This paper introduces a new method for the automatic translation of formal data specifications to voice data-input applications. The objective is to automatically generate Voice XML applications that correspond to the structure of XML DTD specifications. A new formalization called Grammar-DTD (G-DTD) is introduced as an extended version of DTD. G-DTD is obtained by adding semantics to specify valid values of the DTD elements and attributes.The additional semantics are necessary to carry out the automatic translation. This work facilitates the rapid creation of voice applications.