Music analysis considers features such as the key in which the music is composed, how it modulates, tempo changes, how pitches coincide, how phrases reappear and what are common rhythms. Computer analysis converts these features into amplitude and frequency analysis to explain the sound emanating from voices and instruments.
To determine why music sounds the way it does, a number of features must be analyzed before a conclusion can be reached. Computer analysis does this by recording or sampling the piece in question. This occurs via modeling, which is often a musical instrument digital interface (MIDI) sequence which approximates the nuances of the performances that generated the original music.
The two most popular encoding schemes for computer analysis are MIDI and digital alternative representation of music scores (DARMS). MIDI was designed to facilitate performance as well as understand that music is a series of events. DARMS describes the visual appears of the score and details beams, stem lengths, clefs and the like.