Son Huynh Van - Academia.edu (original) (raw)

Papers by Son Huynh Van

4th European Conference on Speech Communication and Technology (Eurospeech 1995)

Conference of the International Speech Communication Association, Aug 27, 2007

We describe the SpeakGoodChinese system that supports beginning students of Mandarin Chinese to p... more We describe the SpeakGoodChinese system that supports beginning students of Mandarin Chinese to produce tones correctly (http://speakgoodchinese.org/). Students pronounce a word spelled in pinyin notation and receive feedback from our system on their production of the tones. The novelty in our approach lies in the use of synthetic reference tone(s) produced from the pinyin notation. Preliminary results indicate a 6% rejection rate for six words, read multiple times, by three reference speakers and less than 15% acceptance rate on incorrectly produced tones on shadowed versions of these words by 8 speakers. With speech from 4 reference speakers collected with a fully functional test application, a rejection rate of less than 15% was achieved.

Interspeech 2007, 2007

In this paper, we evaluate the adequacy of two widely used machine learning algorithms and a comp... more In this paper, we evaluate the adequacy of two widely used machine learning algorithms and a computational linguistic proposal to model L2 perceptual development. The three proposals are, in order, Nearest Neighbor, Naive Bayesian and Stochastic OT and the Gradual Learning Algorithm. We compared the three models' outputs to those of Spanish learners of Dutch who were asked to categorize synthetic stimuli as one of the 12 Dutch vowels. The empirical results of the human learners show that L2 learners differ significantly from native listeners, but also that their perceptual spaces tend to become more native-like with L2 proficiency. The results of the simulations show that all three algorithms are able to model listeners' data to a certain extent but that Stochastic OT and the Gradual Learning Algorithm, i.e. the linguistic model, best reproduces L1 and L2 data. * The first three authors' names are in alphabetical order. Kastelein and Weiand conducted the simulations and Escudero contributed with the empirical data, guidance for simulations, overview of the research, and writing of articles. All four authors contributed to the development of the mathematical analysis, while Van Son and Escudero conducted the statistical analysis.

Interspeech 2007, 2007

It is well known that listeners can ignore disturbances in speech and rely on context to interpol... more It is well known that listeners can ignore disturbances in speech and rely on context to interpolate the message. This fact is used to determine the importance of individual words for projecting Transition Relevance Places, TRPs. Subjects were asked to shadow manipulated pre-recorded dialogs with minimal responses, saying 'ah' when they feel it is appropriate. In these dialogs, at random, of each utterance, either one of the last four words was replaced by white noise (masked condition), or no word was replaced (non masked condition). The reaction times were analyzed for effects of masked words. The presence of masked words, even prominent words, did not affect the response times of our subjects unless the very last word of the utterance was masked. This indicates that listeners are able to seamlessly interpolate the missing words and only need the identity of the last word to determine the exact position of the TRP.

International Journal of Education and Practice, 2019

This study attempts to forecast the results of school students in Vietnam by geographical region ... more This study attempts to forecast the results of school students in Vietnam by geographical region analysis. Results show variations between regions. In the Red River Delta region, it is forecast that in the decade 2015 to 2025, the number of children attending to school will increase slightly. In the Midlands and Northern Mountains region, the average number of children attending school should increase gradually by 2.4 percent over five years. In the North Central and Central Coast regions, student numbers have increased and decreased erratically. In the Central Highlands, there has been an average rate increase of 3.45 percent over five years. In the Southeast region, student numbers at all levels are expected to increase by an average of 3.9 per cent over five years. In the Mekong Delta region, student numbers are expected to increase by one percent over five years but experience a 1.8 percent reduction by 2035. These data will be critical to planning for the education sector in the coming decades. Contribution/Originality: This study contributes to the existing literature by use of a new estimation methodology for analyzing and comparing student numbers at each education level within an appropriate timeframe.

The Journal of Social Sciences Research, 2019

Gender is considered as one of the important factors influencing the development of society, rela... more Gender is considered as one of the important factors influencing the development of society, relating to many fields such as economy, politics, culture, education, etc. For the past years, Vietnam has actively promoted the gender equality, including the adjustment of the gender gap between the number of the children born throughout the years. However, the forecast results of the number of the children and students attending to school in Vietnam showed that there was quite large gender gap between boys and girls. This leads to the imbalance of gender; as a result, the assurance of the conditions for the education development becomes a challenge. Based on the analysis of the forecast results, the paper mentioned the responsibility of the education managers on the issue of the gender gap in the development of education and sustainable development.

