Articulation Battery Test For Tamil Speech Disorder Signals

2021 ◽  
Vol 23 (09) ◽  
pp. 1326-1338
Author(s):  
M Krishnaveni ◽  
◽  
P Subashini ◽  
TT Dhivyaprabha ◽  
◽  
...  
Keyword(s):  
Neural Network ◽  
Vocal Tract ◽  
Acoustic Noise ◽  
Principal Component ◽  
Feature Subset ◽  
Speech Sounds ◽  
Mel Frequency Cepstral Coefficients ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Articulation Disorder

Articulation disorder is referred as difficulty occurs in the pronunciation of specific speech sounds. An irregular coordination of the movement of tongue, lips, palate, jaw, respiratory system, vocal tract, height of the larynx, air flow through nasal leads to the incorrect production of speech sounds. The objective of this paper is to propose a computational model based on Recurrent Neural Network (RNN) algorithm to categorize the phonological patterns of Tamil speech articulation disorder signals into four predefined groups, namely, substitution, omission, distortion and addition. The methodology of the proposed work is described as follows. (1) List of articulation disorder test words suggested by Speech Language Pathologists (SLPs) is selected for this experimental study. (2) Real time speech signals that comprise of Tamil vowels (Uyir eluthukkal) and consonants (Meiyeluthukkal) are collected from people with articulation disorder. (3) Acoustic noise and weak signals are eliminated by applying Low pass filter to acquire the filtered speech signal. (4) Mel-Frequency Cepstral Coefficients (MFCCs) technique is implemented to extract the prominent features from denoised signals. (5) Principal Component Analysis (PCA) method is employed to choose fine-tune feature subset. (6) The refined features are employed to calibrate RNN model for classification. Results show that RNN model achieves 90.25% classification accuracy when compared to other artificial neural network algorithms.

A STUDY OF DIGITAL WATERMARKING RECOGNITION USING ORTHOGONAL CODE SEQUENCES WITH A BACK-PROPAGATION NEURAL NETWORK

2013 ◽  
Vol 37 (3) ◽  
pp. 459-465
Author(s):  
Chih-Ta Yen ◽  
Ing-Jr Ding ◽  
Zong-Wei Lai
Keyword(s):  
Neural Network ◽  
Digital Watermarking ◽  
Median Filter ◽  
Back Propagation ◽  
Back Propagation Neural Network ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Noise Interference ◽  
Low Pass ◽  
Neural Network Algorithm

Digital watermarking is an encryption technology commonly used to protect intellectual property and copyright. In this study, we restored watermarks that had already been affected by noise interference, used the Walsh–Hadamard codes as the watermark identification codes, and applied salt-and-pepper noise and Gaussian noise to destroy watermarks. First method, we used a low-pass filter and median filter to remove noise interferences. The second one, we used a back-propagation neural network algorithm to suppress noises. We removed nearly all noise and recovered the originally embedded watermarks of Walsh–Hadmard codes.

scholarly journals Detection of Computer Graphics Using Attention-Based Dual-Branch Convolutional Neural Network from Fused Color Components

Sensors ◽  
2020 ◽  
Vol 20 (17) ◽  
pp. 4743
Author(s):  
Peisong He ◽  
Haoliang Li ◽  
Hongxia Wang ◽  
Ruimei Zhang
Keyword(s):  
Neural Network ◽  
Computer Graphics ◽  
Convolutional Neural Network ◽  
Video Game ◽  
Detection Methods ◽  
Post Processing ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Generalized Patterns ◽  
Color Components

With the development of 3D rendering techniques, people can create photorealistic computer graphics (CG) easily with the advanced software, which is of great benefit to the video game and film industries. On the other hand, the abuse of CGs has threatened the integrity and authenticity of digital images. In the last decade, several detection methods of CGs have been proposed successfully. However, existing methods cannot provide reliable detection results for CGs with the small patch size and post-processing operations. To overcome the above-mentioned limitation, we proposed an attention-based dual-branch convolutional neural network (AD-CNN) to extract robust representations from fused color components. In pre-processing, raw RGB components and their blurred version with Gaussian low-pass filter are stacked together in channel-wise as the input for the AD-CNN, which aims to help the network learn more generalized patterns. The proposed AD-CNN starts with a dual-branch structure where two branches work in parallel and have the identical shallow CNN architecture, except that the first convolutional layer in each branch has various kernel sizes to exploit low-level forensics traces in multi-scale. The output features from each branch are jointly optimized by the attention-based fusion module which can assign the asymmetric weights to different branches automatically. Finally, the fused feature is fed into the following fully-connected layers to obtain final detection results. Comparative and self-analysis experiments have demonstrated the better detection capability and robustness of the proposed detection compared with other state-of-the-art methods under various experimental settings, especially for image patch with the small size and post-processing operations.

