An application server receives an HTTP request for execution of an instant messaging application operation for a sending party. A selected extensible markup language (XML) document is accessed in response to reception of the HTML request. A first HTML page is generated based on the XML document for recording a voice message by the sending party. A registry is accessed to determine whether a receiving party is available to receive the voice message. A second HTML page is generated for the receiving party having instructions for playing the voice message for the receiving party. Hence instant messaging services may be deployed on a platform that is customizable, scalable, and built upon open standards such as Internet protocol. By preserving multiple sessions, parties involved in instant messaging can suspend their present session or activity for a moment to create a new session to exchange audio messages and when finished, can instantly resume the suspended session.