Sera Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sera Kim has a total of 2 co-authors. Here are the top co-authors by number of shared publications:
- Seok-Pil Lee
Department of Computer Science, Sangmyung University, Seoul 03016, Republic of Korea
2 shared publications - Ji-Young Baek
Department of Computer Science, Graduate School, SangMyung University, Seoul 03016, Republic of Korea
1 shared publication
Latest Publications
- Open Access
- 25 September 2023
- Published by MDPI AG in Electronics
- Vol. 12 (19) , 4034
A BiLSTM–Transformer and 2D CNN Architecture for Emotion Recognition from Speech
The significance of emotion recognition technology is continuing to grow, and research in this field enables artificial intelligence to accurately understand and react to human emotions. This study aims to enhance the efficacy of emotion recognition from speech by using dimensionality reduction algorithms for visualization, effectively outlining emotion-specific audio features. As a model for emotion recognition, we propose a new model architecture that combines the bidirectional long short-term memory (BiLSTM)–Transformer and a 2D convolutional neural network (CNN). The BiLSTM–Transformer processes audio features to capture the sequence of speech patterns, while the 2D CNN handles Mel-Spectrograms to capture the spatial details of audio. To validate the proficiency of the model, the 10-fold cross-validation method is used. The methodology proposed in this study was applied to Emo-DB and RAVDESS, two major emotion recognition from speech databases, and achieved high unweighted accuracy rates of 95.65% and 80.19%, respectively. These results indicate that the use of the proposed transformer-based deep learning model with appropriate feature selection can enhance performance in emotion recognition from speech.
- 25 September 2023
- Published by MDPI AG in Electronics
- Vol. 12 (19) , 4034
Open AccessA BiLSTM–Transformer and 2D CNN Architecture for Emotion Recognition from Speech
The significance of emotion recognition technology is continuing to grow, and research in this field enables artificial intelligence to accurately understand and react to human emotions. This study aims to enhance the efficacy of emotion recognition from speech by using dimensionality reduction algorithms for visualization, effectively outlining emotion-specific audio features. As a model for emotion recognition, we propose a new model architecture that combines the bidirectional long short-term memory (BiLSTM)–Transformer and a 2D convolutional neural network (CNN). The BiLSTM–Transformer processes audio features to capture the sequence of speech patterns, while the 2D CNN handles Mel-Spectrograms to capture the spatial details of audio. To validate the proficiency of the model, the 10-fold cross-validation method is used. The methodology proposed in this study was applied to Emo-DB and RAVDESS, two major emotion recognition from speech databases, and achieved high unweighted accuracy rates of 95.65% and 80.19%, respectively. These results indicate that the use of the proposed transformer-based deep learning model with appropriate feature selection can enhance performance in emotion recognition from speech.
- Open Access
- 13 February 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (4) , 2378
COVID-19 Detection Model with Acoustic Features from Cough Sound and Its Application
Contrary to expectations that the coronavirus pandemic would terminate quickly, the number of people infected with the virus did not decrease worldwide and coronavirus-related deaths continue to occur every day. The standard COVID-19 diagnostic test technique used today, PCR testing, requires professional staff and equipment, which is expensive and takes a long time to produce test results. In this paper, we propose a feature set consisting of four features: MFCC, Δ2-MFCC, Δ-MFCC, and spectral contrast as a feature set optimized for the diagnosis of COVID-19, and apply it to a model that combines ResNet-50 and DNN. Crowdsourcing datasets from Cambridge, Coswara, and COUGHVID are used as the cough sound data for our study. Through direct listening and inspection of the dataset, audio recordings that contained only cough sounds were collected and used for training. The model was trained and tested using cough sound features extracted from crowdsourced cough data and had a sensitivity and specificity of 0.95 and 0.96, respectively.
- 13 February 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (4) , 2378
Open AccessCOVID-19 Detection Model with Acoustic Features from Cough Sound and Its Application
Contrary to expectations that the coronavirus pandemic would terminate quickly, the number of people infected with the virus did not decrease worldwide and coronavirus-related deaths continue to occur every day. The standard COVID-19 diagnostic test technique used today, PCR testing, requires professional staff and equipment, which is expensive and takes a long time to produce test results. In this paper, we propose a feature set consisting of four features: MFCC, Δ2-MFCC, Δ-MFCC, and spectral contrast as a feature set optimized for the diagnosis of COVID-19, and apply it to a model that combines ResNet-50 and DNN. Crowdsourcing datasets from Cambridge, Coswara, and COUGHVID are used as the cough sound data for our study. Through direct listening and inspection of the dataset, audio recordings that contained only cough sounds were collected and used for training. The model was trained and tested using cough sound features extracted from crowdsourced cough data and had a sensitivity and specificity of 0.95 and 0.96, respectively.