Arani, Hasbiya Diona (2017) Identifikasi Ayat pada Bacaan Menggunakan Metode Dynamic Time Warping Berdasarkan Fitur Mel Frequency Cepstral Coefficient untuk Sistem Tutorial Hafalan Al-Quran. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2913100013-Undergraduate_Theses.pdf - Published Version Download (4MB) | Preview |
Abstract
Penelitian ini menerapkan metode pengenalan sinyal wicara untuk membangun sebuah sistem yang membantu para penghafal Al-Quran untuk menambah atau memelihara hafalannya diluar kelas (tidak didampingi oleh guru). Fitur yang digunakan adalah Mel Frequency Cepstral Coefficient (MFCC) untuk menghasilkan sebuah ciri dari sinyal bacaan. Fitur dari sinyal masukan dibandingkan dengan fitur dari ayat-ayat Al-Quran yang disimpan dalam sistem menggunakan metode Dynamic Time Warping (DTW). Ayat yang paling cocok adalah ayat yang jaraknya paling dekat dengan sinyal masukan. Masukan berupa rekaman surah-surah di Al-Quran yang diterima secara utuh dan berkesinambungan. Pengujian pada penelitian ini dilakukan pada beberapa surah pilihan yaitu Q.S. An-naas:1, Q.S. Al-falaq:1, Q.S. Al-ikhlas:1-4, dan Al-falaq:1. Data dibagi menjadi dua yaitu data uji dan data acuan. Acuan diambil dari rekaman dua laki-laki dan dua perempuan yang sudah diketahui ketepatan bacaannya. Pengujian dilakukan empat pembaca dengan 28 sampel acuan. Hasil pengujian memperlihatkan akurasi 89,29% pada pengujian acuan jenis kelamin campuran dan metode ekstraksi MFCC. Pengujian dengan metode ekstraksi STFT menghasilkan akurasi 82,14%, sedangkan pada pengujian acuan diluar database yaitu 53,57%. =================================================================================This research applies the method of speech signal recognition to build a system that helps the interfere of the Quran to increase or maintain the memorizer outside the class (not accompanied by a teacher). The feature used in this research is Mel Frequency Cepstral Coefficient (MFCC) to produce a characteristic signal of the readings. Features of the input signal compared with the features of some verses of the Quran which are stored in the system using Dynamic Time Warping (DTW). The identification result of verses seen from the closest distance to the input signal. The input signals consist of the readings of several verses of the Quran which were recorded and then recognized continuously. Data are limited by the selection of short verses which are Q.S.A-Naas:1, Q.S.Alfalaq:1, Q.S.Al-ikhlas, and Q.S.Al-falaq:1. Data are divided to the test data and reference data. The Reference data is taken from recording of two men and two women who were already known for the accuracy of the reading.
The testing is done by four readers with 28 reference data. The
result of test show the accuracy 89,29% on composite gender reference data and MFCC feature extraction. The accuracy result of testing with STFT feature extraction is 82,14%, while the accuracy result of testing reference data without the database is 53,57%. The most correct identification verses is Q.S. Al-ikhlas:3 and Q.S.Al-kafirun:1 with percentage is 80,56%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSKom 006.454 Ara i |
| Uncontrolled Keywords: | Speech Recognition; Mel Frequency Cepstral Coefficient (MFCC); Dynamic Time Warping (DTW); Al-Quran |
| Subjects: | Q Science > QA Mathematics > QA404 Fourier series Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General) |
| Divisions: | Faculty of Industrial Technology > Multimedia and Network Engineering > (S1) Undergraduate Theses |
| Depositing User: | Arani Hasbiya Diona |
| Date Deposited: | 11 Jan 2018 08:20 |
| Last Modified: | 05 Mar 2019 03:08 |
| URI: | http://repository.its.ac.id/id/eprint/48664 |
Actions (login required)
 |
View Item |