東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Hearing through the noise: Biologica...

Lee, Tyler Paul.

FindBook

Google Book

Amazon

博客來

Hearing through the noise: Biologically inspired noise reduction.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Hearing through the noise: Biologically inspired noise reduction./
作者:	Lee, Tyler Paul.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2016,
面頁冊數:	104 p.
附註:	Source: Dissertation Abstracts International, Volume: 78-07(E), Section: B.
Contained By:	Dissertation Abstracts International78-07B(E).
標題:	Acoustics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10190273
ISBN:	9781369558234

Hearing through the noise: Biologically inspired noise reduction.
Lee, Tyler Paul.

Hearing through the noise: Biologically inspired noise reduction. - Ann Arbor : ProQuest Dissertations & Theses, 2016 - 104 p.

Source: Dissertation Abstracts International, Volume: 78-07(E), Section: B.

Thesis (Ph.D.)--University of California, Berkeley, 2016.

Vocal communication in the natural world demands that a listener perform a remarkably complicated task in real-time. Vocalizations mix with all other sounds in the environment as they travel to the listener, arriving as a jumbled low-dimensional signal. A listener must then use this signal to extract the structure corresponding to individual sound sources. How this computation is implemented in the brain remains poorly understood, yet an accurate description of such mechanisms would impact a variety of medical and technological applications of sound processing. In this thesis, I describe initial work on how neurons in the secondary auditory cortex of the Zebra Finch extract song from naturalistic background noise. I then build on our understanding of the function of these neurons by creating an algorithm that extracts speech from natural background noise using spectrotemporal modulations. The algorithm, implemented as an artificial neural network, can be flexibly applied to any class of signal or noise and performs better than an optimal frequency-based noise reduction algorithm for a variety of background noises and signal-to-noise ratios. One potential drawback to using spectrotemporal modulations for noise reduction, though, is that analyzing the modulations present in an ongoing sound requires a latency set by the slowest temporal modulation computed. The algorithm avoids this problem by reducing noise predictively, taking advantage of the large amount of temporal structure present in natural sounds. This predictive denoising has ties to recent work suggesting that the auditory system uses attention to focus on predicted regions of spectrotemporal space when performing auditory scene analysis.

ISBN: 9781369558234Subjects--Topical Terms:

879105
Acoustics.

Hearing through the noise: Biologically inspired noise reduction.
LDR:02717nmm a2200301 4500 001 2122526
005 20170922124924.5
008 180830s2016 ||||||||||||||||| ||eng d
020 $a 9781369558234
035 $a (MiAaPQ)AAI10190273
035 $a AAI10190273
040 $a MiAaPQ $c MiAaPQ
100 1 $a Lee, Tyler Paul. $3 3284498
245 1 0 $a Hearing through the noise: Biologically inspired noise reduction.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2016
300 $a 104 p.
500 $a Source: Dissertation Abstracts International, Volume: 78-07(E), Section: B.
500 $a Includes supplementary digital materials.
500 $a Advisers: Frederic Theunissen; Bruno Olshausen.
502 $a Thesis (Ph.D.)--University of California, Berkeley, 2016.
520 $a Vocal communication in the natural world demands that a listener perform a remarkably complicated task in real-time. Vocalizations mix with all other sounds in the environment as they travel to the listener, arriving as a jumbled low-dimensional signal. A listener must then use this signal to extract the structure corresponding to individual sound sources. How this computation is implemented in the brain remains poorly understood, yet an accurate description of such mechanisms would impact a variety of medical and technological applications of sound processing. In this thesis, I describe initial work on how neurons in the secondary auditory cortex of the Zebra Finch extract song from naturalistic background noise. I then build on our understanding of the function of these neurons by creating an algorithm that extracts speech from natural background noise using spectrotemporal modulations. The algorithm, implemented as an artificial neural network, can be flexibly applied to any class of signal or noise and performs better than an optimal frequency-based noise reduction algorithm for a variety of background noises and signal-to-noise ratios. One potential drawback to using spectrotemporal modulations for noise reduction, though, is that analyzing the modulations present in an ongoing sound requires a latency set by the slowest temporal modulation computed. The algorithm avoids this problem by reducing noise predictively, taking advantage of the large amount of temporal structure present in natural sounds. This predictive denoising has ties to recent work suggesting that the auditory system uses attention to focus on predicted regions of spectrotemporal space when performing auditory scene analysis.
590 $a School code: 0028.
650 4 $a Acoustics. $3 879105
650 4 $a Neurosciences. $3 588700
690 $a 0986
690 $a 0317
710 2 $a University of California, Berkeley. $b Neuroscience. $3 3191789
773 0 $t Dissertation Abstracts International $g 78-07B(E).
790 $a 0028
791 $a Ph.D.
792 $a 2016
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10190273