ESP32-S3 Bench Guide
ESP32-S3 N16R8 bench guide: live pinout, lessons, simulators and the bench log.
翻译准备中
尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证Open-source French reading & pronunciation trainer for TCF Canada — close reading, IPA, shadowing, AI pronunciation feedback and spaced repetition. Browser-only PWA.
翻译准备中
取自仓库 README
正在加载完整 README。
查看完整 README ↗Read French closely. Then say it out loud.
I am learning French for TCF Canada, so I built the reading tool I wanted. It runs in your browser with no account. Your learning data is stored on your device; when you use an AI feature, what you ask about goes straight to the provider you pick.
Word and sentence lookups in articles, pronunciation feedback and AI speech use your own key for the provider you pick (local Ollama needs none). Graded readers, reviewed conjugation and the noun-gender reference work with no key.
Tap a word. In a graded reader you see its meaning and the sentence it came from, with no key. In an article you paste or import, the lookup also gives IPA, verb forms and examples; that uses a text AI key (local Ollama needs none).
Open a sentence in a pasted or imported article. It splits into labelled parts, with grammar points, collocations and a shadowing guide beside the original text. Needs a text AI key, or local Ollama with no key.
Say it back. Play the sentence and record yourself. Compare plays the reference first, then your take. The last three takes of each sentence stay on your device. Recording needs no key. Graded readers use the device voice for the reference; elsewhere, choose Device voice in Settings to compare without a key.
src/data/noun-gender/pack.json (reproduce with bun scripts/readme-numbers.ts): -tion is feminine in 3,780 of 3,794 nouns, -age is masculine in 3,091 of 3,105, -ure is feminine in 902 of 1,216; 337 example nouns are reviewed.I built Relire with AI coding agents under my direction. They wrote the code and drafted most of the content. Automated checks cover the code, the data shapes and the conjugation tables against reference data. The French in the graded readers was reviewed by other AI models, and no human teacher has reviewed it yet. A human language review is the release gate for beginners.
Every count in this table comes from this repository; bun scripts/readme-numbers.ts prints them.
| Item | Count | Source |
|---|---|---|
| Bundled graded readers | 3 | src/data/graded-readers/starter.json |
| Sample articles | 3 | src/data/sampleArticles.ts |
| Reviewed verbs in conjugation pack | 19 | src/data/grammar-pack/pack.json |
| Noun-ending tendencies | 17 | src/data/noun-gender/pack.json |
| Noun ending -tion, feminine in the dictionary | 3780 of 3794 | src/data/noun-gender/pack.json |
| Noun ending -age, masculine in the dictionary | 3091 of 3105 | src/data/noun-gender/pack.json |
| Noun ending -ure, feminine in the dictionary | 902 of 1216 | src/data/noun-gender/pack.json |
| Reviewed nouns in the noun-gender pack | 337 | src/data/noun-gender/pack.json |
| Exercise types | 11 | src/types/exercises.ts |
| UI languages | 2 | src/i18n/locales/ |
bun -e 'console.log((await Bun.file("src/data/graded-readers/starter.json").json()).readers.length)'
grep -c 'id: ' src/data/sampleArticles.ts
bun -e 'console.log(Object.keys((await Bun.file("src/data/grammar-pack/pack.json").json()).verbs).length)'
bun -e 'console.log((await Bun.file("src/data/noun-gender/pack.json").json()).rules.length)'
bun -e 'const p=(await Bun.file("src/data/noun-gender/pack.json").json());const r=p.rules.find(v=>v.id==="tion");console.log(`${r.dictionary.agree} of ${r.dictionary.n}`)'
bun -e 'const p=(await Bun.file("src/data/noun-gender/pack.json").json());const r=p.rules.find(v=>v.id==="age");console.log(`${r.dictionary.agree} of ${r.dictionary.n}`)'
bun -e 'const p=(await Bun.file("src/data/noun-gender/pack.json").json());const r=p.rules.find(v=>v.id==="ure");console.log(`${r.dictionary.agree} of ${r.dictionary.n}`)'
bun -e 'console.log((await Bun.file("src/data/noun-gender/pack.json").json()).nouns.filter(n=>n.prov.status==="reviewed").length)'
grep -cE "^\s+'[A-Z_]+'," src/types/exercises.ts
ls src/i18n/locales/*.json | wc -l | tr -d ' '
Open https://royisme.github.io/relire/app/. Today, the graded readers, noun gender and conjugation of the reviewed verbs work without a key. AI lookups, AI speech and pronunciation feedback need a key for the provider you choose, except Ollama and the device voice.
You need Bun. For the AI features you also need a key from one of the providers above (Ollama needs none), for example a free Gemini API key. On first launch, a three-step guide lets you add a key or continue without AI.
git clone https://github.com/royisme/relire.git
cd relire
bun install
bun run dev # http://localhost:5173
bun run build produces a plain static site in dist/; host it anywhere (GitHub Pages, Netlify, an S3 bucket) and install it from Chrome, Edge or Safari so it opens in its own window. What is stored on your device then works offline; anything that calls an AI provider or the optional content service needs a connection, and Ollama needs only your own machine.
Full detail: PRIVACY.md.
React 19, Vite, Tailwind CSS 4 and i18next, with AI providers called directly from the browser. PRODUCT.md says who the app is for, DESIGN.md is the visual system, and AGENTS.md holds the repository rules, checks and commands.
Issues and pull requests are welcome. Please read PRODUCT.md and DESIGN.md before changing the UI, and keep every user-facing string in both src/i18n/locales/en.json and zh.json.
MIT for the code and documentation. The dictionary data files under data/lexicon/ are a work based on Morphalou 3.1 and are licensed under the LGPL-LR; see data/lexicon/NOTICE.md and data/lexicon/LGPL-LR.txt.
If Relire helps your French, a star helps others find it.