Aslam Muhammad

A Speaker Identification System Using MFCC Features with VQ Technique

The performance of speaker identification systems has improved due to recent advances in speech processing techniques but there is still need of improvement in term of text-independent speaker identification and suitable modelling techniques for voice feature vectors. It becomes difficult for person to recognize a voice when an uncontrollable noise adds in to it. In this paper, feature vectors from speech are extracted by using Mel-Frequency Cepstral Coefficients and Vector Quantization technique is implemented through Linde-Buzo-Gray algorithm. Two purposeful speech databases with added noise, recorded at sampling frequencies 8000 Hz and 11025 Hz, are used to check the accuracy of the developed speaker identification system in non-ideal conditions. An analysis is also provided by performing different experiments on the databases that number of vectors in VQ codebook and sampling frequency influence the identification accuracy significantly.

Voice Content Matching System for Quran Readers

In Islamic religion, mistakes in recitation of holy Quran (the sacred book of Muslims) are forbidden. Mistakes can be missing words, verse, misreading Harakat (pronunciations, punctuations, and accents). Thus, a hafiz/reciter who memorizes the holy Quran, needs other hafiz/tutor who listens the recitation and points oral mistakes. Due to the seriously commitment, the availability and expertise of a hafiz are also questionable. A listener can also make mistakes while hearing imputable to environmental interruptions like noise, attention. In order to tackle this issue, we designed, developed, and tested the E-hafiz system. E-hafiz is based on Mel-Frequency Cepstral Coefficient (MFCC) technique to extract voice features from Quranic verse recitation and maps them with the data collected during the training phase. Any mismatch mistake is pointed out. Testing results of short verses of Quran using the E-hafiz system are very encouraging.

Fault Prediction Using Artificial Neural Network and Fuzzy Logic

This paper studies different vehicle fault prediction techniques, using artificial neural network and fuzzy logic based model. With increasing demands for efficiency and product quality as well as progressing integration of automatic control systems in high-cost mechatronics and safety-critical processes, monitoring is necessary to detect and diagnose faults using symptoms and related data. However, beyond protective maintenance services, it is viable to integrate fault prediction services. Thus, we studied different parameters to model a fault prediction service. This service not only helps to predict faults but is also useful to take precautionary measures to avoid tangible and intangible losses.

Text-independent speaker identification using VQ-HMM model based multiple classifier system

Every feature extraction and modeling technique of voice/speech is not suitable in all type of environments. In many real life applications, it is not possible to use all type of feature extraction and modeling techniques to design a single classifier for speaker identification tasks because it will make the system complex. So instead of exploring more techniques or making the system complex it is more reasonable to develop the classifier by using existing techniques and then combine them by using different combination techniques to enhance the performance of the system. Thus, this paper describes the design and implementation of a VQ-HMM based Multiple Classifier System by using different combination techniques. The results show that the developed system by using confusion matrix significantly improve the identification rate.

Remote Patient Monitoring and Healthcare Management Using Multi-agent Based Architecture

Governments in under developing countries face serious problems for offering quality healthcare services at reasonable cost due to rapidly graying population and patients with chronic diseases. Due to currently healthcare resources are not sufficient, many deaths without medication are produced. Thus, this increasing important issue needs to be attended. Our approach to tackle this issue is based on Multi-agent system (MAS) architecture. Our Remote Patient Monitoring (RPM) system monitors patients taking into account physiological and environmental parameters. In this way, RPM provides reliable and low-cost healthcare to elderly, chronically, and acutely people either indoor or outdoor environment. Our hypothesis is that an effective use of healthcare resources, entails reduction of unnecessary hospitalizations, minimizes among others: cost, treatment, and monitoring. However, despite of RPM potential applications, it has not become an integrated part of patient care in heterogeneous environments, mainly due to the lack of knowledge and infrastructure. This paper presents the design and implemented MAS architecture for RPM, identifying barriers and providing perspectives for the future.

Awareness and coordination for web cooperative authoring

This paper presents our approach to design and provide elaborated awareness coordination functions for cooperative production of complex Web shared documents. We designed a Group Awareness Inference Engine (GAIE) that catches working focus of collaborators and then deduces some of their potential interests for communication to enhance coordination and cooperative production.

An Inference Engine for Web Adaptive Cooperative Work

This paper describes the principle of an inference engine that analyzes useful information of actions, performed by cooperating users, to propose modifications of the states and/or the presentation of the shared objects. Using cooperative groupware applications, a group of people may work on the same task while other users may pursue their individual goals using various other applications (cooperative or non-cooperative) with different roles. In such environment, consistency, group awareness and security have essential significance. The work of each user can be observed by capturing their actions and then analyzing them in relation to the history of previous actions. The proposed Adaptive Inference Engine (AIE) behaves as a consumer of application events which analyzes this information on the basis of some predefined rules and then proposes some actions that may be applied within the cooperative environment. In all cases, the user controls the execution of the proposed group awareness actions in his working environment. A prototype of the AIE is developed using the Amaya Web Authoring Toolkit and the PInAS collaborative authoring middleware.

Awareness Elements In Web Based Cooperative Writing Applications

In this paper, we present our practical experience of exploring Web based Cooperative Writing Applications(CWA), a kind of groupware which supports people working in groups to achieve common tasks. While producing cooperatively, participants need structured information about activities of their colleagues and the shared production. Without these features, the cooperative production would be inconsistent and incoherent. We study awareness functionalities integrated into several CWAs on the basis of present and past elements, the kind of communication service, and coordination mechanism. The objective of this study is to investigate the trade-off concerning awareness and to provide evaluation assistance to a group or a member in choosing an application for joint production.

Breakdown Prediction of Different Components of Robotics

In high cost mechatronics and safety-critical systems it becomes very important to ensure failure free functionality. Different software engineering techniques are used to ensure that a robot achieves its goals in any circumstances. However, the mechanical components can breakdown accidentally or due to gradual use, making it difficult to ensure failure free functionality. Thus we describe a general purpose model to predict the breakdowns. As a case study we apply the model to predict the failures of wheels of the robot. Our approach uses artificial neural networks (ANN) and rule based logic.

Feed forward neural network training based interactive shopping for blind

Today shopping markets pay attention towards customer needs and services. Unfortunately the blind and vision impaired person are still incapable to access these environments without reliance. Assistive technology is trying to sway the living style of the blind by introducing support systems for routinely actions like reading, writing, walking, Web surfing, and shopping. However, still the blind have to count on others for personal accessories shopping. To overcome this problem, we designed and developed a feed forward talking accessories selector. Our system is trained using feed forward techniques with a feature level block-based multi-focus image fusion method to provide suggestions regarding product selection, fitness, and color combination, for instance in dress and jewelry. The evaluation of our system takes into account specialists opinions, such that statistical analysis shows similarity between both.