J Amer Soc Inform Sci, 2005

In both human-human and human-machine conversation, an important task for the participants is to ... more In both human-human and human-machine conversation, an important task for the participants is to identify the moment the other participant finishes speaking, giving them the possibility of taking over the turn in talk. In an RT experiment, consistent evidence was found for an intermediate stage in the planning and articulation of elicited minimal responses in the shape of early larynx and glottal movements in laryngograph recordings. Using a simple Response Time model, it is estimated that this intermediate stage occurs at approximately 2/3 of the integration-time needed for the articulation of a response. Impoverished intonation only stimuli were still adequate to elicit minimal responses, but a longer integration-time was required to initiate a response.

Speech Communication, 1993

ABSTRACT

Speaking is generally considered efficient in that less effort is spent articulating more redunda... more Speaking is generally considered efficient in that less effort is spent articulating more redundant items. With efficient speech production, less reduction is expected in the pronunciation of phonemes that are more important (distinctive) for word identification. The importance of a single phoneme in word recognition can be quantified as the information (in bits) it adds to the preceding word onset to narrow down the context corrected lexical search. In our study, segmental information showed to correlate consistently with both duration and spectral reduction in vowels and most consonants. No such correlations were found for stops and only little for nasals. This correlation was found after accounting for speaker and vowel identity, speaking style, lexical stress, modeled prominence, position in the syllable, and position of the phoneme in the word. We conclude that speech is organized for efficiency at the level of the phoneme.

Attacks on cryptographic systems are limited by the available computational resources. A theoreti... more Attacks on cryptographic systems are limited by the available computational resources. A theoretical understanding of these resource limitations is needed to evaluate the security of cryptographic primitives and procedures. This study uses an Attacker versus Environment game formalism based on computability logic to quantify Shannon's work function and evaluate resource use in cryptanalysis. A simple cost function is defined which allows to quantify a wide range of theoretical and real computational resources. With this approach the use of custom hardware, e.g., FPGA boards, in cryptanalysis can be analyzed. Applied to real cryptanalytic problems, it raises, for instance, the expectation that the computer time needed to break some simple 90 bit strong cryptographic primitives might theoretically be less than two years. Comment: 19 pages

Review of Income and Wealth - REV INCOME WEALTH, 2001

At the Institute of Phonetic Sciences (IFA) we have collected a corpus of spoken Dutch of 4 male ... more At the Institute of Phonetic Sciences (IFA) we have collected a corpus of spoken Dutch of 4 male and 4 female speakers, containing informal as well as read speech, plus lists of sentences, words, and syllables taken from the transcribed conversation text, and then spoken in isolation. This pertains to about 5.5 hours of speech. All this material is segmented and labeled at the phoneme level. This information plus all meta data are stored in a relational database which makes all material accessible through SQL. All information is freely available under the GNU General Public License. This material will also be used in INTAS project 915, in which a comparison will be made of phonetic properties in Dutch, Finnish and Russian. As an initial result we will present some durational and spectral data of full and reduced phoneme realizations.

Physics Letters B, 2011

A search for narrow resonances with a mass of at least 1 TeV in the dijet mass spectrum is perfor... more A search for narrow resonances with a mass of at least 1 TeV in the dijet mass spectrum is performed using pp collisions at √ s = 7 TeV corresponding to an integrated luminosity of 1 fb −1 , collected by the CMS experiment at the LHC. No resonances are observed. Upper limits at the 95% confidence level are presented on the product of the resonance cross section, branching fraction into dijets, and acceptance, separately for decays into quark-quark, quark-gluon, and gluon-gluon pairs. The data exclude new particles predicted in the following models at the 95% confidence level: string resonances with mass less than 4.00 TeV, E 6 diquarks with mass less than 3.52 TeV, excited quarks with mass less than 2.49 TeV, axigluons and colorons with mass less than 2.47 TeV, and W bosons with mass less than 1.51 TeV. These results extend previous exclusions from the dijet mass search technique.

dare.uva.nl, 2010

• Use LPC analysis-synthesis to manipulate tracheoesophageal speech • On-line experiment with exp... more • Use LPC analysis-synthesis to manipulate tracheoesophageal speech • On-line experiment with expert judges • Rate perceived intelligibility (7-point scale) • Modeling the source amplitude improved speech most • Regularizing pitch had no effect • Using a fully synthetic voice source decreased intelligibility

