Time-frequency reassigned features for automatic chord recognition

During the last decades, automatic speech emotion recognition (SER) has gained an increased interest by the research community. Specifically, SER aims to recognize the emotional state of a speaker directly from a speech recording. The most prominent approaches in the literature include feature extraction of speech signals in time and/or frequency domain that are successively applied as input into a classification scheme. In this paper, we propose to exploit graph theory and structures as alternative forms of speech representations. We suggest applying the so-called Visibility Graph (VG) theory to represent speech data using an adjacency matrix and extract well-known graph-based features from the latter. Finally, these features are fed into a Support Vector Machine (SVM) classifier in a leave-one-speaker-out, multi-class fashion. Our proposed feature set is compared with a well-known acoustic feature set named the Geneva Minimalistic Acoustic Parameter Set (GeMAPS). We test both approaches on two publicly available speech datasets: SAVEE and EMOVO. The experimental results show that the proposed graph-based features provide better results, namely a classification accuracy of 70% and 98%, respectively, yielding an increase by 29.2% and 60.6%, respectively, when compared to GeMAPS.

Despite early success in automatic chord recognition, recent efforts are yielding diminishing returns while basically iterating over the same fundamental approach. Here, we abandon typical conventions and adopt a different perspective of the problem, where several seconds of pitch spectra are classified directly by a convolutional neural network. Using labeled data to train the system in a supervised manner, we achieve state of the art performance through this initial effort in an otherwise unexplored area. Subsequent error analysis provides insight into potential areas of improvement, and this approach to chord recognition shows promise for future harmonic analysis systems.

To meet the testing requirements of optical automatic target recognition system for loitering munition, based on digital twin, a parallel test system is proposed. Using this parallel test system, single application tests in the design stage of optical automatic target recognition system are carried out first, and then the operational course of action of loitering munition is verified. Through test examples, the analytical method of the technical and tactical capacity of optical automatic target recognition system is introduced. Besides, by carrying out operational course of action verification tests, the effectiveness of designed trajectory planning is verified. With the help of parallel test, more trajectory planning options will be derived to provide reference for actual flight tests.

Human perception of sign language can serve as inspiration for the improvement of automatic recognition systems. Experiments with human signers show that sign language signs contain redundancy over time. In this article, experiments are conducted to investigate whether comparable redundancies also exist for an automatic sign language recognition system. Such redundancies could be exploited, for example, by reserving more processing resources for the more informative phases of a sign, or by discarding uninformative phases. In the experiments, an automatic system is trained and tested on isolated fragments of sign language signs. The stimuli used were similar to those of the human signer experiments, allowing us to compare the results. The experiments show that redundancy over time exists for the automatic recognizer. The central phase of a sign is the most informative phase, and the first half of a sign is sufficient to achieve a recognition performance similar to that of the entire sign. These findings concur with the results of the human signer studies. However, there are differences as well, most notably the fact that human signers score better on the early phases of a sign than the automatic system. The results can be used to improve the automatic recognizer, by using only the most informative phases of a sign as input.

This paper presents an automatic fruit recognition system for classifying and identifying fruit types. The work exploits the fruit shape and color, to identify each image feature. The proposed system includes three phases namely: pre-processing, feature extraction, and classification phases. In the pre-processing phase, fruit images are resized to 90 x 90 pixels in order to reduce their color index. In feature extraction phase, the proposed system uses scale invariant feature transform (SIFT) and shape and color features to generate a feature vector for each image in the dataset. For classification phase, the proposed model applies K-Nearest Neighborhood (K-NN) algorithm classification, and support vector machine (SVM) algorithm of different kinds of fruits. A series of experiments were carried out using the proposed model on a dataset of 178 fruit images. The results of carrying out these experiments demonstrate that the proposed approach is capable of automatically recognize the fruit name with a high degree of accuracy.

Vehicle license plate recognition has been a much studied research area in many countries. Due to the different types of license plates being used, the requirement of an automatic license plate recognition system is rather different for each country. In this paper, an automatic license plate recognition system is proposed for Malaysian vehicles with standard license plates based on image processing, feature extraction and neural networks. The image processing library is developed in-house which we referred to as vision system development platform (VSDP). The Kirsch edge feature extraction technique is used to extract features from the license plates characters which are then used as inputs to the neural network classifier. The neural network model is the standard multilayered perceptron trained using the back-propagation algorithm. The prototyped system has an accuracy of about 91%, however, suggestions to further improve the system are discussed in this paper based on the analysis of the error.

