|
|
Query-by-Example Spoken Term Detection
Fapšo, Michal ; Matoušek, Jindřich (oponent) ; Metze, Florian (oponent) ; Černocký, Jan (vedoucí práce)
This thesis investigates query-by-example (QbE) spoken term detection (STD). Queries are entered in their spoken form and searched for in a pool of recorded spoken utterances, providing a list of detections with their scores and timing. We describe, analyze and compare three different approaches to QbE STD, in various language-dependent and language-independent setups with diverse audio conditions, searching for a single example and five examples per query. For our experiments we used Czech, Hungarian, English and Levantine data and for each of the languages we trained a 3-state phone posterior estimator. This gave us 16 possible combinations of the evaluation language and the language of the posterior estimator, out of which 4 combinations were language-dependent and 12 were language-independent. All QbE systems were evaluated on the same data and the same features, using the metrics: non-pooled Figure-of-Merit and our proposed utterrance-normalized non-pooled Figure-of-Merit, which provided us with relevant data for the comparison of these QbE approaches and for gaining a better insight into their behavior. QbE approaches presented in this work are: sequential statistical modeling (GMM/HMM), template matching of features (DTW) and matching of phone lattices (WFST). To compare the performance of QbE approaches with the common query-by-text STD systems, for language-dependent setups we also evaluated an acoustic keyword spotting system (AKWS) and a system searching for phone strings in lattices (WFSTlat). The core of this thesis is the development, analysis and improvement of the WFST QbE STD system, which after the improvements, achieved similar performance to the DTW system in language-dependent setups.
|
|
|
Efficient Reduction of Finite Automata
Molnárová, Veronika ; Havlena, Vojtěch (oponent) ; Lengál, Ondřej (vedoucí práce)
A finite state automaton is a mathematical model used to describe a machine that performs a computation on the given input over a series of states. In the last century, it has found many uses in different fields of information technology, from video game character behavior to compilers. While each automaton denotes its language, one language can be represented by an infinite number of different automata. As these automata vary in size, to ensure the most efficient work with them, we want to find the smallest one possible. In this thesis, we are going to look at five different types of automata reductions. Firstly, we will talk about three known reduction algorithms, which are the minimization of deterministic automata, the reduction based on a relation of simulation, and the reduction by transformation into a canonical residual automaton. These reductions were implemented in C++ and tested on a sample set of automata to compare their results. Lastly, we looked at the possibility of reducing finite state automata using Boolean satisfiability problem (SAT) and quantified Boolean formula (QBF) solvers. We are presenting a set of rules for each solver for generating a clause in conjunctive normal form (CNF), which can precisely represent the given automaton in Boolean algebra. We used this fact to create a new method of nondeterministic automata reduction.
|
|
|
Query-by-Example Spoken Term Detection
Fapšo, Michal ; Matoušek, Jindřich (oponent) ; Metze, Florian (oponent) ; Černocký, Jan (vedoucí práce)
This thesis investigates query-by-example (QbE) spoken term detection (STD). Queries are entered in their spoken form and searched for in a pool of recorded spoken utterances, providing a list of detections with their scores and timing. We describe, analyze and compare three different approaches to QbE STD, in various language-dependent and language-independent setups with diverse audio conditions, searching for a single example and five examples per query. For our experiments we used Czech, Hungarian, English and Levantine data and for each of the languages we trained a 3-state phone posterior estimator. This gave us 16 possible combinations of the evaluation language and the language of the posterior estimator, out of which 4 combinations were language-dependent and 12 were language-independent. All QbE systems were evaluated on the same data and the same features, using the metrics: non-pooled Figure-of-Merit and our proposed utterrance-normalized non-pooled Figure-of-Merit, which provided us with relevant data for the comparison of these QbE approaches and for gaining a better insight into their behavior. QbE approaches presented in this work are: sequential statistical modeling (GMM/HMM), template matching of features (DTW) and matching of phone lattices (WFST). To compare the performance of QbE approaches with the common query-by-text STD systems, for language-dependent setups we also evaluated an acoustic keyword spotting system (AKWS) and a system searching for phone strings in lattices (WFSTlat). The core of this thesis is the development, analysis and improvement of the WFST QbE STD system, which after the improvements, achieved similar performance to the DTW system in language-dependent setups.
|
host ::
RSS.