Jarvis Voice Module

The dedicated Text-to-Speech (TTS) voice module for the overall JARVIS project.

This repository is a customized adaptation of the highly efficient LuxTTS by YatharthS. It serves as a lightweight, plug-and-play module built to provide high-quality tts for the larger JARVIS ecosystem.

✨ Features

• Plug-and-Play Module: A self-contained, object-oriented module designed to be easily imported and used anywhere in the main JARVIS codebase.
• Auto-Cleanup: Audio files are generated, played synchronously, and automatically deleted to keep the system clean without leaving temp files behind (still need to figure out a way to stream the audio as it is generated)
• High-Quality Voice: Retains LuxTTS's state-of-the-art voice cloning capabilities with crisp 48khz speech generation.
• Speed: Reaches lightning-fast speeds (150x real-time) on GPUs and is heavily optimized for Apple Silicon (MPS).

🚀 Usage

1. Installation

Clone this repository and install the required dependencies:

git clone https://github.com/codewithbro95/JarvisLuxTTS.git
cd JarvisLuxTTS
pip install -r requirements.txt

2. Testing the Module

This repository includes a main.py script to easily test the voice module on the fly:

python main.py

This will run a sample interactive loop where you can type prompts and hear JARVIS speak to verify the module is working correctly.

3. Module Integration

To use this within the larger JARVIS architecture, simply import the TTS engine:

from modules.audio import StreamJarvisTTS

jarvisTTS = StreamJarvisTTS()
jarvisTTS.speak("All systems are online. How may I be of service?")

📜 Acknowledgments

• Adapted from the original LuxTTS repository by YatharthS.
• Core architecture derived from ZipVoice and Vocos.