<![CDATA[Attendance is an essential thing in the learning process. In recent years, technology development has been relatively rapid, one of which is in attendance recording. Attendance registration can now be done using QR-Code, palm, face recognition and digital signature. There are shortcomings, such as the lack of flexibility in the attendance process and the problem of the pandemic being limited by distance. The existence of the teacher's online method makes it challenging to communicate. Based on these problems, this study was presented to build a voice verification model using the Mel-Frequency Cepstral Coefficients (MFCC) and Learning Vector Quantization (LVQ) methods as an alternative attendance. This study uses text-dependent recorded data from 35 speakers. In this study, verification was carried out with the condition of using a mask and without a mask. This study obtained an average margin of similarity score of 80% of the average similarity score of 93.96% native speakers and 67.58% fake speakers. The best test results were obtained at a threshold of 0.85 with an accuracy value of 86.2%, precision of 87.86%, and recall of 84% when using a mask, while without a mask the value of accuracy was 90%, precision was 88.97%, and recall was 91.17%]]>
Copyrights © 2022
