Characteristics Of Audio Signals Research Articles

Time-Encoding-Based Ultra-Low Power Features Extraction Circuit for Speech Recognition Tasks

Current trends towards on-edge computing on smart portable devices requires ultra-low power circuits to be able to make feature extraction and classification tasks of patterns. This manuscript proposes a novel approach for feature extraction operations in speech recognition/voice activity detection tasks suitable for portable devices. Whereas conventional approaches are based on either completely analog or digital structures, we propose a “hybrid” approach by means of voltage-controlled-oscillators. Our proposal makes use of a bank a band-pass filters implemented with ring-oscillators to extract the features (energy within different frequency bands) of input audio signals and digitize them. Afterwards, these data will input a digital classification stage such as a neural network. Ring-oscillators are structures with a digital nature, which makes them highly scalable with the possibility of designing them with minimum length devices. Additionally, due to their inherent phase integration, low-frequency band-pass filters can be implemented without large capacitors. Consequently, we strongly benefit from power consumption and area savings. Finally, our proposal may incorporate the analog-to-digital converter into the structure of the own features extractor circuit to make the full conversion of the raw data when triggered. This supposes a unique advantage with respect to other approaches. The architecture is described and proposed at system-level, along with behavioral simulations made to check whether the performance is the expected one or not. Then the structure is designed with a 65-nm CMOS process to estimate the power consumption and area on a silicon implementation. The results show that our solution is very promising in terms of occupied area with a competitive power consumption in comparison to other state-of-the-art solutions.

Feasibility research on break-out detection using audio signal in drilling film cooling holes by EDM

The electrical discharge machining (EDM) process with inner and outer flush is effective for drilling film cooling holes on the blades of aeroengines and gas turbines. Due to the hollow structure of the blades, the break-out status needs to be accurately detected in order to control the position of tool electrode end to prevent the back strike. Currently, the break-out detection is mainly realized based on the change of electrical signals and tool electrode velocity. The previous research shows that the machining stability of the process can be improved by optimizing the outer flush. However, in this case the break-out status is difficult to be detected using the existing methods. In this research, a novel break-out detection method using audio signal is proposed. The audio signals of the process are firstly collected and analyzed through time and frequency domain methods to provide the basis for algorithm design. Then a filter of the audio signals is designed based on the comparison of original signals collected before and after break-out to reduce the impact of noise. And the feature extraction algorithms are designed to reflect the characteristics of break-out with less dimension. Next a recognition model using artificial neural network is established by the features of audio signals, which are classified as before or after break-out. The recognition model is validated by test data acquired by machining experiments and the results verified that the break-out detection using audio signal is feasible in drilling film cooling holes.

Audio Example Recognition and Retrieval Based on Geometric Incremental Learning Support Vector Machine System

With the fast development of computer and information technology, multimedia data has become the most important form of information media. Auditory information plays an important role in information location, this comes from the fact that it can be difficult to find useful information. Thus audio classification becomes more important in audio analysis as it prepares for content-based audio retrieval. There is quite a bit of research on the topic of audio classification methods, audio feature analysis, and extraction based on audio classification. Many works of literature extract features of audio signals based on time or Fourier transform frequency domain. The emergence of the wavelet theory provides a time-frequency analysis tool for signal analysis. Wavelet transformation is a local transformation of the signal in time and frequency which can effectively extract information from the signal, and perform multi-scale refinement analysis on functions or signals through operations such as stretching and translation instead of the traditional Fourier transformation. In the time-frequency analysis of the signal, the wavelet analysis captures the local time and frequency characters of the signal which can improve the ability of signal analysis. It can also change certain locals of the signal without affecting other aspects of it. In this paper, the frequency domain features are combined with the wavelet domain features. At the same time that the MFCC features are extracted, the discrete wavelet transform is used to extract the features of the wavelet domain. Then the statistical features are extracted for each audio example, and the SVM model is used to realize the different forms of audio classification identification.

