aquiny-test.mp4

aquiny

A CLI assistant that schedules voice-based desktop notifications using natural language.

aquiny "remind me to take a break in 30 minutes"

aquiny parses your prompt with a local LLM (Qwen 2.5 via Ollama), generates a spoken message with KittenTTS, and schedules a desktop notification with audio playback at the right time using systemd timers.

How it works

1. Your natural language prompt is sent to Qwen 2.5:7b running locally on Ollama
2. The LLM extracts a structured reminder (title, message, datetime)
3. KittenTTS generates a .wav voice file from the message
4. A systemd user timer is created to fire at the scheduled time
5. When triggered, notify-send shows a desktop notification and aplay plays the voice

Prerequisites

• Python 3.12
• Ollama running locally with the qwen2.5:7b model pulled
• Linux with systemd, notify-send, and aplay

Installation

ollama pull qwen2.5:7b
git clone https://github.com/omegaui/aquiny.git
cd aquiny
./installer.sh

The installer will:

• Copy files to ~/.local/share/aquiny
• Create a Python 3.12 virtual environment
• Install KittenTTS
• Place the aquiny executable in /usr/bin (requires sudo)

Customization

Important: Open aquiny.py and change the user_name on line 107 to your own name.

structured_data = parse_reminder("Arham", args.prompt)  # <-- change "Arham" to your name

This name is used by the LLM to generate personalized notification messages (e.g. "hey YourName, it's time to rest").

Usage

aquiny "remind me to call Arham at 2 PM today"
aquiny "take a break in 45 minutes"
aquiny "stand up and stretch at 5:30 PM"

License

BSD 3-Clause