| Live lesson | Practice together |
|---|
| |
| AI provider | Text to speech |
|---|
| |
Features
- Structured courses: English A1 → C2, plus Spanish, French, German and Turkish at A1. 50+ steps per level, a harvest basket at the end of each unit and a checkpoint at the end of each level. Each unit has a guidebook in your own language, and a unit or level test lets you skip ahead.
- AI-generated lessons: translate, fill in the blank, match, listen, word bank, error correction and more. Lessons adapt to your recent mistakes, and a harder Mastery version turns a finished lesson gold.
- Live lesson: a video-call style lesson with one of six animated tutors who talks with you, with live captions. You can cut in while the tutor is speaking. Open the practice panel to do exercises together while the tutor follows along.
- Sprigo gets to know you: after a conversation it keeps a few facts about you (interests, goals, what you find hard) and uses them in later lessons. You can see, edit or forget them, or turn memory off.
- Speaking coach: live lessons measure how you speak (response time, pauses, answer length, switching to your own language), the tutor adapts during the call, and your profile shows your progress. No audio is recorded.
- Speaking: speech recognition runs inside the app (whisper.cpp), with no extra service to install. Deepgram is an optional cloud alternative, and Try it lets you check it with your own voice.
- Natural voices: besides the system voice, Piper (fast, local) and Kokoro (most natural, English only) run on your machine and are downloaded on first use.
- Stories and roleplay: chat or call characters in real-life scenes, and get corrections as you go.
- Practice hub: fix your mistakes, speaking and listening practice, your word list, Match Madness and a timed challenge.
- Your garden: a tree that grows with your lessons, stays green while you keep your streak (days of roots) and grows a fruit for every level you pass. Each level has its own landscape, from meadow to summit.
- Gamification: XP, coins, drops, daily quests, a shop, friends and a local league.
- Use it on your phone: turn on Settings → Use on your phone and scan the code to open Sprigo in your phone's browser over Tailscale. Your data, speech recognition and Ollama stay on the computer.
- Private by default: your progress stays in a local SQLite database with daily backups, and API keys are kept in the system keychain. You can move the data folder (for example to a synced folder), or export and import it. Several profiles can share one computer, optionally with a PIN.
- Reminders, light and dark themes, and an interface in English, Turkish, German, French or Spanish.
- Automatic in-app updates.
Install
macOS, with Homebrew:
brew install --cask ozerozdas/tap/sprigo
macOS or Linux (x86_64), without Homebrew:
curl -fsSL https://raw.githubusercontent.com/ozerozdas/sprigo/master/install.sh | sh
Both install the latest release and handle the macOS quarantine flag for you. Windows installers and the other packages (.deb, .rpm) are on the Releases page. Step-by-step instructions per system are in the wiki.
If you install the .dmg by hand: the app is not notarized yet, so macOS blocks it the first time. Remove the quarantine flag once:
xattr -dr com.apple.quarantine /Applications/Sprigo.app
Set up an AI model
On first launch, Sprigo looks for a local model server:
- Ollama (recommended): install and start it. If no model is installed yet, Sprigo offers to download
qwen3:8b (about 5.2 GB) for you. Smaller models run but write noticeably weaker lessons, so Sprigo warns when you pick one under 8B.
- LM Studio: start its local server and load a model.
- Cloud: you can also use OpenAI, Anthropic, Gemini or OpenRouter with your own API key. Pick one in Settings → Advanced → AI provider; each option has a short note on what it is good at.
Model sizes, hardware and the cloud options are explained in AI Model Setup.
Documentation
Community
Contributing
Sprigo is built in the open, and the easiest way to help needs no code: courses are plain YAML files in courses/ that list units, vocabulary and grammar, and the AI writes the exercises. Adding A2 to Spanish, French, German or Turkish, or starting a new language, is a great first contribution. See Writing a Course.
Pick a good first issue, and read CONTRIBUTING.md for setup and workflow. Everyone taking part follows the Code of Conduct.
License
MIT
