ICASSP 2018 accepted our paper: BitWise Sourse Separation (BWSS)
As a recent detour to de-noising speech audio signal, my paper “Bitwise Source Separation on Hashed Spectra: An Efficient Posterior Estimation Scheme Using Partial Rank Order Metrics”, has been accepted by the 2018 International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Calgary, Canada.
BWSS (demo link) is a iteration free, dictionary based, semi-supervised denoising algorithm for non-stationary noises. It achieved 4-15 db noise reduction, with average 7.5 db. (paper link).
Previous Dec 10, 2017
« NIPS workshop on Machine Learning for Audio
« NIPS workshop on Machine Learning for Audio
Feb 16, 2019 Next
Introduction to Dirichlet Process Gaussian Mixture Model »
Introduction to Dirichlet Process Gaussian Mixture Model »