BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20260525T160152EDT-8051aUEB4D@132.216.98.100 DTSTAMP:20260525T200152Z DESCRIPTION: \n\nAbstract\n\nVoice control systems enable people to control their computers by speaking to them. After a review of the state-of-the-a rt in sequence modeling\, speech recognition\, and language understanding using deep learning\, this thesis describes a number of contributions to t he art of voice control. The first contribution is a study of large-scale semi-supervised learning through pseudo-labeling for massively multilingua l speech recognition. The second contribution is a study of the use of aut oregressive models for conditional computation with neural networks\, usin g speech recognition as a test case. The third contribution is a method fo r training end-to-end spoken language understanding models using speech sy nthesis. The fourth contribution is a crowdsourced dataset\, Timers and Su ch\, for spoken language understanding involving numbers\, along with base line experimental results and open-source software infrastructure for usin g the dataset. The fifth contribution is our part in the design and implem entation of SpeechBrain\, an open-source software toolkit for speech proce ssing. Finally\, using some of the tools and techniques developed earlier in the thesis\, we propose a simplified and unified approach to voice cont rol in which the entire traditional pipeline\, composed of an automatic sp eech recognition subsystem\, a natural language understanding subsystem\, and human-programmed control logic\, is subsumed within a single deep neur al network.\n DTSTART:20230515T170000Z DTEND:20230515T190000Z LOCATION:\, Room 603\, McConnell Engineering Building\, CA\, QC\, Montreal\ , H3A 0E9\, 3480 rue University SUMMARY:PhD defence of Loren Lugosch - Deep Neural Networks for Voice Contr ol URL:/ece/channels/event/phd-defence-loren-lugosch-deep -neural-networks-voice-control-348248 END:VEVENT END:VCALENDAR