S. Suzic, Tijana Delic, S. Ostrogonac, Simona Đurić, D. Pekar
Modern text-to-speech systems generally achieve good intelligibility. The one of the main drawbacks of these systems is the lack of expressiveness in comparison to natural human speech. It is very unpleasant when automated system conveys positive and negative message in completely the same way. The introduction of parametric methods in speech synthesis gave possibility to easily change speaker characteristics and speaking styles. In this paper a simple method for incorporating styles into synthesized speech by using style codes is presented. The proposed method requires just a couple of minutes of target style and moderate amount of neutral speech. It is successfully applied to both hidden Markov models and deep neural networks-based synthesis, giving style code as additional input to the model. Listening tests confirmed that better style expressiveness is achieved by deep neural networks synthesis compared to hidden Markov model synthesis. It is also proved that quality of speech synthesized by deep neural networks in a certain style is comparable with the speech synthesized in neutral style, although the neutral-speech-database is about 10 times bigger. DNN based TTS with style codes are further investigated by comparing the quality of speech produced by single-style modeling and multi-style modeling systems. Objective and subjective measures confirmed that there is no significant difference between these two approaches.
{"title":"Style-Code Method for Multi-Style Parametric Text-to-Speech Synthesis","authors":"S. Suzic, Tijana Delic, S. Ostrogonac, Simona Đurić, D. Pekar","doi":"10.15622/sp.60.8","DOIUrl":"https://doi.org/10.15622/sp.60.8","url":null,"abstract":"Modern text-to-speech systems generally achieve good intelligibility. The one of the main drawbacks of these systems is the lack of expressiveness in comparison to natural human speech. It is very unpleasant when automated system conveys positive and negative message in completely the same way. The introduction of parametric methods in speech synthesis gave possibility to easily change speaker characteristics and speaking styles. In this paper a simple method for incorporating styles into synthesized speech by using style codes is presented. The proposed method requires just a couple of minutes of target style and moderate amount of neutral speech. It is successfully applied to both hidden Markov models and deep neural networks-based synthesis, giving style code as additional input to the model. Listening tests confirmed that better style expressiveness is achieved by deep neural networks synthesis compared to hidden Markov model synthesis. It is also proved that quality of speech synthesized by deep neural networks in a certain style is comparable with the speech synthesized in neutral style, although the neutral-speech-database is about 10 times bigger. DNN based TTS with style codes are further investigated by comparing the quality of speech produced by single-style modeling and multi-style modeling systems. Objective and subjective measures confirmed that there is no significant difference between these two approaches.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76213341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Kozachok, S. Kopylov, R. Meshcheryakov, O. Evsutin, L. M. Tuan
{"title":"An approach to a robust watermark extraction from images containing text","authors":"A. Kozachok, S. Kopylov, R. Meshcheryakov, O. Evsutin, L. M. Tuan","doi":"10.15622/SP.60.5","DOIUrl":"https://doi.org/10.15622/SP.60.5","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87324232","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Dual Optimization of Monochrome Images Tone Approximation using Parallel Evolutionarily Genetic Search","authors":"Rudlof Anatolyevich Neydorf, A. Aghajanyan","doi":"10.15622/sp.60.6","DOIUrl":"https://doi.org/10.15622/sp.60.6","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"13 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75439327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yoshinov R., Iliev O. The Structural Way for Binding a Learning Material with Personal Preferences of Learners. Abstract. Learning content creation process requires more than just collection and presentation of set of information. In order to gain knowledge, the learning content should be designed in such a way to meet predefined learning goals. Learning goals determine the entire process of learning. Bloom’s Taxonomy provides a description of a cognitive process with six hierarchical levels, each containing specific learning goal to achieve. It could be adapted into a model by which tutors create learning materials. However, when it comes to productivity of learning, it is important to consider the personalization of the presented content according to the learning style of the individual. This article analyzes the correlation between Bloom’s Taxonomy and Honey & Mumford’s learning cycle, providing a way to bind the structure of learning material to the personal preferences of learners. This novel way of creating learning materials is integrated into a model that is used for automatic generation of personalized learning materials. The effectiveness of the model is further verified through an experiment with real participants. The results of the experiment show promising potential in the way of how a learner’s capabilities may be enriched. However, while experimenting and rest of the work on the model outline some challenges before the model’s application and future work.
