A Review on Speech Synthesis Based on Machine Learning

Text-to-speech synthesis system with Arabic diacritic recognition system

Water is a big challenge not only in India but in many countries of the world. Machine learning and forecasting model has been employed towards water demand and ground water level prediction. But in terms of water scarcity, much less work has been carried out by employing machine learning algorithms like 'artificial neural network' (ANN) and 'grey forecasting' model for forecasting water scarcity and none has focused on historical data like water availability, water consumption for a particular area and stress value for predicting water scarcity. So accordingly, we here have developed a water scarcity prediction system based on historical data by employing 'deep neural networks' which is an advanced form of 'artificial neural networks'. We have also compared 'deep neural network' with existing machine learning algorithms such as support vector machine (SVM), logistic regression and Naive Bayes. From the analysis of algorithms based on dataset, deep neural networks have been found as the best prediction model for water scarcity.

Machine Learning (ML) techniques are pervasive tools in various emerging commercial applications, but have to be accommodated by powerful computer systems to process very large data. Although general-purpose CPUs and GPUs have provided straightforward solutions, their energy-efficiencies are limited due to their excessive supports for flexibility. Hardware accelerators may achieve better energy-efficiencies, but each accelerator often accommodates only a single ML technique (family). According to the famous No-Free-Lunch theorem in the ML domain, however, an ML technique performs well on a dataset may perform poorly on another dataset, which implies that such accelerator may sometimes lead to poor learning accuracy. Even if regardless of the learning accuracy, such accelerator can still become inapplicable simply because the concrete ML task is altered, or the user chooses another ML technique.In this study, we present an ML accelerator called PuDianNao, which accommodates seven representative ML techniques, including k-means, k-nearest neighbors, naive bayes, support vector machine, linear regression, classification tree, and deep neural network. Benefited from our thorough analysis on computational primitives and locality properties of different ML techniques, PuDianNao can perform up to 1056 GOP/s (e.g., additions and multiplications) in an area of 3.51 mm^2, and consumes 596 mW only. Compared with the NVIDIA K20M GPU (28nm process), PuDianNao (65nm process) is 1.20x faster, and can reduce the energy by 128.41x.

ABSTRACTThis paper reviews the activity-travel behaviour literature that employs Machine Learning (ML) techniques for empirical analysis and modelling. Machine Learning algorithms, which attempt to build intelligence utilizing the availability of large amounts of data, have emerged as powerful tools in the fields of pattern recognition and big data analysis. These techniques have been applied in activity-travel behaviour studies since the early ’90s when Artificial Neural Networks (ANN) were employed to model mode choice decisions. AMOS, an activity-based modelling system developed in the mid-’90s, has ANN at its core to model and predict individual responses to travel demand management measures. In the dawn of 2000, ALBATROSS, a comprehensive activity-based travel demand modelling system, was proposed by Arentze and Timmermans using Decision Trees. Since then researchers have been exploring ML techniques like Support Vector Machines (SVM), Decision Trees (DT), Neural Networks (NN), Bayes Classifiers, and more recently, Ensemble Learners to model and predict activity-travel behaviour. A large number of publications over the years and an upward trend in the number of published articles over time indicate that Machine Learning is a promising tool for activity-travel behaviour analysis and prediction. This article, first of its kind in the literature, reviews these studies and explores the trends in activity-travel behaviour research that apply ML techniques. The review finds that mode choice decisions have received wide attention in the literature on ML applications. It was observed that most of the studies identify the lack of interpretability as a serious shortcoming in ML techniques. However, very few studies have attempted to improve the interpretability of the models. Further, some studies report the importance of feature engineering in ML-based studies, but very few studies adopt feature engineering before model development. Spatiotemporal transferability of models is another issue that has received minimal attention in the literature. In the end, the paper discusses possible directions for future research in the area of activity-travel behaviour modelling using ML techniques.

Embedded systems technology is undergoing a phase of transformation owing to the novel advancements in computer architecture and the breakthroughs in machine learning applications. The areas of applications of embedded machine learning (EML) include accurate computer vision schemes, reliable speech recognition, innovative healthcare, robotics, and more. However, there exists a critical drawback in the efficient implementation of ML algorithms targeting embedded applications. Machine learning algorithms are generally computationally and memory intensive, making them unsuitable for resource-constrained environments such as embedded and mobile devices. In order to efficiently implement these compute and memory-intensive algorithms within the embedded and mobile computing space, innovative optimization techniques are required at the algorithm and hardware levels. To this end, this survey aims at exploring current research trends within this circumference. First, we present a brief overview of compute intensive machine learning algorithms such as hidden Markov models (HMM), k-nearest neighbors (k-NNs), support vector machines (SVMs), Gaussian mixture models (GMMs), and deep neural networks (DNNs). Furthermore, we consider different optimization techniques currently adopted to squeeze these computational and memory-intensive algorithms within resource-limited embedded and mobile environments. Additionally, we discuss the implementation of these algorithms in microcontroller units, mobile devices, and hardware accelerators. Conclusively, we give a comprehensive overview of key application areas of EML technology, point out key research directions and highlight key take-away lessons for future research exploration in the embedded machine learning domain.

