In recent years, the IoE has become a network of smart devices having applications in smart grids, smart homes, and power plants. Due to enormous sensing nodes in IoE, the replacement of batteries is frequent. For this reason, IoE gadgets... more
In recent years, the IoE has become a network of smart devices having applications in smart grids, smart homes, and power plants. Due to enormous sensing nodes in IoE, the replacement of batteries is frequent. For this reason, IoE gadgets with a productive and low-power analog to digital converters (ADC) as a key block have special technical significance and advantage. The ‘ADC’ is the vital component of the communication system and signal processing. Several techniques are recommended to enhance the ADC performance. The extensive research in past has revealed that for parameters like low power and small area, successive approximation register (SAR) ADC is preferred. The signal-to-noise ratio (SNR) of high-speed SAR ADCs is mainly dominated by comparator noise and usually limited to 50 to 60dB. The power consumption increases exponentially to suppress comparator noise in a limited comparison time to improve SNR. This paper, has attempted to discuss the comparative performance of var...
Research Interests:
The field of image processing, image quality assessment is a fundamental and challenging problem with many interests in a variety of applications. Dynamic monitoring, adjusting image quality, optimizing algorithms and parameter settings... more
The field of image processing, image quality assessment is a fundamental and challenging problem with many interests in a variety of applications. Dynamic monitoring, adjusting image quality, optimizing algorithms and parameter settings of image processing systems are benchmarking in image processing system and algorithms. In this paper using such techniques we are going to analysis the detection of diseases like tumour for the ease in medical use.
Research Interests:
Interleaver plays key role in preserving the performance of turbo encoding systems. Under high dimensional data transmission, the interleaver design often goes complex. This paper presents a hybrid meta-heuristic search algorithm by... more
Interleaver plays key role in preserving the performance of turbo encoding systems. Under high dimensional data transmission, the interleaver design often goes complex. This paper presents a hybrid meta-heuristic search algorithm by combining renowned Genetic Algorithm (GA) and Group Search Optimizer (GSO) in the name of hybrid GSO (HGSO). The HGSO is emphasized to operate in high dimensional space so that the interleaver design is expected to be robust under high dimensional data transmission. The hybridization embodies the mutation operator of GA in the GSO scanning process. This improves the exploration process of GSO to enable faster convergence. Experiments are conducted at higher order data bits and the performance of HGSO is demonstrated. A statistical report is prepared from the observed results to illustrate the reliability of the outcome accomplished by HGSO over the other methods.
Speech signal processing is one of the most interesting areas to work with. Speech signal consists of scads of information like gender categorization, various voice features, emotion characteristics etc. Every speaker and speech has its... more
Speech signal processing is one of the most interesting areas to work with. Speech signal consists of scads of information like gender categorization, various voice features, emotion characteristics etc. Every speaker and speech has its own special signal characteristics. This review is a proposed approach which is an analysis of various methods for detecting the thyroid diseases. Basically, thyroid is butterfly-shaped gland present in the lower anterior of the neck. This controls the metabolism of the body. Different methods used for detecting thyroid disease differentiating into two ways i.e. Biomedical Analysis and Signal Processing. In this review, diverse bio-medical methods like voice self-perceptual assessment, statistical analysis, complication analysis and surgical analysis has been explained. In signal processing method, CAD (Computer Aided Diagnosis) and Acoustic Voice Analysis approaches are found. The main approach of this review is to illustrate the various methods and approaches available to detect thyroid disorders.
Research Interests: Computer Science and IEEE
It is known that one of the essential building blocks of turbo codes is the interleaver and its design using random, semi-random (S-Random) and deterministic permutations. In this paper, two new types of turbo code interleavers, Enhanced... more
It is known that one of the essential building blocks of turbo codes is the interleaver and its design using random, semi-random (S-Random) and deterministic permutations. In this paper, two new types of turbo code interleavers, Enhanced Block S-Random (EBSR) interleaver is proposed. The design algorithm for the new interleaver is described in depth, and the simulation results are compared to the new interleaver with different existing interleavers based on the BER (Bit Error Rate) performances of the turbo codes. Through the simulation, we find a better performance of the EBSR interleaver than random and practical interleavers. INDEX TERMS—Interleaver, semi random, turbo codes, weight distribution. I.