In this paper, an automatic facial expression recognition system is presented. When a face image is input, two inner canthi are detected as the reference points for searching the expression features extracted from the contour and displacement of eyebrows, eyes, and mouth. Our feature extraction method can reduce the partial influence of shadows and noises. Finally, the expression features are used as the input to an Elman neural network of classifiers. The results on the JAFFE facial database show an average recognition accuracy of 84.7% in seven expressions by automatic canthi detection and 92.2% by manual canthi detection.

In recent years, with the popularity of the Internet, digital media has become an indispensable part of people's life. Image acquisition equipment is widely used in daily life and work. However, due to technical problems, it can not meet the performance requirements of users. At the same time, a large number of data and information are difficult to be effectively processed, which also reduces the image quality, reduces the storage space and increases the image distortion. Therefore, it is necessary to develop new algorithms to improve these defects. Therefore, this paper designs and develops an automatic digital media image recognition system based on mean clustering algorithm. Firstly, this paper describes the concept and characteristics of digital media. Then, this paper also studies the application of mean clustering algorithm. Based on this, an automatic digital media image recognition system is designed, and the performance of the system is tested by simulation. Finally, the test results show that the algorithm can accurately identify media images. In general, the performance of the digital media image automatic recognition system based on mean clustering algorithm can meet the needs of target users.

With the increasing improvement of living standards, healthy long-distance running has become one of the current ways of aerobic physical exercise. Long-distance running is beneficial to many bodily functions such as neck, spine, and heart, but inappropriate long-distance running can easily damage the joints of the human body. In order to make more fitness athletes run with a healthy and reasonable posture, this paper studies the automatic recognition and correction system of running movements based on computer vision technology. Based on the analysis of human motion detection methods and human motion recognition methods, an automatic recognition and correction system for running actions is designed. In order to understand the accuracy of the system in the recognition of wrong actions, this paper compares the designed system with the traditional system. The results show that the system designed in this paper has good accuracy and can control the deviation within a reasonable range.

Many applications reported in artificial neural networks are associated with military problems. This paper reviews concepts associated with the processing of military data to find and recognize targets -- automatic target recognition (ATR). A general-purpose automatic target recognition system does not exist. The work presented here is demonstrated on military data, but it can only be considered proof of principle until systems are fielded and proven `under- fire.' ATR data can be in the form of nonimaging one-dimensional sensor returns, such as ultra-high range-resolution radar returns (UHRR) for air-to-air automatic target recognition and vibration signatures from a laser radar for recognition of ground targets. The ATR data can be two-dimensional images. The most common ATR images are infrared, but current systems must also deal with synthetic aperture radar (SAR) images. Finally, the data can be three-dimensional, such as sequences of multiple exposures taken over time from a nonstationary world. Targets move, as do sensors, and that movement can be exploited by the ATR. Hyperspectral data, which are views of the same piece of the world looking at different spectral bands, is another example of multiple image data; the third dimension is now wavelength and not time. ATR system design usually consists of four stages. The first stage is to select the sensor or sensors to produce the target measurements. The next stage is the preprocessing of the data and the location of regions of interest within the data (segmentation). The human retina is a ruthless preprocessor. Physiologically motivated preprocessing and segmentation is demonstrated along with supervised and unsupervised artificial neural segmentation techniques. The third design step is feature extraction and selection: the extraction of a set of numbers which characterize regions of the data. The last step is the processing of the features for decision making (classification). The area of classification is where most ATR related neural network research has been accomplished. The relation of neural classifiers to Bayesian techniques is emphasized along with the more recent use of feature sequences to enhance classification. © (1995) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

The use of classification techniques for machine health monitoring and fault diagnosis has been popular in recent years. System response in form of time series data can be used to identify type of defect, severity of defect etc. However, a central issue with time series classification is that of identifying appropriate features for classification. In this paper, we explore a new feature set based on a delay differential equations (DDEs). DDEs have been used recently for extracting features for classification but have never been used to classify system responses. The Duffing oscillator and Van der Pol–Duffing (VDP-D) oscillator are used as dynamic systems, and the responses are classified into self-similar groups. Responses with the same period should belong to the same group. Misclassification rate is used as an indicator of the efficacy of the feature set. The proposed feature set is compared to a statistical feature set, a power spectral coefficient feature set and a wavelet coefficient feature set. In work described in this paper, a density estimation algorithm called DBSCAN is used as the classification algorithm. The proposed DDE-based feature set is found to be significantly better than the other feature sets for the classifying responses generated by the Duffing system. The wavelet and the power spectral coefficient data sets are not found to be significantly better than the statistical feature set for the Duffing system. None of the feature sets tested are discerning enough on the VDP-D system.

