This report describes all attempt to determine the number and nature of the basic ways in which voices are perceived to differ from each other by a typical listener. In the experiment, 32 listeners described their perceptions of 16 voices by means of a semantic‐differential rating form. Analysis of variance was performed to determine the contributions of speakers, listeners, and various situational parameters to the variance of ratings on each item. Factor‐analytic techniques were employed to determine the dimensionality of the speaker effect, the listener effect, and the effect of the interaction of speakers and listeners. Four factors, labeled clarity, roughness, magnitude, and animation were found to account for an average of 88% of the variance in mean ratings given speakers on each of the 49 items. Six dimensions were found to account for the common‐factor variance in constant errors associated with listeners. Five dimensions were found to account for the common factor variance of the observed interaction of speakers and listeners.