|
|
Codec Detection from Speech
Jon, Josef ; Matějka, Pavel (referee) ; Černocký, Jan (advisor)
Tato práce se zabývá detekcí kodeků z komprimovaného řečového signálu. Cílem bylo zjistit, jaké charakteristiky rozlišují jednotlivé kodeky a následně vytvořit prostředí vhodné pro experimenty s různými typy a konfiguracemi klasifikátorů. Použity byly Support vector machines a především neuronové sítě, které byly vytvořeny pomocí nástroje Keras. Hlavním přínosem této práce je experimentální část, ve které je analyzován vliv různých parametrů neuronové sítě. Po nalezení nejvhodnější kombinace parametrů dosáhla síť přesnosti klasifikace přes 98% na testovací sadě obsahující data z 6 kodeků.
|
|
|
Sound Creation Using VST
Švec, Michal ; Schimmel, Jiří (referee) ; Černocký, Jan (advisor)
This diploma thesis deals with digital sound synthesis. The main task was to design and implement new sound synthesizer. Created tool uses different approaches to the sound synthesis, so it can be described as a hybrid. Instrument design was inspired by existing audio synthesizers. For implementation, C++ language and VST technology from Steinberg are used. As an extension, a module, that can process voice or text input and then build a MIDI file with melody (which can be interpreted with using any synthesizer) was designed and implemented. For this module, Python language is used. For the synthesizer, a simple graphical user interface was created.
|
|
|
Time-Domain Neural Network Based Speaker Separation
Peška, Jiří ; Černocký, Jan (referee) ; Žmolíková, Kateřina (advisor)
A thesis is about the usage of convolutional neural networks for automatic speech separation in an acoustic environment. The goal is to implement the neural network by following a TasNet architecture in the PyTorch framework, train this network with various values of hyper-parameters, and to compare the quality of separations based on the size of the network. In contrast to older architectures that transformed an input mixture into a time-frequency representation, this architecture uses a convolutional autoencoder, which transforms input mixture into a non-negative representation optimized for a speaker extraction. Separation is achieved by applying the masks, which are estimated in the separation module. This module consists of stacked convolutional blocks with increasing dilation, which helps with modeling of the long-term time dependencies in processed speech. Evaluation of the precision of the network is measured by a signal to distortion (SDR) metric, by a perceptual evaluation of speech quality (PESQ), and the short-time objective intelligibility (STOI). The Wall Street Journal dataset (WSJ0) has been used for training and evaluation. Trained models with various values of hyper-parameters enable us to observe the dependency between the size of the network and SDR value. While smaller network after 60 epochs of training reached 10.8 dB of accuracy, a bigger network reached 12.71 dB.
|
|
|
High Level Analysis of the Psychotherapy Sessions
Polok, Alexander ; Karafiát, Martin (referee) ; Matějka, Pavel (advisor)
This work focuses on analyzing psychotherapy sessions within the DeePsy research project. This work aims to design and develop features that model the session dynamics, which can reveal seemingly subtle nuances. The mentioned features are automatically extracted from the source recording using neural networks. They are further processed, compared across sessions, and displayed graphically, creating a document that acts as a feedback document about the session for the therapist. Furthermore, this assistive tool can help therapists to professionally grow and to provide better psychotherapy in the future. A relative improvement in voice activity detection of 37.82% was achieved. The VBx diarization system was generalized to converge to two speakers with a minimum relative error rate degradation of 0.66%. An automatic speech recognition system has been trained with a 17.06% relative improvement over the best available hybrid model. Models for sentiment classification, type of therapeutic interventions, and overlapping speech detection were also trained.
|
|
|
Learning the Face Behind a Voice
Zubalík, Petr ; Mošner, Ladislav (referee) ; Plchot, Oldřich (advisor)
The main goal of this thesis is to design and implement a system that will be able to generate a face based on the speech of a given person. This problem is solved using a system composed of three convolutional neural network models. The first one is based on the ResNet architecture and is used to extract features from speech recordings. The second model is a fully convolutional neural network which converts the extracted features into the styles which form a base for the final facial image. These styles are then passed as an input to the StyleGAN generator, which creates the resulting face. The proposed system is implemented in the Python programming language using the PyTorch framework. The last chapter of the thesis discusses some of the most significant experiments performed to fine-tune and test the developed system.
|
|
|
Analysis of Interview Audio
Polok, Alexander ; Plchot, Oldřich (referee) ; Matějka, Pavel (advisor)
The aim of this thesis is the analysis of psychotherapeutic sessions. Classifiers describing the therapy are extracted from the audio recordings. These are then aggregated, compared with other sessions, and graphically presented in a report summarizing the conversation. In this way, therapists are provided with feedback that can serve for professional growth and better psychotherapy in the future.