Dialect Identification is the process of identifies the dialects of particular standard language. The Telugu Language is one of the historical and important languages. Like any other language Telugu also contains mainly three dialects Telangana, Costa Andhra and Rayalaseema. The research work in dialect identification is very less compare to Language identification because of dearth of database. In any dialects identification system, the database and feature engineering play vital roles because of most the words are similar in pronunciation and also most of the researchers apply statistical approaches like Hidden Markov Model (HMM), Gaussian Mixture Model (GMM), etc. to work on speech processing applications. But in today's world, neural networks play a vital role in all application domains and produce good results. One of the types of the neural networks is Deep Neural Networks (DNN) and it is used to achieve the state of the art performance in several fields such as speech recognition, speaker identification. In this, the Deep Neural Network (DNN) based model Multilayer Perceptron is used to identify the regional dialects of the Telugu Language using enhanced Mel Frequency Cepstral Coefficients (MFCC) features. To do this, created a database of the Telugu dialects with the duration of 5h and 45m collected from different speakers in different environments. The results produced by DNN model compared with HMM and GMM model and it is observed that the DNN model provides good performance.

A machine learning approach for classifying healthy and infarcted patients using heart rate variabilities derived vector magnitude

Impact of pixel-based machine-learning techniques on automated frameworks for delineation of gross tumor volume regions for stereotactic body radiation therapy

Modelling of soil moisture retention curve using machine learning techniques: Artificial and deep neural networks vs support vector regression models

The work presented in this paper is about Text-to-Speech (TTS) synthesis and, more particularly, about statistical speech synthesis using the Hidden Markov Models (HMM). The main objective of this work is to study the functioning of the HMM-based speech synthesis system (HTS) and the implementation of this method to create a system that produces understandable speech output for a given Arabic text. We have done a brief description of the statistical parametric speech synthesis based on HMM, the steps followed to implement this method for Arabic language. Finally, for the evaluations of the system are based on subjective mean opinion score and objective tests. Regarding the intelligibility, naturalness aspects (listening) and the quality (Perceptual Evaluation of Speech Quality (PESQ)).

Vaccine design and development against viral diseases is multifaceted. Classical vaccines (live-attenuated vaccines, inactivated vaccines, subunit, recombinant, polysaccharide, conjugate vaccines, and toxoid vaccines) are often less effective for many viral diseases. Hence, short peptide (10–20 amino acid residues) vaccine components exploiting T-cell mediated immunity have been recognized as alternative solutions. This involves the specific binding of short antigen peptides to allele (gene variant) specific host human leukocyte antigens (HLA). Allele-specific HLA typing among different ethnic groups has gained momentum in recent years through advances in sequencing (Next Generation Sequencing (NGS)), High Performance Computing (HPC), machine learning techniques such as Artificial Neural Networks (ANN) and Support Vector Machine (SVM) techniques. More than 20,000 HLA alleles have been typed, defined, named, and made available at the IMGT®/HLA database (Immuno Polymorphism Database – ImMunoGeneTics Database/Human Leukocyte Antigen) for public access. Identification of short peptide antigens capable of binding specifically to the human host HLA alleles is now possible using computer-aided HLA-peptide binding prediction methods. This is achieved using three dimensional HLA structure based molecular modeling, and known HLA-peptide binding data enabled machine learning techniques like ANN and SVM. The former provides broad coverage across HLA alleles and the later offers high accuracy with high specificity for limited HLA alleles. Thus, the combined use of structural features, molecular modeling, machine learning techniques, and other applied mathematical models including Quantitative matrices (QM), Bayesian Networks (BN), and Hidden Markov Models (HMM) help in the effective design of short peptide vaccine components and immune therapeutics for the prevention and control of diseases caused by viruses. Hence, we outline recent advances in HLA-peptide binding prediction for short peptide vaccine design.

This paper is concerned with various machine Learning (ML) techniques to estimate the crop yield based on various factors, followed by an averaging technique. Multiple models are utilised by this ensemble machine learning technique called Bagging, taking weak learners as the base model to make the prediction. The pertinent ML algorithms that support crop yield prediction are Support Vector machines (SVM), Random trees (RT), Linear Regression (LR), K Nearest Neighbour (KNN), Random Forest (RF), Multiple Regression (MR), Support Vector Regression (SVR), Artificial Neural Networks (ANN) and Deep Neural Networks (DNN). Of these, in this paper, Classification and Regression Tree (CART), Ridge Regression (RR), eXtreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN) were implemented along with Bagging as the ensemble technique. The output of the entity models was thus averaged to give a final prediction.