Research Interests:
Research Interests:
Research Interests:
This paper describes a method for automatic classification of different human emotions obtained using Electroencephalograph (EEG) signals. The human brain is a complex system. The superimposition of the diverse processes in the brain is... more
This paper describes a method for automatic classification of different human emotions obtained using Electroencephalograph (EEG) signals. The human brain is a complex system. The superimposition of the diverse processes in the brain is recognized through EEG signals. Electroencephalographic measurements are commonly used in medical applications and in the research areas to study and analyse different disorders in the brain functioning. EEG signals indicate changes in the state of brain. Data acquisition is done for different emotions with the help of ADinstruments’ power lab instrument. In this research work, we have collected real life EEG signals using Ground Truth Method. Our proposed system consists of four steps, viz., Data Acquisition, Pre-processing, Feature extraction and Classification. The subjects were stimulated for different emotions such as Sad and Happy. The signals are pre-processed and used to calculate statistical features which will be given to the classifier. Th...
Functional magnetic resonance imaging (fMRI) is widely used to identify neural correlates of cognitive tasks. Statistical features like Correlation and Entropy are extracted from fMRI images of human brain of emotional and normal state of... more
Functional magnetic resonance imaging (fMRI) is widely used to identify neural correlates of cognitive tasks. Statistical features like Correlation and Entropy are extracted from fMRI images of human brain of emotional and normal state of mind. The features are based on texture properties of fMRI images. The features so extracted are classified using GMM and kNN classifiers to help distinguish between normal and emotional state of human brain. KeywordsCorrelation, Entropy, GMM, kNN.
This paper describes the statistical analysis of EEG signals. EEG examination is carried out and compared between controlled healthy and Mild cognitive impairment (MCI) patients which may further develop dementia or Alzheimer disease. The... more
This paper describes the statistical analysis of EEG signals. EEG examination is carried out and compared between controlled healthy and Mild cognitive impairment (MCI) patients which may further develop dementia or Alzheimer disease. The statistical techniques provide the comparative analysis of EEG signal. The correct evaluation of EEG provides the extraction of valuable information which is important clinically. Also, extracting significant features from EEG is an important task for classification between various patients. The analysis of EEG data provides correct frequency rhythms. The relative Power spectral density values by Auto Regressive-Burg process cleared that; associated with the control group, the relative PSD is improved in the theta rhythmic range while expressively reduced in the alpha-2 rhythmic range.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
In this paper we propose the support vector machine classifier for the purpose of classifying Human Emotions using Electroencephalogram (EEG). EEG signal consists of different brain waves reflecting brain activity according to the... more
In this paper we propose the support vector machine classifier for the purpose of classifying Human Emotions using Electroencephalogram (EEG). EEG signal consists of different brain waves reflecting brain activity according to the electrode placement and the functioning of the brain. Audio Visual stimuli are given to the human volunteers and emotions in the form of EEG signals are invoked, using the hardware and software set-up. Emotions invoked here are the basic emotions named as Angry, Happy, Neutral and Sad. The signal is then pre-processed, its wavelet decomposition is done using Sub-band decomposition algorithm, and statistical parameters of wavelet coefficients are calculated. Support Vector Machine is used to classify the features of query samples into their class of emotion after training. Output of the Support Vector Machine is the class of Emotion of a Query Sample.
