Research institute for
signals, systems and
computational intelligence

Speech separation and dereverberation

In this line of research we address several speech and audio problems, related to sound source separation, reverberation reduction, denoising, and other related problems, using machine learning based methods

Objetive

To develop advanced signal processing and machine learning algorithms capable of producing high quality audio signals from samples aquired in adverse conditions.

Director

Leandro Di Persia

Publications

Publications in 2019
Penalized nonnegative representations for speech separation   sinc_ISD19a.pdf
F. Ibarrola, R. Spies, L. Di Persia
Proceedings of VII Congreso de Matemática Aplicada, Computacional e Industrial – 2019
Switching divergences for spectral learning in blind speech dereverberation   sinc_ISD19.doi sinc_ISD19.pdf
F. Ibarrola, R. Spies, L. Di Persia
IEEE/ACM Transactions on Audio, Speech, and Language Processing – 2019
Publications in 2018
A Bayesian approach to convolutive nonnegative matrix factorization for blind speech dereverberation   sinc_IDS18.doi sinc_IDS18.pdf
F. Ibarrola, L. Di Persia, R. Spies
Signal Processing, Volume 151, page 89–98 – 2018
Hybrid speech enhancement with wiener filters and deep LSTM Denoising Autoencoders   sinc_CGDR18.doi sinc_CGDR18.pdf
M. Coto-Jimenez, J. Goddard, L. Di Persia, H. L. Rufiner
2018 IEEE International Work Conference on Bioinspired Intelligence (IWOBI) – 2018
Publications in 2017
Blind speech dereverberation using convolutive nonnegative matrix factorization with mixed penalization   sinc_SDI17.pdf
F. Ibarrola, L. Di Persia, R. Spies
Proceedings of VI MACI – 2017
On the use of convolutive nonnegative matrix factorization with mixed penalization for blind speech dereverberation   sinc_IDS17a.doi sinc_IDS17a.pdf
F. Ibarrola, L. Di Persia, R. Spies
2017 XLIII Latin American Computer Conference (CLEI), page 1–4 – 2017
Publications in 2010
Denoising and Recognition using Hidden Markov Models with Observation Distributions Modeled by Hidden Markov Trees   sinc_MDT10.doi po.http sinc_MDT10.pdf
D. H. Milone, L. Di Persia, M. E. Torres
Pattern Recognition, Number 43, page 1577–1589 – april 2010
Publications in 2009
Machine Learning Techniques for Audio Blind Source Separation applied to Automatic Speech Recognition   sinc_Di 09.pdf
L. Di Persia
PhD thesis from Universidad Nacional del Litoral – march 2009
Pseudoanechoic blind source separation with improved Wiener postfilter   sinc_DMY09a.pdf
L. Di Persia, D. H. Milone, Masuzo Yanagida
XIII Reunión de Trabajo en Procesamiento de la Información y Control (RPIC 2009), page 18–23 – September 2009
Publications in 2008
Perceptual evaluation of blind source separation for robust speech recognition   sinc_DMRY08.doi sinc_DMRY08.pdf
L. Di Persia, D. H. Milone, H. L. Rufiner, Masuzo Yanagida
Signal Processing, Volume 88, Number 10, page 2578–2583 – 2008
Publications in 2007
Objective quality evaluation in blind source separation for speech recognition in a real room   sinc_DYRM07.pdf
L. Di Persia, Masuzo Yanagida, H. L. Rufiner, D. H. Milone
Signal Processing, Volume 87, Number 8, page 1951–1965 – 2007
Publications in 2006
Convolutive Blind Source Separation with Wiener post-filtering for robust Speech Recognition   sinc_DMY06b.pdf
L. Di Persia, D. H. Milone, Masuzo Yanagida
7th Argentine Symposium on Computing Technology – 2006
Separation of mixed speech signals of short duration using Wiener filter as postprocessing for Frequency-Domain ICA   sinc_HDOY06.pdf
Akihiro Handa, L. Di Persia, Kenko Ohta, Masuzo Yanagida
IPSJ SIG Notes, Volume 2006, Number 12, page 1-6 – 2006
Publications in 2004
Separation of Mixed Speech in Reverberant Environment Using Independent Component Analysis   sinc_HTDY04.pdf
Akihiro Handa, Kensuke Tsutsumi, L. Di Persia, Masuzo Yanagida
Spring Meeting of Japanese Society of Information Processing – 2004