Percussion-based bolt looseness monitoring using intrinsic multiscale entropy analysis and BP neural network

In this paper, a novel percussion-based bolt looseness monitoring approach using intrinsic multiscale entropy analysis and back propagation (BP) neural network is proposed. The percussion-caused audio signals of bolt connection are decomposed by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to obtain intrinsic mode functions (IMFs). The IMFs are in order of high-to-low instantaneous frequencies and contain underlying dynamical characteristics of audio signals. Multiscale sample entropy (MSE) is improved by smoothed coarse graining process, and the proposed improved multiscale sample entropy (IMSE) values of certain IMFs are adopted as condition indicators in bolt looseness monitoring. The intrinsic multiscale entropy analysis consisting of CEEDMAN and IMSE extracts underlying dynamical characteristics during percussion-caused audio signal processing to identify bolt looseness conditions. The condition indicators, namely IMSE values at smallest scale factors, are employed as input of BP neural network for training and testing, to achieve accurate and stable bolt looseness condition monitoring. The effectiveness and superiority of the proposed approach have been validated by theoretical derivation and practical experimental researches, and the adaptivity and robustness of the proposed approach are also illustrated. The results of the research in this paper demonstrate the proposed approach is promising in practical applications of bolt looseness monitoring.

Trends in audio signal feature extraction methods

Audio signal processing algorithms generally involves analysis of signal, extracting its properties, predicting its behaviour, recognizing if any pattern is present in the signal, and how a particular signal is correlated to another similar signals. Audio signal includes music, speech and environmental sounds. Over the last few decades, audio signal processing has grown significantly in terms of signal analysis and classification. And it has been proven that solutions of many existing issues can be solved by integrating the modern machine learning (ML) algorithms with the audio signal processing techniques. The performance of any ML algorithm depends on the features on which the training and testing is done. Hence feature extraction is one of the most vital part of a machine learning process. The aim of this study is to summarize the literature of the audio signal processing specially focusing on the feature extraction techniques. In this survey the temporal domain, frequency domain, cepstral domain, wavelet domain and time-frequency domain features are discussed in detail.

A Preliminary Study of the Automatic Classification of the Site of Airway Collapse in OSA patients Using Snoring Signals.

In this study we investigated if audio signals may carry information related to the site of obstruction of the upper airway. The information regarding the site of collapse could improve obstructive sleep apnoea (OSA) treatment by allowing more individualized or structure-specific therapy. In this preliminary study, we developed an algorithm for automatically determining the site of collapse in 13 OSA patients through snoring analysis. Audio was recorded with a ceiling mounted microphone, simultaneously with full-night polysomnography during sleep. The surrogate measure of the site of airway collapse was identified by manual analysis of the nasal pressure signal. We extracted various time and frequency features of audio signal to classify the signal into "lateral wall", "palate" and "tongue base" related collapse. The classification was carried out with a Gaussian mixture model classifier. Performance of the proposed model showed that it can achieve an overall accuracy of 78.9±0.96% with specificity and PPV of (89.3±0.81%, 78±1.5%) (73.2±1.3% , 83.2±1.8%) (61.3±2.5%, 71.8±1.3%) for lateral wall, palate and tongue base related collapse respectively. Our preliminary results suggest that the audio signal may be helpful in identifying the site of obstruction and therefore maybe a useful tool for deciding appropriate therapy.

An audio attention computational model based on information entropy of two channels and exponential moving average

The main down-top attention model usually uses some characteristics of audio signal to extract the auditory saliency map at present. But existing audio attention computational model based image saliency mostly doesn’t consider the continuity and attenuation mechanism of the human brain on paying attention to some occurred events in our real environment. To address these issues, we propose a model based on information entropy of two channels and exponential moving average (EMA) to simulate the processing of sound signals from the basilar membrane to the cochlear nucleus, and get the local information entropy of image and audio channel, finally we use the relevant EMA methods to calculate the auditory attention map. Some experimental results on artificial audio signal and real-world audio signal belonging to a public corpus show that the proposed model not only can detect the attention event, but also reflect the mechanism of human brain, and the experimental results of talk show in real environment are better.

