This project provides a pattern for an "always-on" AI assistant, integrating Deepseek V3 for reasoning, RealtimeSTT for speech-to-text, and Typer for command execution. It's designed for engineers and power users seeking a voice-controlled, interactive AI agent capable of performing tasks.

How It Works

The assistant operates with a modular architecture. A "Brain" (Deepseek V3 or Ollama models) processes prompts, while "Ears" (RealtimeSTT) handle voice input and "Mouth" (ElevenLabs or local TTS) provide audio output. A "Job" (prompt file) defines the assistant's task, and "Active Memory" (scratchpad) stores dynamic variables and conversation history. The Typer Assistant specifically leverages a command template file and a scratchpad to generate and execute shell commands based on natural language input.

Quick Start & Requirements

• Install: uv run python main_base_assistant.py chat or uv run python main_typer_assistant.py awaken --typer-file commands/template.py --scratchpad scratchpad.md --mode execute
• Prerequisites: Python 3.11, Deepseek API Key, ElevenLabs API Key. RealtimeSTT and faster-whisper are recommended for STT.
• Resources: Requires API keys for Deepseek and ElevenLabs.

Highlighted Details

• Integrates Deepseek V3, RealtimeSTT, and Typer for a comprehensive AI assistant experience.
• Supports both a base chat interface and a Typer-based command execution mode.
• Utilizes a scratchpad for active memory and dynamic variable management.
• Offers flexibility in Speech-to-Text (RealtimeSTT, faster-whisper, Whisper) and Text-to-Speech (ElevenLabs, local).

Maintenance & Community

No specific information on maintainers, community channels, or roadmap is provided in the README.

Licensing & Compatibility

The README does not specify a license. Compatibility for commercial use or closed-source linking is not detailed.

Limitations & Caveats

The project is presented as a "pattern" and may require significant configuration and integration effort. Specific details on error handling, robustness, and advanced customization are not elaborated upon.