WaveSpeedAI's free tools run in your browser with the help of the open-source software, models and fonts below. We thank their authors. Each is used under its own license; follow the links for the license text and source code. Components under the GNU LGPL are used unmodified, as separate files loaded at runtime, so they can be replaced; their source code is available from the linked projects. Questions: support@wavespeed.ai.
Reading, converting and writing video and audio in the video and audio tools (compressor, converters, merger), subtitle burn-in, GIF maker and screen recorder
License: MPL-2.0
MP3 output in the video and audio tools
License: LGPL-2.0-or-later (encoder) / MPL-2.0 (wrapper)
Source: https://lame.sourceforge.io/
Used unmodified as a separate WebAssembly module loaded at runtime. MP3 encoding powered by LAME (https://lame.sourceforge.io/).
AAC sound in MP4/M4A output where the browser has no AAC encoder
License: LGPL-2.1-or-later (encoder) / MPL-2.0 (wrapper)
Source: https://ffmpeg.org/
Only FFmpeg’s AAC encoder, built without GPL components, used unmodified as a separate WebAssembly module loaded at runtime.
FLAC output in the audio converter
License: BSD-3-Clause (encoder) / MPL-2.0 (wrapper)
Source: https://github.com/xiph/flac
Reading STEP, IGES and BREP files in the 3D converter and 3D viewer
License: LGPL-2.1 (with the Open CASCADE exception)
Source: https://github.com/kovacsv/occt-import-js
Used unmodified, loaded at runtime from cdn.jsdelivr.net as a separate file; our code only calls its public interface. Open CASCADE source: https://dev.opencascade.org/
Passport photo maker; face enhancer (face outlines)
License: Apache-2.0
Source: https://ai.google.dev/edge/mediapipe/solutions/vision/face_landmarker
Converted by us to ONNX (half precision) and hosted at huggingface.co/wavespeed/public-misc-models.
Running the speech recognition model in the subtitle generator
License: Apache-2.0
Running the AI models of the subtitle generator, AI image detector, image eraser, Segment Anything, face enhancer and face swap tools
License: MIT
AI image detector
License: Apache-2.0
Source: https://huggingface.co/Ateeqq/ai-vs-human-image-detector
Modified: converted to ONNX and quantized to 4-bit weights; hosted at huggingface.co/wavespeed/public-misc-models with its licence. Base model: https://huggingface.co/google/siglip2-base-patch16-224
AI music detector: the spectral “fakeprint” features and linear classifier
License: MIT
Source: https://github.com/lofcz/ai-music-detector
Method and feature extraction reimplemented in TypeScript; the weights were retrained by WaveSpeed AI and are hosted at huggingface.co/wavespeed/public-misc-models with the MIT licence.
AI music detector: part of the training data for its weights
License: CC BY 4.0
Source: https://huggingface.co/datasets/Kukedlc/suno-ai-music-dataset
Used to train the detector’s weights; no audio from the dataset ships with the site.
Subtitle generator
License: MIT (OpenAI Whisper weights)
Running the upscaling models in the image and video upscalers
License: Apache-2.0
Image and video upscalers
License: MIT
Image eraser
License: Apache-2.0
Segment Anything tool
License: Apache-2.0
Background remover
License: Apache-2.0
Source: https://github.com/xuebinqin/DIS
Image colorizer
License: Apache-2.0
Face enhancer
License: MIT
Face enhancer
License: Apache-2.0
Reading and drawing PDFs in PDF to JPG
License: Apache-2.0
Writing PDFs in Image to PDF
License: MIT
Image vectorizer
License: Apache-2.0
3D viewer, 3D converter and image to STL
License: MIT (three.js) / Apache-2.0 (Draco)
Barcode generator
License: MIT
DXF viewer
License: MIT
Turning text into outlines in the logo maker
License: MIT
Icons, including the logo maker’s icon picker
License: ISC
Making ZIP downloads in the free tools
License: MIT
QR code generator
License: MIT
Text in the logo maker, collage, subtitles, image to DXF and font generator (Montserrat, Poppins, Inter, Archivo Black, Pacifico, Great Vibes, Noto Sans Math and others)
License: SIL Open Font License 1.1
Source: https://openfontlicense.org/
Font files are used unmodified (Noto Sans Math as a subset; it has no Reserved Font Name).