Ireland – PhD in Speaker Tracking at Trinity College Dublin

Overview

University:
Trinity College Dublin

Country:
Ireland

Fields:
Electronic Engineering, Computer Science, Signal Processing, Machine Learning, Linguistics

Are you passionate about advancing the intersection of speech science, engineering, and artificial intelligence, and eager to contribute to the future of human-computer interaction? If you are motivated to unravel the complexities of communication in multi-speaker environments and develop innovative solutions for speaker tracking in real-world conversations, this fully funded PhD opportunity at Trinity College Dublin could be the perfect next step in your academic journey.

Trinity College Dublin, officially known as the University of Dublin, is Ireland’s oldest and most prestigious university, founded in 1592. Situated in the vibrant heart of Dublin, Trinity boasts a rich heritage of academic excellence and a reputation for world‑class research. The School of Engineering at Trinity is internationally recognized for cutting‑edge work across electrical, electronic, and computer engineering disciplines. As a member of the European Union and a thriving hub for technology and innovation, Ireland offers a dynamic, multicultural environment for graduate study.

Trinity’s campus is renowned for its historic architecture, state‑of‑the‑art facilities, and a lively student community, making it an ideal setting for ambitious scholars to thrive both academically and personally.

The focus of this PhD project is on “Speaker Tracking in Complex Conversations.” In the context of modern communication, accurately identifying and predicting active speakers in multi‑participant conversations is a significant challenge, especially when considering the diverse modalities involved—such as articulation, facial movements, eye gaze, head nods, back channels, and gestures. Current active speaker‑tracking systems primarily leverage audio and facial data, but predicting the next speaker in dynamic, multi‑speaker environments requires a more sophisticated, multimodal approach.

This research aims to incorporate knowledge around the H&H (Hyper‑ and Hypo‑articulation) theory of speech and multimodality in human communication. By gaining deeper insights into how humans exploit multimodal cues in conversation, the project seeks to develop novel neural architectures for more intelligent and context‑aware speaker tracking. The outcomes of this research have far‑reaching implications, from enhancing human‑computer interaction and automated transcription services to improving accessibility technologies and advancing the field of conversational AI.

This PhD studentship is part of the Speech Space project, funded by Research Ireland Frontiers and led by Professor Naomi Harte. The Speech Space initiative brings together a multidisciplinary team of engineers and linguists working at the intersection of speech science and engineering. The selected PhD student will benefit from a collaborative, intellectually stimulating environment and will have access to the resources and expertise of Prof.

Harte’s research group. The group is known for its diversity, collegiality, and commitment to sharing ideas and supporting one another’s growth.

The funding package includes coverage of EU/Non‑EU tuition fees, a tax‑free stipend of €25,000 per annum for four years, as well as support for equipment and conference travel. The position is fully in‑person, fostering direct engagement and collaboration within the research group and the broader Trinity College Dublin community.

• – A primary or Master’s degree in Electronic Engineering or a closely related discipline, with a strong interest in multidisciplinary research.
• – Eligibility to meet Trinity College Dublin’s postgraduate admission requirements, as outlined at https://(Use the "Apply for this Job" box below)..
• – Demonstrated skills in coding, machine learning, deep learning, and signal processing, along with a willingness to learn new tools and methodologies.
• – Prior experience with speech‑based interaction is…