MIT License Copyright (c) 2026 tiny-finglish contributors 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. tiny-finglish Copyright (c) 2026 tiny-finglish contributors This product includes software and data developed by third parties. -------------------------------------------------------------------------- Lilak, Persian Spell Checking Dictionary -------------------------------------------------------------------------- data/lexicon/fa-stems.bin is derived from the Lilak fa-IR dictionary. Copyright 2015 Mostafa Sedaghat Joo Licensed under the Apache License, Version 2.0 https://www.apache.org/licenses/LICENSE-2.0 https://github.com/b00f/lilak The derived artifact contains the dictionary's stem list, normalized to the canonical Persian form used throughout this project, deduplicated, sorted and front-coded. Affix rules are not applied. No modification of meaning is made. Obtained via the npm package `dictionary-fa@2.0.0`, which redistributes the same dictionary under the same license. See data/provenance/lexicon.json for the upstream SHA-256 and byte counts. -------------------------------------------------------------------------- HomoRich G2P Persian -------------------------------------------------------------------------- The shipped word-frequency and vowel tables (dist/data/frequency.bin, dist/data/vowels.bin), the optional bigram table and part of the model's training data are derived from the Persian side of HomoRich G2P Persian, dedicated to the public domain under CC0 1.0. Credit is given though not required: https://huggingface.co/datasets/MahtaFetrat/HomoRich-G2P-Persian Mahta Fetrat Qharabagh, Zahra Dehghanian and Hamid R. Rabiee, "Fast, Not Fancy: Rethinking G2P with Rich Data and Rule-Based Models", arXiv:2505.12973, 2025. See data/provenance/frequency.json and data/provenance/vowels.json. -------------------------------------------------------------------------- finglish-dataset -------------------------------------------------------------------------- The evaluation sets in data/gold/ and data/dev/ (in this repository, not in the npm package) are derived from finglish-dataset, whose Persian side comes from Mozilla Common Voice Persian (CC0): https://github.com/mmahdibarghi/finglish-dataset MIT License Copyright (c) 2021 mohammad mahdi barghi 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. See data/provenance/gold.json and data/provenance/dev.json. -------------------------------------------------------------------------- Deliberate exclusions -------------------------------------------------------------------------- The following widely used Persian resources are NOT included, because their licensing is incompatible with an MIT-licensed package. They are listed here so the exclusion is a recorded decision rather than an oversight: * Hazm `words.dat` — the repository is MIT, but its annotations derive from the Peykare/Bijankhan corpus, which is research/non-commercial. * `behnam/persian-words-frequency` — CC BY-SA 3.0. ShareAlike would infect a shipped frequency table, which is exactly the artifact it would apply to. * Virastyar, perstem, PersianStemmer, Vajehdan — GPL/AGPL. * Several large GitHub word lists (240k/494k/700k entries) — no license file. * `Arshia82sbn/Finglish-To-Persian-Dataset-Large` — tagged Apache-2.0, but built in part from GPLv3 Virastyar rules over a corpus with no stated provenance. See docs/open-items.md for the full survey.