Antanas Lipeika
Antanas Lipeika has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
Latest Publications
- 1 January 2010
- Published by Vilnius University Press in Informatica
- Vol. 21 (3) , 361-374
Optimization of Formant Feature Based Speech Recognition
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: Optimization of Formant Feature Based Speech Recognition, Authors: Antanas Lipeika , The paper deals with the use of formant features in dynamic time warping based speech recognition. These features can be simply visualized and give a new insight into understanding the reasons of speech recognition errors. The formant feature extraction method, based on the singular prediction polynomials, has been applied in recognition of isolated words. However, the speech recognition performance depends on the order of singular prediction polynomials, whether symmetric or antisymmetric singular prediction polynomials are used for recognition and as well on the fact even or odd order of these polynomials is chosen. Also, it is important to know how informative separate formants are, how the speech recognition results depend on other parameters of the recognition system such as: analysis frame length, number of the formants used in recognition, frequency scale used for representation of formant features, and the preemphasis filter parameters. Properly choosing the processing parameters, it is possible to optimize the speech recognition performance. The aim of our current investigation is to optimize formant feature based isolated word recognition performance by varying processing parameters of the recognition system as well as to find improvements of the recognition system which could make it more robust to white noise. The optimization experiments were carried out using speech records of 111 Lithuanian words. The speech signals were recorded in the conventional room environment (SNR = 30 dB). Then the white noise was generated at a predefined level (65 dB, 60 dB and 55 dB) and added to the test utterances. The recognition performance was evaluated at various noise levels. The optimization experiments allowed us to improve considerably the performance of the formant feature based speech recognition system and made the system more robust to white noise.
- 1 January 2010
- Published by Vilnius University Press in Informatica
- Vol. 21 (3) , 361-374
Optimization of Formant Feature Based Speech Recognition
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: Optimization of Formant Feature Based Speech Recognition, Authors: Antanas Lipeika , The paper deals with the use of formant features in dynamic time warping based speech recognition. These features can be simply visualized and give a new insight into understanding the reasons of speech recognition errors. The formant feature extraction method, based on the singular prediction polynomials, has been applied in recognition of isolated words. However, the speech recognition performance depends on the order of singular prediction polynomials, whether symmetric or antisymmetric singular prediction polynomials are used for recognition and as well on the fact even or odd order of these polynomials is chosen. Also, it is important to know how informative separate formants are, how the speech recognition results depend on other parameters of the recognition system such as: analysis frame length, number of the formants used in recognition, frequency scale used for representation of formant features, and the preemphasis filter parameters. Properly choosing the processing parameters, it is possible to optimize the speech recognition performance. The aim of our current investigation is to optimize formant feature based isolated word recognition performance by varying processing parameters of the recognition system as well as to find improvements of the recognition system which could make it more robust to white noise. The optimization experiments were carried out using speech records of 111 Lithuanian words. The speech signals were recorded in the conventional room environment (SNR = 30 dB). Then the white noise was generated at a predefined level (65 dB, 60 dB and 55 dB) and added to the test utterances. The recognition performance was evaluated at various noise levels. The optimization experiments allowed us to improve considerably the performance of the formant feature based speech recognition system and made the system more robust to white noise.