Classification of needle-EMG resting potentials by machine learning.

The diagnostic importance of audio signal characteristics in needle electromyography (EMG) is well established. Given the recent advent of audio-sound identification by artificial intelligence, we hypothesized that the extraction of characteristic resting EMG signals and application of machine learning algorithms could help classify various EMG discharges. Data files of 6 classes of resting EMG signals were divided into 2-s segments. Extraction of characteristic features (384 and 4,367 features each) was used to classify the 6 types of discharges using machine learning algorithms. Across 841 audio files, the best overall accuracy of 90.4% was observed for the smaller feature set. Among the feature classes, mel-frequency cepstral coefficients (MFCC)-related features were useful in correct classification. We showed that needle EMG resting signals were satisfactorily classifiable by the combination of feature extraction and machine learning, and this can be applied to clinical settings. Muscle Nerve 59:224-228, 2019.

A Relative Phase Based Audio Integrity Protection Method: Model and Strategy

Audio oriented integrity protection should consider the characteristics of audio signals based on the combination of audio application scenarios. However, in the current popular network interaction environment, traditional verification based solutions can no longer work. A kind of integrity protection scheme for audio business should be redesigned in these new scenarios. In this context, a method of audio integrity protection based on relative phase (RP-AIP) is proposed. Through the design of integrity object (I.O.), the integrity of audio can be abstracted as the completeness and accuracy of it. The I.O. is bound to the audio signal in a uniformly and randomly embedded manner, and the embedding rules are controlled by the relative phase characteristic of the audio itself. The model and strategy of RP-AIP are illustrated, and the corresponding process and algorithm are demonstrated as well. Simulation experiments illustrate the feasibility of the proposed solution and indicate the superiority of its performance.

Genre Classification for Brazilian Music using Independent and Discriminant Features

Digital music files are largely available both online and in private local collections. These databases may comprise hundreds or thousands of files, which in some cases may not carry tagged information about their content, making the search for the desired audio files very time consuming. An important task in this context is to organize the available database according to the prevailing musical genre. The purpose of this work is to develop an automatic music genre classification system able to identify international music genres (i.e. pop, rock, classic, soul, funk) and also typical Brazilian rhythms such as Samba, Forr\'o and Brazilian Popular Music. The proposed signal processing chain comprises two stages. Initially, audio signal features are computed and their relevance for music genre identification estimated. Independent component analysis is applied to reduce mutual redundancy among the audio attributes. In the following, different classifiers based on neural networks and support vector machines are applied for music genre identification. The proposed system efficiency is evaluated using an experimental dataset.

Prediction of Beck Depression Inventory (BDI-II) Score Using Acoustic Measurements in a Sample of Iium Engineering Students

Psychiatrist currently relies on questionnaires and interviews for psychological assessment. These conservative methods often miss true positives and might lead to death, especially in cases where a patient might be experiencing suicidal predisposition but was only diagnosed as major depressive disorder (MDD). With modern technology, an assessment tool might aid psychiatrist with a more accurate diagnosis and thus hope to reduce casualty. This project will explore on the relationship between speech features of spoken audio signal (reading) in Bahasa Malaysia with the Beck Depression Inventory scores. The speech features used in this project were Power Spectral Density (PSD), Mel-frequency Ceptral Coefficients (MFCC), Transition Parameter, formant and pitch. According to analysis, the optimum combination of speech features to predict BDI-II scores include PSD, MFCC and Transition Parameters. The linear regression approach with sequential forward/backward method was used to predict the BDI-II scores using reading speech. The result showed 0.4096 mean absolute error (MAE) for female reading speech. For male, the BDI-II scores successfully predicted 100% less than 1 scores difference with MAE of 0.098437. A prediction system called Depression Severity Evaluator (DSE) was developed. The DSE managed to predict one out of five subjects. Although the prediction rate was low, the system precisely predict the score within the maximum difference of 4.93 for each person. This demonstrates that the scores are not random numbers.

