Voice is a dominant component of everyday speech in all languages. The possibility is examined that its use may have evolved so that its timing in connected speech is ideal from the point of view of information theory-with voicing taking up 50% of the total speaking time. Initial measurements have been made of voice timing proportions using Laryngograph (EGG) signals as the basis of timing analyses. The results of these analyses for data from two groups of speakers are reported: single native speakers of each of 8 different languages; and 56 speakers of British English. The average 51% and 52% voice timing proportions that were found closely approximate the ideal of 50%. Implications of this finding for voice evolution are briefly discussed.