TOWARDS NEURAL EMULATION OF VOLTAGE-CONTROLLED OSCILLATOR

This code repository is for the article Towards Neural Emulation of Voltage-Controlled Oscillator, in Proceedings of the International Conference on Digital Audio Effects (DAFx25).

This repository contains all the necessary utilities to use our architectures. Find the code located inside the "./Code" folder, and the weights of pre-trained models inside the "./Weights" folder

Visit our companion page with audio examples

Folder Structure

./
├── Code
├── fig
└── Weights
    ├── 

Contents

1. Datasets
2. How to Train and Run Inference
3. VST Download

Datasets

Datasets are available here

Our architectures were evaluated on three analog waveshapes:

• Triangle
• Square
• Sawtooth

How To Train and Run Inference

This code relies on Python 3.9 and TensorFlow. First, install Python dependencies:

cd ./code
pip install -r requirements.txt

To train models, use the starter.py script. Ensure you have loaded the dataset into the chosen datasets folder

Available options:

• --model_save_dir - Folder directory in which to store the trained models [str] (default ="./models")
• --data_dir - Folder directory in which the datasets are stored [str] (default="./datasets")
• --datasets - The names of the datasets to use. [ [str] ] (default=[" "] )
• --epochs - Number of training epochs. [int] (default=60)
• --model - The name of the model to train ('TCN', 'RNN', 'GRU', 'LSTM') [str] (default=" ")
• --batch_size - The size of each batch [int] (default=512)
• --units = The hidden layer size (number of units) of the network. [ [int] ] (default=64)
• --input_size - The number of samples to use as input [int] (default=96)
• --model_internal_dim - The number of samples to be output by the compression layer (only apply to RNN, GRU, and LSTM models) [int] (default=4)
• --kernel_size - The size of the kernel in the TCN layers (only applies to the TCN model) [int] (default=3)
• --learning_rate - the initial learning rate [float] (default=3e-4)
• --only_inference - When True, skips training and runs only inference on the pre-model. When False, runs training and inference on the trained model. [bool] (default=False)

Example training case:

cd ./code/

python starter.py --datasets OSCMonoSquare --model LSTM --epochs 500

To only run inference on an existing pre-trained model, use the "only_inference". In this case, ensure you have the existing model and dataset (to use for inference) both in their respective directories with corresponding names.

Example inference case:

cd ./code/

python starter.py --datasets OSCMonoSquare --model LSTM --only_inference True

VST Download

Coming soon...

Bibtex

If you use the code included in this repository or any part of it, please acknowledge its authors by adding a reference to these publications:

@inproceedings{simionatoVCO_2025,
	author = {Simionato Riccardo and Fasciani Stefano},
	title = {Towards Neural Emulation of Voltage-Controlled Oscillator},
	booktitle = {Proceedings of the International Conference on Digital Audio Effects (DAFx25)},
	year = {2025},
    address = {Ancona, Italy},
}