Gregor Rozinaj

Application of immersive technologies for education: State of the art

Existing multimedia systems used in education mostly address only two senses by using two communication channels (visual and audio) of the five human senses (sight, hearing, taste, smell, and touch), limiting the potential efficiency of learning. This paper presents a survey on existing technical opportunities for the development of an immersive learning environment. Four components of the immersive environment - visual, audio, olfactory, and haptic are described and discussed in the paper. In particular 3D displays, head mounted devices, 3D sound systems, olfactory displays, haptic devices, and interaction devices are presented.

Smart AppStore: Expanding the Frontiers of Smartphone Ecosystems

Smart AppStore offers five important features for todays smartphone users: biometric authentication, multilevel authorization, gesture recognition and navigation, user-tailored reputation scores, and identity management. The Web extra at http://youtu.be/15KPOUp5H_A is a video demonstration of Smart AppStore, which offers five important features for todays smartphone users: biometric authentication, multilevel authorization, gesture recognition and navigation, user-tailored reputation scores, and identity management.

Building accurate and robust HMM models for practical ASR systems

In this article the relevant training aspects for building robust and accurate HMM models for large vocabulary recognition system are discussed and adjusted, namely: speech features, training steps, and the tying options for context dependent (CD) phonemes. As the basis for building HMM models the well known MASPER training scheme is assumed. First the incorporation of the voicing information and its effect on the classical extraction methods like MFCC and PLP will be shown together with the derivative features, where the relative error reductions are up to 50%. Next the suggested enhancement of the standard training procedure by introducing garbled speech models will be presented and tested on real data. As it will be shown it brings more than a 5% drop in the error rate. Finally, the options for tying states of CD phonemes using decision trees and phoneme classification will be adjusted, tested, and explained.

A hybrid pitch period estimation method based on HNM model

Pitch period estimation (also called fundamental frequency estimation) is widely needed in speech processing for many purposes. In our system for prosodic modification of speech, the pitch period estimation is used as a basis for frame length detection. The pitch period estimation method used in the system is a hybrid method that is based on YIN fundamental frequency estimation algorithm and a method for fundamental frequency detection on magnitude of the speech signal. The experiments show, that the method is useful in sinusoidal modeling domain, as in other domains, too.

Forward masking phenomenon in concatenative speech synthesis

The approach described in the paper tries to get more knowledge to the concatenative text-to-speech system design. The knowledge is based on masking phenomenon of the inner ear, particularly of its temporal (forward) masking properties. Designing such knowledge-based system is suggested to use in the unit selection-based speech synthesis, as contemporary a prominent technique in concatenative synthesis, which utilizes a big speech corpus. The more prosodic variability the corpus captures, the more natural a synthetic voice sounds and there are more possibilities to occur a forward masking events during concatenation of selected candidate units from the corpus.

Speech detection in the noisy environment using wavelet transform

In this article we present speech detection systems based on Daubechie, Coiflet and Symlet wavelet transforms respectively. For each a selection of the most eligible levels of signal decomposition for the corrupted speech detection problem was made. Using those levels the distinction between noise and corrupted signal can be amplified as far as 100 times. Tests were accomplished using a set of Slovak words artificially noised to several SNR by white WSS noise.

ZCPA features for speech recognition

In this article we present implementation, modifications and optimization of zero-crossing peak amplitude (ZCPA) speech feature extraction method into Slovak speech recognition system. ZCPA features are closely mimicking the human auditory system in the time domain, and thus they should be more robust against common noises. Except the basic configuration several modifications have been suggested, implemented and evaluated. Furthermore, optimization of settings on a real system using professional database and MASPER training procedure have been found and compared to classical features presented by MFCC and PLP in different scenarios and noise conditions.

MABox - Multimodal Microphone Array Algorithm Development System

In this work a design and a realization of multimodal microphone array algorithm development system which is proposed to develop new microphone array algorithm named MABox is presented. This device incorporates microphone array with four microphones, ADC cards and development software. Microphones are integrated in a separate directional box pointed to the speaker, the box is connected via USB and analog line to the computer. The development software environment allows us to test new beamforming algorithm. This system runs in real-time what allows us to change the structure of algorithm and its parameters.

Facial features detection for 3D face modeling

This paper presents the detection algorithm developed for the specific application of 3D face modeling. The main goal of this algorithm developed under VC++ is to localize accurately some specific and needed points of facial features, to allow a 3D face creation. Instead of using some predefined 3D face model, the program process on two photos of a real subject, one from front view and the other from profile. It permits us to obtain the three coordinates for each localized point on the face. A data normalization allows face modeling using the openGL engine.

New approach of fast ICT and MICT algorithms development

New approach of fast ICT (Integer Cosine Transform) and MICT (Modified ICT) algorithms development is described in the paper. Both-ICT and MICT-transforms are the integer approximations of DCT. As the properties of ICT and MICT are close to those of DCT and their transform kernel is based on integers their computational complexity is less than in the case of DCT. Based on this DCT can be replaced by ICT and MICT, for example in transform coding of images (JPEG, MPEG, ITU-T H.261, etc.).