SISU

Overview

SISU (Super Ingenious Sound Upscaler) is a project that uses GANs (Generative Adversarial Networks) to make low-quality audio better. The goal is to take not-so-good-sounding audio and turn it into high-quality, clear audio.

Structure of the Project

• dataset: This folder has some sample audio files for testing.
• models:
  • generator.py: This file has the code for the part that improves the audio.
  • discriminator.py: This file has the code for the part that checks if the audio is good or not.
• training:
  • training.py: This script is used to teach the computer how to improve the audio.

Using the Project

1. Set Up:

   • Make sure you have Python installed (version 3.8 or higher).
   • Install needed packages: pip install -r requirements.txt
   • Install current version of PyTorch (CUDA/ROCm/What ever your device supports)

2. Prepare Audio Data:

   • Put your audio files in the dataset/good folder.

3. Train the Model:

   • Run the training script: python training.py

4. Generate Better Audio:

   • After training, you can use the generator to make your audio sound better.

License

This project is open-source and licensed under the GPLv3 License. For details, see the LICENSE file.