Dual Attention in Time and Frequency Domain for Voice Activity Detection

Cited 5 time in webofscience Cited 0 time in scopus
  • Hit : 121
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorLee, Joohyungko
dc.contributor.authorJung, Youngmoonko
dc.contributor.authorKim, Hoi-Rinko
dc.date.accessioned2020-12-01T23:50:28Z-
dc.date.available2020-12-01T23:50:28Z-
dc.date.created2020-11-28-
dc.date.created2020-11-28-
dc.date.issued2020-10-29-
dc.identifier.citationInterspeech Conference, pp.3670 - 3674-
dc.identifier.issn2308-457X-
dc.identifier.urihttp://hdl.handle.net/10203/277890-
dc.description.abstractVoice activity detection (VAD) is a challenging task in low signal-to-noise ratio (SNR) environment, especially in non-stationary noise. To deal with this issue, we propose a novel attention module that can be integrated in Long Short-Term Memory (LSTM). Our proposed attention module refines each LSTM layer's hidden states so as to make it possible to adaptively focus on both time and frequency domain. Experiments are conducted on various noisy conditions using Aurora 4 database. Our proposed method obtains the 95.58 % area under the ROC curve (AUC), achieving 22.05 % relative improvement compared to baseline, with only 2.44 % increase in the number of parameters. Besides, we utilize focal loss for alleviating the performance degradation caused by imbalance between speech and non-speech sections in training sets. The results show that the focal loss can improve the performance in various imbalance situations compared to the cross entropy loss, a commonly used loss function in VAD.-
dc.languageEnglish-
dc.publisherISCA-
dc.titleDual Attention in Time and Frequency Domain for Voice Activity Detection-
dc.typeConference-
dc.identifier.wosid000833594103162-
dc.identifier.scopusid2-s2.0-85098195958-
dc.type.rimsCONF-
dc.citation.beginningpage3670-
dc.citation.endingpage3674-
dc.citation.publicationnameInterspeech Conference-
dc.identifier.conferencecountryCC-
dc.identifier.conferencelocationVirtual-
dc.identifier.doi10.21437/Interspeech.2020-997-
dc.contributor.localauthorKim, Hoi-Rin-
Appears in Collection
EE-Conference Papers(학술회의논문)
Files in This Item
There are no files associated with this item.
This item is cited by other documents in WoS
⊙ Detail Information in WoSⓡ Click to see webofscience_button
⊙ Cited 5 items in WoS Click to see citing articles in records_button

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0