scholarly journals Spatio-temporal prediction of suspended sediment concentration in the coastal zone using an artificial neural network and a numerical model

2012 ◽  
Vol 14 (3) ◽  
pp. 574-584 ◽  
Author(s):  
B. Bhattacharya ◽  
T. van Kessel ◽  
D. P. Solomatine
Keyword(s):  
Neural Network ◽  
Time Series ◽  
Artificial Neural Network ◽  
Numerical Model ◽  
Series Data ◽  
Ann Model ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Temporal Prediction ◽  
Artificial Neural

A problem of predicting suspended particulate matter (SPM) concentration on the basis of wind and wave measurements and estimates of bed shear stress done by a numerical model is considered. Data at a location at 10 km offshore from Noordwijk in the Dutch coastal area is used. The time series data have been filtered with a low pass filter to remove short-term fluctuations due to noise and tides and the resulting time series have been used to build an artificial neural network (ANN) model. The accuracy of the ANN model during both storm and calm periods was found to be high. The possibilities to apply the trained ANN model at other locations, where the model is assisted by the correctors based on the ratio of long-term average SPM values for the considered location to that for Noordwijk (for which the model was trained), have been investigated. These experiments demonstrated that the ANN model's accuracy at the other locations was acceptable, which shows the potential of the considered approach.

scholarly journals DIGITAL OSCILLATOR CIRCUIT WITH AN EXTENDED REPETITION PERIOD OF A PSEUDO-RANDOM PULSE SEQUENCE

2020 ◽  
Vol 69 (1) ◽  
pp. 210-214
Author(s):  
A. Zaurbek ◽  
◽  
D.Z. Dzhuruntaev ◽  
Keyword(s):  
Acoustic Noise ◽  
Pulse Sequence ◽  
Linear Feedback ◽  
Pseudorandom Sequence ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Generator Circuit ◽  
Random Pulse ◽  
Low Pass ◽  
Encryption Algorithms

In this paper, we consider the issue of upgrading the circuit of a digital generator of a pseudo-random pulse sequence, which can be used to create cryptographic encryption algorithms. The need to modernize the digital generator circuit is associated with an increase in the number of pseudorandom pulse train sequences generated at its output and with pseudorandom intervals between them. To achieve this, a small number of additional elements are included in the circuit of a digital pseudorandom sequence of pulses based on a five-digit shift register with linear feedback. Based on the modernized circuit of a digital generator of a pseudorandom sequence of pulses and an active secondorder Slenlen-Key RC low-pass filter, a digital acoustic noise generator is constructed, which, unlike the prototype, has a truly random output signal over a period of ~ 4 * (2N - 1), subject to circuit simplicity.

Research and Application of BP Neural Network in Specific Power Harmonic Detection

2012 ◽  
Vol 588-589 ◽  
pp. 379-383
Author(s):  
Xiao Dan Wang ◽  
Li Xin Ma ◽  
Yue Xiao Wang ◽  
Shu Juan Yuan
Keyword(s):  
Neural Network ◽  
Bp Neural Network ◽  
Reactive Power ◽  
Supply Voltage ◽  
Harmonic Component ◽  
Specific Power ◽  
Nonlinear Loads ◽  
Solution Quality ◽  
Pass Filter ◽  
Low Pass Filter

APF (Active Power Filter) is widely used in power system harmonic control and reactive power compensation, has been proven as an effective method to overcome various power quality issues such as unbalanced source current, large reactive power harmonic and neutral currents due to the proliferation of nonlinear loads. Optimizing the performance of APF using conventional ip-iq detection method based on instantaneous reactive power theory is quite difficult because of the complex coordinate conversion, what’s more, the presence of low-pass filter will cause a certain delay. This paper proposes the implementation of BP neural network to extract specific harmonic, it can optimize the APF performance for load compensation under distorted supply voltage condition and sudden load fluctuation. Weight adjustment using the BFGS quasi-Newton algorithm, which can accurately detect the fundamental and harmonic component of the phase amplitude . Matlab simulation results demonstrate that the performance of BP neural network algorithm is superior compared to conventional method, in terms of both convergence rate and solution quality.