{"title":"The Structural Way for Binding a Learning Material with Personal Preferences of Learners","authors":"R. Yoshinov, O. Iliev","doi":"10.15622/sp.60.7","DOIUrl":"https://doi.org/10.15622/sp.60.7","url":null,"abstract":"Yoshinov R., Iliev O. The Structural Way for Binding a Learning Material with Personal Preferences of Learners. Abstract. Learning content creation process requires more than just collection and presentation of set of information. In order to gain knowledge, the learning content should be designed in such a way to meet predefined learning goals. Learning goals determine the entire process of learning. Bloom’s Taxonomy provides a description of a cognitive process with six hierarchical levels, each containing specific learning goal to achieve. It could be adapted into a model by which tutors create learning materials. However, when it comes to productivity of learning, it is important to consider the personalization of the presented content according to the learning style of the individual. This article analyzes the correlation between Bloom’s Taxonomy and Honey & Mumford’s learning cycle, providing a way to bind the structure of learning material to the personal preferences of learners. This novel way of creating learning materials is integrated into a model that is used for automatic generation of personalized learning materials. The effectiveness of the model is further verified through an experiment with real participants. The results of the experiment show promising potential in the way of how a learner’s capabilities may be enriched. However, while experimenting and rest of the work on the model outline some challenges before the model’s application and future work.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"246 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75101276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Methods of sonar signal processing to solve the sensing bottom surface problem","authors":"A. S. Mironov, E. Fomina","doi":"10.15622/sp.59.6","DOIUrl":"https://doi.org/10.15622/sp.59.6","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"109 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78394386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Learning Prosodic Stress from Data in Neural Network based Text-to-Speech Synthesis","authors":"M. Secujski, S. Ostrogonac, S. Suzic, D. Pekar","doi":"10.15622/sp.59.8","DOIUrl":"https://doi.org/10.15622/sp.59.8","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"48 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73780873","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
M. Peregudov, A. Steshkovoy, Aleksey Aleksandrovich Boyko
{"title":"Probabilistic Random Multiple Access Procedure Model to the CSMA/CA Type Medium","authors":"M. Peregudov, A. Steshkovoy, Aleksey Aleksandrovich Boyko","doi":"10.15622/sp.59.4","DOIUrl":"https://doi.org/10.15622/sp.59.4","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"57 8 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77628742","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Teilans, A. Romānovs, Y. Merkuryev, Pjotrs Dorogovs, A. Kleins, S. Potryasaev
Nowadays, the systems developed to integrate real physical processes and virtual computational processes — the cyber-physical systems (CPS), are used in multiple areas of industry and critical national infrastructure, such as manufacturing, medicine, traffic management and security, automotive engineering, industrial process control, energy saving, ecological management, industrial robots, technical infrastructure management, distributed robotic systems, protection target systems, nanotechnology and biological systems technology. With wide use, the level of IT and cyberrisks increases drastically and successful attacks against the CPS will lead to unmanageable and unimaginable consequence. Thus, the need in well-designed risk assessment system of CPS is clear and such system can provide an overall view of CPS security status and support efficient allocations of safeguard resources. The nature of CPS differs from IT mainly with the requirement for real-time operations, thus, traditional risk assessment method for IT system can be adopted in CPS. Design of a unified modelling language based domain specific language described in this paper achieves synergy from in IT industry widely used UML modelling technique and the domain specific risk management extensions. As a novelty for UML modelling, especially for simulation purposes, the presented DSL is enriched by a set of stochastic attributes of modelled activities. Such stochastic attributes are usable for further implementation of discrete-event system simulators.
{"title":"Assessment of Cyber Physical System Risks with Domain Specific Modelling and Simulation","authors":"A. Teilans, A. Romānovs, Y. Merkuryev, Pjotrs Dorogovs, A. Kleins, S. Potryasaev","doi":"10.15622/SP.59.5","DOIUrl":"https://doi.org/10.15622/SP.59.5","url":null,"abstract":"Nowadays, the systems developed to integrate real physical processes and virtual computational processes — the cyber-physical systems (CPS), are used in multiple areas of industry and critical national infrastructure, such as manufacturing, medicine, traffic management and security, automotive engineering, industrial process control, energy saving, ecological management, industrial robots, technical infrastructure management, distributed robotic systems, protection target systems, nanotechnology and biological systems technology. With wide use, the level of IT and cyberrisks increases drastically and successful attacks against the CPS will lead to unmanageable and unimaginable consequence. Thus, the need in well-designed risk assessment system of CPS is clear and such system can provide an overall view of CPS security status and support efficient allocations of safeguard resources. The nature of CPS differs from IT mainly with the requirement for real-time operations, thus, traditional risk assessment method for IT system can be adopted in CPS. Design of a unified modelling language based domain specific language described in this paper achieves synergy from in IT industry widely used UML modelling technique and the domain specific risk management extensions. As a novelty for UML modelling, especially for simulation purposes, the presented DSL is enriched by a set of stochastic attributes of modelled activities. Such stochastic attributes are usable for further implementation of discrete-event system simulators.","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"16 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86069690","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Parallel Linear Generator of Multivalued Pseudorandom Sequences with Operation Errors Control","authors":"D. Samoylenko, M. Eremeev, O. Finko, S. Dichenko","doi":"10.15622/SP.59.2","DOIUrl":"https://doi.org/10.15622/SP.59.2","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"26 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86064936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The Synthesis of Self-Checking Combinational Devices on the Basis of Codes with the Effective Symmetrical Error Detection","authors":"D. Efanov","doi":"10.15622/sp.59.3","DOIUrl":"https://doi.org/10.15622/sp.59.3","url":null,"abstract":"","PeriodicalId":53447,"journal":{"name":"SPIIRAS Proceedings","volume":"55 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2018-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73268431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：