|
|
Rozpoznávání řeči pomocí KALDI
Plátek, Ondřej ; Jurčíček, Filip (advisor) ; Peterek, Nino (referee)
The topic of this thesis is to implement efficient decoder for speech recognition training system ASR Kaldi (http://kaldi.sourceforge.net/). Kaldi is already deployed with decoders, but they are not convenient for dialogue systems. The main goal of this thesis to develop a real time decoder for a dialogue system, which minimize latency and optimize speed. Methods used for speeding up the decoder are not limited to multi-threading decoding or usage of GPU cards for general computations. Part of this work is devoted to training an acoustic model and also testing it in the "Vystadial" dialogue system. Powered by TCPDF (www.tcpdf.org)
|
|
|
Finding the answer in the answers
Záhumenský, Jakub ; Vidová Hladká, Barbora (advisor) ; Bojar, Ondřej (referee)
Title : Searching for the answer in answers Author : Jakub Záhumenský Contact : zahumensky.jakub@gmail.com Department : Institute of Formal and Applied Linguistics Supervisor : Mgr. Barbora Vidová Hladká, Ph.D. Contact on supervisor : hladka@ufal.mff.cuni.cz Abstract : We design a question-answering system Interviewer that enables users to fictionally (virtually) interview this person by asking questions as similar as possible to questions that journalists have already asked. The interviews with a given person posted on the web are being collected as a corpus of (question, answer) pairs. The user asks his/her question and the Interviewer system searches questions in the corpus to provide the answer that belongs to the most similar question. Matching questions is based on the frequency analysis and on the applications coming from natural language processing, namely tagging and parsing. We work with the interviews with Vaclav Havel posted on his personal page.
|
|
|
Mobile personal assistants
Techl, Jan ; Sigmund, Tomáš (advisor) ; Labský, Martin (referee)
This thesis focuses on analysis, definition and description of mobile personal assistants as a phenomenon emerging in past few years. Mobile personal assistants are first mentioned in the context of computational linguistics and information needs, which is one of the motivations to use them. Main interest of this thesis is an introduction of the core technologies for the natural language communication between the assistant and its user, followed by an introduction of host environments and possible usage. The thesis also presents the limitations and risks resulting from using them, which are in some ways affecting their usability. Beside the analysis the main focus is on the design and implementation of the natural language understanding (NLU) system, which can be used in particular personal assistant application. This system is implemented as a web service and consists of an annotation scheme with a set of components. The results show that the system architecture and tools used are suitable solution for the construction of a basic NLU system, which has been created and which is in the compliance with the requested parameters. It is still difficult task to achieve high precision, which depends on many factors including the amount of training data, which was very small in this case. However, the resulting application is a solid starting point for its further development and extensions.
|
guest ::
