Spectral Flatness Measure Research Articles

Dynamic Detection of Vigorous Worm in Infected Host

The Internet has developed to give many benefits to mankind. The access to information being one of the most important. Worms cause major security threats to the Internet. Worms are software components that are capable of infecting a computer and then using that computer to infect another computer. The cycle is repeated, and the population of worm-infected computers grows rapidly. Smart worms cause most important security threats to the Internet. The ability of smart worms spread in an automated fashion and can flood the internet in a very short time. A new class of smart worms, referred to as Camouflaging worm (C-Worm in short). C-worm is different from traditional worm. C-worm intelligently manipulate its scan traffic volume over time. Motivated by observations, we designed a novel Spectrum Based detection Scheme to detect the C-worm. This scheme uses Power Spectral Density (PSD) distribution of the scan traffic volume and its corresponding Spectral Flatness Measure (SFM) to distinguish the C-worm traffic from background traffic. This scheme is effectively detecting not only the C-worm but traditional worms as well. The goal is to prevent, detect and delete the smart worms as well as traditional worms.

Prefilter Design for Improved Spectral Flatness of Physical Pseudorandom Perturbations

System identification techniques have been applied to model human sensorimotor systems for research and diagnostic purposes. Accurate human modeling via system identification requires informative experimental data, which further require informative perturbations applied to the human. Previously, designing ideal perturbations for system identification has been extensively studied. On the other hand, the physical realization of perturbations often produces distorted perturbations, resulting in less accurate system identification and incorrect conclusions about the target dynamic system. However, such issues associated with an uncompensated physical realization process have not been fully resolved in the literature. A notable application is the design and use of the pseudorandom sequence (PRS). This article considers the physical realization of PRS perturbations. The impact of zero-order-hold (ZOH) on perturbation quality in terms of the previously proposed band-limited spectral flatness measure (SFM) is identified. To compensate for the spectral flatness degradation due to ZOH, a reference prefilter designed by the Parks–McClellan algorithm is incorporated into the motion control design. A relationship between the prefilter performance and band-limited SFM is established, based on which a prefilter design procedure that guarantees an SFM lower bound of the physically realized perturbations is proposed. Experimental results demonstrated that the proposed prefilter design method effectively improves the SFM of the physical perturbations.

Broad phonetic feature classifier for real-time hearing aid processing

Current Hearing Aid (HA) devices compensate for hearing loss by amplifying sounds based on instantaneous energy estimates in specific frequency bands. The dynamics of speech are incorporated using attack and release times—a crude approximation of the rich temporal dynamics in natural speech. Recent advances in silicon enables one to consider more complex sound processing approaches with minimal impact on the HA battery life. We envisage a broad phonetic feature classifier, followed by amplification strategies optimized for specific temporal dynamics across each segment of natural speech. In this work we present a broad phonetic classifier for ‘vocalic’, ‘non-vocalic’ and ‘mixed’ speech segments. The classifier is based on (i) Interpretable models- unsupervised learning methods such as Gaussian Mixture Models followed by supervised learning based Radial Basis Function networks; and (ii) Deep Neural Networks with supervised learning. The frontend feature extractor consists of MFCC, delta MFCC features, spectral flatness measure and segmental energy. The training and test sets are generated using speech files and the corresponding phone labels from the TIMIT dataset. We present our findings with respect to the classifier's performance and deployability for real time HA processing.

A Spectral-Based Approach for BCG Signal Content Classification.

This paper has two objectives: the first is to generate two binary flags to indicate useful frames permitting the measurement of cardiac and respiratory rates from Ballistocardiogram (BCG) signals—in fact, human body activities during measurements can disturb the BCG signal content, leading to difficulties in vital sign measurement; the second objective is to achieve refined BCG signal segmentation according to these activities. The proposed framework makes use of two approaches: an unsupervised classification based on the Gaussian Mixture Model (GMM) and a supervised classification based on K-Nearest Neighbors (KNN). Both of these approaches consider two spectral features, namely the Spectral Flatness Measure (SFM) and Spectral Centroid (SC), determined during the feature extraction step. Unsupervised classification is used to explore the content of the BCG signals, justifying the existence of different classes and permitting the definition of useful hyper-parameters for effective segmentation. In contrast, the considered supervised classification approach aims to determine if the BCG signal content allows the measurement of the heart rate (HR) and the respiratory rate (RR) or not. Furthermore, two levels of supervised classification are used to classify human-body activities into many realistic classes from the BCG signal (e.g., coughing, holding breath, air expiration, movement, et al.). The first one considers frame-by-frame classification, while the second one, aiming to boost the segmentation performance, transforms the frame-by-frame SFM and SC features into temporal series which track the temporal variation of the measures of the BCG signal. The proposed approach constitutes a novelty in this field and represents a powerful method to segment BCG signals according to human body activities, resulting in an accuracy of 94.6%.

