CINXE.COM

<!DOCTYPE html> <html xmlns="http://www.w3.org/1999/xhtml"> <head> <title>Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition</title>  <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <meta http-equiv="X-UA-Compatible" content="ie=edge"> <meta name="title" content="Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition"> <meta name="description" content="The performance of various acoustic feature extraction methods has been compared in this work using Long Short-Term Memory (LSTM) neural network in a Bangla speech recognition system. The acoustic features are a series of vectors that represents the speech signals. They can be classified in either words or sub word units such as phonemes. In this work, at first linear predictive coding (LPC) is used as acoustic vector extraction technique. LPC has been chosen due to its widespread popularity. Then other vector extraction techniques like Mel frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) have also been used. These two methods closely resemble the human auditory system. These feature vectors are then trained using the LSTM neural network. Then the obtained models of different phonemes are compared with different statistical tools namely Bhattacharyya Distance and Mahalanobis Distance to investigate the nature of those acoustic features"/> <meta name="keywords" content="LSTM, Perceptual linear prediction, Mel frequency cepstral coefficients, Bhattacharyya Distance, Mahalanobis Distance"/>   <meta name="dc.title" content="Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition"> <meta name="citation_author" content="Nahyan Al Mahmud"> <meta name="citation_author" content="Shahfida Amjad Munni"> <meta name="dc.type" content="Article"> <meta name="dc.source" content="The International Journal of Multimedia & Its Applications (IJMA), Vol 12, No.5"> <meta name="dc.date" content="2020/10/30"> <meta name="dc.identifier" content="10.5121/ijma.2020.12501"> <meta name="dc.publisher" content="AIRCC Publishing Corporation"> <meta name="dc.rights" content="http://creativecommons.org/licenses/by/3.0/"> <meta name="dc.format" content="application/pdf"> <meta name="dc.language" content="en"> <meta name="dc.description" content="The performance of various acoustic feature extraction methods has been compared in this work using Long Short-Term Memory (LSTM) neural network in a Bangla speech recognition system. The acoustic features are a series of vectors that represents the speech signals. They can be classified in either words or sub word units such as phonemes. In this work, at first linear predictive coding (LPC) is used as acoustic vector extraction technique. LPC has been chosen due to its widespread popularity. Then other vector extraction techniques like Mel frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) have also been used. These two methods closely resemble the human auditory system. These feature vectors are then trained using the LSTM neural network. Then the obtained models of different phonemes are compared with different statistical tools namely Bhattacharyya Distance and Mahalanobis Distance to investigate the nature of those acoustic features."/> <meta name="dc.subject" content="LSTM"> <meta name="dc.subject" content="Perceptual linear prediction"> <meta name="dc.subject" content="Mel frequency cepstral coefficients"> <meta name="dc.subject" content="Bhattacharyya Distance"> <meta name="dc.subject" content="Mahalanobis Distance">   <meta name="prism.publicationName" content="The International Journal of Multimedia & Its Applications (IJMA)"> <meta name="prism.publicationDate" content="2020/10/30"> <meta name="prism.volume" content="12"> <meta name="prism.number" content="5"> <meta name="prism.section" content="Article"> <meta name="prism.startingPage" content="1">   <meta name="citation_journal_title" content="The International Journal of Multimedia & Its Applications (IJMA)"> <meta name="citation_publisher" content="AIRCC Publishing Corporation"> <meta name="citation_authors" content="Nahyan Al Mahmud and Shahfida Amjad Munni"> <meta name="citation_title" content="Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition"> <meta name="citation_online_date" content="2020/10/30"> <meta name="citation_volume" content="12"> <meta name="citation_issue" content="5"> <meta name="citation_firstpage" content="1"> <meta name="citation_author" content="Nahyan Al Mahmud"> <meta name="citation_author" content="Shahfida Amjad Munni"> <meta name="citation_doi" content="10.5121/ijma.2020.12501"> <meta name="citation_abstract_html_url" content="https://aircconline.com/abstract/ijma/v12n5/12520ijma01.html"> <meta name="citation_pdf_url" content="https://aircconline.com/ijma/V12N5/12520ijma01.pdf">   <meta property="og:site_name" content="AIRCC" /> <meta property="og:type" content="article" /> <meta property="og:url" content="https://aircconline.com/abstract/ijma/v12n5/12520ijma01.html"/> <meta property="og:title" content="Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition"> <meta