RISMANTO, MUHAMMAD ELIO PHILLO (2025) SPEECH EMOTION RECOGNITION SYSTEM USING CONVOLUTIONAL NEURAL NETWORK METHOD. Tugas Akhir thesis, Informatics.
![[img]](http://eprints.uty.ac.id/style/images/fileicons/text.png) |
Text
5210411291_Muhammad Elio Phillo Rismanto_Abstrak_Informatika.pdf Download (12kB) |
Abstract
Speech Emotion Recognition (SER) is a technological field that explores how machines interpret emotional patterns in audio data using various methods and acoustic features. Despite its potential, SER remains underutilized due to the inherent difficulty of enabling machines to accurately perceive emotions. This study applies and enhances the Convolutional Neural Network (CNN) method to develop a high-accuracy emotion prediction model. The system utilizes a combination of spectrograms, MFCCs, spectral contrast, tonnetz, and chroma features extracted from 3,232 audio samples sourced from the RAVDESS, Indo Wave Sentiment, and SAVEE datasets, encompassing eight emotional categories: angry, calm, disgusted, fearful, happy, neutral, sad, and surprised. Data augmentation techniques were employed, including noise injection, pitch shifting, and time shifting, to simulate real-world audio conditions. The model was trained by tuning parameters such as learning rate, weight decay, optimization method, number of epochs, and batch size. The best-performing CNN configuration used RMSprop optimization, a learning rate of 0.001, weight decay of 0.5, 150 epochs, and a batch size of 32, yielding an overall accuracy of 57.50%. While this suggests that the model effectively memorizes patterns but struggles with generalizing to unseen data, the ROC curve score of 0.88 indicates a strong ability to distinguish between emotional classes in audio input. Keywords: CNN, SER, RAVDESS, SAVEE
| Item Type: | Thesis (Skripsi, Tugas Akhir or Kerja Praktek) (Tugas Akhir) |
|---|---|
| Subjects: | T Technology > T Technology (General) |
| Divisions: | Fakultas Sains Dan Teknologi > S1 Informatika |
| Depositing User: | Kaprodi S1 Informatika UTY |
| Date Deposited: | 16 Jul 2025 08:42 |
| Last Modified: | 16 Jul 2025 08:42 |
| URI: | http://eprints.uty.ac.id/id/eprint/18180 |
Actions (login required)
 |
View Item |
