Time filter

Source Type

Stokic M.,Life Activities Advancement Center | Stokic M.,Institute for Experimental Phonetics and Speech Pathology | Milovanovic D.,University of Belgrade | Ljubisavljevic M.R.,United Arab Emirates University | And 3 more authors.
Experimental Brain Research | Year: 2015

The objective of this preliminary study was to quantify changes in complexity of EEG using fractal dimension (FD) alongside linear methods of spectral power, event-related spectral perturbations, coherence, and source localization of EEG generators for theta (4–7 Hz), alpha (8–12 Hz), and beta (13–23 Hz) frequency bands due to a memory load effect in an auditory–verbal short-term memory (AVSTM) task for words. We examined 20 healthy individuals using the Sternberg’s paradigm with increasing memory load (three, five, and seven words). The stimuli were four-letter words. Artifact-free 5-s EEG segments during retention period were analyzed. The most significant finding was the increase in FD with the increase in memory load in temporal regions T3 and T4, and in parietal region Pz, while decrease in FD with increase in memory load was registered in frontal midline region Fz. Results point to increase in frontal midline (Fz) theta spectral power, decrease in alpha spectral power in parietal region—Pz, and increase in beta spectral power in T3 and T4 region with increase in memory load. Decrease in theta coherence within right hemisphere due to memory load was obtained. Alpha coherence increased in posterior regions with anterior decrease. Beta coherence increased in fronto–temporal regions. Source localization delineated theta activity increase in frontal midline region, alpha decrease in superior parietal region, and beta increase in superior temporal gyrus with increase in memory load. In conclusion, FD as a nonlinear measure may serve as a sensitive index for quantifying dynamical changes in EEG signals during AVSTM tasks. © 2015, Springer-Verlag Berlin Heidelberg.


Grozdic D.T.,University of Belgrade | Grozdic D.T.,Life Activities Advancement Center | Jovicic S.T.,University of Belgrade | Jovicic S.T.,Life Activities Advancement Center | And 2 more authors.
12th Symposium on Neural Network Applications in Electrical Engineering, NEUREL 2014 - Proceedings | Year: 2015

The differences between normal speech and whisper, particularly in terms of their acoustic characteristics, are serious problem of ASR (Automatic Speech Recognition) systems. This paper presents the preliminary results of the new way of speech signal pre-processing, which is based on inverse filtering. This method of signal pre-processing improves whisper recognition with ANNs (Artificial Neural Networks). The ANNs showed high capabilities in speech and whisper recognition in matched train/test scenarios, with the average recognition accuracy of 99.8%. However, the recognition scores in mismatched train/test scenarios were highly degraded. Because of their practical significance, the mismatched train/test scenarios were analyzed in detail in this research. Particularly, the speech/whisper scenario is important. This scenario corresponds to real life situation when speaker is in front of ASR system and from speech switches to whisper. The use of inverse filter enhanced whisper recognition by 9.48%, which in this scenario amounts 70.25%. © 2014 IEEE.


Subotic M.,Life Activities Advancement Center | Saric Z.,Life Activities Advancement Center | Jovicic S.T.,University of Belgrade
Annals of Biomedical Engineering | Year: 2012

Transient otoacoustic emission (TEOAE) is a method widely used in clinical practice for assessment of hearing quality. The main problem in TEOAE detection is its much lower level than the level of environmental and biological noise. While the environmental noise level can be controlled, the biological noise can be only reduced by appropriate signal processing. This paper presents a new two-probe preprocessing TEOAE system for suppression of the biological noise by adaptive filtering. The system records biological noises in both ears and applies a specific adaptive filtering approach for suppression of biological noise in the ear canal with TEOAE. The adaptive filtering approach includes robust sign error LMS algorithm, stimuli response summation according to the derived non-linear response (DNLR) technique, subtraction of the estimated TEOAE signal and residual noise suppression. The proposed TEOAE detection system is tested by three quality measures: signal-to-noise ratio (S/N), reproducibility of TEOAE, and measurement time. The maximal TEOAE detection improvement is dependent on the coherence function between biological noise in left and right ears. The experimental results show maximal improvement of 7 dB in S/N, improvement in reproducibility near 40% and reduction in duration of TEOAE measurement of over 30%. © 2011 Biomedical Engineering Society.


