HanziFlow

A comprehensive ecosystem of 6 interconnected tools designed for Mandarin Chinese study, all running locally on Apple Silicon.

Architecture

| Tool | Purpose | Stack | |------|---------|-------| | Writer | Character practice & dictionary | Next.js, Hanzi Writer, CC-CEDICT | | Server | Backend API & data persistence | Python, FastAPI, SQLite (WAL) | | OCR | Screen text recognition | PaddleOCR v5, Tkinter, Ollama | | Echo | Audio transcription pipeline | mlx-whisper, yt-dlp, Ollama | | Talk | Real-time conversation | mlx-whisper, Kokoro TTS, Sherpa-ONNX | | HSK Pipeline | Vocabulary ETL | BeautifulSoup, SQLite, 11,545 entries |

Key Principles

• 100% Local: No cloud APIs. All inference runs on-device via Ollama and MLX
• Memory-Aware: Pipeline design for 16GB RAM constraint
• Protocol-Based: Each tool follows a defined interface for composability