TxtSays local TTS model notice

Supertonic 3 FP16
- Original model: Supertone/supertonic-3, TTS version v1.7.3
  https://huggingface.co/Supertone/supertonic-3
- Original model copyright: Copyright (c) 2026 Supertone Inc.
- FP16 derivative package: Kyumdroid/supertonic-3-quant
  https://huggingface.co/Kyumdroid/supertonic-3-quant/tree/d9bfd0a9384b42a76f9256eafdd3d8a5cc4706ba
- Pinned package revision: d9bfd0a9384b42a76f9256eafdd3d8a5cc4706ba
- License: BigScience Open RAIL-M (see /tts/LICENSE-SUPERTONIC.txt)
- Kyumdroid modification: converted all four ONNX models to FP16 internal
  weights with FP32 inputs and outputs. The upstream full unicode indexer and
  ten F1-F5/M1-M5 voice-style files are included unchanged.
- Delivery: the browser downloads the pinned FP16 package directly from the
  Kyumdroid Hugging Face Git repository. TxtSays does not proxy the model files.
- Runtime: synthesis uses eight vector-estimator steps. TxtSays does not apply
  the former W4-specific DeepFilterNet or EQ post-processing chain.
- Disclosure: audio produced with this local model is AI-synthesized.
- No endorsement: the Supertonic and Supertone names identify the upstream
  source and do not imply endorsement of TxtSays.