Bilibajkic R.,Life Activities Advancement Center | Saric Z.,Life Activities Advancement Center | Jovicic S.T.,Life Activities Advancement Center | Jovicic S.T.,University of Belgrade | And 2 more authors.
Computer Speech and Language | Year: 2016

Stridence as a form of speech disorder in Serbian language is manifested by the appearance of an intense and sharp whistling. Its acoustic characteristics significantly affect the quality of verbal communication. Although various forms of stridence manifestation are successfully diagnosed by speech therapists, there is a need for the automatic detection and evaluation of stridence. In this paper, an algorithm for stridence detection using Patterson's auditory model is presented. The algorithm consists of three processing stages. In the first stage spectral analysis and masking effects are applied using Paterson's auditory model. In the second stage a contour of spectral peaks that best fits characteristic features of the stridence is selected in the time-frequency (TF) representation of the signal obtained by Patterson's auditory model. In the third stage hypothesis testing is performed with three decisions: D0 - no stridence, D1 - stridence, and D2 - unable to decide. The reliability of stridence detection is tested on the speech corpus of 16 speakers without stridence (with correct speech), 16 speakers without stridence but with some other speech sound disorders, and 16 speakers with stridence. Test results show high correspondence of subjective measures and automatic detection. © 2015 Elsevier Ltd. All rights reserved.


Galic J.,University of Belgrade | Galic J.,University of Banja Luka | Jovicic S.T.,University of Belgrade | Jovicic S.T.,Life Activities Advancement Center | Markovic B.,Cacak Technical College
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

This paper presents results on whispered speech recognition of isolated words with Whi-Spe database, in speaker dependent mode. Word recognition rate is calculated for all speakers, four train/test scenarios, three values of mixture components, with modeling of context independent monophones, context dependent triphones and whole words. As a feature vector, Mel Frequency Cepstral Coefficients was used. The HTK, toolkit for building Hidden Markov Models, was used to implement isolated word recognizer. The best obtained results in match scenarios showed nearly equal recognition rate of 99.86% in normal speech recognition, and 99.90% in whispered speech recognition. Specifically, in mismatch scenarios, the best achieved recognition rate was 64.80% for training on part of normally phonated speech and testing on whispered speech and, in the opposite case, with training on whispered speech, the normal speech recognition was 74.88%. © Springer International Publishing Switzerland 2014.


Bilibajkic R.,Life Activities Advancement Center | Saric Z.,Life Activities Advancement Center | Jovicic S.T.,University of Belgrade
Proceedings of Forum Acusticum | Year: 2011

In this work, an algorithm for automatic speech segmentation on sub-phonemic segments is presented. This algorithm is meant to be used in the speech pathology assessment. The specific conditions of the speech pathology assessment assume cooperation of the examinees who are asked to pronounce predetermined stimuli-words. Hence, the segmentation algorithm may exploit a-priory knowledge about pronounced word that highly improves accuracy of the segmentation. The proposed algorithm determines boundaries of the sub phonemic units using the constrained segmentation. In the training phase of the algorithm the specific model of each word that is to be segmented has to be determent. The word model used in this paper includes phonemic segments structure, parametric description of the each phoneme based on auditory model and time duration limits of each phoneme. Each word is characterized by several models for different gender, age and possible speech pathology. Pronounced word is segmented using the dynamic time warping (DTW) algorithm with constraints applied on each of the word model. The decision on segment boundaries is made using the method of K nearest neighbors. To make algorithm independent of examinees diversity such as age, gender, speech disorder, assessment of the referent word models is done on the learning word group using clustering process. The algorithm is tested on data base records of examinees of different ages, both genders and with different forms of speech pathology.


