Dot - Voice-to-Notes Telegram Bot

🎤 Transform voice messages into structured markdown notes and actionable tasks. Develop as experiment with Claude Code!

What is Dot?

Dot is your AI companion that listens to your voice messages on Telegram and transforms them into organized notes for your second brain. Perfect for capturing ideas while driving, walking, or anytime you prefer speaking over typing.

Features

• 🤖 Telegram Bot Integration: Send voice messages directly to your bot
• 🇮🇹 Italian Transcription: Native support for Italian language
• 📝 Smart Note Generation: AI-powered transformation into structured markdown
• ✅ Task Extraction: Automatically identify actionable items from your recordings
• 🗂️ Obsidian Compatible: Generate notes ready for your second brain

Quick Start

Prerequisites

• Rust 1.70+ installed
• Telegram account
• CMake (required for building whisper.cpp)
• ffmpeg (required for Telegram voice message support)
• (Optional) Ollama installed for local AI processing (Phase 3)

Installing Dependencies

macOS:

brew install cmake ffmpeg

Windows (with Chocolatey):

choco install cmake ffmpeg

Linux (Ubuntu/Debian):

sudo apt install cmake ffmpeg

Setup

1. Download Whisper model:

cd models
curl -L https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.bin -o ggml-base.bin
cd ..

2. Configure your bot:

cp config.example.toml config.toml
# Edit config.toml with your Telegram bot token

3. Build with hardware acceleration:

M1/M2/M3 Mac:

cargo build --release --features metal

Windows/Linux with NVIDIA GPU:

cargo build --release --features cuda

CPU-only (any system):

cargo build --release --features cpu

4. Run:

# Mac
cargo run --features metal

# Windows NVIDIA
cargo run --features cuda

# CPU
cargo run --features cpu

Configuration

See config.example.toml for all available options.

Key settings:

• Telegram bot token
• Transcription service (Whisper API, local, etc.)
• AI model (Ollama local or cloud API)
• Output directories

Project Status

• ✅ Phase 1: Telegram Bot Foundation (Complete & Tested)
• ✅ Phase 2: Audio Transcription (Complete & Tested)
• ✅ Phase 3: AI Note Generation (Complete & Tested) 🎉
• 🔴 Phase 4: Task Extraction (Optional - Not Started)

System Fully Functional! 🚀

Current Features:

• 🤖 Telegram bot with Italian responses
• 🎤 Voice message transcription (Italian)
• ✨ LLM-based transcription cleanup (fixes errors)
• 🧠 AI-powered note generation (Ollama)
• 📝 Structured markdown notes with frontmatter
• 🏷️ Automatic tag suggestions
• 💾 Save notes to files (Obsidian-compatible)
• 🚀 Metal/CUDA acceleration support
• 🌐 Local + Remote Ollama support (LAN)
• 📱 Commands: /start, /help, /status

See where-are-we.md for detailed development status and roadmap.

Development

See ClaudePrompts.md for development guidelines.

License

MIT