Rhythm in Speech and Music
Rhythms organize time, akin to how patterns and contours structure space. In both speech and music, rhythmic elements help listeners segment content. Our research explores how the brain utilizes these temporal cues to decode continuous auditory information into distinct units. In speech, we focus on transforming complex sound sequences into semantically clear units, emphasizing sentence-level prosody (Teng et al., 2020). In music, we study how basic elements like notes and beats form higher-level structures, like musical phrases (Teng, Larrouy-Maestri, & Poeppel, 2021). Our work has implications for diagnosing and treating speech and language disorders, cognitive rehabilitation in conditions like Parkinson's, and developing rhythm-based therapies to improve motor coordination and cognitive function.
Teng, X., Larrouy-Maestri, P., & Poeppel, D. (2021). Segmenting and Predicting Musical Phrase Structure Exploits Neural Gain Modulation and Phase Precession. BioRxiv, 2021-07.
Chang, A., Teng, X., Assaneo, F., & Poeppel, D. (2022). Amplitude modulation perceptually distinguishes music and speech. PsyArXiv
Teng, X., Ma, M., Yang, J., Blohm, S., Cai, Q., & Tian, X. (2020). Constrained structure of ancient Chinese poetry facilitates speech content grouping. Current Biology, 30(7), 1299-1305.
Beyond and Within Rhythm: Active Sensing and Embodied Intelligence
In addition to exploring the role of rhythms in speech and music, our research aims to identify broader cognitive and neural principles governing rhythm processing. We posit that both the brain and body leverage external (e.g., poetic meter) and internal (e.g., neural oscillations, heartbeat) rhythms for proactive sensory engagement and internal communication. This aligns with the concept of 'active sensing,' where organisms intentionally interact with their environment to efficiently sample sensory data. Our focus is on how external and internal rhythms help the brain anticipate and process incoming information. These insights allow us to derive computational principles, extending the applicability of our research to sensory deficit diagnosis and robotics.Rhythm in Tradition and Culture
We aim for our research to enrich local society by deepening our understanding of ourselves, our culture, and our heritage. Traditional Chinese art forms like Kun Opera employ coherent auditory and visual rhythms for emotional and aesthetic expression. Unlike classic Western art, which has been extensively investigated, traditional Chinese art remains largely unexplored. Leveraging the methodologies and theoretical frameworks developed in our research, we delve into these artistic rhythms. Specifically, we study the interaction of auditory and visual rhythms in traditional art forms, identify distinct quantitative traits with signal processing and AI, and assess how audiences engage emotionally and aesthetically with these rhythms.Teng, X., Ma, M., Yang, J., Blohm, S., Cai, Q., & Tian, X. (2020). Constrained structure of ancient Chinese poetry facilitates speech content grouping. Current Biology, 30(7), 1299-1305.
Rhythm in Artificial Neural Network and Tools
We employ recurrent neural networks (RNNs) to simulate and analyze rhythms, aiming to elucidate underlying mechanisms and develop analytical tools. Computational modeling enables us to examine the functional goals of observed rhythms, while perturbing rhythms in artificial systems generates predictions for real-world scenarios. Additionally, analyzing artificial rhythms contributes to the development of tools for experimental data assessment. Specifically, we train RNNs on behavioral tasks that parallel our experiments. We then dissect the trained networks to comprehend how rhythms are generated and their functional roles (Teng and Zhang, 2021). Furthermore, we use RNN-driven agents to interact with simulated environments, exploring how rhythms facilitate interactions between the brain, body, and environment.Teng, X., & Zhang, R. Y. (2021). Sequential Temporal Anticipation Characterized by Neural Power Modulation and in Recurrent Neural Networks. bioRxiv, 2021-10.
Chen, X., Teng, X., Chen, H., Pan, Y., & Geyer, P. (2024). Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX. Biomedical Signal Processing and Control, 87, 105475.