Research Interests:
In this paper we propose the support vector machine classifier for the purpose of classifying Human Emotions using Electroencephalogram (EEG). EEG signal consists of different brain waves reflecting brain activity according to the... more
In this paper we propose the support vector machine classifier for the purpose of classifying Human Emotions using Electroencephalogram (EEG). EEG signal consists of different brain waves reflecting brain activity according to the electrode placement and the functioning of the brain. Audio Visual stimuli are given to the human volunteers and emotions in the form of EEG signals are invoked, using the hardware and software set-up. Emotions invoked here are the basic emotions named as Angry, Happy, Neutral and Sad. The signal is then pre-processed, its wavelet decomposition is done using Sub-band decomposition algorithm, and statistical parameters of wavelet coefficients are calculated. Support Vector Machine is used to classify the features of query samples into their class of emotion after training. Output of the Support Vector Machine is the class of Emotion of a Query Sample.
Research Interests:
This paper designs a Multi-scale Spectral transformation technique for Voice Conversion. The proposed algorithm uses Spectral transformation technique designed using multi-resolution wavelet feature set and a Neural Network to generate a... more
This paper designs a Multi-scale Spectral transformation technique for Voice Conversion. The proposed algorithm uses Spectral transformation technique designed using multi-resolution wavelet feature set and a Neural Network to generate a mapping function between source and target speech. Dynamic Frequency Warping technique is used for aligning source and target speech and Overlap-Add method is used for minimizing the distortions that occur in the reconstruction process. With the use of Neural Network, mapping of spectral parameters between source and target speech has been achieved more efficiently. In this paper, the mapping function is generated in three different ways, using three types of Neural Networks namely, Feed Forward Neural Network, Generalized Regression Neural Network and Radial Basis Neural Network. Results of all three Neural Networks are compared using execution time requirements and Subjective analysis. The main advantage of this approach is that it is speech as well as speaker independent algorithm.
Research Interests:
This paper presents a novel emotion transformation scheme of speech signal which is text independent and speaker independent. Speech signals as many other signals are inherently multi-scale in nature, owing to contributions from events... more
This paper presents a novel emotion transformation scheme of speech signal which is text independent and speaker independent. Speech signals as many other signals are inherently multi-scale in nature, owing to contributions from events occurring with different localizations in time and frequency. Therefore, emotion dependent spectral parameters those characterized by single scale features, approximate the vocal tract, but produce artefacts during speech signal reconstruction. In this paper, multi-resolution spectral transformation technique of Discrete Wavelet Packet Decomposition has been used along with the use of Artificial Neural Network for generation of transform function. This paper specifically carries out transformation of Neutral emotion to Angry, Happy and Sad emotions. The transform function is generated in three different techniques, using three types of Artificial Neural Networks (ANNs), namely, Feed Forward Neural Network (FFNN), Generalized Regression Neural Network (GRNN) and Radial Basis Network (RBN). Results of all the three ANNs are compared using both objective as well as subjective analysis.
Research Interests:
This paper presents a novel emotion transformation scheme of speech signal which is text independent and speaker independent. Speech signals as many other signals are inherently multi-scale in nature, owing to contributions from events... more
This paper presents a novel emotion transformation scheme of speech signal which is text independent and speaker independent. Speech signals as many other signals are inherently multi-scale in nature, owing to contributions from events occurring with different localizations in time and frequency. Therefore, emotion dependent spectral parameters those characterized by single scale features, approximate the vocal tract, but produce artefacts during speech signal reconstruction. In this paper, multi-resolution spectral transformation technique of Discrete Wavelet Packet Decomposition has been used along with the use of Artificial Neural Network for generation of transform function. This paper specifically carries out transformation of Neutral emotion to Angry, Happy and Sad emotions. The transform function is generated in three different techniques, using three types of Artificial Neural Networks (ANNs), namely, Feed Forward Neural Network (FFNN), Generalized Regression Neural Network (GRNN) and Radial Basis Network (RBN). Results of all the three ANNs are compared using both objective as well as subjective analysis.
