|
|
Statistical Natural Language Processing Methods in Music Notation Analysis
Libovický, Jindřich ; Peterek, Nino (advisor) ; Mareček, David (referee)
The thesis summarizes the research in application of statistical methods of computational linguistics in music processing and explains theoretical background of these applications. In the second part methods of symbolic melody extraction are explored. A corpus of approxi- mately 400 hours of melodies of different music styles was created. A melody model using the language modeling techniques was trained on this corpus. In the third part of the thesis the model is used for an attempt to develop an alternative method of audio melody extraction which uses the melody model instead of commonly used heuristics and rules. The chosen ap- proach works well only on simple input data and produces worse results than the commonly used methods on the MIREX contest data. On the other hand, the experiments help to understand the conceptual between the pitch frequency development - the physical melody - and the melody perceived on an abstract level in the symbolic notation - the symbolic melody. 1
|
|
|
Design and Implementation of Sound Recognizer of Particular Grasshopper Species
Schwarz, Jan ; Peterek, Nino (advisor) ; Hlaváčová, Jaroslava (referee)
Biologists asked us to create a system that recognizes particular grasshopper species from stridulation records. Currently we recognize five grasshopper species which can be seen in the Czech Republic using a free available toolkit for speech recognition called HTK. In addition to the acoustic model itself we also created web sites, which would analyse a stridulation record and then save the result for subsequent utilization. The current model is based only on a limited amount of training records, but its results are satisfactory. The web sites also serve as a gathering system; consequently, it is possible to further extend and improve the model.
|
|
|
Voice command for a TV set
Černý, Patrik ; Straňák, Pavel (advisor) ; Peterek, Nino (referee)
Title: Voice command for a TV set Author: Patrik Černý Department: Institute of Formal and Applied Linguistics Supervisor: Mgr. Pavel Straňák, Ph.D. Abstract: A goal of this thesis is to create television voice control intended for poeple with speech and movement disorder. This is achieved by interconnecting computer and television. Voice control is based on well-known dynamic time warping algorithm. It has been shown, that due to high and frequent changes in sound intensity the voice control of television is quite a complex task. The word recognition success rate of the final application is not very high, but for the purpose sufficient. Because of application design, program can be easily extended by techniques, that can improve recognition effectivity. Keywords: voice control, word recognition, dynamic time warping, television 1
|
|
|
Unsegmented speech retrieval
Češka, Pavel ; Peterek, Nino (referee) ; Pecina, Pavel (advisor)
In this work I search through interviews of Czech witnesses of the holocaust from the MALACH project to find relevant parts of these testimonies. Audio records of these interviews are automatically recognized by a system for an automatic speech recognition. Automatically recognized texts are then lemmatized and tagged. In this work I present a script which generates parametrizable collections of documents from these preprocessed texts. The task of unsegmented speech retrieval is then reformulated to a task of information retrieval in this collections of documents. In this work, I describe many experiments which examine the influence of different retrieval techniques on retrieval results on this data collection. Mainly, I study an influence of a morphological normalization (lemmatization), different types of IR systems (TF-IDF model, Okapi model and Indri model), blind relevance feedback, stopword list based on frequencies of terms and part-of-speech categories. I also place emphasis on various values of length and overlap parameters of generated documents. The results of these experiments are verified on test data. Audio records, outputs from automatic speech recognition system and topics for information retrieval are not part of this work due to legal grounds.
|
|
|
Error detection in speech recognition
Tobolíková, Petra ; Peterek, Nino (referee) ; Hajič, Jan (advisor)
This thesis tackles the problem of error detection in speech recognition. First, principles of recent approaches to automatic speech recognition are introduced. Various deficiencies of speech recognition that cause imperfect recognition results are outlined. Current known methods of "confidence score" computation are then described. The next chapter introduces three machine learning algorithms which where employed in the error detection methods implemented in this thesis: logistic regression, artificial neural networks and decision trees. This machine learning methods use certain attributes of the recognized words as input variables and predict an estimated confidence score value. The open source software "R" has been used throughout, showing the usage of the aforementioned methods. These methods have been tested on Czech radio and TV broadcasts. The results obtained by those methods are compared using ROC curves, standard errors and possible (oracle) WER reduction. Programming documentation of the code used in the implementation is enclosed as well. Finally, efficient word attributes for error detection are summarized.
|
|
| |
|
|
Distributed Sytem for Verification of Properties of Natural Numbers
Tomisová, Martina ; Peterek, Nino (referee) ; Mírovský, Jiří (advisor)
The result of my work is a system for distributed verification of properties of natural numbers. It has two parts - server and client. These communicate via HTTP protocol. The clients perform the computation, the server distributes the work (numbers) and gather results (properties of the given numbers). The input of one computation should be one natural number, as well as the result (output). The distribution can be used for verification of a given property for several natural numbers. Particular jobs can be added to the client as plugins. Two examples of plugins are a part of the work. The first one is very simple and shows how to create plugins. The second example searches for prime numbers (and has it's own arithmetics library for long numbers) - the server can distribute (possibly big) numbers, the client will verify whether a given number is a prime number.
|
|
|
Speech Interface for Corpus Annotation Tools
Přikryl, Leoš ; Hajič, Jan (advisor) ; Peterek, Nino (referee)
The thesis considers design and implementation of the interface for the corpus annotation tools used at the Institute of Formal and Applied Linguistics (TrEd and its additional modules) in the natural language (speech). Already existing modules for automatic speech recognition from the University of West Bohemia in Pilsen are used.
|
|
|
Tools and Data for Analysis of Spoken Czech and its Prosody
Peterek, Nino ; Hajičová, Eva (advisor) ; Kopeček, Ivan (referee) ; Psutka, Josef (referee)
This work describes our steps towards prosody models of spoken Czech language. After a characterisation and discussion of recent prosody definitions and of area of prosody applications, we present the central point of the work, development of an easy-accessible and user-friendly research environment Dialogy.Org, supporting exploration of Czech prosody and its automatic analysis and modelling. Powered by TCPDF (www.tcpdf.org)
|
|
|
The past, present, and future of the DIALOG corpus
Kaderka, Petr ; Havlík, Martin ; Svobodová, Zdeňka ; Peterek, Nino ; Havlová, Eva K. ; Klímová, Jana ; Kubáčková, Patricie
The DIALOG corpus is a special corpus of spoken Czech, consisting of video recordings and transcripts of television discussions. The working form of the corpus contains more than two million words. In the introductory section of this paper, we discuss the motivation that led the researchers from the Czech Language Institute, Academy of Sciences of the Czech Republic to collect and analyze dialogical speech, and we also present an overview of the publications based on work with the corpus. In the second section, we provide information about turning the collected material into an electronic linguistic corpus and about the basic characteristics of the working version of the DIALOG corpus and its first public version, known as DIALOG 0.1 (http://ujc.dialogy.cz). In the concluding section, we present the anticipated schedule for releasing the corpus for public access; we also indicate some currently relevant areas of research that can benefit from using the DIALOG corpus.
|
guest ::
