Ears 👂

Learn languages from what you hear. A self-directed language learning app that builds your vocabulary from real content you consume - radio, videos, podcasts, anything.

Philosophy

No courses. No teachers. No fixed curriculum. Just you and real content.

1. Capture - Record any Swedish audio playing on your computer
2. Transcribe - AI converts speech to text
3. Build Vocabulary - Words extracted, frequencies tracked
4. Learn - AI-powered flashcards, explanations, and conversation practice
5. Progress - Your dictionary grows with what you actually hear

Architecture

┌─────────────────────────────────────────┐
│          Tauri Desktop App              │
│  ┌───────────────────────────────────┐  │
│  │       Svelte + TailwindCSS        │  │
│  │   Dashboard │ Vocab │ Learn │ Chat│  │
│  └───────────────────────────────────┘  │
│                   ▼                     │
│  ┌───────────────────────────────────┐  │
│  │          FastAPI Backend          │  │
│  │   Whisper │ LM Studio │ Edge-TTS  │  │
│  └───────────────────────────────────┘  │
│                   ▼                     │
│  ┌───────────────────────────────────┐  │
│  │             SQLite DB             │  │
│  │   Vocab │ Transcripts │ Progress  │  │
│  └───────────────────────────────────┘  │
└─────────────────────────────────────────┘

Setup

Prerequisites

• Python 3.11+
• Node.js 18+
• Rust (for Tauri desktop build)
• LM Studio (for AI features)

Backend

cd backend
pip install -r requirements.txt

Frontend

cd frontend
npm install

Usage

1. Start Backend

cd backend
python app.py

Backend runs on http://localhost:8000

2. Start Frontend (Development)

cd frontend
npm run dev

Frontend runs on http://localhost:5173

3. Or Run as Desktop App

cd frontend
npm run tauri dev

Recording Audio

cd backend
python recorder.py

1. Select audio device (use Stereo Mix or similar for system audio)
2. Press Enter to start recording
3. Press Enter to stop
4. Type 'y' to transcribe immediately

Building Vocabulary

After transcribing:

cd backend
python vocabulary.py build    # Extract words from transcripts
python vocabulary.py stats    # View statistics
python vocabulary.py top 50   # See most frequent words

Features

• Dashboard - Overview of your progress
• Vocabulary - Browse all words, filter by status, get AI explanations
• Learn - Flashcard-style review with TTS pronunciation
• AI Chat - Practice conversation with local LLM
• Recordings - Manage and transcribe audio files

Configuration

Edit backend/config.py:

WHISPER_MODEL = "small"      # tiny/base/small/medium
WHISPER_DEVICE = "cpu"       # cpu or cuda
WHISPER_LANGUAGE = "sv"      # Language code
LM_STUDIO_BASE_URL = "http://localhost:1234/v1"

Tech Stack

• Frontend: Svelte, TailwindCSS, TypeScript
• Backend: FastAPI, Python
• Desktop: Tauri (Rust)
• STT: faster-whisper
• TTS: edge-tts (Microsoft voices)
• LLM: LM Studio (local)
• Database: SQLite

License

MIT