Please express your interest in participation as soon as possible email@example.com
|Workshop opening, The view of SJU
|MALORCA and its forerunners
The presentation gives an overview of the Horizon 2020 SESAR funded project MALORCA. The AcListant® project has shown that command recognition error rates below 2% possible. MALORCA proposes a general, cheap and effective solution to adapt speech recognition system to new environments by automatically learning from recorded speech and radar data.
Austro Control and ANS CR present why they believe that Automatic Speech Recognition has high relevance for ATM and controller productivity. They present their contributions to MALORCA project and next steps.
|Researcher perspective and challenges
MALORCA develops generic basic acoustic, language and command predictions models. The presentations focusses on the performance of these basic models developed in first year and presents ideas and first results how unsupervised and semi-supervised learning can improve these models i.e. adapt to Prague and Vienna Approach area. Radar data and controller pilot voice communication data recorded on daily basis are used for this purpose.
The software prototype of the generic basic assistant based speech recognition (ABSR) system is presented. Show case is Prague Approach Area.
|Workshop 1, 2, ... running in parallel
The workshop participants are divided into separate working groups with 8-12 participants. Currently the following topics are planned:
The participant may select the most preferable topic at the beginning of the workshop. Contact the organizers if you want to suggest an additional topic.
|Summary of Working Groups
Each working group has a rapporteur who will summarize the main finding of each working group for the stakeholder participants.
|Workshops Wrap-up, Q&A and Workshop End