This Letter to the editor is a comment on an article in the JASA Forum by William Yost. It pertains to a reference he made to prior work that indicates the possibility that the human threshold of hearing in the most sensitive region may be limited by Brownian motion of the air molecules, also called thermal noise. I present additional data and my own observations that support this contention.
I enjoyed reading Bill Yost's (2023) recent essay on silence and sound, especially the discussion of the fact that human hearing sensitivity might be so good as to be limited by the noise of Brownian motion of air molecules. This is a topic I have previously explored in the introductory chapter to the AIHA Noise Manual (Tufts and Berger, 2022).
Dr. Yost correctly quoted Sivian and White's (1933) assertion that Brownian motion, also referred to as thermal noise, places a limit on the physiological sensitivity of the human ear, in its range of greatest aural sensitivity. However, when I reviewed the Green and Dai (1991) reference that he cites as coming to a similar conclusion, I found that in their abstract (the citation is to an ASA presentation for which I do not have the actual notes) they concluded, “Our ability to hear weak auditory signals is impressive, but the fundamental limitation is not the thermal agitation of molecules of the air.” Following I share my own thoughts on this issue.
Harris (1968) explored Brownian motion as a limiting factor to human hearing. He computed free-field Brownian noise as 98 dB below 1 dyn/cm2 for a 1000-Hz-wide bandwidth centered at 3000 Hz. A value of 1 dyn/cm2 = 0.1 Pa = 74.0 dB SPL, which would put Brownian noise at –24 dB SPL. He stated that the minimum audible sound field for binaural listening was –6 dB SPL [which agrees closely with the 3000-Hz free-field threshold in ISO 389-7:2019 (2019)], and concluded as did Green and Dai (1991) that Brownian motion “was not the limiting factor in the threshold of hearing.” This seems correct for the average listener since Brownian noise would be 18 dB below their minimum audible sound field. Harris (1968) went on to explore Brownian noise in the cochlear partition; and depending upon his assumptions for coupling of the hair cells, his estimates of the noise levels varied by 55 dB. More recent work by Altoè and Christopher (2024), Sasmal and Grosh (2018), and van Netten (2003) suggests that there are further complexities that influence such computations; and Oxenham and Shera (2003) discuss issues of the sharpness of cochlear tuning curves near threshold.
However, returning to the more straightforward question of Brownian noise at about –24 dB SPL, there are still a few issues to be considered. When we account for listener variability around the average threshold value [according to ISO 28961:2012 (2012), the 5th percentile's hearing is about 8 dB better than the median hearing threshold], then a 5th percentile listener's threshold is about −14 dB, or within 10 dB of Harris's calculated Brownian noise level and a 2nd percentile listener's threshold would be closer still. Since Berger and Killion (1989) have shown that effective masking levels as much as 6 dB below threshold can lead to a 1 dB threshold elevation, it is plausible that the most sensitive listeners could have thresholds limited by Brownian noise in the most sensitive region of human hearing, a finding in agreement with the observation of Sivian and White (1933).
For those interested in further exploration of the noise of Brownian motion, there is a fascinating video posted online by Microsoft at https://news.microsoft.com/stories/building87/audio-lab.php that documents their quest to create the quietest place on Earth for a measurement for the Guiness Book of World Records back in the era of MS Cortana (approximately 2014–2020). They achieved –23 dBA.
AUTHOR DECLARATIONS
Conflict of Interest
I have no conflicts of interest.
