Filters








14 Hits in 10.1 sec

Non-negative tensor factorisation of modulation spectrograms for monaural sound source separation

Tom Barker, Tuomas Virtanen
2013 Interspeech 2013   unpublished
This paper proposes an algorithm for separating monaural audio signals by non-negative tensor factorisation of modulation spectrograms.  ...  The method overcomes the limitation of conventional non-negative matrix factorisation algorithms to utilise the redundancy of sounds in frequency.  ...  Conclusions This paper has proposed a novel algorithm for monaural sound source separation which used the modulation spectrogram as a feature in non-negative tensor factorisation.  ... 
doi:10.21437/interspeech.2013-236 fatcat:wj6almndxrel5kwl4pv2fxohn4

Semi-supervised non-negative tensor factorisation of modulation spectrograms for monaural speech separation

Tom Barker, Tuomas Virtanen
2014 2014 International Joint Conference on Neural Networks (IJCNN)  
A mixture signal is separated through factorisation of a feature-tensor representation, based on the modulation spectrogram.  ...  Specifically, the proposed semi-supervised method outperformed both semi-supervised non-negative matrix factorisation and blind non-negative modulation spectrum tensor factorisation.  ...  Non-negative matrix factorisation (NMF) and non-negative tensor factorisation (NTF) produce state-of-the-art blind single source separation [5] , [6] , [7] , [8] .  ... 
doi:10.1109/ijcnn.2014.6889522 dblp:conf/ijcnn/BarkerV14 fatcat:dckn45sdh5d7thpjtdjlrzsdui

Estimation Of Vocal Duration In Monaural Mixtures

Anders Elowsson, Ragnar Schön, Matts Höglund, Elias Zea, Anders Friberg
2014 Proceedings of the SMC Conferences  
We wish to thank Derry FitzGerald for kindly providing source code and Emilia Gómez for introducing us to Melodia during the summer school.  ...  Acknowledgments This study was initiated during the SMC summer school of 2013. We wish to thank our fellow group member Hama Biglari for his valuable contribution.  ...  In order to further improve the separation performance and to remove some artifacts, FitzGerald and Gainza have also added a tensor factorisation-based separation and a non-negative partial cofactorisation  ... 
doi:10.5281/zenodo.850804 fatcat:tlgwm4entbhf3oefdpexzxrdty

Non-Negative Matrix Factorization Based Algorithms to Cluster Frequency Basis Functions for Monaural Sound Source Separation

Rajesh Jaiswal
2013
The Non negative Matrix Factorisation For musical analysis, non-negative matrix factorization (NMF) has been [23] shown to be a useful decomposition of audio spectrograms.  ...  In recent years, many factorisation techniques, such as Non-negative Matrix Factorisation (NMF) [23] of magnitude spectrograms have been proposed to separate out sources from spectrograms [24, 28, 25  ... 
doi:10.21427/d7v894 fatcat:2rtg5f5lwnc6rixuaqzquba6ny

A virtual auditory environment for investigating the auditory signal processing of realistic sounds

Sylvain Favrot, Jörg M. Buchholz
2008 Journal of the Acoustical Society of America  
Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ...  One such technique, that has been particularly successful, is non-negative matrix factorisation ͑NMF͒.  ...  Beam-forming analyses were performed on four-microphone linear arrays placed just within the interior of a pine forest to separate signal arrivals by direction from an impulsive sound source in an open  ... 
doi:10.1121/1.2936003 fatcat:ontu6yamdvgbnooet5fe36fm74

Analysis of room transfer function and reverberant signal statistics

Eleftheria Georganti, John Mourjopoulos, Finn Jacobsen
2008 Journal of the Acoustical Society of America  
Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ...  One such technique, that has been particularly successful, is non-negative matrix factorisation ͑NMF͒.  ...  Beam-forming analyses were performed on four-microphone linear arrays placed just within the interior of a pine forest to separate signal arrivals by direction from an impulsive sound source in an open  ... 
doi:10.1121/1.2935346 fatcat:lxgnqr6tozajhge3ydet3xomam

Amplitude modulation depth discrimination in hearing‐impaired and normal‐hearing listeners

Stephan D. Ewert, Jutta Volmer, Torsten Dau, Jesko Verhey
2008 Journal of the Acoustical Society of America  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1121/1.2935715 fatcat:qsqohnv63fdsjgegqb2h5s7r7q

Cross‐spectral synergy and consonant identification

Thomas Ulrich Christiansen, Steven Greenberg
2008 Journal of the Acoustical Society of America  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1121/1.2935680 fatcat:znwocm35unasharfwbcap5omim

The perceptual flow of phonetic feature processing

Steven Greenberg, Thomas Ulrich Christiansen
2008 Journal of the Acoustical Society of America  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1121/1.2935993 fatcat:l2ix2ph575gh7l27xtjq264qkm

Temporal suppression and augmentation of click-evoked otoacoustic emissions

Sarah Verhulst, James M. Harte, Torsten Dau
2008 Hearing Research  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1016/j.heares.2008.09.008 pmid:18926894 fatcat:bijdlpxpzfcsrjeefayvvu4pl4

Temporal suppression and augmentation of click‐evoked otoacoustic emissions

Sarah Verhulst, James M. Harte, Torsten Dau
2008 Journal of the Acoustical Society of America  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1121/1.2935694 fatcat:z6yidzqqendznbmrczb3woaiv4

Across‐frequency processes involved in auditory detection of coloration

Jörg M. Buchholz, Paris Kerketsos
2008 Journal of the Acoustical Society of America  
Examples are presented for a room with a diffuse sound field, and for elongated rooms where the sound field is strongly non-diffuse, in the case of both diffuse and specular reflections. 3760 Enclosures  ...  Annually over 5000 construction sites now use this system approach for separating walls and floors for attached houses and apartments.  ...  Bayesian inference in hierarchical non-negative matrix factorisation models of musical sounds.  ... 
doi:10.1121/1.2935741 fatcat:acaporn5qvgb5coauovqkbv6lq

Deep learning for speech separation

Alfredo Zermini
2020
Speech source separation aims to estimate one or more individual sources from mixtures of multiple sound sources, e.g. speech, noise and music.  ...  While humans have an innate ability to separate sources in a sound mixture, this is not a trivial task for computers.  ...  Non-negative Matrix Factorisation (NMF) Non-negative Matrix Factorisation (NMF) aims to decompose a non-negative matrix X ∈ R R×T into the multiplication of two non-negative matrices A ∈ R R×G and S ∈  ... 
doi:10.15126/thesis.00854147 fatcat:lojqk6jpzfcl5ax3ax3rdlu22u

Blind Source Separation: New Proof of Bounded Component Analysis and Nonnegative Matrix Factorization Algorithms for Monaural Audio [article]

Holly Gao
2020
This algorithm was extended using the sinusoidal model for monaural musical sound separation [96] .  ...  Moreover, NMF has been extended to nonnegative tensor factorization (NTF) by a multiple channel filterbank in the modulation spectrogram (MS) domain [9] .  ... 
doi:10.26182/5f3b7ca5bba0b fatcat:t56bhknatbcgznspodhvqxubp4