To work on advanced methods for automatic speech recognition with the aim towards building transferable multiple domain end to end recognition systems that are adaptive and can learn over long periods of time.
The position is funded by the Voicebase Centre for Speech and Language Technologies. Salaries are at University of Sheffield grade scales 7.1 - 7.9, and the opening is for 2 years, with the possibility of extension. More details can be found at: https://www.jobs.ac.uk/job/BKA871/research-associate-in-deep-learning-for-speech-recognition/

To work on advanced deep learning methods to build robust and high performance methods for speech recognition that can work in the context of many languages and that allow fast building of recognition systems for new languages.

The position is funded by the Voicebase Centre for Speech and Language Technologies. Salaries are at University of Sheffield grade scales 7.1 - 7.9, and the opening is for 2 years, with the possibility of extension. More details can be found at: https://www.jobs.ac.uk/job/BKA854/research-associate-in-multi-lingual-automatic-speech-recognition/

To work on unified methodologies to extract information from speech that allows to infer a variety of speech properties relating to the emotional state. Methods should be robust and the work should keep further speech attributes in mind. The position is funded by the Innovate UK (UKRI) project MAUDIE and by IIKE in collaboration with Emotech Ltd. Salaries are at University of Sheffield grade scales 7.1 - 7.9, and the opening is for 12 months.

To work on robust methods for direct recognition from the raw waveforms - to achieve robustness against environmental effects. Work should focus on telephony, but further multichannel input is of interest. Funding for 3 years, in the context of the Voicebase Centre for Speech and Language Technologies.

Diarisation is the process of recognising who speaks when, with as little prior knowledge as possible. To this day it is implemented as a multi-stage process. Research should focus on integrating what is now typically 4 stages of processing into an end to end . Funded by UKRI for 3.5 years. Funding details to be announced soon.

To work on methods that allow to increase the productivity of Movie Voice Dubbing methodology through advanced speech technology. The aim is to develop and implement methods that can help media specialists in foreign language dubbing to produce spoken translations that fit perfectly to the visual content. The project work is in conjunction with Zoo Digital who provide access to rich and very large media archives.
The position is funded by the Innovate UK (UKRI) project MAUDIE . Salaries are at University of Sheffield grade scales 7.1 - 7.9, and the opening is for 2.5 years.

To work on novel deep learning methods for building recognition systems that can operate in several languages, methods for adapting recognition systems to languages with minimal data, methods that allow automatic inference of language attributes, or methods of unsupervised adaptation to new languages.
Funding for 3 years, in the context of the Voicebase Centre for Speech and Language Technologies.