Machine Learning (ML) is used to learn a system. One of the purposes of this automatic learning is the construction of new computational models. ML shows success in various areas such as reference systems, brain-computer interface, robotics, and chemistry. Recently, Perturbation Theory (PT) operators and ML techniques have been combined to create powerful PTML (PT+ML) models, which are applied to complex biological systems in predicting drug-protein interaction. As for those target proteins involved in the dopamine pathway, nanotechnology, material science, etc. The PTML models add the values ​​of the operators to the values ​​of . Therefore, we need to calculate the values ​​of the PTOs (Perturbation Theory Operators) in the data processing step. This allows us to carry out a process of merging information with variables and conditions from different sources. Moving Averages (MA), multi-condition MA (MMA), double MA, covariance operators, etc., are some examples of useful PTOs. Then, we can use Multiple Linear Regression (MLR), Linear Discriminant Analysis (LDA), or other linear ML techniques to find the PTML model. In non-linear cases, we can fit PTML models using Artificial Neural Networks (ANN), Support Vector Machines (SVM), Classification Trees, and other ML methods. One of the important applications of ANN is found in the study of chemical reactions as an alternative to classical regression and classification techniques. Therefore, in this master's thesis, PTML-ANN models are developed with the intention of visualizing possible improvements in classical techniques. In this context, this term is presented in a generic way. ANNs are inspired by the biological neural networks of the human brain. They are made up of elements that behave in a similar way to the biological neuron in its most common functions. The ANN aside from "resembling" the brain present a series of characteristics of the brain. For example, ANNs learn from experience, generalize from previous examples to new examples, and abstract the main characteristics of a data series. With which these networks function as interconnected neurons that create stimuli. ANN-type network is considered to be a complex biomolecular network consisting of nodes and edges. These networks are formed by "neurons", where each of them is a function, which will take a certain amount of data and provide an output response. The ANN presents three different types of functions which are: • Input function: It is the sum function obtained by multiplying the input and output data by their weights. • Excitation or activation function: This will take as previous input / output data. The most common types of functions depending on the input / output will be the threshold (sigmoid) or hyperbolic tangent. • Transfer functions: It is the function used to close. The value provided by the trigger functions is the input to the ANN network.

This study examines the use of machine learning (ML) techniques to optimize the basic machining parameters and protrusion dimensions that affect tool shape degeneration in die-sinking electric discharge machining (EDM). The primary objective is to decrease errors and enhance prediction and optimization effectiveness. This study introduces a completely novel tool geometry model aimed at minimizing tool shape degeneration, which, to our knowledge, has not been previously documented in the literature. Additionally, this research represents the first instance of employing ML techniques to generate data for addressing this specific type of problem, further advancing the field of die-sinking EDM. The pivotal machining parameters include discharge current, pulse time and machining depth. Three ML approaches are implemented in this investigation: Artificial Neural Network (ANN), Adaptive-Network-Based Fuzzy Inference System (ANFIS), and Support Vector Machine (SVM). In comparison with experimental outcomes, the ANN technique exhibited superior predictive ability with an coefficient of determination (R 2) of 0.99985 and an Mean Relative Error (MRE) of 0.854%. Four distinct EDM machining scenarios are presented and machining parameters and protrusion dimensions are optimized using the ANN technique to decrease tool shape degeneration. Optimizing the machining parameters and diagonal dimensions of the protrusion substantially reduced tool shape degeneration. This research demonstrates the effectiveness of ANN in optimizing machining parameters and improving tool performance in die-sinking EDM. A significant reduction in total wear area of 66.7% was achieved with a considerably lower time cost through the optimized ANN network. While the study demonstrates promising results, its reliance on specific datasets for training may limit the generalizability of the model to broader machining scenarios.

This paper proposes to select frame-sized speech segments for waveform concatenation speech synthesis using neural network based acoustic models. First, a deep neural network (DNN) based frame selection method is presented. In this method, three DNNs are adopted to calculate target costs and concatenation costs respectively for selecting candidate frames of 5ms length. One DNN is built in the same way as the DNN-based statistical parametric speech synthesis, which predicts target acoustic features given linguistic context inputs. The distance between the acoustic features of a candidate unit and the predicted ones for a target unit is calculated as the target cost. The other two DNNs are constructed to predict the acoustic features at current frame using its context features and the acoustic features of preceding frames. At synthesis time, these two DNNs are employed to calculate the concatenation cost for each candidate frame given its preceding ones. Furthermore, recurrent neural networks (RNNs) with long short-term memory (LSTM) cells are adopted to replace DNNs for acoustic modeling in order to make better use of the sequential information. A strategy of using multi-frame instead of single frame as the basic unit for selection is also presented to reduce the concatenation points within synthetic speech. Experimental results show that our proposed method can achieve better naturalness than the hidden Markov model (HMM)-based frame selection method and the HMM-based parametric speech synthesis method.