- Open Access
- 1 January 2009
- Published by Vilnius University Press in Information & Media
- Vol. 50, 212-216
Formantinių požymių naudojimas kalbai atpažinti
Straipsnyje nagrinėjami formantinių požymių taikymo atpažįstant kalbą klausimai. Nustatyta, kad formantiniai požymiai tam gali būti naudojami, tačiau atpažinimo tikslumas labai priklauso nuo formantinių požymių išskyrimo metodo. Geriausi atpažinimo rezultatai gaunami formantinių požymių išskyrimui naudojant išsigimusius prognozės polinomus. Šie polinomai gali būti skaičiuojami iš lyginės arba nelyginės eilės tiesinės prognozės modelio parametrų. Be to, atpažinimui galima naudoti simetrinius arba antisimetrinius išsigimusius tiesinės prognozės polinomus. Taip pat svarbu ištirti, kaip kalbos atpažinimo rezultatai priklauso ne tik nuo išsigimusių tiesinės prognozės polinomų parinkimo, bet ir kitų atpažinimo sistemos parametrų: analizės kadro ilgio, atpažinimui naudojamų formančių skaičiaus, formantinių požymių vaizdavimui naudojamos dažnių skalės. Tyrimais nustatyta, kad geriausi atpažinimo rezultatai gaunami naudojant dvi arba tris formantes, apskaičiuotas iš simetrinių išsigimusių prognozės polinomų. Tiriant atskirų formančių informatyvumą paaiškėjo, kad didžiausias indėlis į atpažinimą yra antros formantės. Pirmos, trečios ir ketvirtos formančių indėlis maždaug vienodas, bet aukštesnės formantės mažiau atsparios balto triukšmo įtakai. Tiriant analizės kadro ilgio parinkimą nustatyta, kad geriausi atpažinimo rezultatai yra esant 500 atskaitų kadro ilgiui. Atpažinimo rezultatai taip pat gaunami geresni vaizduojant formančių trajektorijas melų skalėje.Investigation of Formant Features in Speech RecognitionAntanas Leonas Lipeika SummaryThe use of formant features in speech recognition is investigated in the paper. It was established that formant features can be used in speech recognition but recognition accuracy depends remarkably on the formant feature extraction method. The best recognition results were obtained when singular prediction polynomials were used for formant feature extraction. These polynomials can be calculated from parameters of linear prediction models of even or odd order. These polynomials can by symmetric or antisymmetric as well. Also it is important to investigate how results of speech recognition depends not only on choice of singular prediction polynomials but although on other parameters of the recognition system: frame length, number of used formants in recognition, frequency scale, used for representation of formant features. During the experiments it was defi ned that the best recognition results were obtained using 2 or 3 formants calculated from symmetric singular prediction polynomials. The experiments have shown that the most informative is the 2-nd formant. Contribution of the 1-st, 3-rd and 4-th formants is approximately similar, but higher formants are less resistant to white noise. Recognition results also depends on analysis frame length and frequency scale. The best results were obtained using 500 data points frame length and Mel frequency scale.
- 1 January 2009
- Published by Vilnius University Press in Information & Media
- Vol. 50, 212-216
Open AccessFormantinių požymių naudojimas kalbai atpažinti
Straipsnyje nagrinėjami formantinių požymių taikymo atpažįstant kalbą klausimai. Nustatyta, kad formantiniai požymiai tam gali būti naudojami, tačiau atpažinimo tikslumas labai priklauso nuo formantinių požymių išskyrimo metodo. Geriausi atpažinimo rezultatai gaunami formantinių požymių išskyrimui naudojant išsigimusius prognozės polinomus. Šie polinomai gali būti skaičiuojami iš lyginės arba nelyginės eilės tiesinės prognozės modelio parametrų. Be to, atpažinimui galima naudoti simetrinius arba antisimetrinius išsigimusius tiesinės prognozės polinomus. Taip pat svarbu ištirti, kaip kalbos atpažinimo rezultatai priklauso ne tik nuo išsigimusių tiesinės prognozės polinomų parinkimo, bet ir kitų atpažinimo sistemos parametrų: analizės kadro ilgio, atpažinimui naudojamų formančių skaičiaus, formantinių požymių vaizdavimui naudojamos dažnių skalės. Tyrimais nustatyta, kad geriausi atpažinimo rezultatai gaunami naudojant dvi arba tris formantes, apskaičiuotas iš simetrinių išsigimusių prognozės polinomų. Tiriant atskirų formančių informatyvumą paaiškėjo, kad