|
| |
|
|
Speech Recognition of Czech Using Finite-State Machines
Podveský, Petr ; Hajič, Jan (advisor) ; Psutka, Josef (referee) ; Krbec, Pavel (referee)
Speech recognition has become a thriving field with many real-life applications. Voice dialing in cell phones, voice control in embedded devices, speech-driven interactive manuals and many other utilities rely on solid speech recognition software. We believe that research in speech recognition can boost performance of many applications related to the area. The thesis concentrates on automatic large-vocabulary continuous-speech recognition of Czech. Czech differs from English in a few aspects. We focus on these differences and propose new language-depended techniques. Namely rich morphology is investigated and its impact on speech recognition is studied. Out-of-vocabulary (OOV) words are identified as one of the major sources deteriorating recognition performace. New language modeling techniques are proposed to alleviate the problem of OOV words. The proposed language models are tested in speech recognition systems on diverse speech corpora. The obtained results validate the original approach to language modeling. Significant overall speech recognition improvement is observed.
|
|
|
Verb Valency Frames Disambiguation
Semecký, Jiří ; Hajič, Jan (advisor) ; Krbec, Pavel (referee) ; Lopatková, Markéta (referee)
Semantic analysis has become a bottleneck of many natural language applications. Machine translation, automatic question answering, dialog management, and others rely on high quality semantic analysis. Verbs are central elements of clauses with strong influence on the realization of whole sentences. Therefore the semantic analysis of verbs plays a key role in the analysis of natural language. We believe that solid disambiguation of verb senses can boost the performance of many real-life applications. In this thesis, we investigate the potential of statistical disambiguation of verb senses. Each verb occurrence can be described by diverse types of information. We investigate which information is worth considering when determining the sense of verbs. Different types of classification methods are tested with regard to the topic. In particular, we compared the Naive Bayes classifier, decision trees, rule-based method, maximum entropy, and support vector machines. The proposed methods are thoroughly evaluated on two different Czech corpora, VALEVAL and the Prague Dependency Treebank. Significant improvement over the baseline is observed.
|
|
| |
|
|
Language Modeling for Speech Recognition of Czech
Krbec, Pavel
In this thesis, I have designed and implernented a new language model for speerh recognition. The innovative part of the language model is the integration of the HMM-tagger component designed by myself. The HMM tagger can be used as a stand-a.lone disambiguation tool and, when combined with the hand written rules, it is currently the best disarnbiguation tool for Czech language in terrns of error rate. I have perforrned a speech recognition experiment on a highly inflectional language (Czech) where I tested the proposed language model. I have shown that the accuracy of the novel language model outperforms other state--of---the--·-art Czech language models. Powered by TCPDF (www.tcpdf.org)
|
|
|
Research Data Analysis Based on the Collection of Dissertation Theses of Charles University in Prague with Regard to Long-term Digital Preservation
Pavlásková, Eliška ; Krbec, Pavel (advisor) ; Ivánek, Jiří (referee) ; Bartošek, Miroslav (referee)
This dissertation thesis focuses on research data and their use in academics from the point of view of long-term preservation. It maps usage of research data at Charles University in Prague, analyses them and lays the foundation for further research. The first part of the text focuses on the theory of long term preservation and describes the most relevant concepts regarding users, storage and structure of research data. The second part is devoted directly to research data. It consists of definition of research data, the short explanation of their importance and sources, and the model of their lifecycle. The pivotal part of the thesis is the description and the results of the research itself. The research was conducted during the year 2015 and was based on a sample of dissertation theses from the collections of Charles University in Prague. Collected data were analysed by the methods of content analysis and grounded theory. Results are presented in two main parts - content analysis results with regard to differences among science, social science and humanities, and qualitative analysis results. Powered by TCPDF (www.tcpdf.org)
|
|
|
Verb Valency Frames Disambiguation
Semecký, Jiří ; Hajič, Jan (advisor) ; Krbec, Pavel (referee) ; Lopatková, Markéta (referee)
Semantic analysis has become a bottleneck of many natural language applications. Machine translation, automatic question answering, dialog management, and others rely on high quality semantic analysis. Verbs are central elements of clauses with strong influence on the realization of whole sentences. Therefore the semantic analysis of verbs plays a key role in the analysis of natural language. We believe that solid disambiguation of verb senses can boost the performance of many real-life applications. In this thesis, we investigate the potential of statistical disambiguation of verb senses. Each verb occurrence can be described by diverse types of information. We investigate which information is worth considering when determining the sense of verbs. Different types of classification methods are tested with regard to the topic. In particular, we compared the Naive Bayes classifier, decision trees, rule-based method, maximum entropy, and support vector machines. The proposed methods are thoroughly evaluated on two different Czech corpora, VALEVAL and the Prague Dependency Treebank. Significant improvement over the baseline is observed.
|
|
| |
|
| |
|
|
Speech Recognition of Czech Using Finite-State Machines
Podveský, Petr ; Hajič, Jan (advisor) ; Psutka, Josef (referee) ; Krbec, Pavel (referee)
Speech recognition has become a thriving field with many real-life applications. Voice dialing in cell phones, voice control in embedded devices, speech-driven interactive manuals and many other utilities rely on solid speech recognition software. We believe that research in speech recognition can boost performance of many applications related to the area. The thesis concentrates on automatic large-vocabulary continuous-speech recognition of Czech. Czech differs from English in a few aspects. We focus on these differences and propose new language-depended techniques. Namely rich morphology is investigated and its impact on speech recognition is studied. Out-of-vocabulary (OOV) words are identified as one of the major sources deteriorating recognition performace. New language modeling techniques are proposed to alleviate the problem of OOV words. The proposed language models are tested in speech recognition systems on diverse speech corpora. The obtained results validate the original approach to language modeling. Significant overall speech recognition improvement is observed.
|
guest ::
RSS feed.