Goutam Saha

Bio: Goutam Saha is an academic researcher from North Eastern Hill University. The author has contributed to research in topics: Speaker recognition & Gene regulatory network. The author has an hindex of 13, co-authored 143 publications receiving 583 citations. Previous affiliations of Goutam Saha include Indian Institute of Technology Kharagpur & West Bengal University of Technology.

Novel speech features for improved detection of spoofing attacks

Abstract: Now-a-days, speech-based biometric systems such as automatic speaker verification (ASV) are highly prone to spoofing attacks by an imposture. With recent development in various voice conversion (VC) and speech synthesis (SS) algorithms, these spoofing attacks can pose a serious potential threat to the current state-of-the-art ASV systems. To impede such attacks and enhance the security of the ASV systems, the development of efficient anti-spoofing algorithms is essential that can differentiate synthetic or converted speech from natural or human speech. In this paper, we propose a set of novel speech features for detecting spoofing attacks. The proposed features are computed using alternative frequency-warping technique and formant-specific block transformation of filter bank log energies. We have evaluated existing and proposed features against several kinds of synthetic speech data from ASVspoof 2015 corpora. The results show that the proposed techniques outperform existing approaches for various spoofing attack detection task. The techniques investigated in this paper can also accurately classify natural and synthetic speech as equal error rates (EERs) of 0% have been achieved.

Wavelet Transform Based Mel-scaled Features for Acoustic Scene Classification.

Abstract: Acoustic scene classification (ASC) is an audio signal processing task where mel-scaled spectral features are widely used by researchers. These features, considered de facto baseline in speech processing, traditionally employ Fourier based transforms. Unlike speech, environmental audio spans a larger range of audible frequency and might contain short high-frequency transients and continuous low-frequency background noise, simultaneously. Wavelets, with a better time-frequency localization capacity, can be considered more suitable for dealing with such signals. This paper attempts ASC by a novel use of wavelet transform based mel-scaled features. The proposed features are shown to possess better discriminative properties than other spectral features while using a similar classification framework. The experiments are performed on two datasets, similar in scene classes but differing by dataset size and length of the audio samples. When compared with two benchmark systems, one based on mel-frequency cepstral coefficients and Gaussian mixture models, and the other based on log mel-band energies and multi-layer perceptron, the proposed system performed considerably better on the test data.

FT-SDN: A Fault-Tolerant Distributed Architecture for Software Defined Network

Abstract: The traditional Software Defined Network (SDN) architecture is based on single controller in the Control Plane. Therefore, network functioning become highly dependent on the performance of the single controller in the Control Plane, which is undesirable for any reliable application. Despite many advantages of SDN, its deployment in the practical field is restricted since reliability and fault-tolerance capabilities of the system are not satisfactory. To overcome these difficulties of SDN, (FT-SDN) architecture has been proposed. The proposed architecture consists of a simple and effective distributed Control Plane with multiple controllers. FT-SDN uses a synchronized mechanism to periodically update the controller’s state within themselves. In case of failure, FT-SDN has the ability to select another working controller based on the distance and delays among different network entities. The performance of the FT-SDN architecture was examined with respect to different specifications in the presence of faults. Experimentation was done in simulation where results were found to be satisfactory.

Deep Learning Methods for Classification of Certain Abnormalities in Echocardiography

Abstract: This article experiments with deep learning methodologies in echocardiogram (echo), a promising and vigorously researched technique in the preponderance field. This paper involves two different kinds of classification in the echo. Firstly, classification into normal (absence of abnormalities) or abnormal (presence of abnormalities) has been done, using 2D echo images, 3D Doppler images, and videographic images. Secondly, based on different types of regurgitation, namely, Mitral Regurgitation (MR), Aortic Regurgitation (AR), Tricuspid Regurgitation (TR), and a combination of the three types of regurgitation are classified using videographic echo images. Two deep-learning methodologies are used for these purposes, a Recurrent Neural Network (RNN) based methodology (Long Short Term Memory (LSTM)) and an Autoencoder based methodology (Variational AutoEncoder (VAE)). The use of videographic images distinguished this work from the existing work using SVM (Support Vector Machine) and also application of deep-learning methodologies is the first of many in this particular field. It was found that deep-learning methodologies perform better than SVM methodology in normal or abnormal classification. Overall, VAE performs better in 2D and 3D Doppler images (static images) while LSTM performs better in the case of videographic images.

Recent Advances in the Synthesis of Cyclobutanes by Olefin [2+2] Photocycloaddition Reactions

Abstract: The [2 + 2] photocycloaddition is undisputedly the most important and most frequently used photochemical reaction. In this review, it is attempted to cover all recent aspects of [2 + 2] photocycloaddition chemistry with an emphasis on synthetically relevant, regio-, and stereoselective reactions. The review aims to comprehensively discuss relevant work, which was done in the field in the last 20 years (i.e., from 1995 to 2015). Organization of the data follows a subdivision according to mechanism and substrate classes. Cu(I) and PET (photoinduced electron transfer) catalysis are treated separately in sections 2 and 4, whereas the vast majority of photocycloaddition reactions which occur by direct excitation or sensitization are divided within section 3 into individual subsections according to the photochemically excited olefin.

The working brain (an introduction to neuropsychology)

Abstract: CLINICAL NEUROSURGERY Edited by Barnes Wordall. Vol. 18. (Pp. 557; illustrated; £8 25.) Churchill Livingstone: Edinburgh. 1971. CLINICAL NEUROSURGERY Edited by G. T. Tindall. Vol. 19. (Pp. 598; illustrated; £12.) Churchill Livingstone: Edinburgh. 1972. The Congress of Neurosurgeons began in 1951 on the initiative of a group of younger neurosurgeons. In the last 20 years its membership has grown from 69 to over 1,000, but it has retained its original intentions and virility by a constitution which ensured that the office bearers and organizers were always young men. Residents in training are encouraged to join, financial concessions make it possible for them to attend meetings, and these are organized as an educational exercise, by inviting established authorities to give lectures on selected topics, chosen to provide a balanced programme. As a result Clinical Neurosurgery is a valuable volume which all neurosurgeons look forward to each year; it is in quite a different class from the usual conference tome, full of brief and unconnected papers of widely varying quality. As the title suggests Clinical Neurosurgery includes excellent clinical reviews but the two recent volumes include also a section of seminars on fundamental research-in volume 18 on coma and sleep, in volume 19 on basic mechanisms of memory. It is also the custom to invite a senior neurosurgical citizen as guest of honour and his two or three papers afford an opportunity for historical and philosophical reflections as well as an experienced perspective on clinical and experimental work. Add to this the refreshing presidential address, from one of the (angry?) young men of neurosurgery and it will be clear that these volumes really do include something of interest for every neurosurgeon, whatever his own interests or prejudices. It is a relief to be able so warmly to recommend these books, when the question posed by so many other books is whether anyone would really want to read them. The most recent volume has a more consistent theme than former ones, and that is 'head injury'. It includes papers on mechanisms as revealed by animal experiment and by a pathologist who visited the scene of the accident before examining the brains of head injury fatalities. There are chapters on engineering and socio-psychological aspects of accident prevention, as well as down to earth clinical accounts of metabolic disorders, testing for acoustic vestibular 36 damage, and aspects of prognosis. This is a significant contribution to the literature on head injury.