Research Interests:
This paper describes the performance of k-NN classifier to classify the different emotions. The human brain is a superimposition of the diverse processes. This complex structure of brain is recognized through EEG signals. EEG signals... more
This paper describes the performance of k-NN classifier to classify the different emotions. The human brain is a superimposition of the diverse processes. This complex structure of brain is recognized through EEG signals. EEG signals indicate the changes in the state of brain. Electroencephalograph (EEG) measurements are commonly used in different research areas under the field of medical. Data acquisition is done for different emotions with the help of ADInsruments' power lab instrument. The real life EEG signals are collected with the help of Ground Truth Method. In this paper, proposed method consists of four steps, viz., acquisition of data, Pre-processing, Feature extraction and Classification. Subjects are stimulated for Sad and Happy emotions. Statistical features are then given to a k-NN classifier. The k Nearest Neighbor classifier gives different accuracy of classification for different combinations of training and testing dataset. The system has been tested on number of subjects to observe the performance of k-NN classifier.
Research Interests:
This paper describes the performance of k-NN classifier to classify the different emotions. The human brain is a superimposition of the diverse processes. This complex structure of brain is recognized through EEG signals. EEG signals... more
This paper describes the performance of k-NN classifier to classify the different emotions. The human brain is a superimposition of the diverse processes. This complex structure of brain is recognized through EEG signals. EEG signals indicate the changes in the state of brain. Electroencephalograph (EEG) measurements are commonly used in different research areas under the field of medical. Data acquisition is done for different emotions with the help of ADInsruments' power lab instrument. The real life EEG signals are collected with the help of Ground Truth Method. In this paper, proposed method consists of four steps, viz., acquisition of data, Pre-processing, Feature extraction and Classification. Subjects are stimulated for Sad and Happy emotions. Statistical features are then given to a k-NN classifier. The k Nearest Neighbor classifier gives different accuracy of classification for different combinations of training and testing dataset. The system has been tested on number of subjects to observe the performance of k-NN classifier.
Research Interests:
In this paper, three different techniques of feature extraction for identification of emotion in speech have been compared. Traditional feature like LPCC (Linear Predictive Cepstral Coefficient) and MFCC (Mel Frequency Cepstral... more
In this paper, three different techniques of feature extraction for identification of emotion in speech have been compared. Traditional feature like LPCC (Linear Predictive Cepstral Coefficient) and MFCC (Mel Frequency Cepstral Coefficient) have been described. Linear features like LFPC which is FFT based have been explained. Finally TEO (Teager Energy Operator) based nonlinear LFPC features in both time and freqnency domain have been proposed and the performance of the proposed system is compared with the traditional features. The comparison of each approach is performed using SUSAS (Speech Under Simulated and Acid Stress) and ESMBS (Emotional Speech of Mandarin and Burmese Speakers) databases. It is observed that proposed system outperforms the traditional systems. Analysis will be carried for identification mainly of the emotion ‘Anger’ in this paper.
Research Interests:
In this paper, three different techniques of feature extraction for identification of emotion in speech have been compared. Traditional feature like LPCC (Linear Predictive Cepstral Coefficient) and MFCC (Mel Frequency Cepstral... more
In this paper, three different techniques of feature extraction for identification of emotion in speech have been compared. Traditional feature like LPCC (Linear Predictive Cepstral Coefficient) and MFCC (Mel Frequency Cepstral Coefficient) have been described. Linear features like LFPC which is FFT based have been explained. Finally TEO (Teager Energy Operator) based nonlinear LFPC features in both time and freqnency domain have been proposed and the performance of the proposed system is compared with the traditional features. The comparison of each approach is performed using SUSAS (Speech Under Simulated and Acid Stress) and ESMBS (Emotional Speech of Mandarin and Burmese Speakers) databases. It is observed that proposed system outperforms the traditional systems. Analysis will be carried for identification mainly of the emotion ‘Anger’ in this paper.
