A method and system for a user to access back-end legacy systems via voice input. The user dials-in to a voice server using a portable access device. Using speech synthesis, the voice server provides a menu of operations from which the user can select. By speaking into the access device, the user selects an operation and provides any additional data needed for the operation. Using speech recognition, the voice server prepares a user request based on the spoken user input, that is forwarded to the systems interface, which may include a protocol server for providing a protocol interface and a transaction server for receiving user requests and generating legacy transactions based on the user requests. The systems interface retrieves information from the legacy systems based on the user request that is forwarded to the voice server, which formats and outputs the information to the access device.