Wijaya, Rayhan Kurnia Alunantara (2024) Prototipe Sistem Rekomendasi Konten Hiburan Anak-Anak Dan Klasifikasi Kelompok Usia Berbasis Suara Pada Perangkat Tinyml (Tiny Machine Learning). Other thesis, Institut Teknologi Sepuluh Nopember.
Text
5027201030-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (4MB) | Request a copy |
Abstract
Era digital menuntut solusi inovatif untuk memastikan anak-anak mengakses konten hiburan yang sesuai usia. Sistem rekomendasi yang digunakan pada berbagai platform konten hiburan tidak dapat mengklasifikasikan usia anak-anak secara akurat dan memberikan konten yang sesuai untuk mereka. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan prototipe sistem rekomendasi dan klasifikasi usia anak-anak berbasis suara menggunakan perangkat Tiny Machine Learning yang dapat menentukan kelompok usia anak-anak secara akurat. Model kunci dalam penelitian ini adalah Hybrid MLP (Hybrid Multilayer Perceptron), dimana digunakan teknik ekstraksi fitur MFCC (Mel-Frequency Cepstral Coefficients) untuk mengolah data suara. Dalam penelitian ini, dataset utama berasal dari Mozilla Common Voice dan dataset suara anak-anak Indonesia yang dikumpulkan secara mandiri. Dilakukan juga perbandingan dengan model-model yang telah dikembangkan sebelumnya pada dataset yang sama dimana model Hybrid MLP berhasil melebihi akurasi model pembanding tersebut. Pengembangan model klasifikasi usia dilakukan pada platform Edge Impulse dengan menggunakan Arduino Nano BLE 33 Sense Lite. Pengujian model dilakukan dengan memanfaatkan kata kunci 'oke' yang bertujuan untuk mengaktivasi dan mengevaluasi respons sistem. Dalam uji coba untuk mengklasifikasikan kelompok usia anak SD dan SMP dengan skenario optimasi paling optimal, Hybrid MLP berhasil menunjukkan tingkat akurasi pelatihan sebesar 97.6% serta akurasi validasi 87.72%. Hasil ini mengindikasikan efektivitas model dalam proses klasifikasi usia dalam kondisi sumber daya komputasi yang terbatas.
==================================================================================================================================
The digital age demands innovative solutions to ensure children access age-appropriate entertainment content. Recommendation systems used on various entertainment content platforms are unable to accurately classify children's age and provide appropriate content for them. Therefore, this research aims to develop a prototype of a voice-based children's age classification and recommendation system using Tiny Machine Learning tools that can accurately determine children's age groups. The key model in this research is Hybrid MLP (Hybrid Multilayer Perceptron), where MFCC (Mel-Frequency Cepstral Coefficients) feature extraction technique is used to process voice data. In this research, the main dataset comes from Mozilla Common Voice and Indonesian children's voice dataset collected independently. Comparisons were also made with previously developed models on the same dataset where the Hybrid MLP model successfully exceeded the accuracy of the comparison model. The age classification model development was conducted on the Edge Impulse platform using Arduino Nano BLE 33 Sense Lite. Model testing was conducted by utilizing the keyword 'oke' which aims to activate and evaluate the system response. In the test to classify the age groups of elementary and junior high school children with the most optimal optimization scenario, Hybrid MLP successfully showed a training accuracy rate of 97.6% and a validation accuracy of 87.72%. These results indicate the effectiveness of the model in the age classification process under conditions of limited computational resources.
Item Type: | Thesis (Other) |
---|---|
Additional Information: | RSTI 006.31 RAY p 2024 |
Uncontrolled Keywords: | Age Classification, Multilayer Perceptron, Recommendation System, TinyML, Voice |
Subjects: | T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis |
Depositing User: | Rayhan Kurnia Alunantara Wijaya |
Date Deposited: | 05 Feb 2024 04:43 |
Last Modified: | 31 Oct 2024 08:42 |
URI: | http://repository.its.ac.id/id/eprint/106099 |
Actions (login required)
View Item |