Michal Kuba

Development of a Reference Platform for Generic Audio Classification

Detection of key sounds, such as applause, laugh, music, environmental noise, etc., is one of the challenges in intelligent management of multimedia information and content understanding. In this paper, we report progress in development of a reference content-based audio classification algorithm that is based on a conventional and widely accepted approach, namely signal parameterization by MFCC followed by GMM classification. Our developed labeled audio database and the conventional classification model should serve as a reference platform for an evaluation of novel, alternative or more advanced methods in audio content analysis.

GA-Based Feature Extraction for Clapping Sound Detection

Automatically extracting semantic content from audio streams can be helpful in many multimedia applications. In this paper, we introduce a framework for automatic feature subspace selection from a common feature vector. The selected features build a new representation which is better suitable for a given learning task and recognition. In order to solve this problem, we propose the GA-based (genetic algorithm) method to improve the representativeness and robustness of the features generic audio recognition task

Influence of nonlinear effects in WDM system with non-equidistant channel spacing using different types of high-order PSK and QAM modulation formats

The objective of this paper is to investigate nonlinear effects in Wavelength-Division Multiplex (WDM) systems in the case when different types of high-order M-PSK and M-QAM modulation formats for various structures of channel spacing are used optical signals. In general, the degradation mechanisms are caused by transmitted optical signals and their impact on each optical channel in WDM can be very different. Therefore, it is suitable to investigate possibilities for channel arrangement from the point of view of equidistant and non-equidistant channel spacing, respectively, what would lead to the suppression of nonlinear effects. In this article we investigate new types of high-order modulation formats that have ability for increasing of spectral efficiency and total improvement of performance of the transmission WDM system. The attention is put on two classes of channel spacing in WDM system, equidistant channel spacing (Δf = 100, 50, 25 and 12.5 GHz) and non-equidistant channel spacing (Δf ≠ const.), respectively. For investigation of signal propagation the numerical model is created. The model is based on mathematical method Symmetrical-Split Step Fourier Method (S-SSFM), which is utilized for solving the coupled nonlinear Schrödingers equations (CNLSE) describing the transmission of signals in multichannel systems. The results of the created numerical model are analyzed, compared to each other and interpreted in a way that leads to the determination of suitable high-order modulation formats and we try to propose the optimal arrangement of optical channels in WDM system. The key issue is to suppress the impact of nonlinearities on modulated signals for each channel with respect to the employed types of digital modulation formats, various system parameters, different types of optical fibers and localization of reference channel in the wavelength area.

Compact representation of speech using 2-d cepstrum – an application to slovak digits recognition

HMM speech recogniser with a small number of acoustic observations based on 2-D cepstrum (TDC) is proposed. TDC represents both static and dynamic features of speech implicitly in matrix form. It is shown that TDC analysis enables a compact representation of speech signals. Thus a great advantage of the proposed model is a massive reduction of speech features used for recognition what lessens computational and memory requirements, so it may be favourable for limited-power ASR applications. Experiments on isolated Slovak digits recognition task show that the method gives comparable results as the conventional MFCC approach. For speech degraded by additive white noise, it reaches better performance than the MFCC method.

Numerical representation of optical sources for coherent lightwave systems

The numerical model of CW-DFB optical source for fiber-optic transmission systems employing coherent detection technique is presented. The developed model is based on numerical solution of quantum-mechanical coupled rate equations. The numerical model is suitable for optical transmission systems for purpose of generation of optical carrier wave on the transmitter side as well as it can be used as local laser for coherent detection on the receiver side. The main attention of proposed optical source model is focused on signal, time-dependent characteristics to determine spectral properties of presented laser, which fully satisfies current requirements on coherent optical transmission systems. The efficiency of numerical model is also discussed. We have showed that by employing the re-sampling technique, both computational effort and accuracy of presented laser model are markedly enhanced.

