Vocal production and its usage in nonhuman primates may share common features with primitive human language. We trained two Japanese monkeys to use a rake-shaped tool to retrieve distant food. After the training, the monkeys spontaneously began vocalizing coo-calls in the tool-using context. We then trained one of the monkeys to vocalize to request food or the tool. Three independent acoustic parameters were measured and each parameter was independently analyzed across conditions using a multiple comparison test. We found that the monkey spontaneously differentiated their coo-calls to ask for either food or tool during the course of this training. This process might involve a change from emotional vocalizations into intentionally controlled ones by associating them with consciously planned tool use. We thus established a novel hypothesis about the origin of voluntary vocal control that could be approached from neurophysiological procedures.