Licences and credits

Parseh is free software, under the GNU General Public License, version 3 or (at your option) any later version — GPL-3.0-or-later. The licence’s text is the file LICENSE, at the top of the Parseh folder, and every source file says so in its first lines:

text

SPDX-License-Identifier: GPL-3.0-or-later

It comes with no warranty. The source is the Parseh folder itself: the server runs from it as it is.

The copyright is Bruno Ursino’s, who made Parseh: Copyright © 2026 Bruno Ursino, with links to his GitHub account and his website, in the README.md’s License section and at the head of the Licences page. The text of LICENSE is the Free Software Foundation’s, unchanged.

In Parseh, the foot of the hub leads to Licences (/licences/), in the browser interface and in the mobile one. That page says all of this, with a link to each licence’s text.

1. What Parseh carries#

These are in the Parseh folder and are not Parseh’s own work. Each keeps its licence, which travels with it.

What	Where	Whose	Licence
Vazirmatn 33.003	`lib/fonts/`	© 2015 The Vazirmatn Project Authors	SIL Open Font License 1.1
Noto Naskh Arabic 2.004	`lib/fonts/`	© 2019–2020 Google LLC	SIL Open Font License 1.1
Noto Nastaliq Urdu	`lib/fonts/`	© 2014 Google Inc.	SIL Open Font License 1.1
Noto Serif Devanagari 2.001	`lib/fonts/`	© 2019 Google Inc.	SIL Open Font License 1.1
TeX Gyre Pagella, Heros and Chorus	`lib/fonts/`; copied by the studio, and into this guide’s pages (Chorus sets no document, only the name Parseh, in the bar of every page of Parseh, in the hub’s title, and in the bar of the guide published on the web)	© B. Jackowski, J. M. Nowacki and the TeX users groups	GUST Font License
MathJax 3.2.2	`lib/mathjax/`	© The MathJax Consortium	Apache License 2.0
The Italian hyphenation patterns, `hyph-it.tex`	`markdown/exlex/assets/hyph/`	© 2008–2011 Claudio Beccari	LaTeX Project Public License 1.3 or later, or MIT
Hazm’s lists of informal Persian words and verbs, as `hazm` rows of the spoken-Persian tables, `fa.spoken.words.tsv` and `fa.spoken.verbs.tsv`	`lib/lang/`	© 2013 Alireza Nourian	MIT
Conjugation rows, and sentences with their machine translations, in the tests’ data	`tests/fixtures/verbs/`, `tests/fixtures/align/`	Wiktionary’s contributors; Tatoeba’s contributors	CC BY-SA 4.0 (Wiktionary); CC BY 2.0 FR (Tatoeba)

The texts sit beside the files:

lib/fonts/OFL.txt is the Open Font License, with the copyright notice of each font under it. lib/fonts/GUST-FONT-LICENSE.txt is the TeX Gyre fonts’ licence. lib/fonts/README.md says which file is which font. The compiled guide carries copies of all three beside its fonts, in html-guide/site/_parseh/fonts/.
Each font file also carries its notice and the name of its licence in its own metadata. So a font packed into an Anki deck goes with its licence.
lib/mathjax/LICENSE is MathJax’s licence.
lib/lang/fa.spoken.HAZM-LICENSE.txt is Hazm’s licence (MIT), with its copyright line, beside the tables it seeded.
hyph-it.tex says in its own header which licences it is under, and carries the MIT text.
tests/fixtures/README.md says where every text the tests read comes from.

The Japanese and Chinese pages are set in the device’s own faces, so none travels with Parseh.

2. What it downloads when you ask#

The dictionaries, the sentences people have translated, the translation models and their engine, the English synonyms and the character components are not in the Parseh folder. Each is downloaded from its source when you ask for it on the reading-help page, and keeps its own licence. Parseh writes that licence into the file it builds, and shows it wherever it shows what the file says. Whose work it is lists every source and its licence.

Speech to text, which has a page of its own, Settings → Speech to text, is fetched the same way: its program (faster-whisper, CTranslate2 and onnxruntime under the MIT licence, PyAV under BSD-3-Clause with the FFmpeg libraries and codecs it carries under their own) from PyPI, and its two models (OpenAI’s Whisper, under the MIT licence) from Hugging Face. The optional exact-word-time networks are likewise fetched only when asked, from Parseh’s hash-pinned per-language Hugging Face repositories; their own Apache-2.0 notices travel with them (Hindi is MIT and Turkish CC-BY-4.0).

The ARASAAC pictograms, which also have a page of their own, Settings → Pictograms (ARASAAC), are fetched from ARASAAC’s own hosts, and are under a licence that asks something of what you make with them: CC BY-NC-SA 4.0. They were drawn by Sergio Palao for ARASAAC and are the property of the Government of Aragón. Name them wherever you use them (the credit is written, word for word, in arasaac/LICENSE-ARASAAC.txt), never use them for commercial purposes, and share what you make with them on the same licence.

The installer fetches micromamba, Python and the packages environment.yml lists. Each of them is under the licence it comes with.

3. What you read with it#

The books, videos, documents and decks are yours, or their authors’. Parseh ships none of them, and its licence says nothing about them.

4. A new font#

A face added for a new language travels with its licence (Adding a language). Record it in three places:

a row in lib/fonts/README.md;
its copyright line in lib/fonts/OFL.txt, or a licence file of its own beside the font, if it is not under the OFL;
an entry in FONTS, in lib/notices.py, which draws the Licences page.

tests/test_licences.py fails until all three name the new font.