Sound Can Help Us See More Clearly

In the field of video action classification, existing network frameworks often only use video frames as input. When the object involved in the action does not appear in a prominent position in the video frame, the network cannot accurately classify it. We introduce a new neural network structure that uses sound to assist in processing such tasks. The original sound wave is converted into sound texture as the input of the network. Furthermore, in order to use the rich modal information (images and sound) in the video, we designed and used a two-stream frame. In this work, we assume that sound data can be used to solve motion recognition tasks. To demonstrate this, we designed a neural network based on sound texture to perform video action classification tasks. Then, we fuse this network with a deep neural network that uses continuous video frames to construct a two-stream network, which is called A-IN. Finally, in the kinetics dataset, we use our proposed A-IN to compare with the image-only network. The experimental results show that the recognition accuracy of the two-stream neural network model with uesed sound data features is increased by 7.6% compared with the network using video frames. This proves that the rational use of the rich information in the video can improve the classification effect.

Fault detection of contactor using acoustic monitoring

Recently, condition monitoring methods using the sound of the machine have attracted attention. Since approaching high voltage equipment increases the risk of electrocuting, non-contact data acquisition is desirable. Most of the research targets of acoustic monitoring are rotating machines and it is not clear whether it is effective for machines that switch between two states, such as contactors and circuit breakers. In this work, several investigations have been carried out on the acoustic condition monitoring of contactor. The Mel-frequency cepstrum coefficients (MFCCs) were obtained from the sound data of the contactors under normal and simulated fault conditions. Support Vector Machine (SVM) was trained with MFCCs and found that it could detect and diagnose contactor faults with high accuracy.

Characterisation of the soundscape of parks and green areas through the triangulation of methodologies proposed by ISO/TS 12913-2

The management of urban spaces and environmental health has been growing in recent years, and the sound aspects were highlighted during the Coronavirus pandemic (COVID-19). Locations that generally showed noises from vehicle traffic presented a diversity of sounds, generally not perceived in everyday situations before the pandemic. Awareness of the sound impacts generated before the pandemic has provided a broad discussion between the scientific community and managers regarding developing tools to improve urban planning and environmental health in cities. This study aims to characterise the soundscape of two parks in Curitiba by triangulating evaluation methodologies proposed in the ISO/TS 12913-2 (2018). Such triangulation included the descriptive analysis of objective and subjective sound data, analysis and elaboration of sound and perception maps, providing a systemic overview of the sonic environment of the investigated parks.

Application of AI for Analysis of Parkinson’s Disease

Parkinson's disease is an issue of the central tactile framework that impacts advancement provoking shudders. The tangible cell is hurt in the frontal cortex causing dopamine levels to drop which prompts the condition. Parkinson's is a reformist ailment that causes degeneration of the frontal cortex, provoking both motor and mental issues. While Dysphonia is a voice issue that causes mandatory fits in the larynx muscle, this is one of its indications. While, Bradykinesia, which is ordinarily described as slowness of improvements, is one of the cardinal signs of Parkinson's sickness (PD). Essential clinical rating scales are used usually to measure bradykinesia in routine clinical practice albeit this kind of examination is uneven. It requires clinical investigation, and it can happen starting from the age of 6. Along these lines, this is a starter study that endeavors to recognize connections between Parkinson's contamination factors for basic unmistakable verification of the sickness. There are 1 million cases in India. It is hence reasonable to acknowledge that there is a connection between a patient's ability to talk/make and the development towards Parkinson's as these limits rot as time propels. The mark of the examination was to survey the features of the sound data and the hour of contorting drawing as an extent of bradykinesia. Henceforth to make strong proof that vocalization data and the handwriting test from a patient can assist with dissecting whether they experience the evil impacts of Parkinson's. As needs be, it is at first anticipated that there is an association between the two. We attempt to run distinctive AI classifiers on the data in wants to show up at a high consistency rate that is facilitated with a reasonable runtime. The dataset managed is procured from a new report by the journal, IEEE Transactions on Biomedical Engineering, of various limits of voice repeat. The actual assessment obtained a consistency speed of 95.58% hence we want to show up at a rate close to this or possibly to beat it.

Acoustic Anomaly Detection of Mechanical Failures in Noisy Real-Life Factory Environments

Anomaly detection without employing dedicated sensors for each industrial machine is recognized as one of the essential techniques for preventive maintenance and is especially important for factories with low automatization levels, a number of which remain much larger than autonomous manufacturing lines. We have based our research on the hypothesis that real-life sound data from working industrial machines can be used for machine diagnostics. However, the sound data can be contaminated and drowned out by typical factory environmental sound, making the application of sound data-based anomaly detection an overly complicated process and, thus, the main problem we are solving with our approach. In this paper, we present a noise-tolerant deep learning-based methodology for real-life sound-data-based anomaly detection within real-world industrial machinery sound data. The main element of the proposed methodology is a generative adversarial network (GAN) used for the reconstruction of sound signal reconstruction and the detection of anomalies. The experimental results obtained in the Malfunctioning Industrial Machine Investigation and Inspection (MIMII) show the superiority of the proposed methodology over baseline approaches based on the One-Class Support Vector Machine (OC-SVM) and the Autoencoder–Decoder neural network. The proposed schematics using the unscented Kalman Filter (UKF) and the mean square error (MSE) loss function with the L2 regularization term showed an improvement of the Area Under Curve (AUC) for the noisy pump data of the pump.