The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
Predicting the performance of an underwater acoustic network (UAN) is a challenging task due to the spatiotemporal variability of the links and its complicated dependence on multiple factors. We present a machine-learning model based on logistic regression (LogR) to capture the spatio-temporal variation in the performance of a UAN. The model captures the effect of environmental factors such as wind...
The conventional automatic speech recognition (ASR) systems employ the GMM-HMM for acoustic modeling and the n-gram for language modeling. Over the last decade, the deep feed-forward neural network (DFNN) has almost replaced the GMM in acoustic modeling. The current ASR systems are predominantly based on the DFNN-HMM acoustic model and the n-gram language model (LM). Owing to better long-term context...
The main indication for auditory training is central auditory processing disorder (CAPD), which inevitably develops in patients with the chronic sensorineural hearing loss as a consequence of auditory deprivation. Patients with CAPD have difficulties with understanding complex signals, especially, speech in background noise. The aim of the study was to create the optimal algorithm of auditory training...
The logging and further analysis of borehole images is a major step in the interpretation of geological events. Natural fractures and beddings are features whose identification is commonly performed using acoustic and electrical borehole imaging tools. Such identification is a tedious task and is made visually by geologists, who must be experts on classification. The correct identification of planar...
Acoustic event detection (AED) is currently a very active research area with multiple applications in the development of smart acoustic spaces. In this context, the advances brought by Internet of Things (IoT) platforms where multiple distributed microphones are available have also contributed to this interest. In such scenarios, the use of data fusion techniques merging information from several sensors...
As part of the 2016 public evaluation challenge on Detection and Classification of Acoustic Scenes and Events (DCASE 2016), the second task focused on evaluating sound event detection systems using synthetic mixtures of office sounds. This task, which follows the ‘Event Detection-Office Synthetic’ task of DCASE 2013, studies the behaviour of tested algorithms when facing controlled levels of audio...
Many machine learning tasks have been shown solvable with impressive levels of success given large amounts of training data and computational power. For the problems which lack data sufficient to achieve high performance, methods for transfer learning can be applied. These refer to performing the new task while having prior knowledge of the nature of the data, gained by first performing a different...
This paper targets on a generalized vocal mode classifier (speech/singing) that works on audio data from an arbitrary data source. Previous studies on sound classification are commonly based on cross-validation using a single dataset, without considering training-recognition mismatch. In our study, two experimental setups are used: matched training-recognition condition and mismatched training-recognition...
Deep neural networks have been widely applied in the field of environmental sound classification. However, due to the scarcity of carefully labeled data, their training process suffers from over-fitting. Data augmentation is a technique that alleviates this issue. It augments the training set with synthetic data that are created by modifying some parameters of the real data. However, not all kinds...
This paper presents improvements in terms of accuracy for shape object classification using a new low complexity method compared to previous implementation [1]. The method is using echoes generated by a JAVA platform capable of emulate sound propagation in a controlled 2D virtual environment [2][3]. Echoes originate from the ultrasonic waves generated inside a virtual environment which contains geometrical...
Sound event detection (SED) in environmental recordings is a key topic of research in machine listening, with applications in noise monitoring for smart cities, self-driving cars, surveillance, bioa-coustic monitoring, and indexing of large multimedia collections. Developing new solutions for SED often relies on the availability of strongly labeled audio recordings, where the annotation includes the...
The classification of different odontocetes using écholocation clicks plays a significant role in tracking and detecting animals for research and protection purposes. Echolocation clicks were detected by an automatic method based on the Teager-Kaiser Energy Operator (TKEO). Then, these clicks were represented by their FFT magnitude spectrum. To reduce the influence of high similar clicks among species,...
Composites are widely used in aviation, aerospace and other fields because of their high specific strength, high specific stiffness and easy molding. However, in the process of using the concentrated stress, heavy shocks may form different degrees of damage. Especially, the internal delamination will reduce the stability and safety of the structure. Based on the analysis of damage location and damage...
Ultrasonic NDE uses high frequency acoustic waves to evaluate materials, and often signal processing is required to detect echoes from defects in the presence of microstructure scattering noise. Scattering noise, also known as clutter, interferes with the flaw signal and cannot be completely eliminated by using classical signal processing methods such as band-pass filtering. In this paper, neural...
Ultrasonic Non-Destructive Evaluation (NDE) uses high frequency acoustic waves to evaluate materials, and often signal processing is required to detect echoes from defects in the presence of micro-structure scattering noise. Scattering noise is known as the clutter. The clutter interferes with the flaw signal and cannot be completely separated from it by using conventional signal processing methods...
In order to train neural networks (NN) for text-to-speech synthesis (TTS), phonetic segmentation must be performed. The most accurate segmentation is performed manually, but the process of creating manual alignments is costly and time-consuming, so automatic procedures are preferable. In this paper, a simple alignment method based on models trained during hidden Markov Model (HMM) based TTS system...
This work presents an embedded hardware architecture for real-time ultrasonic NDE applications that incorporate Hidden Markov Model (HMM) based statistical signal methods. HMM has been successfully used in applications like audio segment retrieval, speech/language recognition and image processing applications. Recently, we proposed a new Hidden Markov Model (HMM) based ultrasonic flaw detection algorithm...
This work presents an embedded hardware architecture for real-time ultrasonic NDE applications that incorporate Hidden Markov Model (HMM) based statistical signal methods. Proposed algorithm is a combination of Discrete Wavelet Transform (DWT) for pre-processing A-scan signals and HMM for classification of the flaw presence. For this study, a MicroZed FPGA with Xilinx Zynq-7020 System-on-Chip (SoC)...
This paper introduces the use of representations based on nonnegative matrix factorization (NMF) to train deep neural networks with applications to environmental sound classification. Deep learning systems for sound classification usually rely on the network to learn meaningful representations from spectrograms or hand-crafted features. Instead, we introduce a NMF-based feature learning stage before...
This paper constructs speech features based on a generative model using a deep latent Gaussian model (DLGM), which is trained using stochastic gradient variational Bayes (SGVB) algorithm and performs efficient approximate inference and learning with a directed probabilistic graphical model. The trained DLGM then generate latent variables based on Gaussian distribution, which is used as new features...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.