Numerical model for DGD estimation in optical transmission system

In this paper we investigate the nonlinear and polarization effects in optical transmission systems and its influence on transmitted pulses. The main attention is focused on fundamental description of refractive index in nonlinear birefringent environment. In general, optical fiber is nonlinear transmission medium. The mutual interaction between transmission medium and optical intensity induces changes in the refractive index resulting to nonlinear effects that can be polarization-dependent in presence of birefringence. The global effect has a significant impact on pulse propagation. The aim of this paper is to present a numerical model, which will be suitable for estimation of DGD (Differential Group Delay) parameter in nonlinear birefringence medium. The DGD is crucial for evaluation of the impact of PMD (Polarization Mode Dispersion) in high-bit-rates fiber-optic system. Our approach opens the novel opportunity for DGD estimation based on numerical model of optical pulse propagation in nonlinear birefringence medium. Numerical model is based on solving NLSE (Nonlinear Schrodinger Equation) through the SSFM (Split-Step Fourier Method). The model is compatible for various input parameters, different kinds of optical fibers and also for different types of modulation formats. The obtained results show the DGD for different system parameters (such as input power and wavelength) and for different fiber polarization characteristics (birefringence and mode coupling). The total pulse broadening is also calculated and illustrates how all degradation effects influence the performance of fiber-optic transmission system.

Modeling of WDM transmission system with high-order phase modulation formats

The objective of this paper is focused on numerical modeling of WDM (Wavelength-Division Multiplex) transmission systems that employ new classes of high-order PSK (Phase-Shift Keying) modulation formats. This paper provides investigation of signals propagation with corresponding degradation mechanisms in physical layer, when high-order phase modulation formats are used for their transmission. The impacts of linear and nonlinear effects on modulated signal in multichannel system are studied. For this purpose, it was created numerical model by solving CNLSE (Coupled Nonlinear Schrödingers Equations) through SSFM (Split-Step Fourier Method). The fundamental characteristics of optical fibers, employed modulation format and WDM system parameters are significant for our investigation, because they play important role for behavior of modulated signals. Our results are analyzed and interpreted by the way of finding out the suitable and optimal system settings for transmission of phase modulated signals.

Numerical investigation of noise chracteristics of telecommunication laser sources for various modulation formats

We present numerical investigation of noise properties of lasers for telecommunication purposes with emphasis on widely used distributed feedback (DFB) lasers and its influence on multi-level modulation formats. DFB lasers can be used in optical transmitters with internal and external modulation, as well as in optical receivers employing coherent detection, where they act as local oscillator. The main noise factors influencing signal characteristics of semiconductor-based lasers are intensity and phase noise. These random impairments cause degradation of fundamental output laser characteristics such as power and phase fluctuation, which are directly related to the optical signal-to-noise ratio and the laser linewidth. In case of implementation of new modulation formats into the transmission system, these stochastic processes are of main importance and significantly impact the transmission properties of modulated signals and total performance of fiber-optic transmission systems. Throughout this paper, the noise influence on the multi-level modulation formats is evaluated by effective signal-to-noise ratio (SNR) algorithms in combination with bit-error ratio (BER) formulas for appropriate type of modulation format. The obtained results have shown that the noise of DFB laser is serious restriction for multi-level modulation formats and can be improved by higher power levels, yielding to higher SNR, as required for better values of BER.

Audio patterns searching and retrieval

Pattern searching and retrieval plays important role in task of content-based audio analysis for requirements of media database management or in surveillance systems for detecting significant audio events and keywords. In the paper, we present algorithm for spotting audio patterns in record, using Hidden Markov Models and Viterbi decoder. Also, two types of probability normalization are presented and compared. Performance of proposed algorithm was evaluated on database of slovak spoken numbers.

Polydimethylsiloxane coated optical fiber sensor for detection of organic volatile compounds

A compact optical fiber sensor coated with polydimethylsiloxane for detection of organic volatile compounds (e.g. toluene) has been developed and tested. The sensor consists of two mirrors (two optical fiber ends) facing to each other and a variable length cavity. One fiber used as an input/output fiber, and the other one, used as a reflector are aligned using a PDMS coating and together form Fabry-Perot interferometer. We investigated toluene impact on the spectral characteristic of reflected light power in time. Using FFT we determine the period of interference pattern of measured characteristic depending on the time. We also detect that the change in temperature within range from 20°C to 50°C does not affect the period of interference pattern.