Autoregressive Whitening Filtering of Phonocardiography Signals for Detection of Coronary Artery Disease

Background: Narrowing of the coronary arteries, which defines coronary artery disease (CAD), can be discovered through phonocardiography (PCG) analysis. Increased power of frequencies below 200 Hz in the diastole has been associated with CAD, and can be used to distinguish CAD from NonCAD patients. However, spectral roll off is steep (~40 dB/dec), and spectral leakage might mask the weak CAD-related signal.Methods: PCGs from 1168 subjects, 213 CAD and 955 NonCAD, were pooled from three studies. The average power spectral density (PSD) of diastole segments for NonCAD subjects was found, and an auto-regressive (AR) model of this PSD was constructed. The inverse of the corresponding filter was used for whitening.Results: A single iteration of whitening filtering was insufficient to make the PSD white for 5-1000 Hz. Two iterations of whitening filtering with an order of 6-10 were required to reach a plateau of maximal whitening with a spectral flatness measure close to 1 in the frequency band 5-1000 Hz. The whitening process revealed additional PSD differences between CAD and NonCAD subjects for the mid-diastole segment.Conclusion: Whitening of diastole PCG segments emphasized the difference between CAD and NonCAD patients.

Detection of glottal closure instant and glottal open region from speech signals using spectral flatness measure

This paper proposes an approach using spectral flatness measure to detect the glottal closure instant (GCI) and the glottal open region (GOR) within each glottal cycle in voiced speech. The spectral flatness measure is derived from the instantaneous spectra obtained in the analysis of speech using single frequency filtering (SFF) and zero time windowing (ZTW) methods. The Hilbert envelope of the numerator of group delay (HNGD) spectrum at each instant of time is obtained using the ZTW method. The HNGD spectrum highlights the important (like resonances) spectral characteristics of the vocal tract system at each instant of time. The dynamic characteristics of the vocal tract system can be tracked by the spectral flatness feature of the HNGD spectrum, thus bringing out the characteristics of the vocal tract system when the subglottal region is coupled with the supraglottal region during the open phase of the glottal cycle. The SFF spectra at each instant change significantly at the location of the GCI. The GCIs can be detected using the changes in the spectral flatness information derived from the SFF spectra. The proposed methods of detection of GCI and GOR is compared with several existing methods.

Predictive Control of IM Drive Acoustic Noise

Acoustic behavior of electric drives is an important factor for their application in areas where humans expect certain comfort, such as in electric vehicles. The main problem addressed in this article is high tonality of noise emitted by conventional pulsewidth modulation (PWM) controlled drives. We propose to use an explicit model of the transfer function from the electric variables to the sound pressure with experimentally identified parameters. This model is then used to include acoustic quality into the cost function of the finite control set model predictive control (FCS-MPC). Specifically, we propose to penalize the sound pressure resonant frequencies in the spectrum of the direct flux component of the stator current vector. We study the impact of this penalization on acoustic properties of the drive measured by the sound pressure level and spectral flatness measure on three experimental induction machine (IM) drives of different power and acoustic environment. In comparison with the conventional PWM, random PWM, and FCS-MPC techniques, the proposed approach improves the tonality of the noise in terms of spectral flatness and in high-speed regimes it also improves total sound pressure level by several decibel.

Analysis of the Acoustic Noise Spectrum of Domestic Induction Heating Systems Controlled by Phase-Accumulator Modulators

In domestic induction heating applications, the modulation technique applied to the inverter has a high influence on the acoustic noise emissions. These noise emissions must be avoided, since they may be audible and annoying to the final user. This paper analyzes the acoustic noise emissions that appear when a series half-bridge resonant inverter is operated with a phase-accumulator-based modulator. This modulation technique has the advantage of operating in the frequency domain, and it is compared with the classical pulsewidth modulator regarding the audible noise generated. The frequencies of the tones in the acoustic noise spectrum are theoretically calculated from the parameters of the phase-accumulator-based modulator. The spectral flatness measure is used to quantify the number of cases in which tones are generated by the modulation. Two techniques are applied to the phase-accumulator-based modulator, and their effect is tested. Theoretical results are experimentally verified.

Discrimination Between Broadband Underwater Target Echo and Reverberation Based on Signal Spectral Flatness Feature

Reverberation is the major interference of active detection, discrimination between target echo and reverberation is difficult when the reverberation is very strong. For the distinguishing of underwater target echo with strong interference, the modulation feature of echo and reverberation spectrum are characterized by signal spectral flatness measure, and the relationship between signal spectral flatness and target physical properties are theoretically formed. A method of discrimination between broadband underwater target echo and reverberation that based on signal spectral flatness characteristics difference is proposed and studied. Sea experiment results of complex underwater target broadband acoustic scattering have shown that signal spectral flatness characteristics of underwater target broadband echo and reverberation have obvious differences. Target echo and reverberation can be well distinguished using the proposed method.

