Sigma-Lognormal Modeling of Speech.

Area of Science:

• Biomechanics
• Speech Science
• Neuromuscular Modeling

Background:

• Human movement analysis is crucial across diverse scientific fields.
• Existing speech motor models struggle with inverse approaches, like deriving muscular parameters from speech.
• Handwriting research successfully uses kinematic theory and the Sigma-lognormal model for muscular parameter extraction.

Purpose of the Study:

• To present a simplified speech kinematics-based model for studying, analyzing, and reconstructing complex speech movements.
• To adapt the kinematic theory of rapid human movements and the Sigma-lognormal model for speech analysis.
• To investigate the relationship between extracted speech parameters, aging, and formant proportions.

Main Methods:

• Applied the kinematic theory of rapid human movements and the Sigma-lognormal model to speech.
• Described and parameterized the asymptotic impulse response of speech neuromuscular networks.
• Developed a method for transforming formant data into observable movement parameters.
• Utilized the VTR-TIMIT (English) and Saarbrucken Voice Database (German) for experiments.

Main Results:

• Corroborated a link between extracted kinematic parameters and the aging process.
• Confirmed a relationship between formant proportions and the application of the kinematic theory.
• Demonstrated the model's ability to analyze speech kinematics in diverse populations (age, laryngeal pathologies).

Conclusions:

• The developed speech kinematics model offers a simplified approach to analyzing complex speech movements.
• The findings support the applicability of handwriting movement theories to speech production.
• Results pave the way for innovative developments in speech modeling, analysis, and understanding age-related speech changes.