Research Interests:
Worldwide research is going on to judge the emotional state of a speaker just from the quality of human voice. This paper explores use of supervised neural network to design a classifier that can discriminate between several emotions like... more
Worldwide research is going on to judge the emotional state of a speaker just from the quality of human voice. This paper explores use of supervised neural network to design a classifier that can discriminate between several emotions like happiness, anger, fear, sadness & unemotional state in speech. The results found to be are significant, both in cognitive science and in speech technology. In the current paper, statistics of the pitch like, first and second formants, and Energy and speaking rate are used as relevant features. Different neural network based recognizers are created. Ensembles of such recognizers are used as an important part of decision support system for prioritizing voice messages and assigning a proper agent to response the message. The developed intelligent system can be enhanced to automatically predict and adapt to detect people’s emotional states and also to design emotional robot or computer system.
Research Interests:
Worldwide research is going on to judge the emotional state of a speaker just from the quality of human voice. This paper explores use of supervised neural network to design a classifier that can discriminate between several emotions like... more
Worldwide research is going on to judge the emotional state of a speaker just from the quality of human voice. This paper explores use of supervised neural network to design a classifier that can discriminate between several emotions like happiness, anger, fear, sadness & unemotional state in speech. The results found to be are significant, both in cognitive science and in speech technology. In the current paper, statistics of the pitch like, first and second formants, and Energy and speaking rate are used as relevant features. Different neural network based recognizers are created. Ensembles of such recognizers are used as an important part of decision support system for prioritizing voice messages and assigning a proper agent to response the message. The developed intelligent system can be enhanced to automatically predict and adapt to detect people’s emotional states and also to design emotional robot or computer system.
Research Interests:
This paper presents the analysis of affective speech of human for fatigue detection. Experimental results revealed that, there is a dependence of fatigue on human voice [3]. If a person is carrying fatigue or feeling depressed, then it is... more
This paper presents the analysis of affective speech of human for fatigue detection. Experimental results revealed that, there is a dependence of fatigue on human voice [3]. If a person is carrying fatigue or feeling depressed, then it is clearly reflected from his speech. Current research work on fatigue detection is based on normal human speech. This paper analyses the advantages of affective speech for precise determination of the state of fatigue. The analysis is carried out on the basis of various features such as fundamental frequency, formant frequencies, cepstrum, short-time energy etc.
Research Interests: Computer Science and Acm
This paper presents the analysis of affective speech of human for fatigue detection. Experimental results revealed that, there is a dependence of fatigue on human voice [3]. If a person is carrying fatigue or feeling depressed, then it is... more
This paper presents the analysis of affective speech of human for fatigue detection. Experimental results revealed that, there is a dependence of fatigue on human voice [3]. If a person is carrying fatigue or feeling depressed, then it is clearly reflected from his speech. Current research work on fatigue detection is based on normal human speech. This paper analyses the advantages of affective speech for precise determination of the state of fatigue. The analysis is carried out on the basis of various features such as fundamental frequency, formant frequencies, cepstrum, short-time energy etc.
Research Interests: Computer Science and Acm
Fast Fourier Transform is an essential data processing technique in communication systems and DSP systems. In this brief, we propose high speed and area efficient 64 point FFT processor using Vedic algorithm. To reduce computational... more
Fast Fourier Transform is an essential data processing technique in communication systems and DSP systems. In this brief, we propose high speed and area efficient 64 point FFT processor using Vedic algorithm. To reduce computational complexity and area, we develop FFT architecture by devising a radix-4 algorithm and optimizing the realization by Vedic algorithm. Furthermore, it can be used in decimation in frequency (DIF) and decimation in time (DIT) decompositions. Moreover, the design can achieve very high speed, which makes them suitable for the most demanding applications of FFT. Indeed, the proposed radix-4 Vedic algorithm based architecture requires fewer hardware resources. The synthesis results are same as that of theoretical analysis and it is observed that more than 15% reduction can be achieved in terms of slices count. In addition, the dynamic power consumption can be reduced and speed can be increased by as much as 16% using Vedic algorithm.
