{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/101644"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/101644","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Speech enhancement using deep dilated CNN","abstract":"In recent years, deep learning has achieved great success in speech enhancement. However, there are two major limitations regarding existing works. First, the Bayesian framework is not adopted in many such deep-learning-based algorithms. In particular, the prior distribution for speech in the Bayesian framework has been shown useful by regularizing the output to be in the speech space, and thus improving the performance. Second, the majority of the existing methods operate on the frequency domain of the noisy speech, such as spectrogram and its variations. We propose a Bayesian speech enhancement framework, called BaWN (Bayesian WaveNet), which directly operates on raw audio samples. It adopts the recently announced WaveNet, which is shown to be effective in modeling conditional distributions of speech samples while generating natural speech. Experiments show that BaWN is able to recover clean and natural speech. Multi-channel speech enhancement with ad-hoc sensors has been a challenging task. Speech model guided beamforming algorithms are able to recover natural sounding speech, but the speech models tend to be oversimplified to prevent the inference from becoming too complicated. On the other hand, deep learning based enhancement approaches are able to learn complicated speech distributions and perform efficient inference, but they are unable to deal with variable number of input channels. Also, deep learning approaches introduce a lot of errors, particularly in the presence of unseen noise types and settings. We have therefore proposed an enhancement framework called DeepBeam, which combines the two complementary classes of algorithms. DeepBeam introduces a beamforming filter to produce natural sounding speech, but the filter coefficients are determined with the help of a monaural speech enhancement neural network. Experiments on synthetic and real-world data show that DeepBeam is able to produce clean, dry and natural sounding speech, and is robust against unseen noise.","abstract_html":"In recent years, deep learning has achieved great success in speech enhancement. However, there are two major limitations regarding existing works. First, the Bayesian framework is not adopted in many such deep-learning-based algorithms. In particular, the prior distribution for speech in the Bayesian framework has been shown useful by regularizing the output to be in the speech space, and thus improving the performance. Second, the majority of the existing methods operate on the frequency domain of the noisy speech, such as spectrogram and its variations. We propose a Bayesian speech enhancement framework, called BaWN (Bayesian WaveNet), which directly operates on raw audio samples. It adopts the recently announced WaveNet, which is shown to be effective in modeling conditional distributions of speech samples while generating natural speech. Experiments show that BaWN is able to recover clean and natural speech. Multi-channel speech enhancement with ad-hoc sensors has been a challenging task. Speech model guided beamforming algorithms are able to recover natural sounding speech, but the speech models tend to be oversimplified to prevent the inference from becoming too complicated. On the other hand, deep learning based enhancement approaches are able to learn complicated speech distributions and perform efficient inference, but they are unable to deal with variable number of input channels. Also, deep learning approaches introduce a lot of errors, particularly in the presence of unseen noise types and settings. We have therefore proposed an enhancement framework called DeepBeam, which combines the two complementary classes of algorithms. DeepBeam introduces a beamforming filter to produce natural sounding speech, but the filter coefficients are determined with the help of a monaural speech enhancement neural network. Experiments on synthetic and real-world data show that DeepBeam is able to produce clean, dry and natural sounding speech, and is robust against unseen noise.","abstract_has_math":false,"creators":["Qian, Kaizhi"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Electrical & Computer Engr","degree_department":null,"school":null,"contributors":["Hasegawa-Johnson, Mark"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2018,"date_issued":"2018-09-27T16:28:02Z","date_published":"2018-09-27T16:28:02Z","updated_at":"2026-07-22T22:24:40Z","subjects":["speech enhancement","convolutional neural network","beamforming"],"languages":["en"],"rights":["Copyright 2018 Kaizhi Qian"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/101644","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Hasegawa-Johnson, Mark"]},{"key":"dc:creator","label":"Author","values":["Qian, Kaizhi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2018-09-27T16:28:02Z","2020-09-28T09:15:07Z","2018-05-22","2018-08"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical & Computer Engr"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["speech enhancement","convolutional neural network","beamforming"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2018 Kaizhi Qian"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/101644"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["In recent years, deep learning has achieved great success in speech enhancement. However, there are two major limitations regarding existing works. First, the Bayesian framework is not adopted in many such deep-learning-based algorithms. In particular, the prior distribution for speech in the Bayesian framework has been shown useful by regularizing the output to be in the speech space, and thus improving the performance. Second, the majority of the existing methods operate on the frequency domain of the noisy speech, such as spectrogram and its variations. We propose a Bayesian speech enhancement framework, called BaWN (Bayesian WaveNet), which directly operates on raw audio samples. It