Studying vocal tract configuration and articulation is critical across many fields - from phonetics to speech pathology; singing science to speaker identity - but data used for studying vocal tracts and articulation is typically complex and multimodal. As medical imaging datasets, particularly magnetic resonance imaging (MRI) data, become more widely available, there is an urgent need for researchers to address processing bottlenecks and develop generalisable, fully-automatic workflows for extracting vocal tract and articulation data, alongside robust statistical techniques for subsequent analysis. By bringing together an interdisciplinary group of researchers working on related problems, we hope to foster new collaborative working relationships; share data and methods to avoid duplicating research effort; and identify priorities for future research, such as new methods and participant populations.

Although vocal tract shape and articulation studies appear frequently in regular Interspeech sessions, the focus is usually upon the results rather than the methods. The Interspeech 2026 theme of ‘Understanding Speech Together’ is a perfect opportunity to bring together the diverse group of people developing and using similar methods, to enrich our collective understanding of the vocal articulators.

This special session takes place at 14:30-16:30 on Monday 28th September in Room C3.6.

The session will consist of a 40-minute overview talk followed by a poster session.

Survey talk

Developments and challenges in vocal tract shape and articulation analysis
Vikram Ramanarayanan

Posters

• Acoustic Pharyngometry as an Auditable Anchor for Cross-Speaker EMA Normalization

• Vocal Tract Disparity and Potential Implications for Speaker Recognition

• Towards Language-Agnostic Speech Inversion

• Influence of Vocal Tract Curvature on Speech Acoustics: A Three-Dimensional FEM Analysis

• Morphoacoustic Modeling of a Dynamic 3D Vocal Tract Using MRI-Constrained Deformations and FEM Acoustics

• An Approach to Simultaneous Acquisition of Real-Time MRI Video, EEG, and Surface EMG for Articulatory, Brain, and Muscle Activity During Speech Production

• The ArtComp dataset: Articulatory and Acoustic Measurements of Swedish in Speech with Naturally Manipulated Jaw Position

• Articulatory Analysis of the Mandarin Alveolar–Retroflex Contrast Using Real-Time MRI

• Tongue-Shape Strategies for Standard Mandarin Retroflex Sibilants: A Preliminary Ultrasound and Unsupervised Clustering Study

• Amelia Gully, University of York

• Michael Proctor, Macquarie University

• Tünde Szalay, University of Sydney

• Kirrie Ballard, University of Sydney

• Craig Jin, University of Sydney

• Tharinda Piyadasa, University of Sydney

• Emily Kiff, University of York