4th European Conference on Speech Communication and Technology (Eurospeech 1995)

Conference of the International Speech Communication Association, Aug 27, 2007

We describe the SpeakGoodChinese system that supports beginning students of Mandarin Chinese to p... more We describe the SpeakGoodChinese system that supports beginning students of Mandarin Chinese to produce tones correctly (http://speakgoodchinese.org/). Students pronounce a word spelled in pinyin notation and receive feedback from our system on their production of the tones. The novelty in our approach lies in the use of synthetic reference tone(s) produced from the pinyin notation. Preliminary results indicate a 6% rejection rate for six words, read multiple times, by three reference speakers and less than 15% acceptance rate on incorrectly produced tones on shadowed versions of these words by 8 speakers. With speech from 4 reference speakers collected with a fully functional test application, a rejection rate of less than 15% was achieved.

Interspeech 2007, 2007

In this paper, we evaluate the adequacy of two widely used machine learning algorithms and a comp... more In this paper, we evaluate the adequacy of two widely used machine learning algorithms and a computational linguistic proposal to model L2 perceptual development. The three proposals are, in order, Nearest Neighbor, Naive Bayesian and Stochastic OT and the Gradual Learning Algorithm. We compared the three models' outputs to those of Spanish learners of Dutch who were asked to categorize synthetic stimuli as one of the 12 Dutch vowels. The empirical results of the human learners show that L2 learners differ significantly from native listeners, but also that their perceptual spaces tend to become more native-like with L2 proficiency. The results of the simulations show that all three algorithms are able to model listeners' data to a certain extent but that Stochastic OT and the Gradual Learning Algorithm, i.e. the linguistic model, best reproduces L1 and L2 data. * The first three authors' names are in alphabetical order. Kastelein and Weiand conducted the simulations and Escudero contributed with the empirical data, guidance for simulations, overview of the research, and writing of articles. All four authors contributed to the development of the mathematical analysis, while Van Son and Escudero conducted the statistical analysis.

Interspeech 2007, 2007

It is well known that listeners can ignore disturbances in speech and rely on context to interpol... more It is well known that listeners can ignore disturbances in speech and rely on context to interpolate the message. This fact is used to determine the importance of individual words for projecting Transition Relevance Places, TRPs. Subjects were asked to shadow manipulated pre-recorded dialogs with minimal responses, saying 'ah' when they feel it is appropriate. In these dialogs, at random, of each utterance, either one of the last four words was replaced by white noise (masked condition), or no word was replaced (non masked condition). The reaction times were analyzed for effects of masked words. The presence of masked words, even prominent words, did not affect the response times of our subjects unless the very last word of the utterance was masked. This indicates that listeners are able to seamlessly interpolate the missing words and only need the identity of the last word to determine the exact position of the TRP.

International Journal of Education and Practice, 2019

This study attempts to forecast the results of school students in Vietnam by geographical region ... more This study attempts to forecast the results of school students in Vietnam by geographical region analysis. Results show variations between regions. In the Red River Delta region, it is forecast that in the decade 2015 to 2025, the number of children attending to school will increase slightly. In the Midlands and Northern Mountains region, the average number of children attending school should increase gradually by 2.4 percent over five years. In the North Central and Central Coast regions, student numbers have increased and decreased erratically. In the Central Highlands, there has been an average rate increase of 3.45 percent over five years. In the Southeast region, student numbers at all levels are expected to increase by an average of 3.9 per cent over five years. In the Mekong Delta region, student numbers are expected to increase by one percent over five years but experience a 1.8 percent reduction by 2035. These data will be critical to planning for the education sector in the coming decades. Contribution/Originality: This study contributes to the existing literature by use of a new estimation methodology for analyzing and comparing student numbers at each education level within an appropriate timeframe.

The Journal of Social Sciences Research, 2019

Gender is considered as one of the important factors influencing the development of society, rela... more Gender is considered as one of the important factors influencing the development of society, relating to many fields such as economy, politics, culture, education, etc. For the past years, Vietnam has actively promoted the gender equality, including the adjustment of the gender gap between the number of the children born throughout the years. However, the forecast results of the number of the children and students attending to school in Vietnam showed that there was quite large gender gap between boys and girls. This leads to the imbalance of gender; as a result, the assurance of the conditions for the education development becomes a challenge. Based on the analysis of the forecast results, the paper mentioned the responsibility of the education managers on the issue of the gender gap in the development of education and sustainable development.

