(For USM Staff/Student Only)

EngLib USM > Ω School of Electrical & Electronic Engineering >

Automatic stuttering assessment

Automatic stuttering assessment / Heng Peng Wei
Gagap merupakan sejenis ganguan ucapan di mana dianggarkan 1% daripada populasi sedunia terlibat dengannya. Kegagapan dalam ucapan akan menghalang seseorang daripada berucap bebas. Ini akan menjejaskan kehidupan sosialnya selain status pencapaian akademik dan kerjaya. Penilaian terhadap kegagapan biasanya perlu dijalankan oleh pakar terapi pertuturan dengan melakukan pengiraan terhadap nombor ganguan yang berlaku dalam sebuah ucapan. Pernyataan masalah dalam kajian ini ialah penilaian ini memakan masa yang panjang dan tidak mencapai tahap persetujuan yang tinggi antara satu pakar daripada yang lain. Objektif kajian ini justerunya adalah untuk membina pengelas untuk mengelas pemanjangan and pengulangan dalam ucapan. Ini dijalankan dengan prosedur berikut: 1) Memperoleh data; 2) Pengekstrakan ciri; 3) Pembentukan parameter; 4)Membina pengelas. Pengelas ANN menggunakan arkitektur MLP telah digunakan dalam kajian ini untuk mengelaskan pemanjangan dan pengulangan dalam ucapan dan telah mencapai ketepatan keseluruhan terbaik dengan 74%. _______________________________________________________________________________________________________ Stuttering is a speech disorder disrupting the flow of the speech with around 1% of the population worldwide is affected by it. The presence of stuttering hampered a person confidence to speak and could affect one’s social involvements as well as academic and profession achievements. The assessment for stuttering relies on the counting of stuttering artifacts in a person speech, manually, by speech language pathologist. The problem is that this practice is often too time consuming and has a high rate of disagreement between different judges. The objective of his study was hence to develop classifiers for prolongation and repetition dysfluencies in speech. This was done by: 1) Obtaining the data; 2) Feature Extraction; 3) Parameterization of dysfluencies; 4) Constructing the classifier. ANN networks with MLP architecture were employed in this study to classify speech prolongations and repetitions and achieved the best overall classification accuracy of 74.00%.
Contributor(s):
Heng Peng Wei - Author
Primary Item Type:
Final Year Project
Identifiers:
Accession Number : 875005802
Language:
English
Subject Keywords:
Stuttering; disrupting; speech
First presented to the public:
6/1/2015
Original Publication Date:
3/7/2019
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Electrical & Electronic Engineering
Citation:
Extents:
Number of Pages - 72
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2019-03-07 15:39:53.773
Submitter:
Mohd Jasnizam Mohd Salleh

All Versions

Thumbnail Name Version Created Date
Automatic stuttering assessment1 2019-03-07 15:39:53.773