|
|
Convolutional Networks for Lip Reading
Kadleček, Josef ; Kišš, Martin (referee) ; Hradiš, Michal (advisor)
This thesis deals with current methods for automatic speech recognition and lip reading via neural networks. Furthermore it deals with similarities in the architectures of neural networks for audio and visual data and available datasets in the field of audiovisual automatic speech recognition. The main contribution of this thesis is set of experiments comparing different changes in neural network architecture and its impact on results. The thesis includes an implementation of a system for automatic speech recognition from audio (CER: 12.6 %) and visual (CER: 57,7 %) data. The architectures of both systems are based on features extraction via convolutional networks followed by recurrent layers LSTM, another layer of convolutions and loss function CTC.
|
|
|
Speech Recognition Algorithms in FPGA/DSP
Urbiš, Oldřich ; Herout, Adam (referee) ; Szőke, Igor (advisor)
This master's thesis deals with design of speech recognition algorithms with consideration of target technology, which is platform combinating digital signal processing and field programmable gate array. Algorithms for speech recognition includes: feature extraction of Melfrequency cepstral coefficients, hidden Markov models and their evaluation by Viterbi algorithm.
|
|
|
Language Modeling for Spech Recognition in Czech
Mikolov, Tomáš ; Černocký, Jan (referee) ; Smrž, Pavel (advisor)
This work concerns the problematic of language modeling in automatic speech recognition. Currently widely used techniques for advanced language modeling based on statistical approach are described in the first part of work - class based language models, factored language models and neural network based language models. In the next section, implementation of neural network based language model is described. Results obtained on "Pražský mluvený korpus" and "Brněnský mluvený korpus" corpora (1 170 000 words) are reported, with perplexity reduction around 20%. Also, results obtained after rescoring N-best lists with spontaneous speech are reported, with absolute improvement in accuracy by more than 1%. In the conclusion, possible uses of the work are mentioned, along with possible extensions in the future. Finally, main weaknesses of current statistical language modeling techniques are described.
|
|
|
Keyword Spotting Implementation to Mobil Phone (Symbian 60)
Cipr, Tomáš ; Schwarz, Petr (referee) ; Szőke, Igor (advisor)
Keyword spotting is one of the many applications of automatic speech recognition. Its purpose is determining spots in given utterance in which some of the specified words were spoken. Keyword spotting has a great potential to enhance performance of new applications as well as the existing ones. An example could be a mobile phone voice control. Due to OS Symbian's coming to the market it is even possible for end user to implement a keyword spotting for a mobile phone on his or her own. The thesis describes theoretical prerequisites for keyword spotting and its implementation. Firstly the OS Symbian is presented with respect to the given task. Secondly each step of keyword spotting process is described. Finally the object design of keyword spotter is presented followed by implementation description. The thesis concludes with results review and notes on possible improvements.
|
|
|
New Techniques in Neural Networks Training - Connectionist Temporal Classification
Gajdár, Matúš ; Švec, Ján (referee) ; Karafiát, Martin (advisor)
This bachelor’s thesis deals with neural network and their use in speech recognition. Firstly,there is some theory about speech recognition, afterwards we show theory around neural networks in connection with connectionist temporal classification method. In next chapter we introduce toolkits, which were used for training of neural networks and also experiments done by them to find out impact of connectionist temporal classification method on precisionin phoneme decoding. The last chapter include summarization of work and overall evaluation of experiments.
|
|
|
Voice Sample database design for speech recognition purposes
Grobelný, Petr ; Malý, Jan (referee) ; Pfeifer, Václav (advisor)
Práce se zabývá rozpoznáváním řeči a tvorbou řečové databáze, která bude sloužit jako trénovací a testovací data pro systém rozpoznávání řeči. Daný korpus je navrhnut jako databáze čtené řeči. V teoretické části je čtenář seznámen s pojmem rozpoznávání řeči a je hlouběji uveden do problematiky. Praktická část se skládá z podrobného postupu vytvoření databáze čtené řeči. Samotná databáze je prezentována na přiloženém médiu. V závěru práce je přiložena potřebná dokumentace celé databáze.
|
|
|
Real-time voice command recognition system
Šíbl, Evžen ; Kiac, Martin (referee) ; Přinosil, Jiří (advisor)
The bachelor thesis deals with the development of a system for voice command recognition. The classifier of this system was created using a neural network. In this thesis you will learn about the history and problems of speech recognition. A system has been created that detects a section in a recording containing a speech signal, which then uses the classifier to decide what word from the word table it is. Three models with the same architecture but with different training data were created. These models were then compared with each other. A simple user interface was created for the resulting system.
|
|
|
Switching off of Electrical Tools by Voice
Rozsypálek, Lukáš ; Šebesta, Vladimír (referee) ; Sigmund, Milan (advisor)
In theoretical part Master’s thesis deals with processing acoustic signal before speech recognition. There are described methods of automatic speech recognition and capture attributes. Those methods are short-term energy of signal, short-term autocorrelation analysis, linear predictive analysis etc. In practical part has been created software, which has to switch off of electrical tool, if the keyword “zastav” has been spoken. In the second part of this thesis software has been optimized for work in real conditions (noise generated by electrical tools).
|
|
|
Voice Controled Map of FIT
Huták, Petr ; Grézl, František (referee) ; Szőke, Igor (advisor)
This bachelor's thesis describes the design and developement of navigation system for the Faculty of information technology, Brno university of technology with voice control. It explains the methods used for finding the shortest path on the map, speech recognition and describes existing navigation systems with voice control. The thesis is focused on creating a user interface with voice control. The purpose of voice control is to make usage of this navigation system more accesible and effective in a public place. Speech recognizing is realized with BSAPI library.
|
|
|
Recurrent Neural Networks for Speech Recognition
Nováčik, Tomáš ; Karafiát, Martin (referee) ; Veselý, Karel (advisor)
This master thesis deals with the implementation of various types of recurrent neural networks via programming language lua using torch library. It focuses on finding optimal strategy for training recurrent neural networks and also tries to minimize the duration of the training. Furthermore various types of regularization techniques are investigated and implemented into the recurrent neural network architecture. Implemented recurrent neural networks are compared on the speech recognition task using AMI dataset, where they model the acustic information. Their performance is also compared to standard feedforward neural network. Best results are achieved using BLSTM architecture. The recurrent neural network are also trained via CTC objective function on the TIMIT dataset. Best result is again achieved using BLSTM architecture.
|
guest ::
