Nexis is a lightweight offline voice assistant written in Python.

The goal of the project is to provide a fast and simple assistant that can run entirely on local hardware without requiring cloud services, subscriptions, or internet access for voice processing.

Nexis uses:

• Piper for text-to-speech
• Vosk for speech-to-text
• JSON-based response training
• A configurable wake-word system
• DDGS for search engine (subject to change)

• Offline speech recognition
• Offline text-to-speech
• Wake-word activation
• Math capabilites
• Search engine
• Configurable voices
• Customizable responses
• Cross-platform Python code
• Lightweight and easy to modify

Nexis continuously listens for a wake word.

When a wake word is detected, the assistant begins listening for a command. The command is processed and matched against its training data before a response is generated and spoken aloud.

All processing happens locally on your device.

• Python 3.11 or newer
• Piper
• Vosk
• A supported Piper voice
• A supported Vosk model
• At least 256mb of memory (For voice models)

Download the source code and place it in a folder of your choice.

pip install piper-tts vosk sounddevice

Additional dependencies may be required depending on your operating system and audio configuration.

Instructions can be found in:

s-t/voice/README.md

Instructions can be found in:

s-t/vosk/README.md

Edit:

CONFIG.json

to select your preferred voice, wake word, and model configuration.

Start the assistant with:

python Main.py

Once running, say the configured wake word followed by your command.

Example:

Hey Nexis
What is your name?

Main.py        Main application
CONFIG.json    Configuration settings
i-o.json       Response training data
LICENSE.txt    License information
s-t/           Speech-to-text and text-to-speech resources

Nexis was created with a few simple goals:

• Work offline
• Be easy to understand
• Be easy to modify
• Have minimal dependencies
• Run on low-end hardware

The project is intentionally simple and is designed to be a foundation that can be expanded with additional features and integrations.

This project is licensed under the Axion Dev License. See LICENSE.txt for details.