WordWise is a desktop English vocabulary learning app for focused practice and long-term retention. It helps learners build a personal word list, practise at their own pace, review mistakes, and see which words are easiest to confuse.
We are now releasing some parts of its core algorithms
The app is designed to work locally. Learning records stay on the device, and optional local language-model support can provide semantic feedback for typed answers.
Open WordWise Web to register or sign in with your existing account. The hosted portal is now v2.0; the desktop downloads below are WordWise v2.0.0 for Windows x64. This repository continues to provide selected desktop source and core algorithms, rather than the hosted service's complete source code.
- Practise English vocabulary with Chinese or French answers.
- Choose a French, Chinese or English interface. The first visit follows your browser language: French → French, Chinese → Chinese, otherwise English. Your saved choice takes priority on later visits.
- Change the interface language under Settings & vocabulary → Language. A compact language selector is also available before login. The language choice sets a default study mode; the sidebar lets you choose the study mode separately.
- French study offers TOEFL, IELTS, Tous les mots and your personal list (Mes mots), with adaptive practice, mistake review, a confusion map, hints and vocabulary import/export.
- Random, sequential, and adaptive practice modes
- Spaced review and mistake-focused practice
- Personal confusion map for recurring mix-ups
- Import and export for personal vocabulary lists
- macOS and Windows desktop packaging support
npm install
npm test
npm startThe native vocabulary engine can be rebuilt locally with Rust when a matching prebuilt module is not available:
npm run build --prefix vocab-coreThe public source snapshot intentionally does not include local model weights, bundled wordlists, audio archives, paid distribution assets, private research material, operator-only license tooling, or release packages. When adding vocabulary or model files, use sources you are permitted to redistribute.
The source includes audio packing, vocabulary database preparation and desktop packaging helpers. Two additional read-only checks inspect the staged public file allowlist and macOS/Windows package file selection:
npm run check:public-files
npm run check:package-filesWe are now provide the online trail version which is accessible in WordWise-Web
- Baidu Netdisk: WordWise v2.0.0 — Windows x64 code: vjra
- Google Drive: WordWise v2.0.0 — Windows x64
Slim downloads the model inside the app; Full includes it and requires the setup EXE and archive to stay together.
If you have any questions, contact to the developer at feedback@wordwise.dpdns.org for one year free trial
WordWise is released under the MIT License. See LICENSE.