A Coded Digital Watermarking Recognition Using Neural Network Method

2013 ◽  
Vol 284-287 ◽  
pp. 2961-2964
Author(s):  
Chih Ta Yen ◽  
Ing Jr Ding ◽  
Zong Wei Lai
Keyword(s):  
Neural Network ◽  
Digital Watermarking ◽  
Median Filter ◽  
Back Propagation ◽  
Back Propagation Neural Network ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Noise Interference ◽  
Low Pass ◽  
Neural Network Algorithm

Digital watermarking is an encryption technology commonly used to protect intellectual property and copyright. Although watermarks possess advantageous secrecy and robustness, environmental interference in the image propagation through the Internet is inevitable and, certainly, human-based image modification can also destroy the watermark. In this study, we restored watermarks that had already been affected by noise interference, used the Walsh-Hadamard codes as the watermark identification codes, and applied salt-and-pepper noise and Gaussian noise to destroy watermarks. First, we used a low-pass filter and median filter to remove noise interferences. Although these filters can suppress noises, watermarked images remain unidentifiable when the noise interferences strongly. Finally, we used a back-propagation neural network algorithm to filter noises, obtaining results that exceeded our expectations. We removed nearly all noise and recovered the originally embedded watermarks of Walsh-Hadmard codes.

scholarly journals EEMD and Multiscale PCA-Based Signal Denoising Method and Its Application to Seismic P-Phase Arrival Picking

Sensors ◽  
2021 ◽  
Vol 21 (16) ◽  
pp. 5271
Author(s):  
Kang Peng ◽  
Hongyang Guo ◽  
Xueyi Shang
Keyword(s):  
Wavelet Decomposition ◽  
Hankel Matrix ◽  
Principal Component ◽  
Signal Denoising ◽  
Complex Signals ◽  
Denoising Method ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Soft Thresholding ◽  
Adaptive Decomposition

Signal denoising is one of the most important issues in signal processing, and various techniques have been proposed to address this issue. A combined method involving wavelet decomposition and multiscale principal component analysis (MSPCA) has been proposed and exhibits a strong signal denoising performance. This technique takes advantage of several signals that have similar noises to conduct denoising; however, noises are usually quite different between signals, and wavelet decomposition has limited adaptive decomposition abilities for complex signals. To address this issue, we propose a signal denoising method based on ensemble empirical mode decomposition (EEMD) and MSPCA. The proposed method can conduct MSPCA-based denoising for a single signal compared with the former MSPCA-based denoising methods. The main steps of the proposed denoising method are as follows: First, EEMD is used for adaptive decomposition of a signal, and the variance contribution rate is selected to remove components with high-frequency noises. Subsequently, the Hankel matrix is constructed on each component to obtain a higher order matrix, and the main score and load vectors of the PCA are adopted to denoise the Hankel matrix. Next, the PCA-denoised component is denoised using soft thresholding. Finally, the stacking of PCA- and soft thresholding-denoised components is treated as the final denoised signal. Synthetic tests demonstrate that the EEMD-MSPCA-based method can provide good signal denoising results and is superior to the low-pass filter, wavelet reconstruction, EEMD reconstruction, Hankel–SVD, EEMD-Hankel–SVD, and wavelet-MSPCA-based denoising methods. Moreover, the proposed method in combination with the AIC picking method shows good prospects for processing microseismic waves.

scholarly journals Comparison and Performance Evaluation of Low Pass Filter Design for various samples using Artificial Neural Network

2016 ◽  
Vol 4 (3) ◽  
pp. 74-79
Author(s):  
Suchi Sharma ◽  
Anjana Goen
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Filter Design ◽  
Back Propagation Algorithm ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Feed Forward Back Propagation ◽  
Window Method ◽  
Low Pass ◽  
Artificial Neural

To design any type of filters complex calculation is needed. But with the help of window method, it become simple. In this paper a Low pass filter is designed by window method with the help of Artificial Neural Network. Here, Hann window and Blackman window is used to design filter and Feed Forward Back Propagation algorithm has been taken for neural network. In this paper different data sets for cut off frequency are consider for training and testing purpose to get best results.

Sign in / Sign up

Export Citation Format

Share Document