Aug 9, 2021
ELITR’s Speech Team Updated
Since January 2019, the EU project ELITR has been working on technologies improving communication between people. Breaking the language barrier in speech is one of our missions, saving people’s time by automatic meeting summarization is our long term vision towards which ELITR is making the first concrete steps by organizing AutoMin 2021, the first shared task in automatic creation of meeting minutes.
ELITR research partners are Charles University (CUNI, the coordinator), University of Edinburgh (UEDIN) and Karlsruhe Institute of Technology (KIT). Two companies ensure the research results are tested in practical settings: PerVoice (PV, the integrator) and alfaview® (AV, a remote conferencing company).
Recently, the KIT research team has been updated, taking Christian Huber and Tuan Nam Nguyen. Building upon the existing low-latency fully neural KIT system for automatic speech recognition, the new colleagues will focus on domain adaptation and inspection of the learned ASR model, especially aimed at improving their multilingual and adaptation capabilities.
ELITR has successfully tested the speech transcription and translation technologies at a high-profile conference organized in April 2021, as will be reported at the MT-Summit 2021 workshop Automatic Spoken Language Translation in Real-World Settings by Ondřej Bojar on August 20. Another such a demo event is planned later this year with the help of alfaview®, who are not only the user partner in ELITR but who are also responsible for the dissemination of the project results. alfaview® has already presented the ELITR project at many internal and external trade fairs, congresses and workshops in the past and will expand its web presence via further blog articles, etc. in this context.
To learn more about ELITR goals and progress, make sure to check out our website and blog: https://elitr.eu/blog.