Modeling Laryngeal Muscle Activation Noise for Low-Order Physiological Based Speech Synthesis

Rodrigo Manríquez, Sean D. Peterson, Pavel Prado, Patricio Orio, Matías Zañartu


Physiological-based synthesis using low order lumped-mass models of phonation have been shown to mimic and predict complex physical phenomena observed in normal and pathological speech production, and have received significant attention due to their ability to efficiently perform comprehensive parametric investigations that are cost prohibitive with more advanced computational tools. Even though these numerical models have been shown to be useful research and clinical tools, several physiological aspects of them remain to be explored. One of the key components that has been neglected is the natural fluctuation of the laryngeal muscle activity that affects the configuration of the model parameters. In this study, a physiologically-based laryngeal muscle activation model that accounts for random fluctuations is proposed. The method is expected to improve the ability to model muscle related pathologies, such as muscle tension dysphonia and Parkinson’s disease. The mathematical framework and underlying assumptions are described, and the effects of the added random muscle activity is tested in a well-known body-cover model of the vocal folds with acoustic propagation and interaction. Initial simulations illustrate that the random fluctuations in the muscle activity impact the resulting kinematics to varying degrees depending on the laryngeal configuration.


 DOI: 10.21437/Interspeech.2017-1722

Cite as: Manríquez, R., Peterson, S.D., Prado, P., Orio, P., Zañartu, M. (2017) Modeling Laryngeal Muscle Activation Noise for Low-Order Physiological Based Speech Synthesis. Proc. Interspeech 2017, 1378-1382, DOI: 10.21437/Interspeech.2017-1722.


@inproceedings{Manríquez2017,
  author={Rodrigo Manríquez and Sean D. Peterson and Pavel Prado and Patricio Orio and Matías Zañartu},
  title={Modeling Laryngeal Muscle Activation Noise for Low-Order Physiological Based Speech Synthesis},
  year=2017,
  booktitle={Proc. Interspeech 2017},
  pages={1378--1382},
  doi={10.21437/Interspeech.2017-1722},
  url={http://dx.doi.org/10.21437/Interspeech.2017-1722}
}