Sound becomes inaudible if it consists of frequencies lower than 20 Hz, i.e. the human hearing threshold, and is called infrasound. Low frequency acoustic signals were first discovered after the eruption of the Krakatoa (Indonesia) in 1883. Due to its low frequency content, this infrasound traveled up to seven times around the globe while reaching altitudes over 100 kilometer. The ability to detect explosions with infrasound resulted in substantial scientific and societal interest during World War I and the era of atmospheric nuclear testing. This interest diminished as nuclear tests were confined to the underground under the Limited Test Ban Treaty in 1963. Recently, with the signature of the Comprehensive Nuclear-Test-Ban Treaty, infrasound gained renewed attention as it is being used as a verification technique. This thesis describes the complete sequence of measuring, processing and interpreting infrasound data. A microbarometer was developed and deployed in arrays to measure infrasound. Array processing and statistical detection techniques are applied to extract signals of interest from the continuous recordings. Knowledge on the atmospheric propagation appeared essential in identifying infrasonic sources, like meteors, volcanoes, sea waves and accidental chemical explosions. A large amount of coherent infrasound is continuously being detected from both natural and man-made sources, i.e. the inaudible symphony. Applications are foreseen in acoustic remote sensing where infrasound can be used as passive probe for the upper atmosphere. Non-acoustic phenomena, like gravity waves, can also be detected with the methods described in this thesis.