My PhD thesis concerned separating the audio corresponding to the instruments in an orchestral music mixture. This allows for interesting applications such as re-creating the experience of the concert in virtual reality applications.
My research interests span a diverse range of topics: machine learning, signal processing, fairness and explainability of machine learning models, music information retrieval. I participated in interdisciplinary projects, such as HUMAINT, Shake-it and PHENICX.
Check out my latest deep learning repository in python. I am committed to the principles of research reproducibility. Most of the code is made available through github, along with links to the dataset and instructions on how to replicate experiments.
High frequency magnitude spectrogram reconstruction for music mixtures using convolutional autoencoders
autoencoders in frequency domain yet again ... Read More ›