DeepACE is a comprehensive repository that implements ´cochlear implant audio signal processing techniques using deep learning. This project offers dual-framework implementations in both PyTorch and TensorFlow, providing flexibility for researchers and developers to experiment and compare performance across these popular deep learning platforms.

DeepACE is designed for tasks such as audio mixture separation, enhancement, and cochlear implant simulation. The repository includes:

• Complete Implementations: Training, testing, and utility modules in both PyTorch and TensorFlow.
• Custom Loss Functions and Modules: Tailored layers and loss functions designed specifically for audio processing.
• Flexible Data Handling: Custom dataset classes and collate functions for efficient loading and pre-processing of audio data.
• Reproducibility: Utility scripts for setting seeds, logging training progress, and saving model checkpoints.

• Dual-Framework Support: Use PyTorch or TensorFlow, or compare both side-by-side.
• End-to-End Pipeline: Includes everything from dataset preparation and model training to evaluation and testing.
• Customizable Configurations: Easily modify model parameters and training settings via YAML configuration files.
• State-of-the-Art Techniques: Implements advanced techniques such as Squeeze-and-Excitation, cumulative layer normalization, and low-rank augmentations.

[1] Gajecki T, Zhang Y, Nogueira W. A Deep Denoising Sound Coding Strategy for Cochlear Implants. IEEE Trans Biomed Eng. 2023 Sep;70(9):2700-2709. doi: 10.1109/TBME.2023.3262677. Epub 2023 Aug 30. PMID: 37030808.

git clone https://github.com/yourusername/DeepACE.git
cd DeepACE