An AI voice assistant working fully locally. It's also able to perform tasks.
Following tasks are implemented:
- Taking notes and setting reminders for them.
- Ollama
- PyTorch
- RealtimeSTT
- pyttsx3
- plyer (for reminders)
First, you need to install Ollama: https://ollama.com/download
Then, Install necessary PyPI packages using:
pip install RealtimeSTT pyttsx3 plyer
To install the gpu version of PyTorch:
pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121
Download the app folder and Run app.py
script in background:
python app.py &
You can configure the models and parameters at the beginning of the script:
# --------------------------
# Configuration
# --------------------------
LLM_MODEL = "phi3.5:3.8b"
WHISPER_MODEL = 'small.en'
TTS_RATE = 170
TTS_VOICE_INDEX = 0
Command for taking notes is:
Take me note
- Say
cancel
if you changed your mind. - Use
yes
andno
to give positive and negative responses. - The format to give the day, month and the year information:
<day> <month name> <year>
for example, 2025-06-29 ->29 June 2025
- The format to give the clock information:
<hour>.<minute>
for example, 1:05 PM ->13.5