Some San Francisco Bay Area residents from Marin County to the Presidio have noticed a sustained, series of high-pitched tones. The sound reached a new peak volume, and recordings of the eerie noise spread across social media in early June,.
The sound is due to high northwest winds blowing through the slats of the Golden Gate bridge’s newly-installed sidewalk railing. The new slats were thinner than the ones in the previous railing. The purpose of the new design is to make the span more aerodynamically stable on gusty days.
The sound is most likely an Aeolian tone, a noise produced when wind blows over a sharp edge, resulting in tiny harmonic vortices in the air.
The modification of the Golden Gate Bridge railing is the most recent and most audible element of a multi-phase retrofit that has been underway since 1997. Following the magnitude 6.9 Loma Prieta Earthquake in 1989, The Golden Gate Bridge, Highway, & Transportation District began to prepare the iconic bridge for the wind and earthquake loads that it may encounter in the future. The design maximum wind speed is 100 mph.
A sample sound file of the Bridge’s tones is taken from: https://youtu.be/8PnuOf33jN8
It was converted to mp3 format using an online utility. Golden_Gate_singing.mp3
The mp3 file was then called into Matlab and processed using the Vibrationdata tools.
Spectral peaks occur at 354, 398, 439 & 481 Hz. The spacing is nearly uniform with an average separation of 42.3 Hz. The nearest musical notes are F, G, A & B, respectively.
The maximum individual peak occurs at 439 Hz. Higher frequency peaks occur at 880, 1051 & 1160 Hz.
Note that these frequencies should vary with wind speed per the Strouhal number. The formation of the “vortex street” also depends on the Reynold’s number.
____
Vortex shedding behind a circular cylinder. In this animation, the flow on the two sides of the cylinder are shown in different colors, to show that the vortices from the two sides alternate. Courtesy, Cesareo de La Rosa Siqueira.
____
See also: Golden Gate Bridge Wind Tunnel Testing
– Tom Irvine
Vibrationdata 
That’s interesting! The pattern at the higher frequencies matches some of the peaks in the lower frequencies at 1 Octave up. I wonder how well it goes with the singing tubes that connect to the bay.
Awesome little breakdown, thanks for sharing!