didžiausias indėlis į atpažinimą yra antros formantės. Pirmos, trečios ir ketvirtos formančių indėlis maždaug vienodas, bet aukštesnės formantės mažiau atsparios balto triukšmo įtakai. Tiriant analizės kadro ilgio parinkimą nustatyta, kad geriausi atpažinimo rezultatai yra esant 500 atskaitų kadro ilgiui. Atpažinimo rezultatai taip pat gaunami geresni vaizduojant formančių trajektorijas melų skalėje.Investigation of Formant Features in Speech RecognitionAntanas Leonas Lipeika SummaryThe use of formant features in speech recognition is investigated in the paper. It was established that formant features can be used in speech recognition but recognition accuracy depends remarkably on the formant feature extraction method. The best recognition results were obtained when singular prediction polynomials were used for formant feature extraction. These polynomials can be calculated from parameters of linear prediction models of even or odd order. These polynomials can by symmetric or antisymmetric as well. Also it is important to investigate how results of speech recognition depends not only on choice of singular prediction polynomials but although on other parameters of the recognition system: frame length, number of used formants in recognition, frequency scale, used for representation of formant features. During the experiments it was defi ned that the best recognition results were obtained using 2 or 3 formants calculated from symmetric singular prediction polynomials. The experiments have shown that the most informative is the 2-nd formant. Contribution of the 1-st, 3-rd and 4-th formants is approximately similar, but higher formants are less resistant to white noise. Recognition results also depends on analysis frame length and frequency scale. The best results were obtained using 500 data points frame length and Mel frequency scale.
- 1 January 2008
- Published by Vilnius University Press in Informatica
- Vol. 19 (2) , 213-226
On the Use of the Formant Features in the Dynamic Time Warping Based Recognition of Isolated Words
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: On the Use of the Formant Features in the Dynamic Time Warping Based Recognition of Isolated Words, Authors: Antanas Lipeika, Joana Lipeikienė , A possibility to use the formant features (FF) in the user-dependent isolated word recognition has been investigated. The word recognition was performed using a dynamic time-warping technique. Several methods of the formant feature extraction were compared and a method based on the singular prediction polynomials has been proposed for the recognition of isolated words. Recognition performance of the proposed method was compared to that of the linear prediction coding (LPC) and LPC-derived cepstral features (LPCC). In total, 111 Lithuanian words were used in the recognition experiment. The recognition performance was evaluated at various noise levels. The experiments have shown that the formant features calculated from the singular prediction polynomials are more reliable than the LPC and LPCC features at all noise levels.
- 1 January 2008
- Published by Vilnius University Press in Informatica
- Vol. 19 (2) , 213-226
On the Use of the Formant Features in the Dynamic Time Warping Based Recognition of Isolated Words
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: On the Use of the Formant Features in the Dynamic Time Warping Based Recognition of Isolated Words, Authors: Antanas Lipeika, Joana Lipeikienė , A possibility to use the formant features (FF) in the user-dependent isolated word recognition has been investigated. The word recognition was performed using a dynamic time-warping technique. Several methods of the formant feature extraction were compared and a method based on the singular prediction polynomials has been proposed for the recognition of isolated words. Recognition performance of the proposed method was compared to that of the linear prediction coding (LPC) and LPC-derived cepstral features (LPCC). In total, 111 Lithuanian words were used in the recognition experiment. The recognition performance was evaluated at various noise levels. The experiments have shown that the formant features calculated from the singular prediction polynomials are more reliable than the LPC and LPCC features at all noise levels.
