THE DETAILS

Technology and licenses

For 31 languages, TTS Lines runs Supertone’s Supertonic 3 model in your browser. It is supplied under the OpenRAIL-M license, including restrictions on use of the model and its output.

Hebrew uses the BlueTTS 2.5 ONNX model and ReNikud to turn unpointed Hebrew into pronunciation data. Both run in your browser; their model files download from Hugging Face on first use. BlueTTS inference code is under the MIT license. ReNikud is by Maxim Melichov, Yakov Kolani, and Morris Alper under CC BY 4.0. Our browser inference code adapts their published Python examples.

The four Hebrew voice presets use references from LibriTTS-R, published by Google LLC under CC BY 4.0. The references are used to supply voice styles, not spoken scripts.

The browser inference code is adapted from Supertone’s MIT-licensed web example. Read its license notice.

MP3 export uses lamejs under LGPL-3.0. Read the package notice, the combined LGPL-3.0 and GPL-3.0 terms, or the encoder’s source repository. Other browser components include ONNX Runtime Web, JSZip, Phosphor Icons, and the DM Sans and Newsreader fonts.

For the rules governing your use of the voice model, read our Terms. For details on downloads and the contact form, read our Privacy page.