property="og:description" content="The performance of various acoustic feature extraction methods has been compared in this work using Long Short-Term Memory (LSTM) neural network in a Bangla speech recognition system. The acoustic features are a series of vectors that represents the speech signals. They can be classified in either words or sub word units such as phonemes. In this work, at first linear predictive coding (LPC) is used as acoustic vector extraction technique. LPC has been chosen due to its widespread popularity. Then other vector extraction techniques like Mel frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) have also been used. These two methods closely resemble the human auditory system. These feature vectors are then trained using the LSTM neural network. Then the obtained models of different phonemes are compared with different statistical tools namely Bhattacharyya Distance and Mahalanobis Distance to investigate the nature of those acoustic features."/>   <meta name="google-site-verification" content="t8rHIcM8EfjIqfQzQ0IdYIiA9JxDD0uUZAitBCzsOIw" /> <meta name="yandex-verification" content="e3d2d5a32c7241f4" />  <style type="text/css"> .imagess { height:90px; text-align:left; margin:0px 5px 2px 8px; float:right; border:none; } a{ color:white; text-decoration:none; } ul li a{ font-weight:bold; color:#000; list-style:none; text-decoration:none; size:10px;} #button{ float: left; font-size: 17px; margin-left: 10px; height: 28px; width: 100px; background-color: #1e86c6; } </style> <link rel="stylesheet" type="text/css" href="../main.css" /> </head> <body> <div id="wap"> <div id="page"> <div id="top"> <table width="100%" cellspacing="0" cellpadding="0" > <tr><td colspan="3" valign="top"><img src="../ijma.gif" /></td></tr> </table> </div> <div id="menu"> <a href="http://airccse.org/journal/ijma.html">Home</a> <a href="http://airccse.org/journal/Editorialboard.html">Editorial</a> <a href="http://airccse.org/journal/Papersub.html">Submission</a> <a href="http://airccse.org/journal/ijma_index.html">Indexing</a> <a href="http://airccse.org/journal/Specialissue.html">Special Issue</a> <a href="http://airccse.org/journal/Contact.html">Contacts</a> <a href="http://airccse.org" target="_blank">AIRCC</a></div> <div id="content"> <div id="left"> <h2>Volume 12, Number 5</h2> <h4 style="text-align:center;height:auto;"><a>Qualitative Analysis of PLP in LSTM for Bangla Speech Recognition</a></h4> <h3>  Authors</h3> <p class="#left">Nahyan Al Mahmud<sup>1</sup> and Shahfida Amjad Munni<sup>2</sup>, <sup>1</sup>Ahsanullah University of Science and Technology, Bangladesh, <sup>2</sup>Cygnus Innovation Limited, Bangladesh </p> <h3>  Abstract</h3> <p class="#left right" style="text-align:justify">The performance of various acoustic feature extraction methods has been compared in this work using Long Short-Term Memory (LSTM) neural network in a Bangla speech recognition system. The acoustic features are a series of vectors that represents the speech signals. They can be classified in either words or sub word units such as phonemes. In this work, at first linear predictive coding (LPC) is used as acoustic vector extraction technique. LPC has been chosen due to its widespread popularity. Then other vector extraction techniques like Mel frequency cepstral coefficients (MFCC) and perceptual linear prediction (PLP) have also been used. These two methods closely resemble the human auditory system. These feature vectors are then trained using the LSTM neural network. Then the obtained models of different phonemes are compared with different statistical tools namely Bhattacharyya Distance and Mahalanobis Distance to investigate the nature of those acoustic features.</p> <h3>  Keywords</h3> <p class="#left right" style="text-align:justify">LSTM, Perceptual linear prediction, Mel frequency cepstral coefficients, Bhattacharyya Distance, Mahalanobis Distance.</p><br> <button type="button" id="button"><a target="blank" href="/ijma/V12N5/12520ijma01.pdf">Full Text</a></button>   <button type="button" id="button"><a href="http://www.airccse.org/journal/ijma_current20.html">Volume 12</a></button> <br><br><br><br><br> </div> <div id="right"> <div class="menu_right"> <ul><li id="id"><a href="http://airccse.org/journal/ijma.html">Scope & Topics</a></li> <li><a href="http://airccse.org/ethics.html">Ethics</a></li> <li><a href="http://airccse.org/journal/ijma_archive.html">Archives</a></li> <li><a href="http://airccse.org/journal/ijma_cited.html">Most Cited Articles</a></li> <li><a href="http://airccse.org/journal/ijma.pdf" title="">Download leaflet</a></li> <li><a href="http://airccse.org/faq.html" target="_blank">FAQ</a></li> </ul> </div><br /> <p align="center"> </p> <p align="center"> </p> </div> <div class="clear"></div> <div id="footer"><table width="100%" ><tr><td height="25" colspan="2"><br /><p align="center">© AIRCC Publishing Corporation</p></td></table> </div> </div> </div> </div> </body> </html>