- Open Access
- 1 January 2008
- Published by Vilnius University Press in Information & Media
- Vol. 42 (43) , 201-206
Formantinių požymių išskyrimo metodai
Straipsnyje nagrinėjami formantinių požymių išskyrimo metodai. Formantinių požymių išskyrimas remiasi spektro pikų radimu apskaičiuotame iš tiesinės prognozės modelio parametrųų spektre. Formantini ų požymių išskyrimo patikimumas priklauso nuo tiesinės prognozės modelio parametrų vertinimo metodo. Anksčiau tiesinės prognozės modelio parametrų vertinimui naudojome autokoreliacinį metodą, kuris neužtikrindavo patikimo formantinių požymių išskyrimo. Todėl, siekiant padidinti formantini ų požymių išskyrimo patikimumų, ieškoma geresnio tiesinės prognozės modelio parametrų vertinimo metodo. Autokoreliacinis tiesinės prognozės modelio parametrų vertinimo metodas lyginamas su kovariaciniu, Burg, Marple metodais ir modifikuotu Split Levinson algoritmu. Tyrimais nustatyta, kad pagal formančių trajektorijų išskyrimą kovariacinis, Burg, Marple tiesinės prognozės modelio parametrų vertinimo metodai iš esmės nesiskiria nuo autokoreliacinio, o modifikuotu Split Levinson algoritmu gauname daug patikimesnius formančių trajektorijų įverčius.Formant feature extraction methodsAntanas Leonas Lipeika SummaryFormant feature extraction is investigated in the paper. Extraction of formant features is based on calculating frequency positions of spectral peaks. The spectrum has been calculated from parameters of linear prediction model. Reliability of formant feature extraction depends on the method used for linear prediction model parameter estimation. The autocorrelation method previously used for linear prediction model parameter estimation was not reliable enough for formant feature extraction. Therefore we were looking for more reliable method of linear prediction model parameter estimation. The previously used autocorrelation method was compared with covariance, Burg, Marple methods and the modified Split Levinson algorithm. It was concluded, that autocorrelation, covariance, Burg and Marple methods are similar from the point of view of formant feature extraction. The modified Split Levinson algorithm provides the best formant feature estimates.
- 1 January 2008
- Published by Vilnius University Press in Information & Media
- Vol. 42 (43) , 201-206
Open AccessFormantinių požymių išskyrimo metodai
Straipsnyje nagrinėjami formantinių požymių išskyrimo metodai. Formantinių požymių išskyrimas remiasi spektro pikų radimu apskaičiuotame iš tiesinės prognozės modelio parametrųų spektre. Formantini ų požymių išskyrimo patikimumas priklauso nuo tiesinės prognozės modelio parametrų vertinimo metodo. Anksčiau tiesinės prognozės modelio parametrų vertinimui naudojome autokoreliacinį metodą, kuris neužtikrindavo patikimo formantinių požymių išskyrimo. Todėl, siekiant padidinti formantini ų požymių išskyrimo patikimumų, ieškoma geresnio tiesinės prognozės modelio parametrų vertinimo metodo. Autokoreliacinis tiesinės prognozės modelio parametrų vertinimo metodas lyginamas su kovariaciniu, Burg, Marple metodais ir modifikuotu Split Levinson algoritmu. Tyrimais nustatyta, kad pagal formančių trajektorijų išskyrimą kovariacinis, Burg, Marple tiesinės prognozės modelio parametrų vertinimo metodai iš esmės nesiskiria nuo autokoreliacinio, o modifikuotu Split Levinson algoritmu gauname daug patikimesnius formančių trajektorijų įverčius.Formant feature extraction methodsAntanas Leonas Lipeika SummaryFormant feature extraction is investigated in the paper. Extraction of formant features is based on calculating frequency positions of spectral peaks. The spectrum has been calculated from parameters of linear prediction model. Reliability of formant feature extraction depends on the method used for linear prediction model parameter estimation. The autocorrelation method previously used for linear prediction model parameter estimation was not reliable enough for formant feature extraction. Therefore we were looking for more reliable method of linear prediction model parameter estimation. The previously used autocorrelation method was compared with covariance, Burg, Marple methods and the modified Split Levinson algorithm. It was concluded, that autocorrelation, covariance, Burg and Marple methods are similar from the point of view of formant feature extraction. The modified Split Levinson algorithm provides the best formant feature estimates.
