Two Fully-funded PhD Studentships in Spoken Language Technologies

Two Fully-funded PhD Studentships in Spoken Language Technologies, University of Sheffield, UK Deadline for applications: 13 April 2025. Home and International students may apply. Regardless of your fees status (Home or International), all fees will be paid (in addition to a full stipend). Speech and Language Technologies (SLTs) are a range of Artificial Intelligence (AI) approaches for analysing, producing, modifying or responding to spoken and written language. SLTs are underpinned by a number of fundamental research fields including acoustics, signal processing, speech processing, natural language processing, computational linguistics, mathematics, machine learning, physics, psychology, and computer science. We are seeking two candidates to each work on an interdisciplinary SLT research project covering both fields of speech and language research on one of the following topics: Accessible Democracy: UK Houses of Parliament and cross-party Select Committees are at the core of UK democracy. Making the proceedings of these bodies accessible to citizens and journalists is key to holding politicians accountable. This research aims to develop technologies to provide access to the rich linguistic and paralinguistic information in parliamentary audio recordings. Helping journalists to identify newsworthy events is one of the example objectives, alongside more standard tasks such as search, creating alerts or summarisation. Analytics of conversations: Spoken conversations are complex and difficult to understand for AI systems. While the words spoken are of obvious importance, paralinguistic information often plays an essential role for a satisfactory and efficient exchange. In practice only goal oriented metrics are used to assess the quality of an exchange, which are not helpful to describe a wide range of conversations such as interviews, story telling or even examinations. Modelling of the participants' knowledge and state as well as paralinguistic signalling and perception should be used to research novel methods to interpret and understand conversations. Evolving communication in embodied agents: Spoken and written language have developed in the course of human evolution and can be viewed as key species-wide adaptations that have enabled us to better survive on our planet. Modelling the development of language in artificial agents with sensory apparatus that are embedded in a physical environment is an exciting research methodology that promises both deeper understanding of human languages and their origins, as well as insights into how to build more effective autonomous agents. This research will build on the state of the art in this area.

The University of Sheffield is a leading research-intensive university with a reputation for excellent research, world-class facilities and close links with industry.