Aphasia assessment plays a crucial role in rehabilitating people with aphasia (PWA), including classifying healthy individuals, identifying subtypes of aphasia, and assessing severity. This study explores unsupervised clustering methods in aphasic speech data, comparing five feature extraction approaches and four clustering algorithms. Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN), Agglomerative clustering, K-means, and Gaussian Mixture Models. The focus is on distinguishing between fluent and non-fluent aphasia and subtypes such as Broca's, Wernicke's, and anomic aphasia. The feature extraction methods include Mel-Frequency Cepstral Coefficients (MFCCs), OpenL3 embeddings, their combination, and a fusion of MFCC with chroma features, spectral contrast, zero-crossing rate, and a novel deep convolutional autoencoder-based (CAE) model. These methods are evaluated for their ability to capture key characteristics of aphasic speech, with the clustering results assessed using intrinsic evaluation metrics. Our findings highlight the effectiveness of different feature extraction methods, with CAE emerging as the most effective approach in distinguishing aphasic speech categories and improving clustering performance. This research highlights the potential of unsupervised learning in aphasia research and offers insight into automated diagnosis and speech therapy tools, contributing to enhanced aphasia assessment and rehabilitation.

In this study, an automatic birdsong recognition system based on syllable features was developed. In this system, after syllable segmentation, three syllable features, namely mean, QI and QE, were computed from the MFCCs of each syllable aims at capturing variations in time as well as amplitude transitions of the MFCC sequences. With the advantages of the fuzzy c-mean (FCM) clustering algorithm and the linear discriminant analysis (LDA), the presented feature vector was used to construct an automatic birdsong recognition system applied to a birdsong database with 420 bird species.KeywordsMFCCsyllablelinear discriminant analysistransition matrix

Automatic emotion recognition is a challenging task since emotion is communicated through different modalities. Deep Convolution Neural Networks (DCNN) and transfer learning have shown success in automatic emotion recognition using different modalities. However significant improvement in accuracy is still required for practical applications. Existing methods are still not effective in modelling the temporal relationships within emotional expressions or in identifying the salient features from different modes and fusing them to improve accuracies. In this paper, we present an automatic emotion recognition system using audio and visual modalities. VGG19 models are used to capture frame level facial features followed by a Long Short Term Memory (LSTM) to capture their temporal distribution at a segment level. A separate VGG19 model captures auditory features from Mel Frequency Cepstral Coefficients (MFCC). The extracted auditory and visual features are fused together and a Deep Neural Network (DNN) with attention is used in classification using majority voting. Voice Activity Detection (VAD) on the audio stream improves performance by reducing the outliers in learning. The system is evaluated using Leave One Subject Out (LOSO) and K-fold cross-validation and our system outperforms state of the art methods on two challenging databases.

Optical Character Recognition (OCR) research includes computer vision, artificial intelligence, and pattern recognition. Character recognition has garnered a lot of attention in the last decade due to its broad variety of uses and applications, including multiple-choice test data, business documents (e.g., ID cards, bank notes, passports, etc.), and automatic number plate recognition. This paper introduces an automatic recognition system for printed numerals. The automatic reading system is based on extracting local statistical and geometrical features from the text image. Those features are represented by eight vectors extracted from each digit. Two of these features are local statistical (A, A th), and six are local geometrical (P1, P2, P3, P4, P5, and P6). Thus, the database created consists of 1120 statistical and geometrical features. For the purpose of recognition, the features of the test image are compared with the features of all the images saved in the database depending on the value of the Minimum Distance (MD). All digits (0–9) were identified with 100% accuracy. The average computational time required to recognize a numeral at any font size is 0.06879 seconds.