Speech Emotion Classification Analysis using Short-term Features

Speech is an auditory signal produced from the human speech production system used to express ourselves. In this era, speech signals are also used in biometric identification technologies and interacting with machines, so that it can give different response. Emotion recognition is not a new topic and researches and applications exist using different methods to extract specific features from the speech signals. This paper presents a classification analysis of emotional human speech only with short term processing features of the speech signals using artificial neural network based approach. Speech rate, pitch and energy are the most basic features of speech signal but they still have significant differences between emotions such as angry and sad. The most common way to analyze the speech emotion is to extract important features which are related to different emotion states from the voice signal. In the speech pre-processing phase, the samples of four basic types of emotional speeches sad, angry, happy, and neutral are used. Then feed those extracted short term features into the input end of the classifier and obtained different emotions at the output end. 23 short term audio signal features of 40 samples of two frames are selected and extracted from the speech signals to analyze the human emotions. These derived data along with their related emotion target matrix are fed to test and design the classifier using artificial neural network pattern recognition algorithm. The confusion matrix is generated to analyze the performance results. The overall correctly classified results for two times trained network is 73.8 %, while increasing the training times to ten, 95 % of the emotions are correctly classified. The accuracy of the neural network system is improved by multiple times of training. The overall system provides a reliable performance and correctly classifies more than 85 % for the new non-trained dataset.

Classification of the fundamental frequency identification methods

A significant role in the formation of the voice characteristics of the information signals given to the speaker system represents a language group, which is responsible for basic voice parameters such as pitch frequency of the signal, information frames and properties frequency range, volume, frequency and rate of speech, intonation, etc. The main criteria features of voice were identified. The basic characteristics of audio signals generated by a person of a certain language group were analyzed. The methods and models to identify the fundamental frequency, which further gave the opportunity to determine the period of the fundamental tone were analyzed. In the current research, the determination of the fundamental frequency, there are problems such as the complexity of the implementation methods, and low probability of error in determination process; low resistance methods to external interference. Methods of determination of fundamental frequency with the plane of their implementation were classified.

Classifying type of vehicles on the basis of data extracted from audio signal characteristics

The aim of this study is to find and optimize a feature vector for an automatic recognition of the type of vehicles, extracted form an audio signal. First, the influence of weather-based conditions of road surface on spectral characteristic of the audio signal recorded from a passing vehicle in close proximity to the road is discussed. Next, parameterization of the recorded audio signal is performed. For that purpose, the MIRtoolbox, designed for music parameter extraction, is used to obtain a vector of parameters. Correlation analyses are performed to check whether extracted parameters enable to separate selected types of vehicle-associated noise, e.g.: car, truck and motorcycle. Behrens-Fisher statistics is used to find the most suitable parameters that may be contained in the optimized feature vector. The last step is to build a decision system that allows for the automatic classification of a vehicle type. The results of automatic classification of prepared vehicle-noise related samples are shown and ...

Design and Implementation of Sound Searching Robots in Wireless Sensor Networks.

A sound target-searching robot system which includes a 4-channel microphone array for sound collection, magneto-resistive sensor for declination measurement, and a wireless sensor networks (WSN) for exchanging information is described. It has an embedded sound signal enhancement, recognition and location method, and a sound searching strategy based on a digital signal processor (DSP). As the wireless network nodes, three robots comprise the WSN a personal computer (PC) in order to search the three different sound targets in task-oriented collaboration. The improved spectral subtraction method is used for noise reduction. As the feature of audio signal, Mel-frequency cepstral coefficient (MFCC) is extracted. Based on the K-nearest neighbor classification method, we match the trained feature template to recognize sound signal type. This paper utilizes the improved generalized cross correlation method to estimate time delay of arrival (TDOA), and then employs spherical-interpolation for sound location according to the TDOA and the geometrical position of the microphone array. A new mapping has been proposed to direct the motor to search sound targets flexibly. As the sink node, the PC receives and displays the result processed in the WSN, and it also has the ultimate power to make decision on the received results in order to improve their accuracy. The experiment results show that the designed three-robot system implements sound target searching function without collisions and performs well.