|
|
|
Time-Domain Neural Network Based Speaker Separation
Peška, Jiří ; Černocký, Jan (referee) ; Žmolíková, Kateřina (advisor)
A thesis is about the usage of convolutional neural networks for automatic speech separation in an acoustic environment. The goal is to implement the neural network by following a TasNet architecture in the PyTorch framework, train this network with various values of hyper-parameters, and to compare the quality of separations based on the size of the network. In contrast to older architectures that transformed an input mixture into a time-frequency representation, this architecture uses a convolutional autoencoder, which transforms input mixture into a non-negative representation optimized for a speaker extraction. Separation is achieved by applying the masks, which are estimated in the separation module. This module consists of stacked convolutional blocks with increasing dilation, which helps with modeling of the long-term time dependencies in processed speech. Evaluation of the precision of the network is measured by a signal to distortion (SDR) metric, by a perceptual evaluation of speech quality (PESQ), and the short-time objective intelligibility (STOI). The Wall Street Journal dataset (WSJ0) has been used for training and evaluation. Trained models with various values of hyper-parameters enable us to observe the dependency between the size of the network and SDR value. While smaller network after 60 epochs of training reached 10.8 dB of accuracy, a bigger network reached 12.71 dB.
|
|
|
Clustering a load balancing serveru pro zpracování řeči
Trnka, Miroslav
This paper deals with the possibilities for load balancing and clustering of an existing server for speech processing. The paper analyzes problems of load balancing and clustering. There are also described the concepts of network programming and options for I/O processing. A new design of a load balancer is created, fully customized for the needs of speech processing server. This newly designed load balancer is implemented and thoroughly tested.
|
|
|
Non-Parallel Voice Conversion
Brukner, Jan ; Plchot, Oldřich (referee) ; Černocký, Jan (advisor)
Cílem konverze hlasu (voice conversion, VC) je převést hlas zdrojového řečníka na hlas cílového řečníka. Technika je populární je u vtipných internetových videí, ale má také řadu seriózních využití, jako je dabování audiovizuálního materiálu a anonymizace hlasu (například pro ochranu svědků). Vzhledem k tomu, že může sloužit pro spoofing systémů identifikace hlasu, je také důležitým nástrojem pro vývoj detektorů spoofingu a protiopatření. Modely VC byly dříve trénovány převážně na paralelních (tj. dva řečníci čtou stejný text) a na vysoce kvalitních audio materiálech. Cílem této práce bylo prozkoumat vývoj VC na neparalelních datech a na signálech nízké kvality, zejména z veřejně dostupné databáze VoxCeleb. Práce vychází z moderní architektury AutoVC definované Qianem et al. Je založena na neurálních autoenkodérech, jejichž cílem je oddělit informace o obsahu a řečníkovi do samostatných nízkodimenzionýálních vektorových reprezentací (embeddingů). Cílová řeč se potom získá nahrazením embeddingu zdrojového řečníka embeddingem cílového řečníka. Qianova architektura byla vylepšena pro zpracování audio nízké kvality experimentováním s různými embeddingy řečníků (d-vektory vs. x-vektory), zavedením klasifikátoru řečníka z obsahových embeddingů v adversariálním schématu trénování neuronových sítí a laděním velikosti obsahového embeddingu tak, že jsme definovali informační bottle-neck v příslušné neuronové síti. Definovali jsme také další adversariální architekturu, která porovnává původní obsahové embeddingy s embeddingy získanými ze zkonvertované řeči. Výsledky experimentů prokazují, že neparalelní VC na nekvalitních datech je skutečně možná. Výsledná audia nebyla tak kvalitní případě hi fi vstupů, ale výsledky ověření řečníků po spoofingu výsledným systémem jasně ukázaly posun hlasových charakteristik směrem k cílovým řečníkům.
|
|
|
Voice Conversion
Brukner, Jan ; Plchot, Oldřich (referee) ; Černocký, Jan (advisor)
Thesis deals with voice converion. Method, where we want to modify speech parameters of source speaker into that of a target speaker. At the beginning of thesis is described Voice Conversion Challenge (VCC), where participants tried to build better voice conversion systems. In the next part are analysed components of baseline system used in VCC. Modifications which could improve quality of converted voice are proposed. Then is briefly described implementation if these modifications and results are analysed. In the end is part dedicated to further improvements of voice conversion.
|
guest ::
