Featured Research
Artificial Intelligence, Speech Technology and Human Communication
My research combines AI, speech science, cognitive psychology, phonetics and data science to understand how humans learn speech and interact with increasingly sophisticated synthetic voices.
- AI-supported pronunciation learning
- Voice cloning and text-to-speech evaluation
- EEG and Frequency-Following Responses (FFR)
- Individual differences in L2 speech learning
- Machine learning and predictive modelling
My research combines Artificial Intelligence, Speech Technology, Cognitive Science, Applied Linguistics and Data Science. I investigate why some learners achieve exceptional pronunciation outcomes while others face persistent challenges, and how emerging AI technologies can be used to create more personalised and effective language-learning experiences.
I work across AI voice cloning, speech synthesis, pronunciation learning, EEG and Frequency-Following Responses (FFRs), machine learning, and human interaction with synthetic speech. My goal is to bridge cutting-edge research and real-world applications that improve communication, learning, and accessibility.
I welcome opportunities for interdisciplinary collaboration, consultancy, invited talks, supervision, and partnerships involving AI, speech technology, language learning, and human communication.
About
I investigate how humans learn, perceive and interact with speech in an increasingly AI-driven world. As a Research Associate at the University of Cambridge, I combine speech science, artificial intelligence, cognitive psychology and data science to understand why some individuals become highly successful language learners while others face persistent challenges. My research sits at the intersection of phonetics, second language acquisition, artificial intelligence, speech technology and data science.
I combine behavioural experiments, acoustic analysis, EEG/FFR methods, statistical modelling and machine learning to understand how learners acquire new speech patterns and how humans evaluate increasingly realistic synthetic voices.
Research themes
AI voice cloning and TTS
Evaluation of synthetic voices, text-to-speech systems, prosody, naturalness, similarity and human-AI communication.
Pronunciation and perception
Research on second-language speech perception, production, high variability phonetic training and learner variability.
Cognition and auditory processing
Auditory acuity, attention, phonological short-term memory, declarative memory, procedural learning and L2 experience.
EEG and FFR
Neural encoding of speech and its relationship with perception, production and pronunciation-learning success.
Speech analysis
Formant dynamics, VOT, prosody, speech timing and automated acoustic-feature extraction.
Predictive modelling
Mixed-effects models, machine learning, Python, R, SQL, PyTorch, scikit-learn, LightGBM and SHAP.
Latest publications and research outputs
馃 Flagship Project: SoundCoach
AI-Powered Personalised Pronunciation Learning
SoundCoach is an adaptive pronunciation-learning platform that combines speech technology, artificial intelligence and second-language acquisition research. Unlike conventional pronunciation tools, SoundCoach tailors feedback using auditory processing, attention, memory and language-learning profiles to create personalised learning pathways.
Flagship Projects and Research
SoundCoach
SoundCoach is an adaptive AI-supported pronunciation-training platform that combines speech technology, artificial intelligence and second-language acquisition research. It delivers personalised feedback tailored to each learner, combining artificial intelligence, speech technology, learner performance data, and individual learning needs to create adaptive and highly personalised pronunciation-training pathways. The platform continuously adapts its feedback to support more effective, engaging, and efficient language learning.
EEG/FFR speech learning
This research uses EEG and Frequency-Following Responses to investigate neural mechanisms underlying second-language speech learning. By combining neural recordings with behavioural measures of perception and production, the project examines how auditory processing, cognition and language experience shape pronunciation-learning success.
AI voice cloning and prosody
Research on synthetic speech, voice cloning, naturalness, speaker similarity and the acoustic/prosodic determinants of successful text-to-speech output.
L2 vowel and consonant learning
Experimental work on speech perception, production, segmental contrasts, dynamic spectral cues and high variability phonetic training.
Teaching and supervision
I teach and support students in phonetics, phonology, applied linguistics, speech sciences, language assessment and research methods. I also advise dissertation and research projects involving experimental design, quantitative analysis, acoustic phonetics and second language acquisition.
Data science and technical skills
Python, C++, R, MATLAB, SQL, HTML, CSS, JavaScript and Docker.
PyTorch, scikit-learn, LightGBM, SHAP, model evaluation and predictive analytics.
Praat, Parselmouth, openSMILE, Montreal Forced Aligner, Whisper-based pipelines and acoustic feature extraction.
Contact and profiles
lb983@cam.ac.uk
Profile page
Research group page
0000-0002-1435-0242
Google Scholar profile