Analisis Efektivitas Pengenalan Perintah Suara Menggunakan Algoritma Machine Learning untuk Mendukung Pembelajaran Digital di Wilayah Bengkulu

Putra, Fherdi Hunata
Albar, Ahmad Rere
Kurniawan, Andre
Okta, Leo
Akbar, Anggi Sunata Nur
Fitriah, Fitriah
Muntahanah, Muntahanah

Mestaka: Jurnal Pengabdian Kepada Masyarakat
Vol. 4 No. 4 (2025): Agustus 2025

This study aims to evaluate the effectiveness of a voice command recognition system based on Machine Learning algorithms in supporting digital learning at Panti Asuhan Kasih Sayang, Bengkulu City. Four models were tested: DNN, LSTM, CNN, and Transformer. The Transformer model demonstrated the best performance with an accuracy of 93.4%, precision of 91.8%, recall of 92.1%, and F1-score of 92.0%, followed by LSTM with 90.2% accuracy, and DNN as the baseline with 84.7%. The system was tested in real-time with an average latency of 320ms and handled up to 28 voice commands per minute. The use of this system increased participants’ post-test scores by an average of 18 points and enhanced learning engagement by 85%. Qualitative data from interviews and focus group discussions indicated that the system supported students who struggle with text-based interfaces and was perceived as interactive and easy to use. These findings show that ASR based on Machine Learning is effective in low-resource settings and supports more inclusive learning aligned with the principles of Society 5.0.