Acoustics of breath noises in human speech (Werner et al., 2023)

Purpose: Breathing is ubiquitous in speech production, crucial for structuring speech, and a potential diagnostic indicator for respiratory diseases. However, the acoustic characteristics of speech breathing remain underresearched. This work aims to characterize the spectral properties of human inhalation noises in a large speaker sample and explore their potential similarities with speech sounds. Speech sounds are mostly realized with egressive airflow. To account for this, we investigated the effect of airflow direction (inhalation vs. exhalation) on acoustic properties of certain vocal tract (VT) configurations.

Method: To characterize human inhalation, we describe spectra of breath noises produced by human speakers from two data sets comprising 34 female and 100 male participants. To investigate the effect of airflow direction, three-dimensional–printed VT models of a male and a female speaker with static VT configurations of four vowels and four fricatives were used. An airstream was directed through these VT configurations in both directions, and their spectral consequences were analyzed.

Results: For human inhalations, we found spectra with a decreasing slope and several weak peaks below 3 kHz. These peaks show moderate (female) to strong (male) overlap with resonances found for participants inhaling with a VT configuration of a central vowel. Results for the VT models suggest that airflow direction is crucial for spectral properties of sibilants, /ç/, and /i:/, but not the other sounds we investigated. Inhalation noise is most similar to /ə/ where airflow direction does not play a role.

Conclusions: Inhalation is realized on ingressive airflow, and inhalation noises have specific resonance properties that are most similar to /ə/ but occur without phonation. Airflow direction does not play a role in this specific VT configuration, but subglottal resonances may do. For future work, we suggest investigating the articulation of speech breathing and link it to current work on pause postures.

Supplemental Material S1. Unnormalized inhalation spectra averaged by sex for female (100–4500 Hz) and male participants (100–8000 Hz).

Supplemental Material S2. Inhalation spectra averaged by speaker for the male data (100–4500 Hz).

Supplemental Material S3. Inhalation spectra averaged by speaker for the female data (100–4500 Hz).

Werner, R., Fuchs, S., Trouvain, J., Kürbis, S., Möbius, B., & Birkholz, P. (2023). Acoustics of breath noises in human speech: Descriptive and three-dimensional modeling approaches. Journal of Speech, Language, and Hearing Research. Advance online publication. https://doi.org/10.1044/2023_JSLHR-23-00112

Publisher Note: This article is part of the Special Issue: Select Papers From the 8th International Conference on Speech Motor Control.