Supertonic text processor and inference adaptation
Source: https://github.com/supertone-oss-archive/supertonic
Source revision: 1e9799e964ea4c0dad7cde993b65c3c813a7b373
Original code license: MIT (see LICENSE).

VariableVista changes (October 2026): retained the Unicode processor;
reimplemented single-speaker inference in reader-worker.mjs with explicit
tensor disposal, bounded passages, cancellation and cached downloads.

Unmodified Supertonic 3 model and F1 preset voice:
https://huggingface.co/supertone-oss-archive/supertonic-3
Revision: aafc6e32416a594460b32413efc49d7fe4ce6d46
Model license: OpenRAIL-M (see MODEL-LICENSE and terms.html).
Model use is subject to paragraph 5 and Attachment A of MODEL-LICENSE.
The upstream project is archived. Updates require explicit review and testing.

ONNX Runtime Web 1.22.0: https://github.com/microsoft/onnxruntime
Distributed unchanged from the npm onnxruntime-web package (MIT).
See ../onnxruntime/LICENSE and ../onnxruntime/ThirdPartyNotices.txt.