- 1 January 2007
- Published by Vilnius University Press in Informatica
- Vol. 18 (3) , 395-406
Framework for Choosing a Set of Syllables and Phonemes for Lithuanian Speech Recognition
This paper describes a framework for making up a set of syllables and phonemes that subsequently is used in the creation of acoustic models for continuous speech recognition of Lithuanian. The target is to discover a set of syllables and phonemes that is of utmost importance in speech recognition. This framework includes operations with lexicon, and transcriptions of records. To facilitate this work, additional programs have been developed that perform word syllabification, lexicon adjustment, etc. Series of experiments were done in order to establish the framework and model syllable- and phoneme-based speech recognition. Dominance of a syllable in lexicon has improved speech recognition results and encouraged us to move away from a strict definition of syllable, i.e., a syllable becomes a simple sub-word unit derived from a syllable. Two sets of syllables and phonemes and two types of lexicons have been developed and tested. The best recognition accuracy achieved 56.67% ±0.33. The speech recognition system is based on Hidden Markov Models (HMM). The continuous speech corpus LRN0 was used for the speech recognition experiments.
- 1 January 2007
- Published by Vilnius University Press in Informatica
- Vol. 18 (3) , 395-406
Framework for Choosing a Set of Syllables and Phonemes for Lithuanian Speech Recognition
This paper describes a framework for making up a set of syllables and phonemes that subsequently is used in the creation of acoustic models for continuous speech recognition of Lithuanian. The target is to discover a set of syllables and phonemes that is of utmost importance in speech recognition. This framework includes operations with lexicon, and transcriptions of records. To facilitate this work, additional programs have been developed that perform word syllabification, lexicon adjustment, etc. Series of experiments were done in order to establish the framework and model syllable- and phoneme-based speech recognition. Dominance of a syllable in lexicon has improved speech recognition results and encouraged us to move away from a strict definition of syllable, i.e., a syllable becomes a simple sub-word unit derived from a syllable. Two sets of syllables and phonemes and two types of lexicons have been developed and tested. The best recognition accuracy achieved 56.67% ±0.33. The speech recognition system is based on Hidden Markov Models (HMM). The continuous speech corpus LRN0 was used for the speech recognition experiments.
- 15 April 2006
- Published by Vilnius University Press in Informatics in Education
- Vol. 5 (1) , 87-96
Animation Tools of CAS for Dynamic Exploration of Mathematics
The scope of the paper is animation facilities of computer algebra systems (CAS). Animation offers opportunities for visualization of complex mathematical concepts, provides convincing demonstration of ideas and influence of quantities or parameters, helps to generate hypothesis, encourages exploration. Animation can be used to demonstrate many mathematical concepts that are difficult to explain verbally or to show with static pictures. Using animation allows students to explore, experiment and visualize mathematics as a dynamic process. But CAS creates only opportunities. The problem remains for users to realize this potential. So features of CAS such as ease of use, convenience of procedures are important for teaching and learning. The paper deals with animation features of the three most popular CAS - Maple, Matlab, Mathcad and their usefulness in education. The results of practical use of the three CAS in teaching animation procedures are discussed and students' opinion about animation tools of CAS is presented.
- 15 April 2006
- Published by Vilnius University Press in Informatics in Education
- Vol. 5 (1) , 87-96
Animation Tools of CAS for Dynamic Exploration of Mathematics
The scope of the paper is animation facilities of computer algebra systems (CAS). Animation offers opportunities for visualization of complex mathematical concepts, provides convincing demonstration of ideas and influence of quantities or parameters, helps to generate hypothesis, encourages exploration. Animation can be used to demonstrate many mathematical concepts that are difficult to explain verbally or to show with static pictures. Using animation allows students to explore, experiment and visualize mathematics as a dynamic process. But CAS creates only opportunities. The problem remains for users to realize this potential. So features of CAS such as ease of use, convenience of procedures are important for teaching and learning. The paper deals with animation features of the three most popular CAS - Maple, Matlab, Mathcad and their usefulness in education. The results of practical use of the three CAS in teaching animation procedures are discussed and students' opinion about animation tools of CAS is presented.
