Biologically-inspired auditory artificial intelligence for speech recognition in multi-talker environments

Simple item page
dc.contributor.authorGrasse, Lukas Walter Neufeld
dc.contributor.supervisorTata, Matthew S.
dc.contributor.supervisorLuczak, Artur
dc.date.accessioned2020-12-23T22:23:28Z
dc.date.available2020-12-23T22:23:28Z
dc.date.issued2020
dc.degree.levelMastersen_US
dc.description.abstractUnderstanding speech in the presence of distracting talkers is a difficult computational problem known as the cocktail party problem. Motivated by auditory processing in the human brain, this thesis developed a neural network to isolate the speech of a single talker given binaural input containing a target talker and multiple distractors. In this research the network is called a Binaural Speaker Isolation FFTNet or BSINet for short. To compare the performance of BSINet to human participant performance on recognizing the target talker's speech with a varying number of distractors, a "cocktail party" dataset was designed and made available online. This dataset also enables the comparison of network performance to human participant performance. Using the Word-Error-Rate metric for evaluation, this research finds that BSINet performs comparably to the human participants. Thus BSINet provides significant advancement for solving the challenging cocktail party problem.en_US
dc.description.sponsorshipThe research was funded by an NSERC Canada Discovery Grant, a Government of Alberta Centre for Autonomous Systems in Strengthening Future Communities grant, a MITACS Globalink Award, a NSERC CGS-M Award, and a AITF Graduate Student Scholarship.en_US
dc.identifier.urihttps://hdl.handle.net/10133/5815
dc.language.isoen_USen_US
dc.proquest.subject0317en_US
dc.proquest.subject0800en_US
dc.proquest.subject0984en_US
dc.proquestyesYesen_US
dc.publisherLethbridge, Alta. : University of Lethbridge, Dept. of Neuroscienceen_US
dc.publisher.departmentDepartment of Neuroscienceen_US
dc.publisher.facultyArts and Scienceen_US
dc.relation.ispartofseriesThesis (University of Lethbridge. Faculty of Arts and Science)en_US
dc.subjectSpeech Recognitionen_US
dc.subjectDenoisingen_US
dc.subjectSpeaker Isolationen_US
dc.subjectCocktail Party Problemen_US
dc.subjectAuditory selective attentionen_US
dc.subjectNeural networks (Computer science)en_US
dc.subjectSpeech perceptionen_US
dc.subjectAutomatic speech recognitionen_US
dc.subjectDirectional hearingen_US
dc.subjectAuditory perceptionen_US
dc.subjectDissertations, Academicen_US
dc.titleBiologically-inspired auditory artificial intelligence for speech recognition in multi-talker environmentsen_US
dc.typeThesisen_US
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
GRASSE_LUKAS_MSC_2020.pdf
Size:
545.42 KB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
3.25 KB
Format:
Item-specific license agreed upon to submission
Description:
Download
Collections
Arts and Science, Faculty of
University of Lethbridge Theses