Markovic B.,Cacak Technical College | Jovicic S.T.,University of Belgrade | Jovicic S.T.,Life Activities Advancement Center | Galic J.,University of Banja Luka | And 2 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

This paper presents creation of a whispered speech database Whi-Spe for Serbian language. The database has been collected in order to investigate how well the whisper is used by humans in intelligible verbal communication and how well whispered information can be used in human-computer communication. The database consists of 50 isolated words. They are generated by ten speakers (five male and five female). Each of them pronounced this vocabulary ten times in two modes: normal and whispered. So, the database contains 5.000 pairs of normal/whispered pronunciations. Database evaluation was performed by an analysis of specific manifestations in whispered articulation. Finally, the preliminary results in whispering recognition by using of HMM, ANN and DTW techniques are presented. © 2013 Springer-Verlag.


Adamovic T.,Life Activities Advancement Center | Adamovic T.,Institute for Experimental Phonetics and Speech Pathology | Kosanovic R.,>>Zvezdara<< Clinical Hospital Center | Kosanovic R.,University of Belgrade | And 5 more authors.
Collegium Antropologicum | Year: 2015

The longitudinal study was conducted in order to establish whether the success rate of reflexes related to maintaining balance at birth is in correlation with the success rate of maintaining balance in early childhood, as well as to examine the correlation of a certain level of speech and language development with the ability of maintaining balance at birth and at the age of 5. The main study group included 54 children of both genders, aged 5.0 to 5.4, whose balance ability and speech and language status were evaluated based on the battery of standardized tests, whereas the group of reflexes related to the function of the vestibular sense was clinically tested on the 3rd day upon birth, within the same sample of children. The data at birth and at the age of 5 were recorded by means of a digital camera, then scored and statistically and descriptively processed. The research results indicated a statistically significant correlation between the achieved level of balance ability in the newborns and five-year-olds, as well as between balance skills and a certain level of speech and language development in children at the age of 5. The importance of this research lies in new knowledge in the domain of maturation of vestubular function immediately after birth, given that this segment of physiology of a newborn has not so far been processed in such a way, as well as in the recognition of function of the vestibular sense as another parametre of a child’s maturation. © 2015, Croatian Anthropological Society. All rights reserved.


Bala G.,University of Novi Sad | Adamovic T.,Life Activities Advancement Center | Adamovic T.,Institute for Experimental Phonetics and Speech Pathology | Madic B.,University of Novi Sad | Popovic B.,University of Novi Sad
Collegium Antropologicum | Year: 2015

The aim of this study was to determine whether acute physical exercise may increase the ability to quickly solve basic mathematical operations in young children. In this way, the children acquired the means to activate a larger area of the brain when necessary. The research sample of 38 preschool and 18 schoolchildren was tested in basic mathematical operations before and after physical exercise. The results showed that children’s computational performance was enhanced significantly during exercise and remained stable after relaxation part of their physical training. © 2015, Croatian Anthropological Society. All rights reserved.


PubMed | Life Activities Advancement Center
Type: Journal Article | Journal: Annals of biomedical engineering | Year: 2012

Transient otoacoustic emission (TEOAE) is a method widely used in clinical practice for assessment of hearing quality. The main problem in TEOAE detection is its much lower level than the level of environmental and biological noise. While the environmental noise level can be controlled, the biological noise can be only reduced by appropriate signal processing. This paper presents a new two-probe preprocessing TEOAE system for suppression of the biological noise by adaptive filtering. The system records biological noises in both ears and applies a specific adaptive filtering approach for suppression of biological noise in the ear canal with TEOAE. The adaptive filtering approach includes robust sign error LMS algorithm, stimuli response summation according to the derived non-linear response (DNLR) technique, subtraction of the estimated TEOAE signal and residual noise suppression. The proposed TEOAE detection system is tested by three quality measures: signal-to-noise ratio (S/N), reproducibility of TEOAE, and measurement time. The maximal TEOAE detection improvement is dependent on the coherence function between biological noise in left and right ears. The experimental results show maximal improvement of 7 dB in S/N, improvement in reproducibility near 40% and reduction in duration of TEOAE measurement of over 30%.

Loading Life Activities Advancement Center collaborators
Loading Life Activities Advancement Center collaborators