Combining the evidences of temporal and spectral enhancement techniques for improving the performance of Indian language identification system in the presence of background noise

Language Identification has gained significant importance in recent years, both in research and commercial market place, demanding an improvement in the ability of machines to distinguish between languages. Although methods like Gaussian mixture models, hidden Markov models and neural networks are used for identifying languages the problem of language identification in noisy environments could not be addressed so far. This paper addresses the performance of automatic language identification system in noisy environments. A comparative performance analysis of speech enhancement techniques like minimum mean squared estimation, spectral subtraction and temporal processing, with different types of noise at different SNRs, is presented here. Though these individual enhancement techniques may not yield good performance with different types of noise at different SNRs, it is proposed to combine the evidences of all these techniques to improve the overall performance of the system significantly. The language identification studies are performed using IITKGP-MLILSC (IIT Kharagpur-Multilingual Indian Language Speech Corpus) databases which consists of 27 languages.     

Exploring the costs of technical debt management – a case study

Technical debt is a metaphor for delayed software maintenance tasks. Incurring technical debt may bring short-term benefits to a project, but such benefits are often achieved at the cost of extra work in future, analogous to paying interest on the debt. Currently technical debt is managed implicitly, if at all. However, on large systems, it is too easy to lose track of delayed tasks or to misunderstand their impact. Therefore, we have proposed a new approach to managing technical debt, which we believe to be helpful for software managers to make informed decisions. In this study we explored the costs of the new approach by tracking the technical debt management activities in an on-going software project. The results from the study provided insights into the impact of technical debt management on software projects. In particular, we found that there is a significant start-up cost when beginning to track and monitor technical debt, but the cost of ongoing management soon declines to very reasonable levels.     

On the detection of custom memory allocators in C binaries

Many reverse engineering techniques for data structures rely on the knowledge of memory allocation routines. Typically, they interpose on the system’s malloc and free functions, and track each chunk of memory thus allocated as a data structure. However, many performance-critical applications implement their own custom memory allocators. Examples include webservers, database management systems, and compilers like gcc and clang. As a result, current binary analysis techniques for tracking data structures fail on such binaries. We present MemBrush, a new tool to detect memory allocation and deallocation functions in stripped binaries with high accuracy. We evaluated the technique on a large number of real world applications that use custom memory allocators. We demonstrate that MemBrush can detect allocators/deallocators with a high accuracy which is 52 out of 59 for allocators, and 29 out of 31 for deallocators in SPECINT 2006. As we show, we can furnish existing reverse engineering tools with detailed information about the memory management API, and as a result perform an analysis of the actual application specific data structures designed by the programmer. Our system uses dynamic analysis and detects memory allocation and deallocation routines by searching for functions that comply with a set of generic characteristics of allocators and deallocators.     

A lagrangian propagator for artificial neural networks in constraint programming

This paper discusses a new method to perform propagation over a (two-layer, feed-forward) Neural Network embedded in a Constraint Programming model. The method is meant to be employed in Empirical Model Learning, a technique designed to enable optimal decision making over systems that cannot be modeled via conventional declarative means. The key step in Empirical Model Learning is to embed a Machine Learning model into a combinatorial model. It has been showed that Neural Networks can be embedded in a Constraint Programming model by simply encoding each neuron as a global constraint, which is then propagated individually. Unfortunately, this decomposition approach may lead to weak bounds. To overcome such limitation, we propose a new network-level propagator based on a non-linear Lagrangian relaxation that is solved with a subgradient algorithm. The method proved capable of dramatically reducing the search tree size on a thermal-aware dispatching problem on multicore CPUs. The overhead for optimizing the Lagrangian multipliers is kept within a reasonable level via a few simple techniques. This paper is an extended version of [27], featuring an improved structure, a new filtering technique for the network inputs, a set of overhead reduction techniques, and a thorough experimentation.     

Computing the Ramsey number R(4,3,3) using abstraction and symmetry breaking

The number R(4, 3, 3) is often presented as the unknown Ramsey number with the best chances of being found “soon”. Yet, its precise value has remained unknown for almost 50 years. This paper presents a methodology based on abstraction and symmetry breaking that applies to solve hard graph edge-coloring problems. The utility of this methodology is demonstrated by using it to compute the value R(4, 3, 3) = 30. Along the way it is required to first compute the previously unknown set \(\mathcal {R}(3,3,3;13)\) consisting of 78,892 Ramsey colorings.     

