Welcome to Garvis! π This project leverages the speech-to-text endpoint from Groq, utilizing their cutting-edge LPU (Language Processing Unit) chips, which offer unprecedented speed, faster than traditional GPUs. The goal of Garvis is to provide a near real-time user experience by chunking based on silence and translating instantly, thanks to the high-speed capabilities of Groq's hardware and the LLAMA3 large language model. π
At the moment, Groq API is free but has a limited rate. The techniques I'm using to create real-time results require multiple calls, so you may reach your rate limit quickly. Just keep this in mind. β³
- Real-Time Translation: By chunking input based on silence, Garvis can translate in near real-time, providing an almost instantaneous user experience. β±οΈ
- Asynchronous Core Library: The core of Garvis, named "garvis", is built to be asynchronous, making it adaptable to various user interfaces. π
- Flexible User Interface: Garvis is designed to be user interface agnostic, allowing it to be integrated into any platform. Currently, there are examples for both a desktop application and a terminal interface. π»π₯οΈ
- examples/: Contains example implementations of Garvis.
- desktop/: A simple desktop application showcasing Garvis in action. π₯οΈ
- terminal/: A sample of using the Garvis library in a terminal environment. π₯οΈ
- Python 3.x π
- Dependencies listed in
requirements.txt
- Clone the repository:
git clone https://github.com/unclecode/garvis.git
cd garvis
- Install the required dependencies:
pip install -r requirements.txt
[This part is a work in progress. Please check back soon for more information.] π§
Navigate to the examples/desktop
directory and run the application:
cd examples/desktop
python garvis_app_async.py
Navigate to the examples/terminal
directory and run the terminal example:
cd examples/terminal
python terminal_ui_async.py
Contributions are welcome! If you have suggestions for improvements or encounter any issues, please feel free to open an issue or submit a pull request. π οΈ
This project is licensed under the MIT License - see the LICENSE file for details. π
- Groq for their Speciatex endpoint and LPU chips. π‘
- The developers of the LLAMA3 large language model. π§
- X (UncleCode) - @unclecode π±
Thank you for checking out Garvis! Stay tuned for more updates and examples. π