Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal POSITRON Jurnal Sains Dasar
Agustika, Dyah Kurniawati
Unknown Affiliation
Education Energy Physics

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
FREQUENCY COMPONENT EXTRACTION OF HEARTBEAT CUES WITH SHORT TIME FOURIER TRANSFORM (STFT) Sumarna, Sumarna; Purwanto, Agus; Agustika, Dyah Kurniawati
Jurnal Sains Dasar Vol 5, No 1 (2016): April 2016
Publisher : Faculty of Mathematics and Natural Science, Universitas Negeri Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (431.871 KB) | DOI: 10.21831/jsd.v5i1.12658

Abstract

Abstract Electro-acoustic human heartbeat detector have been made with the main parts : (a) stetoscope (piece chest), (b) mic condenser, (c) transistor amplifier, and (d) cues analysis program with MATLAB. The frequency components that contained in heartbeat. cues have also been extracted with Short Time Fourier Transform (STFT) from 9 volunteers. The results of the analysis showed that heart rate appeared in every cue frequency spectrum with their harmony. The steps of the research were including detector instrument design, test and instrument repair, cues heartbeat recording with Sound Forge 10 program and stored in wav file ; cues breaking at the start and the end, and extraction/cues analysis using MATLAB. The MATLAB program included filter (bandpass filter with bandwidth between 0.01 – 110 Hz), cues breaking with hamming window and every part was calculated using Fourier Transform (STFT mechanism) and the result were shown in frequency spectrum graph. Keywords: frequency components extraction, heartbeat cues, Short Time Fourier Transform
Automatic Recognition of Pelung and Canary Bird Sounds Using Machine Learning and Signal Enhancement Agustika, Dyah Kurniawati; Kadarisman, Nur; Sumarna, Sumarna; Purwanto, Agus
POSITRON Vol 15, No 1 (2025): Vol. 15 No. 1 Edition
Publisher : Fakultas Matematika dan Ilmu Pengetahuan Alam, Univetsitas Tanjungpura

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26418/positron.v15i1.93137

Abstract

Bird sound classification is a valuable tool in ecological monitoring and species identification, particularly for non-invasive assessment in natural environments. However, challenges such as limited labeled data and environmental noise often reduce the reliability of classification models. This study presents a lightweight bird sound classification pipeline that integrates signal preprocessing, audio augmentation, and machine learning to address these issues. Two bird species with distinct vocal characteristics, Pelung (a crossbreed involving Bangkok chickens) and Canary (Serinus canaria), were used as case subjects. A total of 40 original 2-second audio clips were extracted from longer field recordings, then processed through frame-based energy attenuation, bandpass filtering (1"“8 kHz), and RMS normalization. Ten augmentation techniques were applied to each original file to improve generalization, generating 400 augmented files for model training. Feature extraction was performed using 13-dimensional Mel Frequency Cepstral Coefficients (MFCCs), and Principal Component Analysis (PCA) was used to visualize the effect of filtering. Classification was conducted using a Support Vector Machine (SVM) with a radial basis function (RBF) kernel. Results showed that filtering improved classification accuracy from 90% to 95% on the original data. Furthermore, using only augmented data for training and original data for testing yielded 100% classification accuracy, demonstrating excellent generalization. This study highlights the effectiveness of combining adaptive preprocessing and augmentation for reliable bird sound classification under limited and noisy conditions.
Co-Authors AGUS PURWANTO Nur Kadarisman Sumarna Sumarna