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

Linear dynamic model for continuous ...

Ma, Tao.

FindBook

Google Book

Amazon

博客來

Linear dynamic model for continuous speech recognition.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Linear dynamic model for continuous speech recognition./
作者:	Ma, Tao.
面頁冊數:	138 p.
附註:	Source: Dissertation Abstracts International, Volume: 72-06, Section: B, page: 3602.
Contained By:	Dissertation Abstracts International72-06B.
標題:	Computer engineering. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3450275
ISBN:	9781124587981

Linear dynamic model for continuous speech recognition.
Ma, Tao.

Linear dynamic model for continuous speech recognition. - 138 p.

Source: Dissertation Abstracts International, Volume: 72-06, Section: B, page: 3602.

Thesis (Ph.D.)--Mississippi State University, 2011.

This item is not available from ProQuest Dissertations & Theses.

In the past decades, statistics-based hidden Markov models (HMMs) have become the predominant approach to speech recognition. Under this framework, the speech signal is modeled as a piecewise stationary signal (typically over an interval of 10 milliseconds). Speech features are assumed to be temporally uncorrelated. While these simplifications have enabled tremendous advances in speech processing systems, for the past several years progress on the core statistical models has stagnated. Since machine performance still significantly lags human performance, especially in noisy environments, researchers have been looking beyond the traditional HMM approach.

ISBN: 9781124587981Subjects--Topical Terms:

621879
Computer engineering.

Linear dynamic model for continuous speech recognition.
LDR:03177nmm a2200301 4500 001 2075109
005 20161017114307.5
008 170521s2011 ||||||||||||||||| ||eng d
020 $a 9781124587981
035 $a (MiAaPQ)AAI3450275
035 $a AAI3450275
040 $a MiAaPQ $c MiAaPQ
100 1 $a Ma, Tao. $3 1680454
245 1 0 $a Linear dynamic model for continuous speech recognition.
300 $a 138 p.
500 $a Source: Dissertation Abstracts International, Volume: 72-06, Section: B, page: 3602.
500 $a Adviser: Saurabh Prasad.
502 $a Thesis (Ph.D.)--Mississippi State University, 2011.
506 $a This item is not available from ProQuest Dissertations & Theses.
520 $a In the past decades, statistics-based hidden Markov models (HMMs) have become the predominant approach to speech recognition. Under this framework, the speech signal is modeled as a piecewise stationary signal (typically over an interval of 10 milliseconds). Speech features are assumed to be temporally uncorrelated. While these simplifications have enabled tremendous advances in speech processing systems, for the past several years progress on the core statistical models has stagnated. Since machine performance still significantly lags human performance, especially in noisy environments, researchers have been looking beyond the traditional HMM approach.
520 $a Recent theoretical and experimental studies suggest that exploiting frame-to-frame correlations in a speech signal further improves the performance of ASR systems. This is typically accomplished by developing an acoustic model which includes higher order statistics or trajectories. Linear Dynamic Models (LDMs) have generated significant interest in recent years due to their ability to model higher order statistics. LDMs use a state space-like formulation that explicitly models the evolution of hidden states using an autoregressive process. This smoothed trajectory model allows the system to better track the speech dynamics in noisy environments.
520 $a In this dissertation, we develop a hybrid HMM/LDM speech recognizer that effectively integrates these two powerful technologies. This hybrid system is capable of handling large recognition tasks, is robust to noise-corrupted speech data and mitigates the ill-effects of mismatched training and evaluation conditions. This two-pass system leverages the temporal modeling and N-best list generation capabilities of the traditional HMM architecture in a first pass analysis. In the second pass, candidate sentence hypotheses are re-ranked using a phone-based LDM model. The Wall Street Journal (WSJ0) derived Aurora-4 large vocabulary corpus was chosen as the training and evaluation dataset. This corpus is a well-established LVCSR benchmark with six different noisy conditions. The implementation and evaluation of the proposed hybrid HMM/LDM speech recognizer is the major contribution of this dissertation.
590 $a School code: 0132.
650 4 $a Computer engineering. $3 621879
690 $a 0464
710 2 $a Mississippi State University. $b Electrical and Computer Engineering. $3 1044030
773 0 $t Dissertation Abstracts International $g 72-06B.
790 $a 0132
791 $a Ph.D.
792 $a 2011
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3450275