A unified web-based voice messaging system provides voice application control between a web browser and an application server via an hypertext transport protocol (HTTP) connection on an Internet Protocol (IP) network. The application server, configured for executing a voice application defined by XML documents, selects an XML document for execution of a corresponding voice application operation based on a determined presence of a user-specific XML document that specifies the corresponding voice application operation. The application server, upon receiving a voice application operation request from a browser serving a user, determines whether a personalized, user specific XML document exists for the user and for the corresponding voice application operation. If the application server determines the presence of the personalized XML document for a user-specific execution of the corresponding voice application operation, the application server dynamically generates a personalized HTML page having media content and control tags for personalized execution of the voice application operation; however if the application server determines an absence of the personalized XML document for the user-specific execution of the corresponding voice application operation, the application server dynamically generates a generic HTML page for generic execution of the voice application operation. Hence, a user can personalize any number of voice application operations, enabling a web-based voice application to be completely customized or merely partially customized.