Institutional Repository of Vilnius University
Balso įrašų anotavimo automatizavimas /
Abstract
dc:descriptionIn this paper, a graphical interface for managing actions with the HTK toolset is created in order to parallelize the tools and visualize the training process. The tool is then used to create a Lithuanian language phonetization HTK HMM model set and experiments with phoneme splitting and joining are done with analysis into how they affect the accuracy of the model. The model set is trained with 20241 LIEPA project annotated speaker audio clips. The created model is integrated into the Praat annotation program and is compared to other Lithuanian HTK HMM sets. When compared to other HMM sets, the final set has the highest alignment rate, learned strong Gaussian distributions for phonemes and reached a low phoneme alignment error.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Kirda, Donatas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD308337360&prefLang=en_US