Third-Party Notices¶
Mirumoji is licensed under the MIT License
This page acknowledges the third-party software and data Mirumoji depends on and redistributes
License Locations
-
The full license text for every bundled dependency can be found
inside the distributed artifacts -
In the
Backend Docker Images, they ship inside eachPythonpackage's metadata (*.dist-info/) and system package docs (/usr/share/doc/...) -
In The
Frontend Docker Image, they ship inside eachnpmpackage and system package docs (usr/share/doc/...) -
The notices below cover the components whose licenses require explicit attribution
Required Attributions¶
JMdict / JMnedict / KANJIDIC2 / KRADFILE → Dictionary Data¶
-
Mirumoji's dictionary lookups use data derived from the
JMdict,JMnedict,KANJIDIC2, andKRADFILE / RADKFILEfiles, the property of theElectronic Dictionary Research and Development Group (EDRDG), used in conformance with the group'sLicence -
These files are made available under the
Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0)Licence -
This data is accessed via
kotobase, which also includes a similar notice
KanjiVG → Kanji Stroke-Order Diagrams¶
-
The kanji stroke-order drawings are rendered from
KanjiVGdata, copyrightUlrich Apel, released under theCreative Commons Attribution-ShareAlike 3.0licence -
This data is accessed via
kotobase, which also includes a similar notice
Tatoeba → Example Sentences¶
-
Example sentences and their translations come from the
Tatoeba Project, released under theCreative Commons Attribution 2.0 FRlicence -
This data is accessed via
kotobase, which also includes a similar notice
Kanji Alive → Pronunciation Audio¶
-
Kanji pronunciation clips come from the
Kanji Aliveproject of the University of Chicago, released under theCreative Commons Attribution 4.0licence -
This data is accessed via
kotobase, which also includes a similar notice
JmdictFurigana → Furigana Segmentation¶
-
The per-kanji furigana segmentation comes from the
JmdictFuriganaproject, released under theMITlicence -
The underlying dictionary data falls under the
EDRDGlicence above -
This data is accessed via
kotobase, which also includes a similar notice
UniDic → Morphological Analysis Dictionary¶
-
Japanese tokenization on the server uses
fugashiwithUniDic -
UniDicis distributed under aBSD-3-Clauselicence by theNational Institute for Japanese Language and Linguistics (NINJAL) -
Attribution to
NINJALis preserved
NVIDIA CUDA → Base GPU Docker Image¶
-
The GPU server image is built on the
nvidia/cudabase image -
Redistribution and use of that image are governed by the
NVIDIA Deep Learning Container License -
This applies only to the GPU image variant
FFmpeg → Bundled Static Build (GPU / Modal Images)¶
-
The
GPUandModalserver images bundle a staticFFmpegbuild fromBtbN/FFmpeg-Builds, configured asGPLso it includes theNVENC/NVDECencoders and thescale_cudafilter used for hardware-accelerated video conversion -
This build is redistributed under the
GNU GPL v3. The corresponding source is available fromffmpeg.organd the build configuration from theBtbN/FFmpeg-Buildsrepository -
The
CPUimage variants keep the distribution'sLGPLFFmpeginstalled viaapt
Notable Components¶
| Component | Role | License |
|---|---|---|
CTranslate2 |
Whisper Inference Engine | MIT |
faster-whisper |
Transcription | MIT |
kotobase |
Japanese Dictionary Database | MIT |
FFmpeg |
Media Conversion | LGPL / GPL |
MeCab |
Morphological Analyzer (Used By Fugashi) | BSD-3-Clause |
FastAPI |
Python Server Framework | MIT |
Flet |
Desktop GUI | Apache-2.0 |
React |
Frontend Framework | MIT |
Nginx |
Frontend Image Web Server | BSD-2-Clause |
Node.js |
Frontend Build / Runtime | MIT (and others) |
This list is not exhaustive
See each project's repository and the bundled metadata for complete terms