Without instruments each note of a song is one flat tone. instruments gives the voices the shape an 8-bit tracker gives them: lists of numbers read one a frame, a sixtieth of a second, that move a note's volume, its colour (duty) and its pitch while it sounds. A | marks where a list starts again once the note is held. vibrato adds a wobble and slide a glide into each note; arpeggio plays a chord one note a frame, which is how the console did chords.
A voice starts with its instrument, and @name in its notes changes it from there on. Version 3 is version 1 once as an intro and then version 2, in one song with a loop. Version 4 defines nothing: it uses only the engine's instruments, effects written after a single note in brackets ([slide], [vib:5,0.4]), and drum samples on the fifth voice, dpcm.
Space to play, 1 to 4 to compare the versions.