myai

My personal AI assistant

Looking for architecture, feature breakdowns, or optimization notes? Start with docs/system_overview.md. Detailed legacy write-ups now live under docs/archive/.

Installation Instructions

Prerequisites:

• [List required software or dependencies, e.g., Node.js, Java, Docker, etc.]
• Python 3.13.2

Steps:

1. Clone the repository

   git clone https://github.com/ToBeAss/myai.git

2. Create a virtual environment

   Creating a Python virtual environment allows you to manage dependencies separately for different projects, preventing conflicts and maintaining cleaner setups. Real Python, 2024

   python -m venv venv

   For macOS you may need to use python3 -m venv venv.

3. Activate the virtual environment

   • Windows (Command Prompt)

     venv\Scripts\activate

   • macOS

     source venv/bin/activate

4. Install dependencies

   pip install -e ".[dev]"

   For macOS you may need to use pip3 install -e ".[dev]".

   This installs runtime dependencies from requirements.txt and development tools (such as pytest) from pyproject.toml.

5. Create the .env file

   Copy the .env.example file and rename it .env:

   • Windows (Command Prompt)

     copy .env.example .env

   • macOS

     cp .env.example .env

   Open the .env file and insert the API keys.

Warning

The .env file is used to store sensitive information like API keys and database credentials securely. Make sure not to commit it to version control by adding .env to your .gitignore file.

🚀 Performance Optimizations

This voice assistant is highly optimized for low-latency responses:

⚡ Chunked Transcription (NEW!)

Transcribes speech in parallel chunks while you continue speaking.

• 300-1200ms faster response time for multi-phrase commands
• Automatically enabled in main_continuous.py
• See the quick overview in docs/system_overview.md and the archived deep dive in docs/archive/CHUNKED_TRANSCRIPTION_QUICKREF.md

🎯 Other Optimizations

• faster-whisper: 4-5x faster transcription with CTranslate2
• Dynamic silence detection: Adaptive thresholds (750-1250ms)
• Comma-based TTS chunking: Synthesis starts at natural pauses
• Parallel processing: TTS synthesis and playback in separate threads

Combined result: ~2-3 seconds faster than baseline! 🎉

For configuration specifics, see docs/system_overview.md and the archived reference docs/archive/LATENCY_OPTIMIZATION_SETTINGS.md

Environment Management

• Deactivate the virtual environment (when done)

  deactivate

• Run tests

  python -m pytest

  By default, interactive/manual scripts are excluded from this run.

• (Optional) Export Installed Packages

  If you add new dependencies, update the requirements.txt:

  pip freeze > requirements.txt