Variation in Multitrack Mixes: Analysis of Low-level Audio Signal Features

To further the development of intelligent music production tools, towards generating mixes that would realistically be created by a human mix-engineer, it is important to understand what kind of mixes can be created, and are typically created, by human mix-engineers. This paper presents an analysis of 1501 mixes, over 10 different songs, created by mix-engineers. The primary dimensions of variation in the full dataset of mixes were ‘amplitude’, ‘brightness’, ‘bass’ and ‘width’, as determined by feature-extraction and subsequent principal component analysis. The distribution of representative features approximated a normal distribution and this is then used to obtain general trends and tolerance bounds for these features. The results presented here are useful as parametric guidance for intelligent music production systems.

An Associative Memorization Architecture of Extracted Musical Features from Audio Signals by Deep Learning Architecture

In this paper, we develop associative memorization architecture of the musical features from time sequential data of the music audio signals. This associative memorization architecture is constructed by using deep learning architecture. Challenging purpose of our research is the development of the new composition system that automatically creates a new music based on some existing music. How does a human composer make musical compositions or pieces? Generally speaking, music piece is generated by the cyclic analysis process and re-synthesis process of musical features in music creation procedures. This process can be simulated by learning models using Artificial Neural Network (ANN) architecture. The first and critical problem is how to describe the music data, because, in those models, description format for this data has a great influence on learning performance and function. Almost of related works adopt symbolic representation methods of music data. However, we believe human composers never treat a music piece as a symbol. Therefore raw music audio signals are input to our system. The constructed associative model memorizes musical features of music audio signals, and regenerates sequential data of that music. Based on experimental results of memorizing music audio data, we verify the performances and effectiveness of our system.

A Method of Extracting Feature of Phonic Signal Based on the Matrix Analysis

In the speech recognition technology, feature extraction is essential for the system recognition rate, taking amount of strategies to find the better feature vectors are most researchers target. This paper presents a method of extracting feature of audio signal based on the discrete wavelet transform, then decomposed the coefficient matrix by the matrix analysis way, through this method to find a new thinking on the way of extracting feature vector. The method can be achieved in the procedure. The main purpose is to reduce the dimension of feature vector, make the vector briefer, and then reduce the computing complexity in the embedded system. This method can reduce the feature vectors dimension, accelerated the computing velocity.

Sound source localization using compressive sensing-based feature extraction and spatial sparsity

In this paper, we propose a source localization algorithm based on a sparse Fast Fourier Transform (FFT)-based feature extraction method and spatial sparsity. We represent the sound source positions as a sparse vector by discretely segmenting the space with a circular grid. The location vector is related to microphone measurements through a linear equation, which can be estimated at each microphone. For this linear dimensionality reduction, we have utilized a Compressive Sensing (CS) and two-level FFT-based feature extraction method which combines two sets of audio signal features and covers both short-time and long-time properties of the signal. The proposed feature extraction method leads to a sparse representation of audio signals. As a result, a significant reduction in the dimensionality of the signals is achieved. In comparison to the state-of-the-art methods, the proposed method improves the accuracy while the complexity is reduced in some cases.

High payload data-hiding in audio signals based on a modified OFDM approach

This paper presents a high payload data-hiding scheme for audio signals following the OFDM concept. It is based on changing the phase component of the audio signal via a reduced-arc M-order Phase Shift Keying (MPSK) modulator on selected frequency samples of the audio signal. This approach allows the data-hiding payload and perceptual distortion to be controlled by the number of altered frequency components of the audio signal and the severity of this alteration. The data-hiding payload can be estimated a priori independently of the audio signal features. In addition, due to the relation of the proposed scheme with the traditional use of MPSK modulation, the BER of the encoded data is analytically derived, therefore, it is known in advance. This allows the transmission of reliable hidden data, in contrast to previous works. Results obtained with representative test cases show that the proposed algorithm provides an increase of one order of magnitude in data payload compared with state-of-the-art works.