The procedure he used for generating these sounds was the following. He started with doppler velocity data, averaged over the solar disk, so that only modes of low angular degree (l = 0, 1, 2) remained. Subsequent processing removed the spacecraft motion effects, instrument tuning, and some spurious points. Then Kosovichev filtered the data at about 3 mHz to select clean sound waves (and not supergranulation and instrumental noise). Finally, he interpolated over the missing data and scaled the data (speeded it up a factor 42,000 to bring it into the audible human-hearing range (kHz)).
For more audio files, visit the Stanford Experimental Physics Lab Solar Sounds page. http://soi.stanford.edu/results/sounds.html
Credits: A. Kosovichev, Stanford Experimental Physics Lab
Supporting materials can be downloaded from NASA’s Scientific Visualization Studio at: https://svs.gsfc.nasa.gov/13011