- 1 January 2004
- Published by Vilnius University Press in Informatica
- Vol. 15 (4) , 465-474
Development of HMM/Neural Network‐Based Medium‐Vocabulary Isolated‐Word Lithuanian Speech Recognition System
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: Development of HMM/Neural Network‐Based Medium‐Vocabulary Isolated‐Word Lithuanian Speech Recognition System, Authors: Mark Filipovič, Antanas Lipeika , The development of Lithuanian HMM/ANN speech recognition system, which combines artificial neural networks (ANNs) and hidden Markov models (HMMs), is described in this paper. A hybrid HMM/ANN architecture was applied in the system. In this architecture, a fully connected three‐layer neural network (a multi‐layer perceptron) is trained by conventional stochastic back‐propagation algorithm to estimate the probability of 115 context‐independent phonetic categories and during recognition it is used as a state output probability estimator. The hybrid HMM/ANN speech recognition system based on Mel Frequency Cepstral Coefficients (MFCC) was developed using CSLU Toolkit. The system was tested on the VDU isolated‐word Lithuanian speech corpus and evaluated on a speaker‐independent ∼750 distinct isolated‐word recognition task. The word recognition accuracy obtained was about 86.7%.
- 1 January 2004
- Published by Vilnius University Press in Informatica
- Vol. 15 (4) , 465-474
Development of HMM/Neural Network‐Based Medium‐Vocabulary Isolated‐Word Lithuanian Speech Recognition System
Publisher: Vilnius University Institute of Data Science and Digital Technologies, Journal: Informatica, Title: Development of HMM/Neural Network‐Based Medium‐Vocabulary Isolated‐Word Lithuanian Speech Recognition System, Authors: Mark Filipovič, Antanas Lipeika , The development of Lithuanian HMM/ANN speech recognition system, which combines artificial neural networks (ANNs) and hidden Markov models (HMMs), is described in this paper. A hybrid HMM/ANN architecture was applied in the system. In this architecture, a fully connected three‐layer neural network (a multi‐layer perceptron) is trained by conventional stochastic back‐propagation algorithm to estimate the probability of 115 context‐independent phonetic categories and during recognition it is used as a state output probability estimator. The hybrid HMM/ANN speech recognition system based on Mel Frequency Cepstral Coefficients (MFCC) was developed using CSLU Toolkit. The system was tested on the VDU isolated‐word Lithuanian speech corpus and evaluated on a speaker‐independent ∼750 distinct isolated‐word recognition task. The word recognition accuracy obtained was about 86.7%.
- 1 January 2003
- Published by Vilnius University Press in Informatica
- Vol. 14 (4) , 487-496
Word Endpoint Detection Using Dynamic Programming
The paper deals with the use of dynamic programming for word endpoint detection in isolated word recognition. Endpoint detection is based on likelihood maximization. Expectation maximization approach is used to deal with the problem of unknown parameters. Speech signal and background noise energy is used as features for making decision. Performance of the proposed approach was evaluated using isolated Lithuanian words speech corpus.
- 1 January 2003
- Published by Vilnius University Press in Informatica
- Vol. 14 (4) , 487-496
Word Endpoint Detection Using Dynamic Programming
The paper deals with the use of dynamic programming for word endpoint detection in isolated word recognition. Endpoint detection is based on likelihood maximization. Expectation maximization approach is used to deal with the problem of unknown parameters. Speech signal and background noise energy is used as features for making decision. Performance of the proposed approach was evaluated using isolated Lithuanian words speech corpus.