Please use this identifier to cite or link to this item:
https://hdl.handle.net/20.500.11779/1989
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Karamatlı, Ertuğ | - |
dc.contributor.author | Kırbız, Serap | - |
dc.date.accessioned | 2023-10-18T12:06:14Z | |
dc.date.available | 2023-10-18T12:06:14Z | |
dc.date.issued | 2022 | - |
dc.identifier.citation | Karamatlı, E., & Kırbız, S. (2022). MixCycle: Unsupervised Speech Separation via Cyclic Mixture Permutation Invariant Training. IEEE Signal Processing Letters, 29, 2637-2641. | en_US |
dc.identifier.issn | 1070-9908 | - |
dc.identifier.issn | 1558-2361 | - |
dc.identifier.uri | https://hdl.handle.net/20.500.11779/1989 | - |
dc.identifier.uri | https://doi.org/10.1109/LSP.2022.3232276 | - |
dc.description.abstract | We introduce two unsupervised source separation methods, which involve self-supervised training from single-channel two-source speech mixtures. Our first method, mixture permutation invariant training (MixPIT), enables learning a neural network model which separates the underlying sources via a challenging proxy task without supervision from the reference sources. Our second method, cyclic mixture permutation invariant training (MixCycle), uses MixPIT as a building block in a cyclic fashion for continuous learning. MixCycle gradually converts the problem from separating mixtures of mixtures into separating single mixtures. We compare our methods to common supervised and unsupervised baselines: permutation invariant training with dynamic mixing (PIT-DM) and mixture invariant training (MixIT). We show that MixCycle outperforms MixIT and reaches a performance level very close to the supervised baseline (PIT-DM) while circumventing the over-separation issue of MixIT. Also, we propose a self-evaluation technique inspired by MixCycle that estimates model performance without utilizing any reference sources. We show that it yields results consistent with an evaluation on reference sources (LibriMix) and also with an informal listening test conducted on a real-life mixtures dataset (REAL-M). | en_US |
dc.language.iso | en | en_US |
dc.publisher | IEEE | en_US |
dc.rights | info:eu-repo/semantics/openAccess | en_US |
dc.subject | Training | en_US |
dc.subject | Recording | en_US |
dc.subject | Source separation | en_US |
dc.subject | Time-domain analysis | en_US |
dc.subject | Task analysis | en_US |
dc.subject | Optimized production technology | en_US |
dc.subject | Unsupervised learning | en_US |
dc.subject | Blind source separation | en_US |
dc.subject | deep learning | en_US |
dc.subject | self-supervised learning | en_US |
dc.subject | unsupervised learning | en_US |
dc.title | MixCycle: Unsupervised Speech Separation via Cyclic Mixture Permutation Invariant Training | en_US |
dc.type | Article | en_US |
dc.identifier.doi | 10.1109/LSP.2022.3232276 | - |
dc.identifier.scopus | 2-s2.0-85146250664 | en_US |
dc.description.woscitationindex | Science Citation Index Expanded | - |
dc.identifier.wosquality | Q2 | - |
dc.description.WoSDocumentType | article | |
dc.description.WoSInternationalCollaboration | Uluslararası işbirliği ile yapılmayan - HAYIR | en_US |
dc.description.WoSPublishedMonth | Ocak | en_US |
dc.description.WoSIndexDate | 2022 | en_US |
dc.description.WoSYOKperiod | YÖK - 2022-23 | en_US |
dc.identifier.scopusquality | Q1 | - |
dc.relation.publicationcategory | Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı | en_US |
dc.identifier.endpage | 2641 | en_US |
dc.identifier.startpage | 2637 | en_US |
dc.identifier.volume | 29 | en_US |
dc.department | Mühendislik Fakültesi, Endüstri Mühendisliği Bölümü | en_US |
dc.relation.journal | Ieee Signal Processing Letters | en_US |
dc.identifier.wos | WOS:000910559500004 | en_US |
dc.institutionauthor | Kırbız, Serap | - |
item.openairecristype | http://purl.org/coar/resource_type/c_18cf | - |
item.grantfulltext | open | - |
item.languageiso639-1 | en | - |
item.cerifentitytype | Publications | - |
item.fulltext | With Fulltext | - |
item.openairetype | Article | - |
crisitem.author.dept | 02.05. Department of Electrical and Electronics Engineering | - |
Appears in Collections: | Endüstri Mühendisliği Bölümü koleksiyonu Scopus İndeksli Yayınlar Koleksiyonu / Scopus Indexed Publications Collection WoS İndeksli Yayınlar Koleksiyonu / WoS Indexed Publications Collection |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
MixCycle_Unsupervised_Speech_Separation_via_Cyclic_Mixture_Permutation_Invariant_Training.pdf | Full Text- Article | 604.92 kB | Adobe PDF | View/Open |
CORE Recommender
SCOPUSTM
Citations
3
checked on Aug 1, 2024
WEB OF SCIENCETM
Citations
1
checked on Jun 23, 2024
Page view(s)
4
checked on Jun 26, 2024
Google ScholarTM
Check
Altmetric
Items in GCRIS Repository are protected by copyright, with all rights reserved, unless otherwise indicated.