J Amer Soc Inform Sci, 2005

In both human-human and human-machine conversation, an important task for the participants is to ... more In both human-human and human-machine conversation, an important task for the participants is to identify the moment the other participant finishes speaking, giving them the possibility of taking over the turn in talk. In an RT experiment, consistent evidence was found for an intermediate stage in the planning and articulation of elicited minimal responses in the shape of early larynx and glottal movements in laryngograph recordings. Using a simple Response Time model, it is estimated that this intermediate stage occurs at approximately 2/3 of the integration-time needed for the articulation of a response. Impoverished intonation only stimuli were still adequate to elicit minimal responses, but a longer integration-time was required to initiate a response.

Speech Communication, 1993

ABSTRACT

Speaking is generally considered efficient in that less effort is spent articulating more redunda... more Speaking is generally considered efficient in that less effort is spent articulating more redundant items. With efficient speech production, less reduction is expected in the pronunciation of phonemes that are more important (distinctive) for word identification. The importance of a single phoneme in word recognition can be quantified as the information (in bits) it adds to the preceding word onset to narrow down the context corrected lexical search. In our study, segmental information showed to correlate consistently with both duration and spectral reduction in vowels and most consonants. No such correlations were found for stops and only little for nasals. This correlation was found after accounting for speaker and vowel identity, speaking style, lexical stress, modeled prominence, position in the syllable, and position of the phoneme in the word. We conclude that speech is organized for efficiency at the level of the phoneme.

Attacks on cryptographic systems are limited by the available computational resources. A theoreti... more Attacks on cryptographic systems are limited by the available computational resources. A theoretical understanding of these resource limitations is needed to evaluate the security of cryptographic primitives and procedures. This study uses an Attacker versus Environment game formalism based on computability logic to quantify Shannon's work function and evaluate resource use in cryptanalysis. A simple cost function is defined which allows to quantify a wide range of theoretical and real computational resources. With this approach the use of custom hardware, e.g., FPGA boards, in cryptanalysis can be analyzed. Applied to real cryptanalytic problems, it raises, for instance, the expectation that the computer time needed to break some simple 90 bit strong cryptographic primitives might theoretically be less than two years. Comment: 19 pages

Review of Income and Wealth - REV INCOME WEALTH, 2001

At the Institute of Phonetic Sciences (IFA) we have collected a corpus of spoken Dutch of 4 male ... more At the Institute of Phonetic Sciences (IFA) we have collected a corpus of spoken Dutch of 4 male and 4 female speakers, containing informal as well as read speech, plus lists of sentences, words, and syllables taken from the transcribed conversation text, and then spoken in isolation. This pertains to about 5.5 hours of speech. All this material is segmented and labeled at the phoneme level. This information plus all meta data are stored in a relational database which makes all material accessible through SQL. All information is freely available under the GNU General Public License. This material will also be used in INTAS project 915, in which a comparison will be made of phonetic properties in Dutch, Finnish and Russian. As an initial result we will present some durational and spectral data of full and reduced phoneme realizations.

Physics Letters B, 2011

A search for narrow resonances with a mass of at least 1 TeV in the dijet mass spectrum is perfor... more A search for narrow resonances with a mass of at least 1 TeV in the dijet mass spectrum is performed using pp collisions at √ s = 7 TeV corresponding to an integrated luminosity of 1 fb −1 , collected by the CMS experiment at the LHC. No resonances are observed. Upper limits at the 95% confidence level are presented on the product of the resonance cross section, branching fraction into dijets, and acceptance, separately for decays into quark-quark, quark-gluon, and gluon-gluon pairs. The data exclude new particles predicted in the following models at the 95% confidence level: string resonances with mass less than 4.00 TeV, E 6 diquarks with mass less than 3.52 TeV, excited quarks with mass less than 2.49 TeV, axigluons and colorons with mass less than 2.47 TeV, and W bosons with mass less than 1.51 TeV. These results extend previous exclusions from the dijet mass search technique.

dare.uva.nl, 2010

• Use LPC analysis-synthesis to manipulate tracheoesophageal speech • On-line experiment with exp... more • Use LPC analysis-synthesis to manipulate tracheoesophageal speech • On-line experiment with expert judges • Rate perceived intelligibility (7-point scale) • Modeling the source amplitude improved speech most • Regularizing pitch had no effect • Using a fully synthetic voice source decreased intelligibility