Rancang Bangun Pengenalan Suara Sebagai Alat Bantu Keamanan Pejalan Kaki Tunarungu

Haladi, Reza Amalina (2023) Rancang Bangun Pengenalan Suara Sebagai Alat Bantu Keamanan Pejalan Kaki Tunarungu. Other thesis, Intstitut Teknologi Sepuluh Nopember.

Text 07111840007002-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (1MB) | Request a copy

Abstract

Tunarungu adalah kondisi ketika seseorang mengalami gangguan pendengaran atau kehilangan pendengaran secara total atau sebagian. Dalam penelitian ini, dilakukan eksperimen untuk mengatasi masalah tersebut dengan menggunakan teknik ekstraksi Mel-Frequency Cepstral Coefficients (MFCC) dan klasifikasi menggunakan Support Vector Machine (SVM) dengan kernel RBF (Radial Basis Function). Selain itu, digunakan juga algoritma Direction of Arrival (DOA) melalui pemanfaatan perangkat ReSpeaker V2.0. Pada tahap ekstraksi fitur, MFCC digunakan untuk mengubah sinyal audio menjadi bentuk domain frekuensi yang lebih mudah diolah oleh sistem. Kemudian, SVM dengan kernel RBF digunakan untuk mengklasifikasikan suara menjadi dua kategori, yaitu suara alarm (klakson kendaraan dan alarm palang pintu kereta api) dan suara non-alarm. Hasil penelitian menunjukkan bahwa sistem yang diimplementasikan dapat mengenali suara alarm dan suara non-alarm dengan akurasi sempurna, yaitu mencapai 100%. Hasil tersebut dicapai setelah melakukan penyesuaian pada variabel tuning, dengan menggunakan nilai n_mfcc sebanyak 15, durasi rekaman audio selama 3 detik, dan pembagian data uji (test_size) sebesar 20% .
================================================================================================================================
Deafness is a condition when a person has hearing loss or total or partial hearing loss. In this study, experiments were conducted to overcome these problems using Mel-Frequency Cepstral Coefficients (MFCC) extraction techniques and classification using Support Vector Machine (SVM) with RBF (Radial Basis Function) kernels. In addition, the Direction of Arrival (DOA) algorithm is also used through the use of the ReSpeaker V2.0 device. In the feature extraction stage, MFCC is used to convert the audio signal into a form of frequency domain that is easier for the system to process. Then, SVM with RBF kernel is used to classify sounds into two categories: alarm sounds (vehicle horns and train door bar alarms) and non-alarm sounds. The results showed that the implemented system can recognize alarm sounds and non-alarm sounds with perfect accuracy, reaching 100%. These results were achieved after making adjustments to the tuning variables, using an n_mfcc value of 15, an audio recording duration of 3 seconds, and a test data division (test_size) of 20%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deafness, Sounds Recognition, ReSpeaker V2.0, Mel Frequency Cepstral Coefficients, Support Vector Machine, Direction of Arrival, Tunarungu
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7878 Electronic instruments
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Haladi Reza Amalina
Date Deposited:	25 Jul 2023 08:15
Last Modified:	25 Jul 2023 08:15
URI:	http://repository.its.ac.id/id/eprint/99429

Actions (login required)

View Item