adopts the recently announced WaveNet, which is shown to be effective in modeling conditional distributions of speech samples while generating natural speech. Experiments show that BaWN is able to recover clean and natural speech. Multi-channel speech enhancement with ad-hoc sensors has been a challenging task. Speech model guided beamforming algorithms are able to recover natural sounding speech, but the speech models tend to be oversimplified to prevent the inference from becoming too complicated. On the other hand, deep learning based enhancement approaches are able to learn complicated speech distributions and perform efficient inference, but they are unable to deal with variable number of input channels. Also, deep learning approaches introduce a lot of errors, particularly in the presence of unseen noise types and settings. We have therefore proposed an enhancement framework called DeepBeam, which combines the two complementary classes of algorithms. DeepBeam introduces a beamforming filter to produce natural sounding speech, but the filter coefficients are determined with the help of a monaural speech enhancement neural network. Experiments on synthetic and real-world data show that DeepBeam is able to produce clean, dry and natural sounding speech, and is robust against unseen noise.","Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2020-08-01","The student, Kaizhi Qian, accepted the attached license on 2018-05-22 at 11:31.","The student, Kaizhi Qian, submitted this Thesis for approval on 2018-05-22 at 11:56.","This Thesis was approved for publication on 2018-05-22 at 16:40.","DSpace SAF Submission Ingestion Package generated from Vireo submission #12588 on 2018-09-27 at 11:15:47","Made available in DSpace on 2018-09-27T16:28:02Z (GMT). No. of bitstreams: 2 QIAN-THESIS-2018.pdf: 1052365 bytes, checksum: a5061b7d424b4b588ec83697f59276db (MD5) LICENSE.txt: 4208 bytes, checksum: 0677070bb347af75d8256b41853b37cc (MD5) Previous issue date: 2018-05-22","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:28:07Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:30:34Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:31:43Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:34:29Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","U of I Only Restriction Lifted for Item 107741 on 2020-09-28T09:15:07Z."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Speech enhancement using deep dilated CNN"]}]}],"canonical_facts":{"dc:contributor":["Hasegawa-Johnson, Mark"],"dc:creator":["Qian, Kaizhi"],"dc:date":["2018-09-27T16:28:02Z","2020-09-28T09:15:07Z","2018-05-22","2018-08"],"dc:description":["In recent years, deep learning has achieved great success in speech enhancement. However, there are two major limitations regarding existing works. First, the Bayesian framework is not adopted in many such deep-learning-based algorithms. In particular, the prior distribution for speech in the Bayesian framework has been shown useful by regularizing the output to be in the speech space, and thus improving the performance. Second, the majority of the existing methods operate on the frequency domain of the noisy speech, such as spectrogram and its variations. We propose a Bayesian speech enhancement framework, called BaWN (Bayesian WaveNet), which directly operates on raw audio samples. It adopts the recently announced WaveNet, which is shown to be effective in modeling conditional distributions of speech samples while generating natural speech. Experiments show that BaWN is able to recover clean and natural speech. Multi-channel speech enhancement with ad-hoc sensors has been a challenging task. Speech model guided beamforming algorithms are able to recover natural sounding speech, but the speech models tend to be oversimplified to prevent the inference from becoming too complicated. On the other hand, deep learning based enhancement approaches are able to learn complicated speech distributions and perform efficient inference, but they are unable to deal with variable number of input channels. Also, deep learning approaches introduce a lot of errors, particularly in the presence of unseen noise types and settings. We have therefore proposed an enhancement framework called DeepBeam, which combines the two complementary classes of algorithms. DeepBeam introduces a beamforming filter to produce natural sounding speech, but the filter coefficients are determined with the help of a monaural speech enhancement neural network. Experiments on synthetic and real-world data show that DeepBeam is able to produce clean, dry and natural sounding speech, and is robust against unseen noise.","Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2020-08-01","The student, Kaizhi Qian, accepted the attached license on 2018-05-22 at 11:31.","The student, Kaizhi Qian, submitted this Thesis for approval on 2018-05-22 at 11:56.","This Thesis was approved for publication on 2018-05-22 at 16:40.","DSpace SAF Submission Ingestion Package generated from Vireo submission #12588 on 2018-09-27 at 11:15:47","Made available in DSpace on 2018-09-27T16:28:02Z (GMT). No. of bitstreams: 2 QIAN-THESIS-2018.pdf: 1052365 bytes, checksum: a5061b7d424b4b588ec83697f59276db (MD5) LICENSE.txt: 4208 bytes, checksum: 0677070bb347af75d8256b41853b37cc (MD5) Previous issue date: 2018-05-22","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:28:07Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:30:34Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:31:43Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","Embargo set by: Seth Robbins for item 107741 Lift date: 2020-09-27T16:34:29Z Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD system","U of I Only Restriction Lifted for Item 107741 on 2020-09-28T09:15:07Z."],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/101644"],"dc:language":["en"],"dc:rights":["Copyright 2018 Kaizhi Qian"],"dc:subject":["speech enhancement","convolutional neural network","beamforming"],"dc:title":["Speech enhancement using deep dilated CNN"],"dc:type":["text"],"thesis:degree_discipline":["Electrical & Computer Engr"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:40Z"}