciPy	BSD-3-Clause
FastAPI	MIT
Uvicorn	BSD-3-Clause
Pydantic	MIT
Loguru	MIT
huggingface_hub	Apache-2.0
certifi (CA bundle)	MPL-2.0

Adobe® and Premiere Pro® are trademarks of Adobe Inc. Vampro Voice Generator is an independent product, not affiliated with or endorsed by Adobe.

support@vampro.in