Research Interests:
Research Interests: Clinical Psychology, Depression, History Of Emotions, Image Features Extraction, EEG, and 15 moreBrain behavior from statistical analysis of eeg signals, Music and Emotions, Academic Emotions and Motivation, Anxiety, Emotions, EEG Signal Processing, EEG Biofeedback, Feature Extraction, Anxiety and Depression, ANXIETY, History of Emotions, EEG Signal Analysis, Normal, Emotions Social Psychology, and Excited Brain
Fast Fourier Transform is an essential data processing technique in communication systems and DSP systems. In this brief, we propose high speed and area efficient 64 point FFT processor using Vedic algorithm. To reduce computational... more
Fast Fourier Transform is an essential data processing technique in communication systems and DSP systems. In this brief, we propose high speed and area efficient 64 point FFT processor using Vedic algorithm. To reduce computational complexity and area, we develop FFT architecture by devising a radix-4 algorithm and optimizing the realization by Vedic algorithm. Furthermore, it can be used in decimation in frequency (DIF) and decimation in time (DIT) decompositions. Moreover, the design can achieve very high speed, which makes them suitable for the most demanding applications of FFT. Indeed, the proposed radix-4 Vedic algorithm based architecture requires fewer hardware resources. The synthesis results are same as that of theoretical analysis and it is observed that more than 15% reduction can be achieved in terms of slices count. In addition, the dynamic power consumption can be reduced and speed can be increased by as much as 16% using Vedic algorithm.
Research Interests:
Research Interests:
Research Interests: Clinical Psychology, Depression, History Of Emotions, Image Features Extraction, EEG, and 15 moreBrain behavior from statistical analysis of eeg signals, Music and Emotions, Academic Emotions and Motivation, Anxiety, Emotions, EEG Signal Processing, EEG Biofeedback, Feature Extraction, Anxiety and Depression, ANXIETY, History of Emotions, EEG Signal Analysis, Normal, Emotions Social Psychology, and Excited Brain
In this research, the emotions and the patterns of EEG signals of human brain will be studied. The aim of this research is to study the analysis of the changes in the brain signals in the domain of different emotions. The observations can... more
In this research, the emotions and the patterns of EEG signals of human brain will be studied. The aim of this research is to study the analysis of the changes in the brain signals in the domain of different emotions. The observations can be analysed for its utility in the diagnosis of psychosomatic disorders like anxiety and depression in economical way with higher precision.
Research Interests:
Functional magnetic resonance imaging (fMRI) is widely used to identify neural correlates of cognitive tasks. Basic statistical parameters are considered to analyze fMRI images of human brain of emotional and normal state of mind. These... more
Functional magnetic resonance imaging (fMRI) is widely used to identify neural correlates of cognitive tasks. Basic statistical parameters are considered to analyze fMRI images of human brain of emotional and normal state of mind. These basic statistical parameters are mean, standard deviation, variance, kurtosis, skewness etc. We are calculating these parameters, and along with that we are making various combinations of these parameters like mean and skewness, mean and kurtosis etc. SPM8 is also used as a software tool to analyze the fMRI images. Analysis is done on the reference database. We are using Wavelet and PCA based proposed method for classification of two states by deciding threshold value.
Research Interests:
Voice transformation is an application in speech processing which involves modification of speech signal of source speaker to a target speaker. This paper explains transformation of Neutral speech of target speaker into Angry, Happy and... more
Voice transformation is an application in speech processing which involves modification of speech signal of source speaker to a target speaker. This paper explains transformation of Neutral speech of target speaker into Angry, Happy and Sad emotions. In this algorithm modeling of speech is performed using Discrete Wavelet Transform (DWT) for the emotions of Source and Target speakers. Linear least square method is used to find the best-fit nth order polynomial for the given set of data points to build a transformation function. Finally the transformed signal is generated using Least Mean Square (LMS), Normalized Least Mean Square (NLMS) and Recursive Least Square (RLS) adaptive filtering techniques and the results of the 3 adaptive filters is compared. The results are evaluated by Subjective evaluation by carrying out a formal listening test. Objective evaluation is performed by calculating the parameters of transformed speech using Mean Square Error (MSE), Euclidean Distance (ED) and Correlation Coefficient (CC).
