@article{ART003258719},
author={Younguk Yun},
title={A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={10},
pages={53-60}
TY - JOUR
AU - Younguk Yun
TI - A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 10
PB - The Korean Society Of Computer And Information
SP - 53
EP - 60
SN - 1598-849X
AB - In this paper, we propose TinyAlcoCNN, a lightweight deep learning model designed to non-invasively detect alcohol consumption based on voice data. The proposed model adopts a 2D-CNN architecture that takes Mel-spectrograms as input and is trained on approximately 40,000 Korean voice samples. To support real-time applications, the dataset was preprocessed using the Whisper API for automatic segmentation. Experimental results demonstrate that TinyAlcoCNN achieves a training accuracy of 0.9982 and an inference accuracy of 1.000, while maintaining efficiency with approximately one million parameters and 13.9 million FLOPs. These results confirm both the effectiveness and computational efficiency of the model. This study highlights the feasibility of voice-based alcohol detection and suggests potential for broader applications, including personalized services, through multilingual expansion and integration with mobile systems.
KW - Alcohol Detection;Lightweight Deep Learning;CNN;Mel-Spectrogram;Edge Computing
DO -
UR -
ER -
Younguk Yun. (2025). A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features. Journal of The Korea Society of Computer and Information, 30(10), 53-60.
Younguk Yun. 2025, "A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features", Journal of The Korea Society of Computer and Information, vol.30, no.10 pp.53-60.
Younguk Yun "A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features" Journal of The Korea Society of Computer and Information 30.10 pp.53-60 (2025) : 53.
Younguk Yun. A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features. 2025; 30(10), 53-60.
Younguk Yun. "A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features" Journal of The Korea Society of Computer and Information 30, no.10 (2025) : 53-60.
Younguk Yun. A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features. Journal of The Korea Society of Computer and Information, 30(10), 53-60.
Younguk Yun. A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features. Journal of The Korea Society of Computer and Information. 2025; 30(10) 53-60.
Younguk Yun. A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features. 2025; 30(10), 53-60.
Younguk Yun. "A Lightweight CNN Model for Alcohol-intoxicated Detection Using Mel-Spectrogram Voice Features" Journal of The Korea Society of Computer and Information 30, no.10 (2025) : 53-60.
