Detection thresholds of infrasound complex tones

The growing number of infrasound sources (infrasound: frequencies below 20 Hz) in our environment is calling society's attention to infrasound exposure. Together with the Department of Experimental Audiology at the Otto von Guericke University Magdeburg, PTB has participated in the project titled "Infrasound and its relevance for audible sound". The project is funded by the Deutsche Forschungsgemeinschaft (DFG – German Research Foundation) and deals with the question as to how human 'hear' infrasound. Within the scope of this project, signals comprising several infrasound components (multi-frequency tone complexes) were used to investigate how the number of components influences the perceptibility of the signal (detection threshold). This allows conclusions as to the perception of complex signals to be drawn. Such signals are much more similar to those occurring in our real environment than single-frequency signals (pure tones).

Infrasound with frequencies f < 20 Hz is outside the conventional frequency range of audible sound. Several studies have, however, shown that sound may be perceived down to 2 Hz, provided that the sound pressure level is sufficiently high (see, e.g., [1] and [2]). The corresponding psychoacoustic measurements are usually carried out with pure tones. However, no direct conclusions can be drawn from such measurements as to the actual perception when more than one infrasound frequency component is present – as is the case with real environmental noises. In the present study, we therefore investigated how the number of infrasound components influences the detection threshold of a complex tone.

As early as 60 years ago, Green [3] suggested three mechanisms for processing multi-frequency-tone complexes in conventional audio-frequency range between 20 Hz and 20 kHz:

1. Dominance of one of the components: If one component of a complex tone dominates over all the other ones, the detection threshold of the complex tone should then be exactly as high as that of the dominant component presented as a pure tone. In this case, the level at threshold would not change.
2. Integration of information: If the information of the components of a complex tone is integrated, this indicates that the components lie spectrally so far apart that each component excites a different critical band. In this case, the level at threshold would decrease by approx. 1.5 dB for a two-tone complex and by approx. 2.4 dB for a three-tone complex.
3. Integration of the intensity: If the spectral intensities of the individual components of the complex tone add up, this indicates that the components are processed within one critical band. Here, the level at threshold would be smaller by approx. 3 dB for a two-tone complex and by approx. 4.8 dB for a three-tone complex.

With a group of 14 test subjects, we investigated whether one of these mechanisms can explain the detection thresholds of complex tones consisting of two and three different infrasound components. Figure 1 shows the group medians (full dots) and the associated interquartile ranges (vertical error bars) of the levels at the detection threshold of a 4 Hz pure tone (blue), of two two-tone complexes (4 Hz and 8 Hz; 4 Hz and 12 Hz; both purple), and of a three-tone complex (4 Hz, 8 Hz and 12 Hz; orange). In addition, highlighted horizontal lines represent the thresholds predicted by the three mechanisms. By definition and independent of the mechanism, the pure-tone signal has a threshold of 0 dB (dark gray solid line).

Fig. 1: Measured and predicted threshold levels in dB for infrasound complex tones (purple and orange) and for a pure tone (blue). The group medians of the measured thresholds are represented by means of full dots; the associated interquartile ranges are represented by means of error bars. The highlighted horizontal lines represent the thresholds predicted by the three mechanisms: the dark gray solid line corresponds to the prediction in the event of the dominance of one component for all complex tones; the dotted lines correspond to the prediction in the event of information integration for the two-tone complex (purple) and the three-tone complex (orange); the dashed lines correspond to the prediction in the event of intensity integration for the two previously described signals (purple and orange).

The results show that the detection threshold of infrasound complex tones is not determined by a dominant component, but that mostly spectral integration takes place (intensity integration). The detection thresholds of the three-tone complex are, however, not entirely compatible with the notion of intensity integration. The decrease is not as steep as would be expected in the event of intensity integration (dashed orange line). This deviation could suggest that this mechanism is somehow limited. In order to clarify this discrepancy, additional experiments that will be based on a broader data range are planned.

Additional information concerning the results of this study is available in our publication [4].

This study is part of the DFG-funded project titled "Infrasound and its relevance for audible sound" (FE 1192/3-1 | VE 373/4-1).

Literature:

[1] H. Møller, C. S. Pedersen: Hearing at low and infrasonic frequencies. Noise and Health 6 (2004) 37–57.

[2] R. Kühler, T. Fedtke, J. Hensel: Infrasonic and low-frequency insert earphone hearing threshold. JASA 137 (2015), EL347–EL353, DOI: 10.1121/1.4916795.

[3] D. M. Green: Detection of multiple component signals in noise. JASA 30 (1958), 904–911, DOI: 10.1121/1.1909400.

[4] B. Friedrich, H. Joost, T. Fedtke, J. L. Verhey: Spectral integration of infrasound at threshold. JASA 147 (2020), EL259–EL263, DOI: 10.1121/10.0000897.

Contact:

Holger Joost, FB 1.6, AG 1.61, E-Mail: holger.joost@ptb.de