arxiv_cl 70% Match Research Paper Speech scientists,Phoneticians,Linguists,Biomedical engineers,Researchers in speech synthesis 2 weeks ago

Dynamical model parameters from ultrasound tongue kinematics

speech-audio › text-to-speech

📄 Abstract

Abstract: The control of speech can be modelled as a dynamical system in which articulators are driven toward target positions. These models are typically evaluated using fleshpoint data, such as electromagnetic articulography (EMA), but recent methodological advances make ultrasound imaging a promising alternative. We evaluate whether the parameters of a linear harmonic oscillator can be reliably estimated from ultrasound tongue kinematics and compare these with parameters estimated from simultaneously-recorded EMA data. We find that ultrasound and EMA yield comparable dynamical parameters, while mandibular short tendon tracking also adequately captures jaw motion. This supports using ultrasound kinematics to evaluate dynamical articulatory models.

Authors (2)

Sam Kirkham

Patrycja Strycharczuk

Submitted

October 21, 2025

arXiv Category

cs.CL

JASA Express Letters 5, 115201 (2025)

arXiv PDF

Key Contributions

This study evaluates the reliability of estimating dynamical speech production parameters (specifically, linear harmonic oscillator parameters) from ultrasound tongue kinematics. It demonstrates that ultrasound yields comparable parameters to Electromagnetic Articulography (EMA) and supports its use for evaluating articulatory models.

Business Value

Enables more accessible and potentially lower-cost research into speech production, leading to advancements in speech synthesis, voice disorders treatment, and human-computer interaction.

Paper Metadata

Innovation Type

Methodology Validation

Deployment Feasibility

High for research purposes. Integration into real-time clinical or commercial applications would require further development.

Limitations Addressed

Addresses the limitations of traditional fleshpoint data (like EMA) by validating ultrasound imaging as a more accessible and comparable alternative for capturing articulatory dynamics.

Performance Gains

Ultrasound and EMA yield comparable dynamical parameters.,Mandibular short tendon tracking adequately captures jaw motion.

Technical Tags

Speech ProductionArticulatory DynamicsUltrasound ImagingElectromagnetic Articulography (EMA)Dynamical Systems ModelingSpeech KinematicsPhoneticsSpeech ScienceParameter EstimationLinear Harmonic Oscillator

Research Topics

Speech Production and AcousticsPhoneticsDynamical Systems TheoryBiomedical Signal ProcessingComputational Linguistics

Methods & Architectures

Ultrasound tongue kinematics measurementElectromagnetic Articulography (EMA) recordingParameter estimation for linear harmonic oscillator modelsComparative analysis of ultrasound vs. EMA data Linear Harmonic Oscillator Model

Applications & Tasks

Speech Synthesis Speech Therapy Phonetics Research Linguistics Human-Computer Interaction Evaluating ultrasound imaging as an alternative to EMA for capturing articulatory dynamicsReliably estimating parameters of dynamical speech models from kinematic data Estimating dynamical model parameters from ultrasound tongue kinematicsComparing ultrasound-derived parameters with EMA-derived parametersValidating ultrasound for articulatory modeling

Related Fields

Speech SciencePhoneticsBiomedical EngineeringDynamical SystemsSignal Processing

Keywords

Speech ProductionUltrasoundEMAArticulatory DynamicsDynamical SystemsSpeech KinematicsPhoneticsParameter EstimationSpeech SynthesisModeling

Academic Context

#Speech Production and Acoustics#Phonetics#Dynamical Systems Theory#Biomedical Signal Processing#Computational Linguistics

Commercial Potential

Potential Products

Advanced speech synthesis enginesDiagnostic tools for speech disordersResearch software for articulatory modeling

Target Industries

TechnologyHealthcare (Speech Therapy)Research & DevelopmentEducation

Use Case Examples

Developing more natural-sounding text-to-speech systemsAnalyzing speech impedimentsImproving voice recognition technology

Competitive Edge

Validates ultrasound as a viable, potentially more accessible, alternative to EMA for studying speech dynamics, opening new avenues for research.

Market Opportunity

Growing market for speech technology, AI voice assistants, and personalized healthcare solutions.

Revenue Models

Licensing of research findingsdevelopment of specialized speech analysis software.

Resource Requirements

Compute Needs

Moderate, for data processing and model parameter estimation.

Data Requirements

Simultaneous ultrasound and EMA recordings of speech.

Deployment Constraints

Ultrasound data acquisition requires specialized equipment and expertise.,Real-time processing for clinical applications may be challenging.

Scalability

The methodology can be scaled to larger participant groups and different speech tasks.

Production Readiness

Maturity Level

Research/Validation

Time to Market

2-4 years for integration into commercial speech technology.

Patent Potential

Low, focused on methodology validation.

View Full Paper Back to Papers