N.O.V.A orbit and star emblemN.O.V.A v2
N.O.V.A v2 / documentation

Quickstart

Set up the local dependencies and run the assistant loop for the first time.

Requirements

  • Python 3.11 or newer
  • Ollama, with a small model pulled — ollama pull llama3.2:3b or qwen2.5:7b if you have more RAM
  • A working microphone; a second audio-loopback source if you also want call recording

Install

git clone https://github.com/RiyaTorgal/N.O.V.A
cd N.O.V.A
git checkout v2
pip install -r requirements.txt
GPU or a cloned voice?

Follow Installation & Environment first. It creates an isolated environment and pins the GPU build of PyTorch.

Run

python main.py

On first run the wake word detector and speech-to-text model download their weights, which can take a few minutes. After that, say the wake word and speak a command.

Tip

Start with a short, distinct wake word during testing — the default is easiest to say clearly in a quiet room.

Next steps

PREVOverviewNEXTInstallation & Environment →