Eun Joo Rhee

This paper presents a sound source distance estimation (SSDE) method using a convolutional recurrent neural network (CRNN). We approach the sound source distance estimation task as an image classification problem, and we aim to classify a given audio signal into one of three predefined distance classes—one meter, two meters, and three meters—irrespective of its orientation angle. For the purpose of training, we create a dataset by recording audio signals at the three different distances and three angles in different rooms. The CRNN is trained using time-frequency representations of the audio signals. Specifically, we transform the audio signals into log-scaled mel spectrograms, allowing the convolutional layers to extract the appropriate features required for the classification. When trained and tested with combined datasets from all rooms, the proposed model exhibits high classification accuracies; however, training and testing the model in separate rooms results in lower accuracies, indicating that further study is required to improve the method’s generalization ability. Our experimental results demonstrate that it is possible to estimate sound source distances in known environments by classification using the log-scaled mel spectrogram.

Research on clouds has an enormous influence on sky sciences and related applications, and cloud classification plays an essential role in it. Much research has been conducted which includes both traditional machine learning approaches and deep learning approaches. Compared with traditional machine learning approaches, deep learning approaches achieved better results. However, most deep learning models need large data to train due to the large number of parameters. Therefore, they cannot get high accuracy in case of small datasets. In this paper, we propose a complete solution for high accuracy of classification of cloud image patches on small datasets. Firstly, we designed a suitable convolutional neural network (CNN) model for small datasets. Secondly, we applied regularization techniques to increase generalization and avoid overfitting of the model. Finally, we introduce a model average ensemble to reduce the variance of prediction and increase the classification accuracy. We experiment the proposed solution on the Singapore whole-sky imaging categories (SWIMCAT) dataset, which demonstrates perfect classification accuracy for most classes and confirms the robustness of the proposed model.