Lyra V2 SoundStream

Introduction

The Lyra V2 SoundStream repository contains model files intended for an eventual web demo. These models are sourced from the official Lyra repository by Google, which focuses on efficient audio coding.

Architecture

The repository currently hosts model files from the Lyra V2 SoundStream. The models are designed to deliver high-quality audio compression and decompression.

Training

There are no specific training instructions provided within this repository. The models are derived from pre-existing files in the official Lyra repository.

Guide: Running Locally

To run the models locally:

1. Clone the repository: git clone https://github.com/google/lyra
2. Navigate to the model coefficients directory: cd lyra/model_coeffs
3. Install required dependencies as per the official Lyra repository instructions.
4. Consider converting models to ONNX for compatibility with different environments.

For optimal performance, it is recommended to use cloud GPU services such as Google Cloud, AWS, or Azure.

License

The content and models in this repository are licensed under the Apache 2.0 License. This allows for free use, modification, and distribution under the terms of the license.