Research Interests:
... subjects by recording five short reactive sentences/words in Indian Regional Language Marathi as: 1. "Magh"; 2."Sangitala"; 3."ho kaa ga."; 4.Sharvari; 5.Asa kasa ... that the... more
... subjects by recording five short reactive sentences/words in Indian Regional Language Marathi as: 1. "Magh"; 2."Sangitala"; 3."ho kaa ga."; 4.Sharvari; 5.Asa kasa ... that the two speakers are not the same 3 It is possible that two speakers are not the same 4 I am unable to decide ...
Research Interests:
This paper develops an algorithm “Discrete Wavelet Transform with Adaptive Filter” (DWTAF) to transform Neutral speech into emotional speech like Angry, Happy or Sad and this is compared with two other emotion transformation algorithms.... more
This paper develops an algorithm “Discrete Wavelet Transform with Adaptive Filter” (DWTAF) to transform Neutral speech into emotional speech like Angry, Happy or Sad and this is compared with two other emotion transformation algorithms. The other two algorithms are “Speech Transformation using Statistical Parameters and Pitch Contours” (STSPPC) and “Speech Transformation using Mel Frequency Cepstral Coefficients (MFCC) and Gaussian Mixture Model (GMM)” (STMG). STSPPC calculates statistical parameters of speech like mean and variance and they are used for transformation of emotional speech. STMG performs DTW (Dynamic Time Warping), extracts MFCC and applies a GMM for speech transformation. The proposed algorithm DWTAF, presents a novel approach to model the speech using Discrete Wavelet Transform (DWT) and emotional speech is generated by filtering the speech using an adaptive filter with Least Mean Square (LMS) algorithm. For this purpose a database of 400 English language sentences has been created with 10 sentences uttered by 10 female speakers with 4 emotions each and a comparison between the three transformation algorithms is carried out using Subjective and Objective evaluation tests. The parameters used for Objective evaluation of transformed speech are Segmental SNR (Seg SNR), Log-Likelihood Ratio (LLR) and Weighted Spectral Slope (WSS).
Research Interests:
This paper develops an algorithm “Discrete Wavelet Transform with Adaptive Filter” (DWTAF) to transform Neutral speech into emotional speech like Angry, Happy or Sad and this is compared with two other emotion transformation algorithms.... more
This paper develops an algorithm “Discrete Wavelet Transform with Adaptive Filter” (DWTAF) to transform Neutral speech into emotional speech like Angry, Happy or Sad and this is compared with two other emotion transformation algorithms. The other two algorithms are “Speech Transformation using Statistical Parameters and Pitch Contours” (STSPPC) and “Speech Transformation using Mel Frequency Cepstral Coefficients (MFCC) and Gaussian Mixture Model (GMM)” (STMG). STSPPC calculates statistical parameters of speech like mean and variance and they are used for transformation of emotional speech. STMG performs DTW (Dynamic Time Warping), extracts MFCC and applies a GMM for speech transformation. The proposed algorithm DWTAF, presents a novel approach to model the speech using Discrete Wavelet Transform (DWT) and emotional speech is generated by filtering the speech using an adaptive filter with Least Mean Square (LMS) algorithm. For this purpose a database of 400 English language sentences has been created with 10 sentences uttered by 10 female speakers with 4 emotions each and a comparison between the three transformation algorithms is carried out using Subjective and Objective evaluation tests. The parameters used for Objective evaluation of transformed speech are Segmental SNR (Seg SNR), Log-Likelihood Ratio (LLR) and Weighted Spectral Slope (WSS).