QuietTake Audio & video tools

Open source licences

The noise removal runs in your browser, which means we send you the engine itself. Here is what it is made of, and the notices its licences require us to pass along.

DeepFilterNet

The noise removal is DeepFilterNet 3. We compile its libDF library to WebAssembly ourselves and load its published model weights at runtime.

The model weights come from the models/ directory of that same repository and are covered by the same notice. We did not train them, and we have not modified them. 

The MIT License (MIT)

Copyright (c) 2026 QuietTake
Copyright (c) 2021 Hendrik Schröter

Permission is hereby granted, free of charge, to any person obtaining a copy of
this software and associated documentation files (the "Software"), to deal in
the Software without restriction, including without limitation the rights to
use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
the Software, and to permit persons to whom the Software is furnished to do so,
subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

MediaPipe

Face anonymisation uses Google MediaPipe Tasks Vision and its published full-range face detector model. Both the WebAssembly runtime and model execute in the visitor's browser.

On-device transcription

On-device audio transcription bundles the quantized Whisper base model from onnx-community/whisper-base, plus the Transformers.js and ONNX Runtime Web runtimes. The model and runtime files are served from /transcribe-model/ with their corresponding notices.

Mediabunny

Video subtitle export uses Mediabunny 1.56.0 by Vanilagy, distributed under the Mozilla Public License 2.0. We bundle the library without modifying its source files. The corresponding source code and notices are available here, and the project is maintained at github.com/Vanilagy/mediabunny.

Video inpainting

The browser video eraser uses STTN for video inpainting under the MIT license, and RapidOCR with PP-OCR text detection. The original model source and notices are included under services/video-erase/third_party. Converted model weights and ONNX Runtime Web are downloaded for local browser inference.

Fonts

This site serves its own fonts rather than loading them from a font CDN. All three are licensed under the SIL Open Font License, Version 1.1.

Sample media

The video and before-and-after audio on the noise removal and bleeping tools come from real aircraft-maintenance interviews recorded on a flight line and published by DVIDS, not something staged for a demo. DVIDS marks the source recording as public domain.

We ran the noise-removal excerpt through the strong preset. We did not loudness-match the two tracks: the cleaned one measures about 1.2 LU quieter, because removing noise removes energy, and that is what your own file will do too.

Voice changer sample

The voice demo is an 11.5-second interview with Jim Bridenstine from Artemis Media Resource-Interview Soundbites, published by NASA. Source use follows the NASA media usage guidelines. The original picture and speech share the same timeline. A uniform 5 dB gain was applied to the source before making the original preview and all ten tool preset exports. Only MP3 encoding follows each export. No endorsement by NASA or the speaker is implied. Recording and processing details.

This site

Built with Astro. The site code is ours; the parts above are not, and that is the point of this page.