Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2023
- Goals
- Record
- c83d3635-7fb6-4c6d-b643-ae5dbc54808d
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted an experimental activity on the automatic categorization of bird sound patterns. The activity involved the listed researchers as participants and examined a method for automatically categorizing bird sounds. The activity was conducted in the context that Indonesia is home to several endangered species.
The experiment used publicly available bird data to obtain bird sound patterns. The source explains that bird sounds contain noise and that features may be inappropriate, creating a need for data processing to reduce noise and select valuable features. The activity used YAMNet sound classification network data processing and feature selection incorporating Particle Swarm Optimization. These processes were used to decrease noise in bird voice data and choose appropriate features. The steps were completed before and after the gammatone frequency cepstral coefficient approach was used as feature extraction.
The resulting data were processed using the KNN classification method to obtain classification performance results. The experimental findings demonstrated bird identification recognition performance of 78.33%, which was 1.27% better than the prior study. The source also states that this finding outperformed previous research that used only a dimensionality reduction-based classification algorithm without selecting the most important features. Through this activity, UDINUS undertook research involving bird sound data processing, feature selection, feature extraction, and classification. The activity contributes evidence concerning an experimental approach to bird sound pattern categorization in the stated context of endangered species.