Recorded with an ultrasonic microphone and slowed down, a male mouse encountering female scent produces long bouts of calls, typically between about 30 and 110 kilohertz. The calls are made of discrete syllable types arranged in repeated sequences, and the ordering is not random, which is the basis for the term song.
The function appears to be courtship. Females show preferences in playback tests, and song varies with strain, individual and social context. How much of the structure is learned remains contested: mice are not classical vocal learners, and some studies have found the songs largely innate with limited plasticity, while others report subtle context-dependent change.
The production mechanism was itself unclear until recently. Rather than the vocal-fold vibration used for audible sound, the ultrasonic calls appear to be generated by an air jet striking the inner wall of the larynx, a whistle-like mechanism previously described mainly in machines rather than in animals.