Incorporation of Manner of Articulation Constraint in LSTM for Speech Recognition

The variants of recurrent neural networks such as long short-term memory (LSTM) and gated recurrent unit are successful in sequence modelling such as automatic speech recognition. However, the decoded sequence is prune to have false substitutions, insertions and deletions. In our work, we investigate the outcome of the hidden layers in LSTM trained on TIMIT dataset. We found interestingly that the first hidden layer was capturing information related to some broad manners of articulation. The successive hidden layers try to cluster among the broad manners of articulation. We detected two broad manners of articulation, namely sonorants (vowels, semi-vowels, nasals) and obstruents (fricatives, stops, affricates) by exploiting the spectral flatness measure (SFM) on the linear prediction coefficients. We define a additional gate called manner of articulation gate that is high if the broad manners of articulation of tth frame are same as that of $$(t+1)$$ th frame. The manner of articulation detection is embedded at the output of the activation gate of LSTM at the first hidden layer. By doing so, the sonorants being substituted as obstruents are minimized at the output layer. The proposed method decreased the phone error rates by 0.7% when evaluated on the core test set of the TIMIT.

Robust Voice Activity Detection Algorithm based on Long Term Dominant Frequency and Spectral Flatness Measure

Robust Voice Activity Detection Algorithm based on Long Term Dominant Frequency and Spectral Flatness Measure

Robust spectrum sensing based on spectral flatness measure

The authors investigate the spectral flatness measure (SFM)-based spectrum sensing technique for cognitive radios. This scheme exploits the fact that under Gaussian noise, the noise-only observations have flattened spectrum, i.e. more white, when compared with that of the observations containing the incumbent or primary signal; hence, an increased SFM under the null hypothesis. Under the null hypothesis, the authors derive the asymptotic distribution of the test statistic, and the asymptotically optimal detection threshold, with a constraint on the probability of false-alarm. Furthermore, the authors show that this test is robust to the noise variance uncertainty (NVU) and is related to a test based on the entropy in the observed sequence. Through extensive Monte-Carlo simulations, the authors show that the test based on SFM performs better than the existing energy detector and the blind detector, under realistic signal and fading models, all in the presence of NVU. The authors also highlight the practical utility of this technique based on experimental results.

Speech Enhancement Using Spectral Flatness Measure Based Spectral Subtraction

Speech Enhancement Using Spectral Flatness Measure Based Spectral Subtraction

Design and Development of Humiliating of C-Worm Utilizing an Arbitrary Visualizing

A worm is a malicious self-replicating programs, it is intended to feast via computer links. The computer worms are one method of malware beside with Trojans and Viruses. The active worms impersonate main refuge threats to that an Internet. This is the ability of active worms to continuously propagate in the computers on the Internet as an automated fashion. The Active worms changes through their circulation, and thus, pose large tasks to preserve alongside them. This manuscript, to explore an original class of active worms, denoted to as Camouflaging Worm (Abbreviated to as C-Worm) respectively. The C-Worm means different from traditional worms because of its ability to intelligently manipulate its scan traffic size over time respectively. So, the C-Worm camouflages its circulation from prevailing worm detection systems constructed on investigating the circulation traffic produced by worms respectively. To analyze characteristics of the C- Worm and conduct a comprehensive comparison between its non-worm traffic and traffic respectively. To detect that these double kinds of traffic are barely different in the time sphere. Though, their difference is clear in the incidence sphere, due to the periodic devious environment of the C-Worm respectively. Interested by these clarifications, it designs a novel spectrum-constructed pattern to detect the C-Worm. This pattern consumes the Power Spectral Density (PSD) circulation of the scan traffic size and its corresponding Spectral Flatness Measure (SFM) respectively, to distinguish the C-Worm traffic from related traffic. Consuming a complete group of detection metrics and real-world suggestions as contextual traffic, it directs widespread implementation estimates on these planned spectrumconstructed detection pattern. This implementation data evidently establishes that this pattern can efficiently detect the C-Worm circulation. Besides, to display the simplification of this spectrum-based pattern in efficiently detecting not only the C-Worm, although conventional worms also.

Real Time Suppression of Howling Noise in Public Address System

Howling noise is a common phenomenon in a public address system. It is built up due to the acoustic coupling between the speaker system and the microphone when it creates a positive feedback. Real time implementation of howling noise detection and suppression was implemented using TMS320C6713 DSK starter kit. The whole implementation was done based on direct memory access (DMA) feature of the DSP processor. The method uses the properties of howling noise for efficient detection and has the advantage of suppressing the noise. Howling detection is performed based on spectral flatness measure (SFM) of each input speech frame. For frames without howling, the input is passed as such to the output. Howling suppression is performed by making output samples as zero if the presence of howling noise is detected.