A GMM/HMM model for reconstruction of missing speech spectral components for continuous speech recognition

This paper presents a method for reconstructing unreliable spectral components of speech signals using the statistical distributions of the clean components. Our goal is to model the temporal patterns in speech signal and take advantage of correlations between speech features in both time and frequency domain simultaneously. In this approach, a hidden Markov model (HMM) is first trained on clean speech data to model the temporal patterns which appear in the sequences of the spectral components. Using this model and according to the probabilities of occurring noisy spectral component at each states, a probability distributions for noisy components are estimated. Then, by applying maximum a posteriori (MAP) estimation on the mentioned distributions, the final estimations of the unreliable spectral components are obtained. The proposed method is compared to a common missing feature method which is based on the probabilistic clustering of the feature vectors and also to a state of the art method based on sparse reconstruction. The experimental results exhibits significant improvement in recognition accuracy over a noise polluted Persian corpus.     

Emotional speech feature normalization and recognition based on speaker-sensitive feature clustering

In this paper we propose a feature normalization method for speaker-independent speech emotion recognition. The performance of a speech emotion classifier largely depends on the training data, and a large number of unknown speakers may cause a great challenge. To address this problem, first, we extract and analyse 481 basic acoustic features. Second, we use principal component analysis and linear discriminant analysis jointly to construct the speaker-sensitive feature space. Third, we classify the emotional utterances into pseudo-speaker groups in the speaker-sensitive feature space by using fuzzy k-means clustering. Finally, we normalize the original basic acoustic features of each utterance based on its group information. To verify our normalization algorithm, we adopt a Gaussian mixture model based classifier for recognition test. The experimental results show that our normalization algorithm is effective on our locally collected database, as well as on the eNTERFACE’05 Audio-Visual Emotion Database. The emotional features achieved using our method are robust to the speaker change, and an improved recognition rate is observed.     

Study of sub-word acoustical models for <Emphasis Type="Italic">Kannada</Emphasis> isolated word recognition system

The speech recognition system basically extracts the textual information present in the speech. In the present work, speaker independent isolated word recognition system for one of the south Indian language—Kannada has been developed. For European languages such as English, large amount of research has been carried out in the context of speech recognition. But, speech recognition in Indian languages such as Kannada reported significantly less amount of work and there are no standard speech corpus readily available. In the present study, speech database has been developed by recording the speech utterances of regional Kannada news corpus of different speakers. The speech recognition system has been implemented using the Hidden Markov Tool Kit. Two separate pronunciation dictionaries namely phone based and syllable based dictionaries are built in-order to design and evaluate the performances of phone-level and syllable-level sub-word acoustical models. Experiments have been carried out and results are analyzed by varying the number of Gaussian mixtures in each state of monophone Hidden Markov Model (HMM). Also, context dependent triphone HMM models have been built for the same Kannada speech corpus and the recognition accuracies are comparatively analyzed. Mel frequency cepstral coefficients along with their first and second derivative coefficients are used as feature vectors and are computed in acoustic front-end processing. The overall word recognition accuracy of 60.2 and 74.35 % respectively for monophone and triphone models have been obtained. The study shows a good improvement in the accuracy of isolated-word Kannada speech recognition system using triphone HMM models compared to that of monophone HMM models.     

Simulation and overall comparative evaluation of performance between different techniques for high band feature extraction based on artificial bandwidth extension of speech over proposed global system for mobile full rate narrow band coder

This paper addresses a novel approach to investigate, study and simulate computation of high band (HB) feature extraction based on linear predictive coding (LPC) and mel frequency cepstral coefficient (MFCC) techniques. Further, HB features are embedded into encoded bitstream of proposed global system for mobile (GSM) full rate (FR) 06.10 coder using joint source coding and data hiding before being transmitted to receiving terminal. At receiver, HB features are extracted to reproduce HB portion of speech and for the same different extension of excitation techniques are applied and their results evaluated in terms of quality (intelligibility and naturalness) and bandwidth. MATLAB based e-test bench is created for implementing the proposed artificial bandwidth extension (ABE) coder following series of simulations, that are carried out to discover and gain insight about the performance of it using subjective [mean opinion score (MOS)] and objective [perceptual evaluation of speech quality (PESQ)] analysis. The results obtained for both the analyses advocate that proposed ABE coder outperforms proposed GSM FR NB (legacy GSM FR) coder. While the fact remains that, compared to LPC based parameterizations over ABE coder, MFCC parameterization results in higher speech intelligibility which is evident from obtained slightly better PESQ and MOS scores.