Research

Early learning, infant–caregiver interaction and neuropersonalised education technology.

Young children learn language through rich, responsive interaction with caregivers. As more early learning moves onto screens, a key question is what gets lost — and whether technology can be designed to put it back. My research tackles this from the brain upwards.

PhD project

From Infant-Directed Interaction to Neuropersonalised Digital Learning: Enhancing Learning via Neural Entrainment (tentative title).

Infant-directed speech and neural entrainment

I characterise the acoustic properties of infant-directed speech and how children's brain rhythms align (entrain) to it, comparing face-to-face and screen-mediated learning. The findings inform digital education policy.

Neuropersonalised digital avatar

With colleagues at EMPOWER, I'm co-developing a socially responsive digital avatar whose speech and song are tuned to each child's individual brain rhythm (alpha frequency), to test whether personalised timing improves word learning. The avatar paper is accepted at IEEE TALE 2026 and is the subject of a Singapore provisional patent. Related work on neuropersonalised acoustic signals for screen-based learning appeared at the AI for Children workshop, ICLR 2025.

Automated audio-visual speaker diarization

I built a pipeline that combines audio and video to identify who is speaking when in mother–infant interactions, replacing slow manual coding. Presented at the 51st Boston University Conference on Language Development (BUCLD 51), November 2026.

Methods: EEG (EEGLAB) · audio-visual signal processing · machine learning · Python, R, MATLAB · PsychoPy

Earlier research