Erratum to: Efficient voice activity detection algorithm using long-term spectral flatness measure

Erratum to: Efficient voice activity detection algorithm using long-term spectral flatness measure

Automatic Music Mood Classification Based on Timbre and Modulation Features

In recent years, many short-term timbre and long-term modulation features have been developed for content-based music classification. However, two operations in modulation analysis are likely to smooth out useful modulation information, which may degrade classification performance. To deal with this problem, this paper proposes the use of a two-dimensional representation of acoustic frequency and modulation frequency to extract joint acoustic frequency and modulation frequency features. Long-term joint frequency features, such as acoustic-modulation spectral contrast/valley (AMSC/AMSV), acoustic-modulation spectral flatness measure (AMSFM), and acoustic-modulation spectral crest measure (AMSCM), are then computed from the spectra of each joint frequency subband. By combining the proposed features, together with the modulation spectral analysis of MFCC and statistical descriptors of short-term timbre features, this new feature set outperforms previous approaches with statistical significance.

Numerical simulation for optimizing mode shaping and supercontinuum flatness of liquid filled seven-core photonic crystal fibers

A seven-core photonic crystal fiber filled with commercial index-matching liquids is designed to optimize mode shaping and supercontinuum flatness. Numerical simulation of supercontinuum generation in these liquid-filled seven-core PCFs is conducted at 25°C. The definition of spectral flatness measure is used to quantitatively describe SC flatness. Numerical simulations are performed to study the propagation of femtosecond pulse in the liquid-filled seven-core PCFs. Results show that mode shaping and supercontinuum flatness can be easily optimized and modified using the index-matching liquids in seven-core PCF without varying the structure of the air rings around the guiding cores. Simulations also show that 50fs pulses with a center wavelength of 1064nm generate relatively flat SC spectra in the 25cm-long liquid-filled PCF. A flat spectral bandwidth of 400nm (900–1300nm) is achieved with an applied pump power of 30kW. The simulation results demonstrate that using index-matching liquids to fill the inner ring of the seven-core PCF optimizes mode shaping and generates flat SC spectrum in specified wavelength region. Results further demonstrate that the SC flatness increased with increasing PCF dispersion corresponding to pump wavelength, on the premise that generated enough spectrum width, when the pump worked in the normal dispersion region. Temperature barely affects the spectrum flatness, but can affect spectrum broadening.

Broad Phoneme Classification Using Signal Based Features

Speech is the most efficient and popular means of human communication Speech is produced as a sequence of phonemes. Phoneme recognition is the first step performed by automatic speech recognition system. The state-of-the-art recognizers use mel-frequency cepstral coefficients (MFCC) features derived through short time analysis, for which the recognition accuracy is limited. Instead of this, here broad phoneme classification is achieved using features derived directly from the speech at the signal level itself. Broad phoneme classes include vowels, nasals, fricatives, stops, approximants and silence. The features identified useful for broad phoneme classification are voiced/unvoiced decision, zero crossing rate (ZCR), short time energy, most dominant frequency, energy in most dominant frequency, spectral flatness measure and first three formants. Features derived from short time frames of training speech are used to train a multilayer feedforward neural network based classifier with manually marked class label as output and classification accuracy is then tested. Later this broad phoneme classifier is used for broad syllable structure prediction which is useful for applications such as automatic speech recognition and automatic language identification.

Numerical simulation of supercontinuum generation in liquid-filled photonic crystal fibers with a normal flat dispersion profile

A photonic crystal fiber (PCF) filled with commercial index-matching liquids is designed to control the dispersion properties of PCF. Numerical simulation of supercontinuum (SC) generation in these liquid-filled PCFs is then conducted at a temperature of 25°C. The definition of spectral flatness measure (SFM) is introduced to quantitatively describe the SC flatness. Numerical simulations are performed to study the propagation of femtosecond pulse in the liquid-filled PCFs. Results show that using the index-matching liquids in PCF, the dispersion properties of the PCF can be easily engineered without changing in the geometry. Simulations also show that 50fs pulses with a center wavelength of 1060nm generate relatively flat SC spectra in the 25cm-long PCF with two Oil2-filled rings. With an applied pump power of 24kW, a flat (SFM=0.9670) spectral bandwidth of 700nm (900–1400nm) is achieved. Results further demonstrate that using index-matching liquids to fill the PCF inner ring can exactly control its dispersion properties and generate a flat SC spectrum in the specified wavelength region.