Scaling and universality in the human voice

Area of Science:

• Physics of complex systems
• Human speech production
• Bioacoustics

Background:

• Speech is a complex human capability.
• Understanding the physics of speech production is crucial.
• Current models often attribute speech nonlinearities to deterministic chaos.

Purpose of the Study:

• To empirically analyze the statistical properties of human speech.
• To investigate the underlying physical mechanisms of speech production.
• To explore potential impacts on speech synthesis technologies.

Main Methods:

• Analysis of large, multilingual human speech datasets.
• Statistical analysis of energy release and inter-event time distributions.
• Application of Gutenberg-Richter law and renormalization group transformations.

Main Results:

• Speech energy release follows a universal, power-law distribution (Gutenberg-Richter-like law).
• Temporal correlations in speech events are also power-law distributed.
• Speech generation operates near a critical point, driven by physiological mechanisms.

Conclusions:

• The physics of speech production exhibits universal characteristics, independent of language or speaker count.
• Speech complexity likely arises from physiological mechanisms near a critical point.
• Findings offer new perspectives for advancing speech synthesis technology.