The "proper" volume level can be determined by focusing on one sound source in the music that is familiar to you. I use a violin, or sometimes a human voice. I know how loud a volin can play, and how loud a person can sing. It is easy to crank up the volume of the total music program to the point where the violin, for example, is impossibly loud. When you do this it's like puting a a beautiful object under a microscope: all the defects become evident.
Sometimes the mastering of the recording is a problem. For example, in a violin concerto the soloist is sometimes recorded so loudly that the orchestra cannot be brought up to an exciting volume without making the soloist too loud. However, in a multichannel system, where the soloist is in the center channel he can be toned down by reducing gain in that channel relative to the other channels. I often need to do that.
Sometimes the mastering of the recording is a problem. For example, in a violin concerto the soloist is sometimes recorded so loudly that the orchestra cannot be brought up to an exciting volume without making the soloist too loud. However, in a multichannel system, where the soloist is in the center channel he can be toned down by reducing gain in that channel relative to the other channels. I often need to do that.