A musical tone signal processing apparatus configured to freely expand or contract an acoustic image. A setting means may set a second reference localization, a first function that stipulates expansion of the boundary of one end of a direction range, and a second function that stipulates expansion of the boundary of an other end of said direction range. In addition, an output direction indicated by localization information of an extraction signal that exists in the direction range in which the conditions have been set in this manner is shifted by an acoustic image scaling means. The acoustic image in said direction range is expanded or contracted accordingly. In other words, the acoustic image formed by the extraction signal is expanded or contracted by shifting the output direction for each extraction signal extracted from within the direction range.