(No longer available) Topics regarding neural speech synthesis and speaker voice conversion
Organization
NLP Research Group
Abstract
Various practical topics that contribute towards developing neural speech synthesis technology.
As our research focuses on multi-speaker text-to-speech models (check out the prototype at https://www.neurokone.ee) and speaker voice conversion, the thesis would likely be related to one of these topics, but there are other also other aspects regarding speech synthesis that can be explored.
This will involve working with speech datasets, training neural networks, evaluating the results and comparing them to existing works. Having some prior experience training neural models or working with (speech) data would be useful, but is not required.
As our research focuses on multi-speaker text-to-speech models (check out the prototype at https://www.neurokone.ee) and speaker voice conversion, the thesis would likely be related to one of these topics, but there are other also other aspects regarding speech synthesis that can be explored.
This will involve working with speech datasets, training neural networks, evaluating the results and comparing them to existing works. Having some prior experience training neural models or working with (speech) data would be useful, but is not required.
Graduation Theses defence year
2020-2021
Supervisor
Liisa Rätsep, Mark Fishel
Spoken language (s)
Estonian, English
Requirements for candidates
Level
Bachelor, Masters
Keywords
Application of contact
Name
Liisa Rätsep
Phone
E-mail
See more