RANCANG BANGUN APLIKASI PANDUAN PERTOLONGAN PERTAMA BERBASIS RNN-CTC DAN NAIVE BAYES DENGAN DUKUNGAN OPERASI BERBASIS SUARA

Anja Bunga Aditya, . (2026) RANCANG BANGUN APLIKASI PANDUAN PERTOLONGAN PERTAMA BERBASIS RNN-CTC DAN NAIVE BAYES DENGAN DUKUNGAN OPERASI BERBASIS SUARA. Skripsi thesis, Universitas Pembangunan Nasional Veteran Jakarta.

	Text ABSTRAK.pdf Download (221kB)
	Text AWAL.pdf Download (928kB)
	Text BAB 1.pdf Restricted to Repository UPNVJ Only Download (593kB)
	Text BAB 2.pdf Restricted to Repository UPNVJ Only Download (714kB)
	Text BAB 3.pdf Restricted to Repository UPNVJ Only Download (2MB)
	Text BAB 4.pdf Restricted to Repository UPNVJ Only Download (2MB)
	Text BAB 5.pdf Download (247kB)
	Text DAFTAR PUSTAKA.pdf Download (228kB)
	Text DAFTAR RIWAYAT HIDUP.pdf Restricted to Repository staff only Download (103kB)
	Text LAMPIRAN.pdf Restricted to Repository UPNVJ Only Download (561kB)
	Text HASIL PLAGIARISME.pdf Restricted to Repository staff only Download (32MB)
	Text ARTIKEL KI.pdf Restricted to Repository staff only Download (712kB)

Abstract

Delays in first aid response within emergency situations are oftenly caused by inefficiency in searching for conventional guides and dependence on internet connectivity. This study implements an Android application named Daruratku with voice-based offline first approach. The system architecture integrates Google Speech API and RNN-CTC model with voice preprocess utilizing Spectral Substraction. User complaint text is classified using TF-IDF and Multinomial Naive bayes into urgent, non-urgent, and ambigous page, with first aid guides stores locally in SQLite. Evaluation results show Naive bayes classification’s accuracy up to 95.83%. In the transcription evaluation, Google Speech API achieved an average WER of 8% for the long sentence scenario, while the local RNN-CTC achiever WER of 88% in the same scenario. In the isolated medical keyword scenario in offline mode, the local RNN-CTC achieved WER of 42.86% and CER of 7.42%, yet was still able to map keywords to the correct guide with a routing accuracy of 100%. Online voice response latency was recorded at 1.86 seconds and offline voice response was recorded at 0.04 seconds, with stable RAM consumption of 35-120 MB. Usability testing using the System Usability Scale (SUS) on participants yielded an average score of 77.5 or acceptable category. The Daruratku application is considered reliable as a fast, voice-based, and offline medical guide access solution for layperson rescuers.

Item Type:	Thesis (Skripsi)
Additional Information:	[No.Panggil: 2210511161] [Pembimbing 1: Neny Rosmawarni] [Pembimbing 2: Muhammad Panji Muslim] [Penguji 1: Ridwan Raafi’udin] [Penguji 2: Kharisma Wiati Gusti]
Uncontrolled Keywords:	Google Speech API, Naive bayes, RNN-CTC, spectral subtraction
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General)
Divisions:	Fakultas Ilmu Komputer > Program Studi Informatika (S1)
Depositing User:	ANJA BUNGA ADITYA
Date Deposited:	08 Sep 2026 06:30
Last Modified:	08 Sep 2026 06:30
URI:	http://repository.upnvj.ac.id/id/eprint/51831

Actions (login required)

View Item