Search Results for author:
Found 2 papers, 1 papers with code
Music Enhancement with Deep Filters: A Technical Report for The ICASSP 2024 Cadenza Challenge
In this challenge, we disentangle the deep filters from the original DeepfilterNet and incorporate them into our Spec-UNet-based network to further improve a hybrid Demucs (hdemucs) based remixing pipeline.
Towards Improving Harmonic Sensitivity and Prediction Stability for Singing Melody Extraction
In deep learning research, many melody extraction models rely on redesigning neural network architectures to improve performance.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
